1 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
3 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
6 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7 Kito Cheng <kito.cheng@gmail.com>
9 * config/nds32/constraints.md: Provide more constraints.
10 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
11 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
12 support constraints usage.
14 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
16 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
18 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
19 Kito Cheng <kito.cheng@gmail.com>
21 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
22 * config/nds32/nds32.opt: Refine the layout.
23 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
24 TARGET_EXT_STRING): Support new options.
25 * config/nds32/nds32.h: Likewise.
26 * config/nds32/nds32.md: Likewise.
27 * config/nds32/nds32-predicates.c: Likewise.
28 * config/nds32/constraints.md: Likewise.
29 * common/config/nds32/nds32-common.c: Likewise.
31 2017-11-16 Julia Koval <julia.koval@intel.com>
34 * config/i386/gfniintrin.h: Add sse check.
35 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
37 2017-11-16 Julia Koval <julia.koval@intel.com>
39 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
40 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
41 (ix86_handle_option): Handle -mavx512vbmi2.
42 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
43 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
44 * config/i386/i386-c.c (__AVX512VBMI2__): New.
45 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
46 (ix86_valid_target_attribute_inner_p): Ditto.
47 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
48 * config/i386/i386.opt (mavx512vbmi2): New option.
49 * doc/invoke.texi: Add new option.
51 [2017-11-16 Julia Koval <julia.koval@intel.com>
53 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
54 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
55 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
56 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
57 _mm512_gf2p8mul_epi8): New intrinsics.
58 * config/i386/i386-builtin-types.def
59 (V64QI_FTYPE_V64QI_V64QI): New type.
60 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
61 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
62 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
63 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
64 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
65 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
67 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
69 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
70 explicitly as a stream of bytes.
72 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
74 * config/rs6000/altivec.h (vec_xst_be): New #define.
75 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
76 and externalize from *altivec_vperm_<mode>_internal.
77 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
79 (XL_BE_V8HI): Likewise.
80 (XL_BE_V4SI): Likewise.
81 (XL_BE_V4SI): Likewise.
82 (XL_BE_V2DI): Likewise.
83 (XL_BE_V4SF): Likewise.
84 (XL_BE_V2DF): Likewise.
85 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
86 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
87 all array entries with these keys: VSX_BUILTIN_VEC_XL,
88 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
89 VSX_BUILTIN_VEC_XST_BE.
90 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
91 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
93 (altivec_init_builtins): Replace conditional calls to def_builtin
94 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
95 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
96 with unconditional calls. Remove calls to def_builtin for
97 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
99 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
100 to define_expand, and add alternate RTL generation for P8.
101 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
103 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
104 add alternate RTL generation for P8.
105 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
106 vsx_ld_elemrev_v16qi.
107 (vsx_st_elemrev_v8hi): Convert define_insn
108 to define_expand, and add alternate RTL generation for P8.
109 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
111 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
112 add alternate RTL generation for P8.
113 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
114 vsx_st_elemrev_v16qi.
116 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
119 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
120 TARGET_AVX512ER check.
121 (ix86_option_override_internal): Set MASK_VZEROUPPER if
122 neither -mzeroupper nor -mno-zeroupper is used and
123 TARGET_EMIT_VZEROUPPER is set.
124 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
125 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
127 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
129 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
130 folding of vector compares.
131 (fold_build_vec_cmp): New helper function.
132 (fold_compare_helper): New helper function.
133 (builtin_function_type): Add compare builtins to the list of functions
134 having unsigned arguments. Cosmetic updates to comment indentation.
135 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
136 the not+eq combination.
138 2017-11-15 Bin Cheng <bin.cheng@arm.com>
140 PR tree-optimization/82726
141 PR tree-optimization/70754
142 * tree-predcom.c (order_drefs_by_pos): New function.
143 (combine_chains): Move code setting has_max_use_after to...
144 (try_combine_chains): ...here. New parameter. Sort combined chains
145 according to position information.
146 (tree_predictive_commoning_loop): Update call to above function.
147 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
149 2017-11-15 Bin Cheng <bin.cheng@arm.com>
151 PR tree-optimization/82726
153 2017-01-23 Bin Cheng <bin.cheng@arm.com>
155 PR tree-optimization/70754
156 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
157 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
158 combined stmt before it if not NULL.
159 (combine_chains): Process refs reversely and compute dominance point
163 2017-02-23 Bin Cheng <bin.cheng@arm.com>
165 PR tree-optimization/79663
166 * tree-predcom.c (combine_chains): Process refs in reverse order
167 only for ZERO length chains, and add explaining comment.
169 2017-11-15 Tamar Christina <tamar.christina@arm.com>
171 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
172 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
173 * doc/invoke.texi (ARM Options): Add armv8.3-a.
175 2017-11-15 Tamar Christina <tamar.christina@arm.com>
177 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
179 2017-11-15 Martin Liska <mliska@suse.cz>
181 * tree-cfg.c (pass_warn_function_return::execute):
182 Compare warn_return_type for greater than zero.
184 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
188 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
189 to return true on Xeon and not on Xeon Phi.
190 (ix86_check_avx256_register): Changed to ...
191 (ix86_check_avx_upper_register): ... this. Add extra check for
192 VALID_AVX512F_REG_OR_XI_MODE.
193 (ix86_avx_u128_mode_needed): Changed
194 ix86_check_avx256_register to ix86_check_avx_upper_register.
195 (ix86_check_avx256_stores): Changed to ...
196 (ix86_check_avx_upper_stores): ... this. Changed
197 ix86_check_avx256_register to ix86_check_avx_upper_register.
198 (ix86_avx_u128_mode_after): Changed
199 avx_reg256_found to avx_upper_reg_found. Changed
200 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
201 (ix86_avx_u128_mode_entry): Changed
202 ix86_check_avx256_register to ix86_check_avx_upper_register.
203 (ix86_avx_u128_mode_exit): Ditto.
204 * config/i386/i386.h: (host_detect_local_cpu): New define.
206 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
208 * config/arm/xgene1.md (xgene1): Split into automatons
209 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
210 (xgene1_f_load): Adjust reservations and/or types.
211 (xgene1_f_store): Likewise.
212 (xgene1_load_pair): Likewise.
213 (xgene1_store_pair): Likewise.
214 (xgene1_fp_load1): Likewise.
215 (xgene1_load1): Likewise.
216 (xgene1_store1): Likewise.
217 (xgene1_move): Likewise.
218 (xgene1_alu): Likewise.
219 (xgene1_simd): Likewise.
220 (xgene1_bfm): Likewise.
221 (xgene1_neon_load1): Likewise.
222 (xgene1_neon_store1): Likewise.
223 (xgene1_neon_logic): Likewise.
224 (xgene1_neon_st1): Likewise.
225 (xgene1_neon_ld1r): Likewise.
226 (xgene1_alu_cond): Added.
227 (xgene1_shift_reg): Likwise.
228 (xgene1_bfx): Likewise.
229 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
231 2017-11-15 Jakub Jelinek <jakub@redhat.com>
234 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
236 (can_widen_mult_without_libcall): New function.
237 (expand_mul_overflow): If only checking unsigned mul overflow,
238 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
239 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
240 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
241 (expand_DIVMOD): Formatting fix.
242 * expmed.h (expand_mult): Add NO_LIBCALL argument.
243 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
244 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
246 PR tree-optimization/82977
247 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
248 constructed temporary to strlen_to_stridx.put.
250 2017-11-15 Martin Liska <mliska@suse.cz>
252 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
253 * configure: Regenerate.
255 2017-11-15 Martin Liska <mliska@suse.cz>
258 * config/sh/sh-mem.cc: Use proper probability for
261 2017-11-14 Jeff Law <law@redhat.com>
263 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
264 the red zone for stack_clash_protection_final_dynamic_probe targets
265 when the total dynamic stack size is zero bytes.
267 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
268 blocks is post order.
270 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
272 * dumpfile.h (TDF_COMPARE_DEBUG): New.
273 * final.c (rest_of_clean_state): Set it for the
274 -fcompare-debug dump.
275 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
276 class when TDF_COMPARE_DEBUG is set.
278 * dwarf2out.c (gen_producer_string): Discard
281 2017-11-15 Joseph Myers <joseph@codesourcery.com>
284 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
285 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
286 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
287 (__TGMATH_REAL_2_3): Remove macros.
288 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
289 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
291 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
293 (remquo): Define using __TGMATH_REAL_3.
295 2017-11-14 Jeff Law <law@redhat.com>
297 * vr-values.c: New file with contents extracted from tree-vrp.c.
298 * Makefile.in (OBJS): Add vr-values.o
299 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
300 (set_value_range, set_and_canonicalize_value_range): Likewise.
301 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
302 (value_range_constant_singleton, symbolic_range_p): Likewise.
303 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
304 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
305 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
306 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
307 (find_case_label_range, find_case_label_index): Likewise.
308 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
309 (range_int_cst_singleton_p, value_inside_range): Likewise.
310 (get_single_symbol): Likewise.
311 (switch_update): Move structure definition here.
312 (to_remove_edges, to_update_switch_stmts): Provide externs.
313 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
314 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
315 (vrp_val_is_min, set_value_range): Likewise.
316 (set_and_canonicalize_value_range, copy_value_range): Likewise.
317 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
318 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
319 (range_is_nonnull, range_int_cst_p): Likewwise.
320 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
321 (get_single_symbol, operand_less_p): Likewise
322 (compare_values_warnv, compare_values): Likewise.
323 (value_inside_range, value_range_constant_singleton): Likewise.
324 (zero_nonzero_bitgs_from_vr): Likewise.
325 (extract_range_from_binary_expr_1): Likewise.
326 (overflow_comparison_p): Likewise.
327 (to_remove_edges, to_update_switch_stmts): Likewise.
328 (find_case_label-index, find_case_label_range): Likewise.
329 (switch_update, set_value_range_to_nonnegative): Remove.
330 (set_value_range_to_truthvalue): Likewise.
331 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
332 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
333 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
334 (find_case_label_ranges, test_for_singularity): Likewise.
335 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
336 (x_vr_values): Move to its remaining use site.
338 2017-11-10 Jeff Law <law@redhat.com>
340 * vr-values.h (VR_INITIALIZER): Move #define here.
341 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
342 * Makefile.in (OBJS): Add tree-evrp.o
343 * tree-vrp.h (assert_info): Move structure definition here.
344 (set_value_range_to_varying): Prototype.
345 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
346 (infer_value_range, register_edge_assert_for): Likewise.
347 (stmt_interesting_for_vrp): Likewise.
348 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
349 (set_value_range_to_varying): No longer static.
350 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
351 (infer_value_range, register_edge_assert_for): Likewise.
353 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
355 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
356 generate the XXBRD instruction.
358 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
359 (rs6000_builtin_type_compatible): Treat _Float128 and long double
360 as being compatible if -mabi=ieeelongdouble.
361 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
362 to setup float128 built-ins with hardware support.
363 (BU_FLOAT128_HW_2): Likewise.
364 (BU_FLOAT128_HW_3): Likewise.
365 (BU_FLOAT128_HW_VSX_1): Likewise.
366 (BU_FLOAT128_HW_VSX_2): Likewise.
367 (scalar_extract_expq): Change float128 built-in functions to
368 accommodate having both KFmode and TFmode functions. Use the
369 KFmode variant as the default.
370 (scalar_extract_sigq): Likewise.
371 (scalar_test_neg_qp): Likewise.
372 (scalar_insert_exp_q): Likewise.
373 (scalar_insert_exp_qp): Likewise.
374 (scalar_test_data_class_qp): Likewise.
375 (sqrtf128_round_to_odd): Delete processing the round to odd
376 built-in functions as special built-in functions, and define them
377 as float128 built-ins. Use the KFmode variant as the default.
378 (truncf128_round_to_odd): Likewise.
379 (addf128_round_to_odd): Likewise.
380 (subf128_round_to_odd): Likewise.
381 (mulf128_round_to_odd): Likewise.
382 (divf128_round_to_odd): Likewise.
383 (fmaf128_round_to_odd): Likewise.
384 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
385 support for KFmode and TFmode xststdcqp calls.
386 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
387 point, switch the built-in handlers for the get/set float128
388 exponent, get float128 mantissa, float128 test built-ins, and the
389 float128 round to odd built-in functions. Eliminate creating the
390 float128 round to odd built-in functions as special built-ins.
391 (rs6000_init_builtins): Eliminate special creation of the float128
392 round to odd built-in functions.
393 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
394 function insns to support both TFmode and KFmode variants.
395 (xsxsigqp_<mode>): Likewise.
396 (xsiexpqpf_<mode>): Likewise.
397 (xsiexpqp_<mode>): Likewise.
398 (xststdcqp_<mode>): Likewise.
399 (xststdcnegqp_<mode>): Likewise.
400 (xststdcqp_<mode>): Likewise.
402 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
404 * tree-ssa-threadupdate.c (compute_path_counts): Remove
405 unused path_in_freq_ptr parameter.
406 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
408 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
410 * ipa-inline.c (edge_badness): Dump sreal frequency.
411 (compute_inlined_call_time): Match natural implementaiton ...
412 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
413 forgotten division by CGRAPH_FREQ_BASE.
415 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
417 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
418 Solaris 11. Update comment.
419 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
421 * config/sol2.h (STARTFILE_SPEC): Likewise.
422 * configure: Regenerate.
424 2017-11-14 Carl Love <cel@us.ibm.com>
426 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
427 le_ and be_ prefixes to swap* variables. Remove
428 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
431 2017-11-14 Jason Merrill <jason@redhat.com>
433 Support GTY((cache)) on hash_map.
434 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
435 (ggc_cache_remove): Override it instead of ggc_mx.
436 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
437 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
438 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
439 (simple_cache_map_traits): Override maybe_mx.
440 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
441 (hash_map): Friend gt_cleare_cache.
442 (gt_cleare_cache): New.
443 * tree.h (tree_cache_traits): New hash_map traits class.
444 (tree_cache_map): New typedef.
446 2017-11-14 Richard Biener <rguenther@suse.de>
448 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
449 paramter and handling.
450 (cleanup_control_flow_bb): Likewise.
451 (cleanup_control_flow_pre): New helper performing a DFS walk
452 to call cleanup_control_flow_bb in PRE order.
453 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
454 via cleanup_control_flow_pre.
456 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
458 * config/aarch64/aarch64-simd.md
459 (aarch64_simd_bsl<mode>_internal): Remove DImode.
460 (*aarch64_simd_bsl<mode>_alt): Likewise.
461 (aarch64_simd_bsldi_internal): New.
462 (aarch64_simd_bsldi_alt): Likewise.
464 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
466 * tracer.c (better_p): Do not compare frequencies.
467 * reg-stack.c (better_edge): Likewise.
468 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
471 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
473 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
474 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
475 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
476 * lto-streamer-in.c (input_function): Use update_max_bb_count.
477 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
478 * predict.c (maybe_hot_frequency_p): Inline to ...
479 (maybe_hot_count_p): ... here; rewrite to counts.
480 (counts_to_freqs): Rename to ...
481 (update_max_bb_count): ... this one.
482 (expensive_function_p): Use counts.
483 (estimate_bb_frequencies): Update.
484 (rebuild_frequencies): Update.
485 * predict.h (counts_to_freqs): Rename to ...
486 (update_max_bb_count): ... this one.
487 * profile.c (compute_branch_probabilities): Add debug info
488 * tree-inline.c (expand_call_inline): Update debug info.
489 (optimize_inline_calls): Use update_max_bb_count..
490 (tree_function_versioning): Use update_max_bb_count..
491 * value-prof.c (gimple_value_profile_transformations):
492 Do not use update_max_bb_count.
494 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
496 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
497 always use frequencies.
499 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
501 * bb-reorder.c: Remove frequencies from comments.
502 (better_edge_p): Use profile counts.
503 (find_traces): Dump profile counts.
504 (rotate_loop): Use profile counts.
505 (find_traces_1_round): Likewise.
506 (connect_better_edge_p): Use counts instead of probabilities for
508 (copy_bb_p): Drop early check for non-0 frequency.
509 (sanitize_hot_paths): Update comments.
511 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
513 * ipa-split.c (struct split_point): Add count.
514 (consider_split): Do not compute incoming frequency; compute incoming
515 count and store it to split_point.
516 (split_function): Set count of the call to split part correctly.
518 2017-11-13 Carl Love <cel@us.ibm.com>
520 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
522 2017-11-13 Tom Tromey <tom@tromey.com>
524 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
526 2017-11-13 Carl Love <cel@us.ibm.com>
528 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
529 Add support for builtins:
530 unsigned int vec_first_{,miss}_match_{,or_eos}index,
531 vector {un,}signed {char,int,short},
532 vector {un,}signed {char,int,short}) arguments.
533 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
534 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
535 Add BU_P9V_AV_2 expansions for the builtins.
536 * config/rs6000/altivec.h (vec_first_match_index,
537 vec_first_mismatch_index, vec_first_match_or_eos_index,
538 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
539 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
540 new extern declaration.
541 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
542 * config/rs6000/vsx.md (first_match_index_<mode>,
543 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
544 first_mismatch_or_eos_index_<mode>): Add define expand.
545 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
546 * doc/extend.texi: Update the built-in documenation file for the new
549 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
551 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
552 into the min/max operations for _Float<N> and _Float<N>X types.
554 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
557 * dwarf2out.c (do_eh_frame): New static variable.
558 (dwarf2out_begin_prologue): Set it.
559 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
561 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
563 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
565 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
567 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
570 * debug.h (dwarf2out_do_eh_frame): Declare.
571 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
572 (dwarf2out_do_frame): Use it.
573 (dwarf2out_do_cfi_asm): Likewise.
574 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
575 (dwarf2out_assembly_start): Likewise.
576 (dwarf2out_begin_prologue): Fix comment.
577 * toplev.c (compile_file): Always call dwarf2out_frame_finish
578 if the target needs either debug or unwind DWARF2 info.
579 * lto-opts.c (lto_write_options): Do not save -fexceptions,
580 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
581 -ftrapping-math, -ftrapv and -fwrapv.
583 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
585 * cgraph.c (cgraph_edge::sreal_frequency): New function.
586 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
587 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
588 (estimate_edge_size_and_time): Likewise.
589 (ipa_merge_fn_summary_after_inlining): Likewise.
590 * ipa-inline.c (cgraph_freq_base_rec): Remove.
591 (compute_uninlined_call_time): Use sreal_frequency.
592 (compute_inlined_call_time): Likewise.
593 (ipa_inline): Do not initialize cgraph_freq_base_rec.
594 * profile-count.c: Include sreal.h.
595 (profile_count::to_sreal_scale): New.
596 * profile-count.h: Forward declare sreal.
597 (profile_count::to_sreal_scale): Declare.
599 2017-11-13 Nathan Sidwell <nathan@acm.org>
601 * diagnostic.c (maybe_line_and_column): New.
602 (diagnostic_get_location_text): Use it.
603 (diagnostic_report_current_module): Likewise.
604 (test_diagnostic_get_location_text): Add tests.
606 2017-11-13 Luis Machado <luis.machado@linaro.org>
608 * doc/md.texi (Specifying processor pipeline description): Fix
609 incorrect latency for the div instruction example.
611 2017-11-13 Jakub Jelinek <jakub@redhat.com>
613 PR tree-optimization/78821
614 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
615 that bit_not_p is the same.
616 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
617 (split_group): Count precisely bit_not_p bits in each statement.
618 (invert_op): New function.
619 (imm_store_chain_info::output_merged_store): Use invert_op to
620 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
621 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
623 2017-11-13 Martin Liska <mliska@suse.cz>
625 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
626 (struct source_info): Likewise.
627 (add_branch_counts): Likewise.
628 (add_line_counts): Likewise.
629 (function_summary): Likewise.
630 (output_intermediate_line): Likewise.
631 (generate_results): Likewise.
633 2017-11-13 Martin Liska <mliska@suse.cz>
635 * gcov.c (struct block_info): Remove typedef for block_t.
636 (struct line_info): Likewise.
637 (line_info::has_block): Likewise.
638 (EXIT_BLOCK): Likewise.
641 (get_cycles_count): Likewise.
642 (process_file): Likewise.
643 (read_graph_file): Likewise.
644 (solve_flow_graph): Likewise.
645 (find_exception_blocks): Likewise.
646 (add_line_counts): Likewise.
647 (accumulate_line_info): Likewise.
648 (output_line_details): Likewise.
650 2017-11-13 Martin Liska <mliska@suse.cz>
652 * gcov.c (struct arc_info): Remove typedef for arc_t.
653 (struct line_info): Likewise.
654 (add_branch_counts): Likewise.
655 (output_branch_count): Likewise.
656 (function_info::~function_info): Likewise.
658 (output_intermediate_line): Likewise.
659 (read_graph_file): Likewise.
660 (solve_flow_graph): Likewise.
661 (find_exception_blocks): Likewise.
662 (add_line_counts): Likewise.
663 (accumulate_line_info): Likewise.
664 (output_line_details): Likewise.
665 (output_function_details): Likewise.
667 2017-11-13 Martin Liska <mliska@suse.cz>
669 * gcov.c (struct function_info): Remove typedef for function_t.
670 (struct source_info): Likewise.
671 (source_info::get_functions_at_location): Likewise.
672 (solve_flow_graph): Likewise.
673 (find_exception_blocks): Likewise.
674 (add_line_counts): Likewise.
675 (output_intermediate_file): Likewise.
676 (process_file): Likewise.
677 (generate_results): Likewise.
678 (release_structures): Likewise.
679 (read_graph_file): Likewise.
680 (read_count_file): Likewise.
681 (accumulate_line_counts): Likewise.
682 (output_lines): Likewise.
684 2017-11-13 Martin Liska <mliska@suse.cz>
686 * gcov.c (function_info::function_info): Remove num_counts
687 and add vector<gcov_type>.
688 (function_info::~function_info): Use the vector.
689 (process_file): Likewise.
690 (read_graph_file): Likewise.
691 (read_count_file): Likewise.
692 (solve_flow_graph): Likewise.
694 2017-11-13 Martin Liska <mliska@suse.cz>
696 * gcov.c (function_info::is_artificial): New function.
697 (process_file): Erase all artificial early.
698 (generate_results): Skip as all artificial are already
701 2017-11-13 Martin Liska <mliska@suse.cz>
703 * gcov.c (read_graph_file): Store to global vector of functions.
704 (read_count_file): Iterate the vector.
705 (process_file): Likewise.
706 (generate_results): Likewise.
707 (release_structures): Likewise.
709 2017-11-13 Jakub Jelinek <jakub@redhat.com>
711 PR tree-optimization/82954
712 * gimple-ssa-store-merging.c
713 (imm_store_chain_info::coalesce_immediate_stores): If
714 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
716 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
718 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
719 Upddate call to ENDIAN_LANE_N.
720 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
721 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
722 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
723 and use aarch64_endian_lane_rtx.
724 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
726 2017-11-12 Tom de Vries <tom@codesourcery.com>
728 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
730 2017-11-12 Tom de Vries <tom@codesourcery.com>
732 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
734 2017-11-12 Tom de Vries <tom@codesourcery.com>
736 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
738 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
739 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
740 * defaults.h (ASM_OUTPUT_LABELREF): Same.
742 2017-11-11 Martin Sebor <msebor@redhat.com>
745 * doc/extend.texi (attribute nonstring): Remove spurious argument.
748 * prefic.c (translate_name): Replace strncpy with memcpy to
749 avoid -Wstringop-truncation.
751 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
753 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
755 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
757 * predict.c (maybe_hot_frequency_p): Do not use cfun.
759 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
761 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
764 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
766 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
767 enable generating XXBRH if the value is in a vector register.
768 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
769 value is in a vector register.
770 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
771 register to register bswap64's instead of doing the GPR sequence
772 used on previous machines.
773 (bswapdi2_xxbrd): New insn.
774 (bswapdi2_reg): Disallow on ISA 3.0.
775 (register to register bswap64 splitter): Do not split the insn on
776 ISA 3.0 systems that use XXBRD.
778 2017-11-10 Martin Sebor <msebor@redhat.com>
781 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
783 (find_subframework_file): Same.
785 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
787 * auto-profile.c (afdo_indirect_call): Drop frequency.
788 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
789 (cgraph_node::create_edge): Drop frequency argument.
790 (cgraph_node::create_indirect_edge): Drop frequency argument.
791 (cgraph_edge::make_speculative): Drop frequency arguments.
792 (cgraph_edge::resolve_speculation): Do not update frequencies
793 (cgraph_edge::dump_edge_flags): Do not dump frequency.
794 (cgraph_node::dump): Check consistency in IPA mode.
795 (cgraph_edge::maybe_hot_p): Use IPA counter.
796 (cgraph_edge::verify_count_and_frequency): Rename to ...
797 (cgraph_edge::verify_count): ... this one; drop frequency checking.
798 (cgraph_node::verify_node): Update.
799 * cgraph.h (struct cgraph_edge): Drop frequency.
800 (cgraph_edge::frequency): New function.
801 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
803 (cgraph_edge::rebuild_edges): Likewise.
804 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
805 (duplicate_thunk_for_node): Do not pass frequency.
806 (cgraph_node::create_clone): Scale only counts.
807 (cgraph_node::create_virtual_clone): Do not pass frequency.
808 (cgraph_node::create_edge_including_clones): Do not pass frequency.
809 (cgraph_node::create_version_clone): Do not pass frequency.
810 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
811 (cgraph_node::expand_thunk): Do not pass frequency.
812 (cgraph_node::create_wrapper): Do not pass frequency.
813 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
815 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
816 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
817 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
818 (gather_caller_stats): Use frequency function.
819 (ipcp_cloning_candidate_p): Use frequency function.
820 (ipcp_propagate_stage): Use frequency function.
821 (get_info_about_necessary_edges): Use frequency function.
822 (update_profiling_info): Update only IPA profile.
823 (update_specialized_profile): Use frequency functoin.
824 (perhaps_add_new_callers): Update only IPA profile.
825 * ipa-devirt.c (ipa_devirt): Use IPA profile.
826 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
827 (dump_ipa_call_summary): Use frequency function.
828 (estimate_edge_size_and_time): Use frequency function.
829 (ipa_merge_fn_summary_after_inlining): Use frequency function.
830 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
831 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
832 (update_noncloned_counts): ... ths one; scale counts only.
833 (clone_inlined_nodes): Do not scale frequency.
834 (inline_call): Do not pass frequency.
835 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
836 (compute_inlined_call_time): Use IPA profile.
837 (want_inline_small_function_p): Use IPA profile.
838 (want_inline_self_recursive_call_p): Use IPA profile.
839 (edge_badness): Use IPA profile.
840 (lookup_recursive_calls): Use IPA profile.
841 (recursive_inlining): Do not pass frequency.
842 (resolve_noninline_speculation): Do not update frequency.
843 (inline_small_functions): Collect max of IPA profile.
844 (dump_overall_stats): Dump IPA porfile.
845 (dump_inline_stats): Dump IPA porfile.
846 (ipa_inline): Collect IPA stats.
847 * ipa-inline.h (clone_inlined_nodes): Update prototype.
848 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
849 (ipa_propagate_frequency): Use frequency function.
850 (ipa_profile): Cleanup.
851 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
852 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
853 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
854 (input_node): Do not stream frequency.
855 (input_edge): Do not stream frequency.
856 (merge_profile_summaries): Scale only IPA profiles.
857 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
858 * predict.c (drop_profile): Do not recompute frequency.
859 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
860 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
861 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
862 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
863 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
864 * tree-inline.c (copy_bb): Do not scale frequency.
865 (expand_call_inline): Do not scale frequency.
866 (tree_function_versioning): Do not scale frequency.
867 * ubsan.c (ubsan_create_edge): Do not pass frequency.
869 2017-11-10 Julia Koval <julia.koval@intel.com>
871 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
872 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
873 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
874 (_mm256_maskz_gf2p8affine_epi64_epi8)
875 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
876 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
877 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
878 (__builtin_ia32_vgf2p8affineqb_v32qi)
879 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
880 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
882 2017-11-10 Tamar Christina <tamar.christina@arm.com>
886 (arm_option_override): Refactor.
887 (arm_option_reconfigure_globals): New.
888 (arm_options_perform_arch_sanity_checks): New.
889 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
891 (arm_options_perform_arch_sanity_checks): Likewise
893 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
895 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
897 * rs6000/rs6000.md ("type" attr): Add qmul.
898 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
899 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
900 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
902 2017-11-10 Martin Sebor <msebor@redhat.com>
905 * builtins.c (compute_objsize): Handle arrays that
906 compute_builtin_object_size likes to fail for. Make extern.
907 * builtins.h (compute_objsize): Declare.
908 (check_strncpy_sizes): New function.
909 (expand_builtin_strncpy): Call check_strncpy_sizes.
910 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
911 -Wstringop-truncation.
912 (gimple_fold_builtin_strncat): Same.
913 * gimple.c (gimple_build_call_from_tree): Set call location.
914 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
915 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
916 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
917 (handle_builtin_strlen): Use strlen_to_stridx.
918 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
920 Use strlen_to_stridx.
921 (pass_strlen::execute): Release strlen_to_stridx.
922 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
923 (-Wstringop-truncation): Document new option.
925 2017-11-10 Martin Liska <mliska@suse.cz>
927 PR gcov-profile/82702
928 * gcov.c (main): Handle intermediate files in a different
930 (get_gcov_intermediate_filename): New function.
931 (output_gcov_file): Remove support of intermediate files.
932 (generate_results): Allocate intermediate file.
933 (release_structures): Clean-up properly fn_end.
934 (output_intermediate_file): Start iterating with line 1.
936 2017-11-10 Jakub Jelinek <jakub@redhat.com>
938 PR tree-optimization/82929
939 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
940 ops_swapped_p non-static data member.
941 (store_immediate_info::store_immediate_info): Clear it.
942 (imm_store_chain_info::coalesce_immediate_stores): If swapping
943 ops set ops_swapped_p.
944 (count_multiple_uses): Handle ops_swapped_p.
946 2017-11-10 Martin Liska <mliska@suse.cz>
948 * coverage.c (coverage_init): Stream information about
949 support of has_unexecuted_blocks.
950 * doc/gcov.texi: Document that.
951 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
952 * gcov.c (read_graph_file): Likewise.
953 (output_line_beginning): Fix a small issue with
956 2017-11-10 Bin Cheng <bin.cheng@arm.com>
958 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
959 reference of trivial component.
961 2017-11-10 Jakub Jelinek <jakub@redhat.com>
964 * gimple-ssa-store-merging.c
965 (pass_store_merging::terminate_all_aliasing_chains): For
966 gimple_store_p stmts also call refs_output_dependent_p.
968 PR rtl-optimization/82913
969 * compare-elim.c (try_merge_compare): Punt if def_insn is not
972 2017-11-09 Jeff Law <law@redhat.com>
974 * vr-values.h: New file with vr_values class.
975 * tree-vrp.c: Include vr-values.h
976 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
977 data objects into the vr_values class.
978 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
979 (get_value_range): Make it a member function within vr_values class.
980 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
981 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
982 (op_with_constant_singleton_value_range): Likewise.
983 (extract_range_for_var_from_comparison_expr): Likewise.
984 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
985 (extract_range_from_binary_expr): Likewise.
986 (extract_range_from_unary_expr): Likewise.
987 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
988 (check_for_binary_op_overflow, extract_range_basic): Likewise.
989 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
990 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
991 (compare_name_with_value, compare_names): Likewise.
992 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
993 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
994 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
995 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
996 (extract_range_from_phi_node): Likewise.
997 (simplify_truth_ops_using_ranges): Likewise.
998 (simplify_div_or_mod_using_ranges): Likewise.
999 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
1000 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
1001 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
1002 (simplify_float_conversion_using_ranges): Likewise.
1003 (simplify_internal_call_using_ranges): Likewise.
1004 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
1005 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
1006 poitner via x_vr_values for calls into gimple folder.
1007 (vrp_initialize_lattice): Make this the vr_values ctor.
1008 (vrp_free_lattice): Make this the vr_values dtor.
1009 (set_vr_value): New function.
1010 (class vrp_prop): Add vr_values data member. Add various member
1011 functions as well as member functions that delegate to vr_values.
1012 (check_array_ref): Make a member function within vrp_prop class.
1013 (search_for_addr_array, vrp_initialize): Likewise.
1014 (vrp_finalize): Likewise. Revamp to avoid direct access to
1015 vr_value, values_propagated, etc.
1016 (check_array_bounds): Extract vrp_prop class instance pointer from
1017 walk info structure. Use it to call member functions.
1018 (check_all_array_refs): Make a member function within vrp_prop class.
1019 Smuggle class instance pointer via walk info structure.
1020 (x_vr_values): New local static.
1021 (vrp_valueize): Use x_vr_values to get class instance.
1022 (vr_valueize_1): Likewise.
1023 (class vrp_folder): Add vr_values data member. Add various member
1024 functions as well as member functions that delegate to vr_values.
1025 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
1026 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
1027 class instance from vr_values. Use it to call member functions.
1028 (vrp_dom_walker): Add vr_values data member.
1029 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
1030 instance via x_vr_values.
1031 (identify_jump_threads): Accept vr_values as argument. Store
1032 it into the walker structure.
1033 (evrp_dom_walker): Add vr_values class data member. Add various
1035 (evrp_dom_walker::try_find_new_range): Use vr_values data
1036 member to access the memory allocator.
1037 (evrp_dom_walker::before_dom_children): Store vr_values class
1038 instance into the vrp_folder class.
1039 (evrp_dom_walker::push_value_range): Rework to avoid direct
1040 access to num_vr_values and vr_value.
1041 (evrp_dom_walker::pop_value_range): Likewise.
1042 (execute_early_vrp): Remove call to vrp_initialize_lattice.
1043 Use vr_values to get to dump_all_value_ranges member function.
1044 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
1045 and simplify_cond_using_ranges_2 via vrp_prop class instance.
1046 Pass vr_values class instance down to identify_jump_threads.
1047 Remove call to vrp_free_lattice.
1048 (debug_all_value_ranges): Remove.
1050 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
1051 (vrp_folder): Likewise.
1053 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
1054 Get it from the existing bitmap instead.
1055 (vrp_intersect_ranges_1): Likewise.
1057 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1059 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1061 (store_immediate_info::store_immediate_info): Add bitnotp argument,
1062 set bit_not_p to it.
1063 (imm_store_chain_info::coalesce_immediate_stores): Break group
1064 if bit_not_p is different.
1065 (count_multiple_uses, split_group,
1066 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
1067 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
1068 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
1069 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
1070 to store_immediate_info ctor.
1072 2017-11-09 Jim Wilson <jimw@sifive.com>
1074 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
1075 (scan_prog_file): Likewise.
1077 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1079 * bb-reorder.c (max_entry_frequency): Remove.
1080 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
1081 connect_traces, push_to_next_round_p): Remove prototypes.
1082 (find_traces_1_round): Use counts only.
1083 (push_to_next_round_p): Likewise.
1084 (find_traces): Likewise.
1085 (rotate_loop): Likewise.
1086 (find_traces_1_round): Likewise.
1087 (connect_traces): Likewise.
1088 (edge_order): Likewise.
1090 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
1092 * config/arm/arm.c (output_return_instruction): Add comments to
1093 indicate requirement for cmse_nonsecure_entry return to account
1094 for the size of clearing instruction output here.
1095 (thumb_exit): Likewise.
1096 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
1097 return in hardfloat mode.
1099 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
1101 * config/rs6000/rs6000.c (machine_function): Add a bool,
1102 "toc_is_wrapped_separately".
1103 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
1104 if it wasn't explicitly set or unset, we are optimizing for speed, and
1105 doing separate shrink-wrapping.
1106 (rs6000_get_separate_components): Enable the TOC component if
1107 saving the TOC register in the prologue.
1108 (rs6000_components_for_bb): Handle the TOC component.
1109 (rs6000_emit_prologue_components): Store the TOC register where needed.
1110 (rs6000_set_handled_components): Mark TOC as handled, if handled.
1111 (rs6000_emit_prologue): Don't save the TOC if that is already done.
1113 2017-11-09 Martin Jambor <mjambor@suse.cz>
1115 * ipa-param-manipulation.c: New file.
1116 * ipa-param-manipulation.h: Likewise.
1117 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
1118 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
1119 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
1120 (ipa_parm_adjustment): Likewise.
1121 (ipa_parm_adjustment_vec): Likewise.
1122 (ipa_get_vector_of_formal_parms): Moved declaration to
1123 ipa-param-manipulation.h.
1124 (ipa_get_vector_of_formal_parm_types): Likewise.
1125 (ipa_modify_formal_parameters): Likewise.
1126 (ipa_modify_call_arguments): Likewise.
1127 (ipa_combine_adjustments): Likewise.
1128 (ipa_dump_param_adjustments): Likewise.
1129 (ipa_modify_expr): Likewise.
1130 (ipa_get_adjustment_candidate): Likewise.
1131 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
1132 ipa-param-manipulation.c.
1133 (ipa_get_vector_of_formal_parm_types): Likewise.
1134 (ipa_modify_formal_parameters): Likewise.
1135 (ipa_modify_call_arguments): Likewise.
1136 (ipa_modify_expr): Likewise.
1137 (get_ssa_base_param): Likewise.
1138 (ipa_get_adjustment_candidate): Likewise.
1139 (index_in_adjustments_multiple_times_p): Likewise.
1140 (ipa_combine_adjustments): Likewise.
1141 (ipa_dump_param_adjustments): Likewise.
1142 * tree-sra.c: Also include ipa-param-manipulation.h
1143 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
1146 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1147 Alan Hayward <alan.hayward@arm.com>
1148 David Sherwood <david.sherwood@arm.com>
1150 * doc/sourcebuild.texi (vect_masked_store): Document.
1152 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1153 Alan Hayward <alan.hayward@arm.com>
1154 David Sherwood <david.sherwood@arm.com>
1156 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
1158 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1159 Alan Hayward <alan.hayward@arm.com>
1160 David Sherwood <david.sherwood@arm.com>
1162 * doc/sourcebuild.texi (vect_variable_length): Document.
1164 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1165 Alan Hayward <alan.hayward@arm.com>
1166 David Sherwood <david.sherwood@arm.com>
1168 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
1170 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1171 Alan Hayward <alan.hayward@arm.com>
1172 David Sherwood <david.sherwood@arm.com>
1174 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
1176 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1177 Alan Hayward <alan.hayward@arm.com>
1178 David Sherwood <david.sherwood@arm.com>
1180 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1181 previously undocumented selectors.
1182 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1184 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1186 * doc/rtl.texi (const_vector): Say that elements can be
1187 const_wide_ints too.
1188 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1189 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1190 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1191 * optabs.c (expand_vector_broadcast): Likewise.
1193 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1194 Alan Hayward <alan.hayward@arm.com>
1195 David Sherwood <david.sherwood@arm.com>
1197 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1198 scaled index even if the unscaled address was invalid.
1199 Don't increase the complexity of using a scale in that case.
1201 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1202 Alan Hayward <alan.hayward@arm.com>
1203 David Sherwood <david.sherwood@arm.com>
1205 * doc/rtl.texi: Rewrite the subreg rules so that they partition
1206 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1207 UNITS_PER_WORD bytes.
1208 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1209 based on REGMODE_NATURAL_SIZE of the inner mode.
1210 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1211 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1212 * expmed.c (lowpart_bit_field_p): Divide the value up into
1213 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1214 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1215 whether something is likely to occupy more than one register.
1217 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1220 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1221 profile_count::adjust_for_ipa_scaling.
1222 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1223 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1225 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1227 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1229 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1230 (split_group): Add total_orig and total_new arguments, estimate the
1231 number of statements related to the store group without store merging
1232 and with store merging.
1233 (imm_store_chain_info::output_merged_store): Adjust split_group
1234 callers, punt if estimated number of statements with store merging
1235 is not smaller than estimated number of statements without it.
1237 (handled_load): Remove has_single_use checks.
1238 (pass_store_merging::process_store): Likewise.
1240 2017-11-09 Richard Biener <rguenther@suse.de>
1242 PR tree-optimization/82902
1243 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1245 2017-11-09 Martin Liska <mliska@suse.cz>
1248 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1251 2017-11-09 Martin Liska <mliska@suse.cz>
1253 PR tree-optimization/82669
1254 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1256 2017-11-09 Martin Liska <mliska@suse.cz>
1258 PR gcov-profile/48463
1259 * coverage.c (coverage_begin_function): Output also end locus
1260 of a function and information whether the function is
1262 * gcov-dump.c (tag_function): Parse and print the information.
1263 * gcov.c (INCLUDE_MAP): Add include.
1264 (INCLUDE_SET): Likewise.
1265 (struct line_info): Move earlier in the source file because
1266 of vector<line_info> in function_info structure.
1267 (line_info::line_info): Likewise.
1268 (line_info::has_block): Likewise.
1269 (struct source_info): Add new member index.
1270 (source_info::get_functions_at_location): New function.
1271 (function_info::group_line_p): New function.
1272 (output_intermediate_line): New function.
1273 (output_intermediate_file): Use the mentioned function.
1274 (struct function_start): New.
1275 (struct function_start_pair_hash): Likewise.
1276 (process_file): Add code that identifies group functions.
1277 Assign lines either to global or function scope.
1278 (generate_results): Skip artificial functions.
1279 (find_source): Assign index for each source file.
1280 (read_graph_file): Read new flag artificial and end_line.
1281 (add_line_counts): Assign it either to global of function scope.
1282 (accumulate_line_counts): Isolate core of the function to
1283 accumulate_line_info and call it for both function and global
1285 (accumulate_line_info): New function.
1286 (output_line_beginning): Fix GNU coding style.
1287 (print_source_line): New function.
1288 (output_line_details): Likewise.
1289 (output_function_details): Likewise.
1290 (output_lines): Iterate both source (global) scope and function
1292 (struct function_line_start_cmp): New class.
1293 * doc/gcov.texi: Reflect changes in documentation.
1295 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1298 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1299 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1300 and similarly for not instead of neg.
1302 2017-11-08 Andi Kleen <ak@linux.intel.com>
1304 * config/i386/i386.opt: Add -mforce-indirect-call.
1305 * config/i386/predicates.md: Check for flag_force_indirect_call.
1306 * doc/invoke.texi: Document -mforce-indirect-call
1308 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1310 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1312 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1313 config/riscv/riscv.c (predict.h): New include.
1314 (riscv_slow_unaligned_access_p): No longer static.
1315 (riscv_block_move_straight): Add require.
1316 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1318 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1321 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1322 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1323 nonimmediate_operand predicate for operand 1 instead of
1326 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1328 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1330 * config/aarch64/constraints.md (Uml): New constraint.
1331 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1334 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1336 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1337 of two vec_duplicates into a vec_concat.
1339 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1341 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1342 Simplify vec_merge of vec_duplicate and vec_concat.
1343 * config/aarch64/constraints.md (Utq): New constraint.
1344 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1347 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1349 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1350 Simplify vec_merge of vec_duplicate and const_vector.
1351 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1353 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1354 mode iterator. Update predicate on operand 1 to
1355 handle non-const_vec constants. Delete constraints.
1356 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1358 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1360 PR tree-optimization/78821
1361 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1363 (store_operand_info::store_operand_info): Initialize it to false.
1364 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1365 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1366 store in the group, and if chain_info is non-NULL, to ignore altogether
1368 (compatible_load_p): Fail if bit_not_p does not match.
1369 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1370 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1371 (pass_store_merging::process_store): Adjust
1372 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1373 call terminate_all_aliasing_chains newly when adding a store into
1374 a chain with non-NULL chain_info.
1376 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1378 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1380 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1382 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1384 (aarch64_layout_frame): Initialise emit_frame_chain.
1385 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1386 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1388 2017-11-08 Martin Liska <mliska@suse.cz>
1390 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1391 of gimple_call_internal_p.
1393 2017-11-07 Tom de Vries <tom@codesourcery.com>
1395 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1397 2017-11-07 Tom de Vries <tom@codesourcery.com>
1399 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1402 2017-11-08 Martin Liska <mliska@suse.cz>
1405 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1407 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1409 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1411 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1414 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1415 (*cmp<mode>_ccz_1): New insn with $k alternative.
1418 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1419 RTX_COMM_COMPARE as commutative as well.
1420 (ix86_binary_operator_ok): Formatting fix.
1421 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1422 *<code><mode>3<mask_name><round_saeonly_name>,
1423 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1424 *<s>mul<mode>3_highpart<mask_name>,
1425 *vec_widen_umult_even_v16si<mask_name>,
1426 *vec_widen_umult_even_v8si<mask_name>,
1427 *vec_widen_umult_even_v4si<mask_name>,
1428 *vec_widen_smult_even_v16si<mask_name>,
1429 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1430 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1431 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1432 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1433 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1434 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1435 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1436 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1437 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1438 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1439 ix86_binary_operator_ok. Formatting fixes.
1440 (*<plusminus_insn><mode>3<mask_name><round_name>,
1441 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1444 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1446 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1447 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1448 eq and ne if TARGET_ISEL.
1449 (cmp): New code_iterator.
1450 (UNS, UNSU_, UNSIK): New code_attrs.
1451 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1452 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
1454 ("eq<mode>3"): ... this.
1455 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1457 ("ne<mode>3"): ... this.
1459 2017-11-07 Julia Koval <julia.koval@intel.com>
1462 * common/config/i386/i386-common.c
1463 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1464 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1465 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1466 * config/i386/i386.opt: Ditto.
1467 * config/i386/i386.c (ix86_target_string): Ditto.
1468 (ix86_option_override_internal): Ditto.
1469 (ix86_init_mpx_builtins): Move MPX to args2.
1470 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
1471 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1472 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1473 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1474 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1475 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1476 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
1478 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
1481 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
1482 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
1483 (zero-extendsidi peephole2): Remove peephole.
1485 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1488 * stor-layout.c: Include attribs.h.
1489 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
1490 explicit lookup of "aligned" attribute.
1492 2017-11-07 Andrew Waterman <andrew@sifive.com>
1494 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
1495 (riscv_expand_block_move): Likewise.
1496 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
1498 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
1499 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
1500 * config/riscv/riscv.c (riscv_block_move_straight): New function.
1501 (riscv_adjust_block_mem): Likewise.
1502 (riscv_block_move_loop): Likewise.
1503 (riscv_expand_block_move): Likewise.
1504 gcc/config/riscv/riscv.md (movmemsi): New pattern.
1506 2017-11-07 Michael Clark <michaeljclark@mac.com>
1508 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
1509 (MUSL_DYNAMIC_LINKER): Likewise.
1511 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
1513 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
1514 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
1517 2017-11-07 Richard Biener <rguenther@suse.de>
1519 * match.pd: Fix build.
1521 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
1522 Jackson Woodruff <jackson.woodruff@arm.com>
1524 PR tree-optimization/71026
1525 * match.pd: Canonicalize negate in division.
1527 2017-11-07 Sudakshina Das <sudi.das@arm.com>
1530 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
1532 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1534 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
1535 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
1537 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1539 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
1540 non-scalar integral types.
1541 * match.pd (negate_expr_p): Handle MINUS_EXPR.
1542 (-(A-B), -(~A)): New transformations.
1544 2017-11-07 Tom de Vries <tom@codesourcery.com>
1546 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
1547 semicolon after "do {} while (0)".
1548 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1549 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1550 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1551 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1552 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1554 2017-11-07 Tom de Vries <tom@codesourcery.com>
1556 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
1557 after "do {} while (0)".
1558 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1559 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1560 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1561 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1562 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1564 2017-11-07 Tom de Vries <tom@codesourcery.com>
1567 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
1569 (arm_rtx_costs_internal): Add missing semicolon after
1570 HANDLE_NARROW_SHIFT_ARITH call.
1572 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1574 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1575 disable isel if it was not set explicitly.
1577 2017-11-06 James Bowman <james.bowman@ftdichip.com>
1579 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
1581 (add_type_attribute) likewise.
1583 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
1585 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
1586 of struct ix86_frame.
1587 (ix86_initial_elimination_offset): Likewise.
1588 (ix86_expand_split_stack_prologue): Likewise.
1590 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
1592 * tree-vrp.h (enum value_range_type): Update stale comment.
1594 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1595 Alan Hayward <alan.hayward@arm.com>
1596 David Sherwood <david.sherwood@arm.com>
1598 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
1599 (aarch64_expand_vec_perm_const): Take the number of units too.
1600 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
1601 (aarch64_expand_vec_perm_const): Likewise.
1602 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
1603 (vec_perm<mode>): Update accordingly.
1605 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1606 Alan Hayward <alan.hayward@arm.com>
1607 David Sherwood <david.sherwood@arm.com>
1609 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
1610 Take the number of units too.
1611 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
1612 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
1613 but check for a vector mode before rather than after the call.
1614 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
1615 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
1616 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
1617 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
1618 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
1619 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
1620 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
1621 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
1622 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
1623 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
1624 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
1625 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
1626 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
1627 (aarch64_sqdmull2_n<mode>): Update accordingly.
1629 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1630 Alan Hayward <alan.hayward@arm.com>
1631 David Sherwood <david.sherwood@arm.com>
1633 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
1634 the number of units too.
1635 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
1636 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
1637 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
1638 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
1639 (vec_store_lanesxi<mode>): Update accordingly.
1641 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1642 Alan Hayward <alan.hayward@arm.com>
1643 David Sherwood <david.sherwood@arm.com>
1645 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
1646 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
1647 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
1648 of units rather than the mode.
1649 * config/aarch64/iterators.md (nunits): New mode attribute.
1650 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
1651 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
1652 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
1653 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
1654 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1655 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
1656 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
1657 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
1658 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
1659 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
1660 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1661 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
1662 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
1663 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
1664 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1665 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1666 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
1667 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
1668 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
1669 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
1670 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
1671 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
1672 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
1673 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1674 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
1675 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
1676 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1677 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
1678 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
1679 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
1680 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
1681 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
1682 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
1683 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
1684 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
1685 (aarch64_simd_vec_setv2di): Likewise.
1687 2017-11-06 Carl Love <cel@us.ibm.com>
1689 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
1691 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
1692 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
1694 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
1696 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
1697 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
1698 Add power 8 macro expansions.
1699 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
1700 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
1701 power 8 instructions. (VSX_XXBR): Add iterator.
1703 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
1705 * config/arm/arm.md (predicable_short_it): Change default to "no",
1706 improve documentation, remove uses that are identical to the default.
1707 (enabled_for_depr_it): Rename to enabled_for_short_it.
1708 * gcc/config/arm/arm-fixed.md (predicable_short_it):
1709 Remove default uses.
1710 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
1711 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
1712 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
1713 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
1715 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
1718 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
1719 float128 helper macros, which are no longer used after deleting
1720 the old 'q' built-in functions, and moving the round to odd
1721 built-in functions to being special built-in functions.
1722 (BU_FLOAT128_2): Likewise.
1723 (BU_FLOAT128_1_HW): Likewise.
1724 (BU_FLOAT128_2_HW): Likewise.
1725 (BU_FLOAT128_3_HW): Likewise.
1726 (FABSQ): Delete old 'q' built-in functions.
1727 (COPYSIGNQ): Likewise.
1728 (SQRTF128_ODD): Move round to odd built-in functions to be
1729 special built-in functions, so that we can handle
1730 -mabi=ieeelongdouble.
1731 (TRUNCF128_ODD): Likewise.
1732 (ADDF128_ODD): Likewise.
1733 (SUBF128_ODD): Likewise.
1734 (MULF128_ODD): Likewise.
1735 (DIVF128_ODD): Likewise.
1736 (FMAF128_ODD): Likewise.
1737 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
1738 built-in names to 'f128'.
1739 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
1740 old 'q' built-in functions, as the machine independent code for
1741 'f128' built-in functions handles this.
1742 (rs6000_expand_builtin): Add expansion for float128 round to odd
1743 functions, keying off on -mabi=ieeelongdouble of whether to use
1744 the KFmode or TFmode variant.
1745 (rs6000_init_builtins): Initialize the _Float128 round to odd
1747 * doc/extend.texi (PowerPC Built-in Functions): Document the old
1748 _Float128 'q' built-in functions are now mapped into the new
1749 'f128' built-in functions.
1751 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
1753 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
1754 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
1756 2017-11-06 David Malcolm <dmalcolm@redhat.com>
1759 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
1760 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
1761 * toplev.c: Include "ipa-fnsummary.h".
1762 (toplev::finalize): Call ipa_fnsummary_c_finalize.
1764 2017-11-06 Jakub Jelinek <jakub@redhat.com>
1766 PR tree-optimization/82838
1767 * gimple-ssa-store-merging.c
1768 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
1769 on a separate gimple_seq which is then appended to seq.
1771 2017-11-06 Jeff Law <law@redhat.com>
1774 * config/i386/i386.c (PROBE_INTERVAL): Remove.
1775 (get_probe_interval): New functions.
1776 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
1777 (ix86_adjust_stack_and_probe): Likewise.
1778 (output_adjust_stack_and_probe): Likewise.
1779 (ix86_emit_probe_stack_range): Likewise.
1780 (ix86_expand_prologue): Likewise.
1782 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1784 PR tree-optimization/82816
1785 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
1786 if the modes of the two types are the same.
1787 (convert_plusminus_to_widen): Likewise.
1789 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1791 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
1793 (usadv16qi): New define_expand.
1794 (usadv8hi): New define_expand.
1796 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
1799 * ipa-inline-transform.c (update_noncloned_frequencies): Always
1801 (inline_transform): Likewise.
1802 * predict.c (counts_to_freqs): Remove useless conditional.
1803 * profile-count.h (profile_count::apply_scale): Move sanity check.
1804 * tree-inline.c (copy_bb): Always scale.
1805 (copy_cfg_body): Likewise.
1807 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
1810 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
1812 (*cmp_ite0): Add enabled_for_depr_it attribute.
1813 (*cmp_ite1): Likewise.
1815 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1817 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
1820 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1822 * tree-vrp.c (vrp_int_const_binop): Return true on success and
1823 return the value by pointer.
1824 (extract_range_from_multiplicative_op_1): Update accordingly.
1825 Return as soon as an operation fails.
1827 2017-11-05 Tom de Vries <tom@codesourcery.com>
1830 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
1831 (DEF_SANITIZER_BUILTIN): ... here.
1832 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
1833 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
1835 2017-11-05 Tom de Vries <tom@codesourcery.com>
1838 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
1840 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
1841 ASM_OUTPUT_BEFORE_CASE_LABEL call.
1842 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
1844 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1845 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1846 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1848 2017-11-05 Tom de Vries <tom@codesourcery.com>
1851 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
1854 2017-11-04 Michael Clark <michaeljclark@mac.com>
1856 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
1857 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
1859 (*addsi3_extended): Likewise.
1860 (*addsi3_extended2): Likewise.
1861 (<optab>si3): Likewise.
1862 (<optab>di3): Likewise.
1863 (<optab><mode>3): Likewise.
1864 (<*optabe>si3_internal): Likewise.
1865 (zero_extendqi<SUPERQI:mode>2): Likewise.
1866 (*add<mode>hi3): Likewise.
1867 (*xor<mode>hi3): Likewise.
1868 (<optab>di3): Likewise.
1869 (*<optab>si3_extend): Likewise.
1870 (*sge<u>_<X:mode><GPR:mode>): Likewise.
1871 (*slt<u>_<X:mode><GPR:mode>): Likewise.
1872 (*sle<u>_<X:mode><GPR:mode>): Likewise.
1874 2017-11-04 Andrew Waterman <andrew@sifive.com>
1876 * config/riscv/riscv.c (riscv_option_override): Conditionally set
1877 TARGET_STRICT_ALIGN based upon -mtune argument.
1879 2017-11-04 Andrew Waterman <andrew@sifive.com>
1881 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
1883 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
1885 config/i386/i386.c (choose_basereg): Use optional scratch
1886 register and add assertion.
1887 (x86_emit_outlined_ms2sysv_save): Use scratch register when
1888 needed, and don't allocate stack.
1889 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
1890 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
1891 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
1893 2017-11-03 Jeff Law <law@redhat.com>
1895 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
1896 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
1897 to probe at the start of a noreturn function.
1899 2017-11-03 Jakub Jelinek <jakub@redhat.com>
1901 PR tree-optimization/78821
1902 * gimple-ssa-store-merging.c: Update the file comment.
1903 (MAX_STORE_ALIAS_CHECKS): Define.
1904 (struct store_operand_info): New type.
1905 (store_operand_info::store_operand_info): New constructor.
1906 (struct store_immediate_info): Add rhs_code and ops data members.
1907 (store_immediate_info::store_immediate_info): Add rhscode, op0r
1908 and op1r arguments to the ctor, initialize corresponding data members.
1909 (struct merged_store_group): Add load_align_base and load_align
1911 (merged_store_group::merged_store_group): Initialize them.
1912 (merged_store_group::do_merge): Update them.
1913 (merged_store_group::apply_stores): Pick the constant for
1914 encode_tree_to_bitpos from one of the two operands, or skip
1915 encode_tree_to_bitpos if neither operand is a constant.
1916 (class pass_store_merging): Add process_store method decl. Remove
1917 bool argument from terminate_all_aliasing_chains method decl.
1918 (pass_store_merging::terminate_all_aliasing_chains): Remove
1919 var_offset_p argument and corresponding handling.
1920 (stmts_may_clobber_ref_p): New function.
1921 (compatible_load_p): New function.
1922 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
1923 if there is overlap and rhs_code is not INTEGER_CST. For
1924 non-overlapping stores terminate group if rhs is not mergeable.
1925 (get_alias_type_for_stmts): Change first argument from
1926 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
1927 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
1928 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
1930 (get_location_for_stmts): Change first argument from
1931 auto_vec<gimple *> & to vec<gimple *> &.
1932 (struct split_store): Remove orig_stmts data member, add orig_stores.
1933 (split_store::split_store): Create orig_stores rather than orig_stmts.
1934 (find_constituent_stmts): Renamed to ...
1935 (find_constituent_stores): ... this. Change second argument from
1936 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
1937 to info structures rather than the statements.
1938 (split_group): Rename ALLOW_UNALIGNED argument to
1939 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
1940 it. Adjust find_constituent_stores caller.
1941 (imm_store_chain_info::output_merged_store): Handle rhs_code other
1942 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
1943 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
1944 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
1945 (mem_valid_for_store_merging): New function.
1946 (handled_load): New function.
1947 (pass_store_merging::process_store): New method.
1948 (pass_store_merging::execute): Use process_store method. Adjust
1949 terminate_all_aliasing_chains caller.
1951 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1953 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
1954 Return true for more constants, symbols and label references.
1955 (aarch64_valid_floating_const): Remove unused function.
1957 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1960 * config/aarch64/aarch64.c (aarch64_layout_frame):
1961 Undo forcing of LR at bottom of frame.
1963 2017-11-03 Jeff Law <law@redhat.com>
1966 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
1967 for int_registers_saved.
1969 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
1970 extracted from tree-ssa-dom.c.
1971 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
1972 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
1973 (record_equivalences_from_incoming_edge): Add additional argument
1974 to single_pred_edge_ignoring_loop_edges call.
1975 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
1976 (uncprop_dom_walker::before_dom_children): Add additional argument
1977 to single_pred_edge_ignoring_loop_edges call.
1978 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
1979 single_pred_edge_ignoring_loop_edges rather than open coding.
1980 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
1982 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
1984 * match.pd (-(-A)): Rewrite.
1986 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1988 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
1989 (rs6000_emit_int_cmove): New declaration.
1990 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
1991 (rs6000_emit_sISEL): Delete.
1992 (rs6000_emit_int_cmove): Make non-static.
1993 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
1994 instead of rs6000_emit_sISEL.
1996 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
1998 * asan.c (create_cond_insert_point): Maintain profile.
1999 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
2001 * basic-block.h (struct basic_block_def): Remove frequency.
2002 (EDGE_FREQUENCY): Use to_frequency
2003 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
2005 (find_traces): Update to use to_frequency.
2006 (find_traces_1_round): Likewise; use only IPA counts.
2007 (bb_to_key): Likewise.
2008 (connect_traces): Use IPA counts only.
2009 (copy_bb_p): Update to use to_frequency.
2010 (fix_up_crossing_landing_pad): Likewise.
2011 (sanitize_hot_paths): Likewise.
2012 * bt-load.c (basic_block_freq): Likewise.
2013 * cfg.c (init_flow): Set count_max to uninitialized.
2014 (check_bb_profile): Remove frequencies; check counts.
2015 (dump_bb_info): Do not dump frequencies.
2016 (update_bb_profile_for_threading): Update counts only.
2017 (scale_bbs_frequencies_int): Likewise.
2018 (MAX_SAFE_MULTIPLIER): Remove.
2019 (scale_bbs_frequencies_gcov_type): Update counts only.
2020 (scale_bbs_frequencies_profile_count): Update counts only.
2021 (scale_bbs_frequencies): Update counts only.
2022 * cfg.h (struct control_flow_graph): Add count-max.
2023 (update_bb_profile_for_threading): Update prototype.
2024 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
2025 (find_many_sub_basic_blocks): Likewise.
2026 * cfgcleanup.c (try_forward_edges): Likewise.
2027 (try_crossjump_to_edge): Likewise.
2028 * cfgexpand.c (expand_gimple_cond): Likewise.
2029 (expand_gimple_tailcall): Likewise.
2030 (construct_init_block): Likewise.
2031 (construct_exit_block): Likewise.
2032 * cfghooks.c (verify_flow_info): Check consistency of counts.
2033 (dump_bb_for_graph): Do not dump frequencies.
2034 (split_block_1): Do not update frequencies.
2035 (split_edge): Do not update frequencies.
2036 (make_forwarder_block): Do not update frequencies.
2037 (duplicate_block): Do not update frequencies.
2038 (account_profile_record): Do not update frequencies.
2039 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
2040 for global heuristics.
2041 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
2042 (expected_loop_iterations_unbounded): Use counts only.
2043 * cfgloopmanip.c (scale_loop_profile): Simplify.
2044 (create_empty_loop_on_edge): Simplify
2046 (duplicate_loop_to_header_edge): Simplify
2047 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
2048 (update_br_prob_note): Take care of removing note when profile
2050 (relink_block_chain): Do not dump frequency.
2051 (rtl_account_profile_record): Use to_frequency.
2052 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
2053 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
2054 (cgraph_update_edges_for_call_stmt_node): Likewise.
2055 (cgraph_edge::verify_count_and_frequency): Update.
2056 (cgraph_node::verify_node): Temporarily disable frequency verification.
2057 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
2058 to_cgraph_frequency.
2059 (cgraph_edge::rebuild_edges): Convert to ipa counts.
2060 * cgraphunit.c (init_lowered_empty_function): Do not initialize
2062 (cgraph_node::expand_thunk): Update profile.
2063 * except.c (dw2_build_landing_pads): Do not update frequency.
2064 * final.c (compute_alignments): Use to_frequency.
2065 (dump_basic_block_info): Do not dump frequency.
2066 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
2067 (dump_gimple_bb_header): Do not dump frequency.
2068 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
2070 * gimple-streamer-in.c (input_bb): Do not stream frequency.
2071 * gimple-streamer-out.c (output_bb): Do not stream frequency.
2072 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
2073 (init_before_recovery): Do not update frequency.
2074 (sched_create_recovery_edges): Do not update frequency.
2075 * hsa-gen.c (convert_switch_statements): Do not update frequency.
2076 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
2077 (ipa_cp_c_finalize): Set max_count to uninitialized.
2078 * ipa-fnsummary.c (get_minimal_bb): Use counts.
2079 (param_change_prob): Use counts.
2080 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
2082 * ipa-split.c (consider_split): Use to_frequency.
2083 (split_function): Use to_frequency.
2084 * ira-build.c (loop_compare_func): Likewise.
2085 (mark_loops_for_removal): Likewise.
2086 (mark_all_loops_for_removal): Likewise.
2087 * loop-doloop.c (doloop_modify): Do not update frequency.
2088 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
2090 * lto-streamer-in.c (input_function): Update count_max.
2091 * omp-expand.c (expand_omp_taskreg): Update count_max.
2092 * omp-simd-clone.c (simd_clone_adjust): Update profile.
2093 * predict.c (maybe_hot_frequency_p): Use to_frequency.
2094 (maybe_hot_count_p): Use ipa counts only.
2095 (maybe_hot_bb_p): Simplify.
2096 (maybe_hot_edge_p): Simplify.
2097 (probably_never_executed): Do not take frequency argument.
2098 (probably_never_executed_bb_p): Do not pass frequency.
2099 (probably_never_executed_edge_p): Likewise.
2100 (combine_predictions_for_bb): Check that profile is nonzero.
2101 (propagate_freq): Do not set frequency.
2102 (drop_profile): Simplify.
2103 (counts_to_freqs): Simplify.
2104 (expensive_function_p): Use to_frequency.
2105 (propagate_unlikely_bbs_forward): Simplify.
2106 (determine_unlikely_bbs): Simplify.
2107 (estimate_bb_frequencies): Add hack to silence graphite issues.
2108 (compute_function_frequency): Use ipa counts.
2109 (pass_profile::execute): Update.
2110 (rebuild_frequencies): Use counts only.
2111 (force_edge_cold): Use counts only.
2112 * profile-count.c (profile_count::dump): Dump new count types.
2113 (profile_count::differs_from_p): Check compatiblity.
2114 (profile_count::to_frequency): New function.
2115 (profile_count::to_cgraph_frequency): New function.
2116 * profile-count.h (struct function): Declare.
2117 (enum profile_quality): Add profile_guessed_local and
2118 profile_guessed_global0.
2119 (class profile_proability): Decrease number of bits to 29;
2120 update from_reg_br_prob_note and to_reg_br_prob_note.
2121 (class profile_count: Update comment; decrease number of bits
2122 to 61. Check compatibility.
2123 (profile_count::compatible_p): New private member function.
2124 (profile_count::ipa_p): New member function.
2125 (profile_count::operator<): Handle global zero correctly.
2126 (profile_count::operator>): Handle global zero correctly.
2127 (profile_count::operator<=): Handle global zero correctly.
2128 (profile_count::operator>=): Handle global zero correctly.
2129 (profile_count::nonzero_p): New member function.
2130 (profile_count::force_nonzero): New member function.
2131 (profile_count::max): New member function.
2132 (profile_count::apply_scale): Handle IPA scalling.
2133 (profile_count::guessed_local): New member function.
2134 (profile_count::global0): New member function.
2135 (profile_count::ipa): New member function.
2136 (profile_count::to_frequency): Declare.
2137 (profile_count::to_cgraph_frequency): Declare.
2138 * profile.c (OVERLAP_BASE): Delete.
2139 (compute_frequency_overlap): Delete.
2140 (compute_branch_probabilities): Do not use compute_frequency_overlap.
2141 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
2142 * sched-ebb.c (rank): Use counts only.
2143 * shrink-wrap.c (handle_simple_exit): Use counts only.
2144 (try_shrink_wrapping): Use counts only.
2145 (place_prologue_for_one_component): Use counts only.
2146 * tracer.c (find_best_predecessor): Use to_frequency.
2147 (find_trace): Use to_frequency.
2148 (tail_duplicate): Use to_frequency.
2149 * trans-mem.c (expand_transaction): Do not update frequency.
2150 * tree-call-cdce.c: Do not update frequency.
2151 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
2152 (gimple_merge_blocks): Likewise.
2153 (gimple_split_edge): Likewise.
2154 (gimple_duplicate_sese_region): Likewise.
2155 (gimple_duplicate_sese_tail): Likewise.
2156 (move_sese_region_to_fn): Likewise.
2157 (gimple_account_profile_record): Likewise.
2158 (insert_cond_bb): Likewise.
2159 * tree-complex.c (expand_complex_div_wide): Likewise.
2160 * tree-eh.c (lower_resx): Update profile.
2161 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
2163 (initialize_cfun): Do not initialize frequencies
2164 (freqs_to_counts): Delete.
2165 (copy_cfg_body): Ignore count parameter.
2166 (copy_body): Update.
2167 (expand_call_inline): Update count_max.
2168 (optimize_inline_calls): Update count_max.
2169 (tree_function_versioning): Update count_max.
2170 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
2171 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
2173 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
2174 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
2175 (try_peel_loop): Likewise.
2176 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
2178 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
2179 (tree_transform_and_unroll_loop): Do not use frequencies
2180 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2181 Use reliable prediction only.
2182 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2183 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2184 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2185 probability scaling.
2186 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2187 not update frequency
2188 (any_remaining_duplicated_blocks): Likewise.
2189 (update_profile): Likewise.
2190 (estimated_freqs_path): Delete.
2191 (freqs_to_counts_path): Delete.
2192 (clear_counts_path): Delete.
2193 (ssa_fix_duplicate_block_edges): Likewise.
2194 (duplicate_thread_path): Likewise.
2195 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2196 * tree-tailcall.c (decrease_profile): Do not update frequency.
2197 (eliminate_tail_call): Likewise.
2198 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2199 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2200 (optimize_mask_stores): Likewise.
2201 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2202 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2203 (ubsan_expand_ptr_ifn): Update profile.
2204 * value-prof.c (gimple_ic): Simplify.
2205 * value-prof.h (gimple_ic): Update prototype.
2206 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2207 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2209 (want_inline_self_recursive_call_p): Likewise.
2210 (resolve_noninline_speculation): Only cummulate defined counts.
2211 (inline_small_functions): Use nonzero_p.
2212 (ipa_inline): Do not access freed node.
2214 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2216 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2217 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2219 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2221 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2222 non-legitimate address.
2224 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2226 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2227 (*lt0_<mode>di, *lt0_<mode>si): New.
2229 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2231 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2232 TARGET_PAIRED_FLOAT.
2234 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2235 Jim Wilson <jim.wilson@linaro.org>
2237 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2238 * config/aarch64/aarch64-tune.md: Regenerated.
2239 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2240 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2242 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2244 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2245 (arc_expand_prologue): Restore blink for millicode.
2246 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2248 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2251 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2252 gen_vec_duplicate after forcing the scalar into a register.
2254 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2256 * combine (try_combine): Print the insns input to try_combine to the
2259 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2262 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2263 Remove second argument from aarch64_process_target_attr call.
2264 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2266 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2268 (aarch64_handle_attr_arch): Remove second argument.
2269 (aarch64_handle_attr_cpu): Ditto.
2270 (aarch64_handle_attr_tune): Ditto.
2271 (aarch64_handle_attr_isa_flags): Ditto.
2272 (aarch64_process_one_target_attr): Ditto.
2273 (aarch64_process_target_attr): Ditto.
2274 (aarch64_option_valid_attribute_p): Remove second argument.
2275 on aarch64_process_target_attr call.
2277 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2279 * diagnostic.c: Include "selftest-diagnostic.h".
2280 (selftest::assert_location_text): New function.
2281 (selftest::test_diagnostic_get_location_text): New function.
2282 (selftest::diagnostic_c_tests): Call it.
2284 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2286 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2287 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2288 (class selftest::test_diagnostic_context): Move to...
2289 * selftest-diagnostic.c: New file.
2290 * selftest-diagnostic.h: New file.
2292 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2294 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2295 offset range for FT32B.
2296 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2297 * config/ft32/ft32.md: Add TARGET_NOPM.
2298 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2299 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2301 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2303 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2305 2017-11-02 Jeff Law <law@redhat.com>
2307 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2308 virtual keyword on FINAL OVERRIDE members.
2310 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2311 virtuals together. Add virtual destructor.
2312 (substitute_and_fold_engine): Similarly.
2314 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2316 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2318 2017-11-02 Richard Biener <rguenther@suse.de>
2320 PR tree-optimization/82795
2321 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2323 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2325 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2327 (gcc_GAS_CHECK_FEATURE): Remove.
2328 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2329 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2330 * configure: Regenerate.
2332 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2334 * config/arc/arc.c (hwloop_optimize): Account for empty
2337 2017-11-02 Richard Biener <rguenther@suse.de>
2340 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2341 Truncate ARRAY_REF index and element size.
2343 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2345 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2347 2017-11-01 Jeff Law <law@redhat.com>
2349 * tree-ssa-ccp.c (ccp_folder): New class derived from
2350 substitute_and_fold_engine.
2351 (ccp_folder::get_value): New member function.
2352 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2353 (ccp_fold_stmt): Remove prototype.
2354 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2355 * tree-ssa-copy.c (copy_folder): New class derived from
2356 substitute_and_fold_engine.
2357 (copy_folder::get_value): Renamed from get_value.
2358 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2359 * tree-vrp.c (vrp_folder): New class derived from
2360 substitute_and_fold_engine.
2361 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2362 (vrp_folder::get_value): New member function.
2363 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2364 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2365 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2366 provide a class interface to folder/substitute routines.
2367 (ssa_prop_fold_stmt_fn): Remove typedef.
2368 (ssa_prop_get_value_fn): Likewise.
2369 (subsitute_and_fold): Remove prototype.
2370 (replace_uses_in): Likewise.
2371 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2372 Renamed from replace_uses_in. Call the virtual member function
2373 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2374 (substitute_and_fold_dom_walker): Remove initialization of
2375 data member entries for calbacks. Add substitute_and_fold_engine
2376 member and initialize it.
2377 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2378 member functions for get_value, replace_phi_args_in c
2379 replace_uses_in, and fold_stmt calls.
2380 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2381 substitute_and_fold. Remove assert. Update ctor call.
2383 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2384 (ssa_prop_visit_phi_fn): Likewise.
2385 (class ssa_propagation_engine): New class to provide an interface
2387 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2389 (ssa_prop_visit_phi): Likewise.
2390 (ssa_propagation_engine::simulate_stmt): Moved into class.
2391 Call visit_phi/visit_stmt from the class rather than via
2392 file scoped static variables.
2393 (ssa_propagation_engine::simulate_block): Moved into class.
2394 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2395 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2396 set file scoped statics for the visit_stmt/visit_phi callbacks.
2397 * tree-complex.c (complex_propagate): New class derived from
2398 ssa_propagation_engine.
2399 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2400 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2401 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2403 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2404 ssa_propagation_engine.
2405 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2406 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2407 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2408 * tree-ssa-copy.c (copy_prop): New class derived from
2409 ssa_propagation_engine.
2410 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2411 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2412 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2413 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2414 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2415 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2416 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2418 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2420 PR rtl-optimization/82778
2421 PR rtl-optimization/82597
2422 * compare-elim.c (struct comparison): Add in_a_setter field.
2423 (find_comparison_dom_walker::before_dom_children): Remove killed
2424 bitmap and df_simulate_find_defs call, instead walk the defs.
2425 Compute last_setter and initialize in_a_setter. Merge definitions
2426 with first initialization for a few variables.
2427 (try_validate_parallel): Use insn_invalid_p instead of
2428 recog_memoized. Return insn rather than just the pattern.
2429 (try_merge_compare): Fix up comment. Don't uselessly test if
2430 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2432 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2435 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2436 Alan Hayward <alan.hayward@arm.com>
2437 David Sherwood <david.sherwood@arm.com>
2439 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2440 aarch64_hard_regno_nregs to get the number of registers
2443 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2444 Alan Hayward <alan.hayward@arm.com>
2445 David Sherwood <david.sherwood@arm.com>
2447 * config/aarch64/constraints.md (Upl): Rename to...
2449 * config/aarch64/aarch64.md
2450 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
2453 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2454 Alan Hayward <alan.hayward@arm.com>
2455 David Sherwood <david.sherwood@arm.com>
2457 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2458 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2461 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2462 Alan Hayward <alan.hayward@arm.com>
2463 David Sherwood <david.sherwood@arm.com>
2465 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2466 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2467 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2469 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2470 of a single element.
2471 * config/aarch64/iterators.md: Add a comment above the permute
2472 unspecs to say that they are generated directly by
2473 aarch64_expand_vec_perm_const.
2474 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
2476 2017-11-01 Nathan Sidwell <nathan@acm.org>
2478 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
2480 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2482 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
2483 and medany code models, and describe what they do.
2485 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2487 Revert accidental duplicate:
2489 * combine.c (can_change_dest_mode): Reject changes in
2490 REGMODE_NATURAL_SIZE.
2492 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
2494 PR rtl-optimization/64682
2495 PR rtl-optimization/69567
2496 PR rtl-optimization/69737
2497 PR rtl-optimization/82683
2498 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
2499 register mentioned in the note, drop the note, unless it came from I3,
2500 in which case it should go to I3 again.
2502 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2504 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
2505 and return false if not.
2506 (clear_bytes_written_by, live_bytes_read): Update accordingly.
2508 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2510 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
2511 range is known to be empty.
2513 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2514 Alan Hayward <alan.hayward@arm.com>
2515 David Sherwood <david.sherwood@arm.com>
2517 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
2518 and CONST_VECTOR_NUNITS instead of computing the number of units from
2519 the byte sizes of the vector and element.
2520 (simplify_binary_operation_1): Likewise.
2521 (simplify_const_binary_operation): Likewise.
2522 (simplify_ternary_operation): Likewise.
2524 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2525 Alan Hayward <alan.hayward@arm.com>
2526 David Sherwood <david.sherwood@arm.com>
2528 * var-tracking.c (INT_MEM_OFFSET): Replace with...
2529 (int_mem_offset): ...this new function.
2530 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
2531 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
2532 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
2535 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2536 Alan Hayward <alan.hayward@arm.com>
2537 David Sherwood <david.sherwood@arm.com>
2539 * lower-subreg.c (interesting_mode_p): New function.
2540 (compute_costs, find_decomposable_subregs, decompose_register)
2541 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
2542 (resolve_clobber, dump_choices): Use it.
2544 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2545 Alan Hayward <alan.hayward@arm.com>
2546 David Sherwood <david.sherwood@arm.com>
2548 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
2550 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2551 Alan Hayward <alan.hayward@arm.com>
2552 David Sherwood <david.sherwood@arm.com>
2554 * alias.c (find_base_value, find_base_term): Only process integer
2555 truncations. Check the precision rather than the size.
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 * machmode.h (is_narrower_int_mode): New function
2562 * optabs.c (expand_float, expand_fix): Use it.
2563 * dwarf2out.c (rotate_loc_descriptor): Likewise.
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 * rtl.h (narrower_subreg_mode): New function.
2570 * ira-color.c (update_costs_from_allocno): Use it.
2572 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2573 Alan Hayward <alan.hayward@arm.com>
2574 David Sherwood <david.sherwood@arm.com>
2576 * optabs-query.h (convert_optab_p): New function, split out from...
2577 (convert_optab_handler): ...here.
2578 (widening_optab_handler): Delete.
2579 (find_widening_optab_handler): Remove permit_non_widening parameter.
2580 (find_widening_optab_handler_and_mode): Likewise. Provide an
2581 override that operates on mode class wrappers.
2582 * optabs-query.c (widening_optab_handler): Delete.
2583 (find_widening_optab_handler_and_mode): Remove permit_non_widening
2584 parameter. Assert that the two modes are the same class and that
2585 the "from" mode is narrower than the "to" mode. Use
2586 convert_optab_handler instead of widening_optab_handler.
2587 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
2588 instead of widening_optab_handler.
2589 * expr.c (expand_expr_real_2): Update calls to
2590 find_widening_optab_handler.
2591 * optabs.c (expand_widen_pattern_expr): Likewise.
2592 (expand_binop_directly): Take the insn_code as a parameter.
2593 (expand_binop): Only call find_widening_optab_handler for
2594 conversion optabs; use optab_handler otherwise. Update calls
2595 to find_widening_optab_handler and expand_binop_directly.
2596 Use convert_optab_handler instead of widening_optab_handler.
2597 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
2598 find_widening_optab_handler and use scalar_mode rather than
2600 (convert_plusminus_to_widen): Likewise.
2602 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2603 Alan Hayward <alan.hayward@arm.com>
2604 David Sherwood <david.sherwood@arm.com>
2606 * machmode.h (fixed_size_mode): New class.
2607 * rtl.h (get_pool_mode): Return fixed_size_mode.
2608 * gengtype.c (main): Add fixed_size_mode.
2609 * target.def (get_raw_result_mode): Return a fixed_size_mode.
2610 (get_raw_arg_mode): Likewise.
2611 * doc/tm.texi: Regenerate.
2612 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
2613 * targhooks.c (default_get_reg_raw_mode): Likewise.
2614 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
2615 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
2616 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
2617 (msp430_get_raw_result_mode): Likewise.
2618 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
2619 * dbxout.c (dbxout_parms): Require fixed-size modes.
2620 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
2621 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2622 * omp-low.c (lower_oacc_reductions): Likewise.
2623 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
2624 (simplify_subreg): Update accordingly.
2625 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
2626 (force_const_mem): Update accordingly. Return NULL_RTX for modes
2627 that aren't fixed-size.
2628 (get_pool_mode): Return a fixed_size_mode.
2629 (output_constant_pool_2): Take a fixed_size_mode.
2631 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2632 Alan Hayward <alan.hayward@arm.com>
2633 David Sherwood <david.sherwood@arm.com>
2635 * doc/rtl.texi (vec_series): Document.
2636 (const): Say that the operand can be a vec_series.
2637 * rtl.def (VEC_SERIES): New rtx code.
2638 * rtl.h (const_vec_series_p_1): Declare.
2639 (const_vec_series_p): New function.
2640 * emit-rtl.h (gen_const_vec_series): Declare.
2641 (gen_vec_series): Likewise.
2642 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
2643 (gen_vec_series): Likewise.
2644 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
2645 * simplify-rtx.c (simplify_unary_operation): Handle negations
2647 (simplify_binary_operation_series): New function.
2648 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
2649 (test_vector_ops_series): New function.
2650 (test_vector_ops): Call it.
2651 * config/powerpcspe/altivec.md (altivec_lvsl): Use
2652 gen_const_vec_series.
2653 (altivec_lvsr): Likewise.
2654 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
2656 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2657 Alan Hayward <alan.hayward@arm.com>
2658 David Sherwood <david.sherwood@arm.com>
2660 * doc/rtl.texi (const): Update description of address constants.
2661 Say that vector constants are allowed too.
2662 * common.md (E, F): Use CONSTANT_P instead of checking for
2664 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
2665 checking for CONST_VECTOR.
2666 * expmed.c (make_tree): Use build_vector_from_val for a CONST
2668 * expr.c (expand_expr_real_2): Check for vector modes instead
2669 of checking for CONST_VECTOR.
2670 * rtl.h (const_vec_p): New function.
2671 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
2672 (unwrap_const_vec_duplicate): Handle them here too.
2674 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2675 David Malcolm <dmalcolm@redhat.com>
2676 Alan Hayward <alan.hayward@arm.com>
2677 David Sherwood <david.sherwood@arm.com>
2679 * rtl.h (vec_duplicate_p): New function.
2680 * selftest-rtl.c (assert_rtx_eq_at): New function.
2681 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
2682 (assert_rtx_eq_at): Declare.
2683 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
2684 * selftest-run-tests.c (selftest::run_tests): Call it.
2685 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
2686 (simplify_unary_operation_1): Recursively handle vector duplicates.
2687 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
2689 (simplify_subreg): Handle subregs of vector duplicates.
2690 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
2691 (selftest::simplify_rtx_c_tests): New functions.
2693 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2694 Alan Hayward <alan.hayward@arm.com>
2695 David Sherwood <david.sherwood@arm.com>
2697 * emit-rtl.h (gen_const_vec_duplicate): Declare.
2698 (gen_vec_duplicate): Likewise.
2699 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
2701 (gen_const_vector): ...here.
2702 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
2703 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
2704 whose elements are all equal.
2705 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
2706 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
2707 (simplify_relational_operation): Likewise.
2708 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
2710 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
2711 (aarch64_expand_vector_init): Likewise.
2712 * config/arm/arm.c (neon_vdup_constant): Likewise.
2713 (neon_expand_vector_init): Likewise.
2714 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
2715 (arm_block_set_unaligned_vect): Likewise.
2716 (arm_block_set_aligned_vect): Likewise.
2717 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
2718 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
2719 (expand_vec_perm_even_odd_pack): Likewise.
2720 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
2721 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
2722 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
2723 gen_const_vec_duplicate.
2724 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
2725 * config/mips/mips.c (mips_gen_const_int_vector): Use
2726 gen_const_vec_duplicate.
2727 (mips_expand_vector_init): Use CONST0_RTX.
2728 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
2729 (define_split): Use gen_const_vec_duplicate.
2730 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
2731 (define_split): Use gen_const_vec_duplicate.
2732 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
2733 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
2734 * config/spu/spu.c (spu_const): Likewise.
2736 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2737 Alan Hayward <alan.hayward@arm.com>
2738 David Sherwood <david.sherwood@arm.com>
2740 * combine.c (can_change_dest_mode): Reject changes in
2741 REGMODE_NATURAL_SIZE.
2743 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
2745 * configure.ac (--enable-libssp): New.
2746 (gcc_cv_libc_provides_ssp): Check for explicit setting before
2747 trying to determine target-specific default. Adjust indentation.
2748 * configure: Regenerated.
2749 * doc/install.texi (Configuration): Expand --disable-libssp
2752 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
2754 config/i386/i386.c (ix86_expand_epilogue): Correct stack
2757 2017-10-31 Martin Jambor <mjambor@suse.cz>
2760 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
2762 2017-10-31 David Malcolm <dmalcolm@redhat.com>
2764 * auto-profile.c (autofdo_source_profile::read): Use
2765 UNKNOWN_LOCATION rather than 0.
2766 * diagnostic-core.h (warning_at_rich_loc): Rename to...
2767 (warning_at): ...this overload.
2768 (warning_at_rich_loc_n): Rename to...
2769 (warning_n): ...this overload.
2770 (error_at_rich_loc): Rename to...
2771 (error_at): ...this overload.
2772 (pedwarn_at_rich_loc): Rename to...
2773 (pedwarn): ...this overload.
2774 (permerror_at_rich_loc): Rename to...
2775 (permerror): ...this overload.
2776 (inform_at_rich_loc): Rename to...
2777 (inform): ...this overload.
2778 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
2779 (diagnostic_n_impl_richloc): Rename to...
2780 (diagnostic_n_impl): ...this rich_location *-based decl.
2781 (inform_at_rich_loc): Rename to...
2782 (inform): ...this, and add an assertion.
2783 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
2784 (warning_at_rich_loc): Rename to...
2785 (warning_at): ...this, and add an assertion.
2786 (warning_at_rich_loc_n): Rename to...
2787 (warning_n): ...this, and add an assertion.
2788 (warning_n): Update location_t-based implementation for removal of
2789 location_t-based diagnostic_n_impl.
2790 (pedwarn_at_rich_loc): Rename to...
2791 (pedwarn): ...this, and add an assertion.
2792 (permerror_at_rich_loc): Rename to...
2793 (permerror): ...this, and add an assertion.
2794 (error_n): Update for removal of location_t-based diagnostic_n_impl.
2795 (error_at_rich_loc): Rename to...
2796 (error_at): ...this, and add an assertion.
2797 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
2798 (driver::do_spec_on_infiles): Likewise.
2799 * substring-locations.c (format_warning_va): Update for renaming
2800 of inform_at_rich_loc.
2802 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2804 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
2805 _Float<N>X built-in functions so that the variant without the
2806 "__builtin_" prefix is only enabled for the GNU C and Objective C
2807 languages when they are in non-strict ANSI/ISO mode.
2808 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
2809 * target.def (floatn_builtin_p): Add a target hook to control
2810 whether _Float<N> and _Float<N>X built-in functions without the
2811 "__builtin_" prefix are enabled, and return true for C and
2812 Objective C in the default hook. Include langhooks.h in
2814 * targhooks.h (default_floatn_builtin_p): Likewise.
2815 * targhooks.c (default_floatn_builtin_p): Likewise.
2816 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
2817 floatn_builtin_p target hook.
2818 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
2820 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
2821 Eric Botcazou <ebotcazou@adacore.com>
2823 PR rtl-optimization/81803
2824 * lra-constraints.c (curr_insn_transform): Also reload the whole
2825 register for a strict subreg no wider than a word if this is for
2826 a WORD_REGISTER_OPERATIONS target.
2828 2017-10-31 Jason Merrill <jason@redhat.com>
2830 * gdbinit.in: Skip over inlines from timevar.h.
2832 2017-10-31 Martin Liska <mliska@suse.cz>
2834 * doc/gcov.texi: Document new option.
2835 * gcov.c (print_usage): Likewise print it.
2836 (process_args): Support the argument.
2837 (format_count): New function.
2838 (format_gcov): Use the function.
2840 2017-10-31 Martin Liska <mliska@suse.cz>
2842 * gcov.c (struct name_map): do not use typedef.
2843 Define operator== and operator<.
2844 (name_search): Remove.
2845 (name_sort): Remove.
2846 (main): Do not allocate names.
2847 (process_file): Add vertical space.
2848 (generate_results): Use std::find.
2849 (release_structures): Do not release memory.
2850 (find_source): Use std::find.
2852 2017-10-31 Martin Liska <mliska@suse.cz>
2854 * gcov.c (struct line_info): Remove it's typedef.
2855 (line_info::line_info): Add proper ctor.
2856 (line_info::has_block): Do not use a typedef.
2857 (struct source_info): Do not use typedef.
2858 (circuit): Likewise.
2859 (get_cycles_count): Likewise.
2860 (output_intermediate_file): Iterate via vector iterator.
2861 (add_line_counts): Use std::vector methods.
2862 (accumulate_line_counts): Likewise.
2863 (output_lines): Likewise.
2865 2017-10-31 Martin Liska <mliska@suse.cz>
2867 * gcov.c (struct source_info): Remove typedef.
2868 (source_info::source_info): Add proper ctor.
2869 (accumulate_line_counts): Use struct, not it's typedef.
2870 (output_gcov_file): Likewise.
2871 (output_lines): Likewise.
2872 (main): Do not allocate an array.
2873 (output_intermediate_file): Use size of vector container.
2874 (process_file): Resize the vector.
2875 (generate_results): Do not preallocate, use newly added vector
2877 (release_structures): Do not release sources.
2878 (find_source): Use vector methods.
2879 (add_line_counts): Do not use typedef.
2881 2017-10-31 Martin Liska <mliska@suse.cz>
2883 * doc/gcov.texi: Document that.
2884 * gcov.c (add_line_counts): Mark lines with a non-executed
2886 (output_line_beginning): Handle such lines.
2887 (output_lines): Pass new argument.
2888 (output_intermediate_file): Print it in intermediate format.
2890 2017-10-31 Martin Liska <mliska@suse.cz>
2892 * color-macros.h: New file.
2893 * diagnostic-color.c: Factor out color related to macros to
2895 * doc/gcov.texi: Document -k option.
2896 * gcov.c (INCLUDE_STRING): Include string.h.
2897 (print_usage): Add -k option.
2898 (process_args): Parse it.
2899 (pad_count_string): New function.
2900 (output_line_beginning): Likewise.
2901 (DEFAULT_LINE_START): New macro.
2902 (output_lines): Support color output.
2904 2017-10-31 Martin Liska <mliska@suse.cz>
2906 PR gcov-profile/82633
2907 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
2908 their interaction with GCOV infrastructure.
2909 * configure.ac: Add -fkeep-{inline,static}-functions to
2911 * configure: Regenerate.
2913 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
2916 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
2918 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
2921 * config/rs6000/rs6000.md (allocate_stack): Force update interval
2922 into a register if it does not fit into an immediate offset field.
2924 2017-10-31 Olivier Hainque <hainque@adacore.com>
2926 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
2928 2017-10-31 Julia Koval <julia.koval@intel.com>
2930 * config.gcc: Add gfniintrin.h.
2931 * config/i386/gfniintrin.h: New.
2932 * config/i386/i386-builtin-types.def
2933 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2934 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2935 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2936 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2937 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2938 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
2939 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
2940 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
2941 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
2942 V64QI_FTYPE_V64QI_V64QI_INT): New types.
2943 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2944 * config/i386/immintrin.h: Include gfniintrin.h.
2945 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
2947 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
2949 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
2951 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2953 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
2954 (arm_ashldi3_1bit): Remove pattern.
2955 (ashrdi3): Remove shift by 1 expansion.
2956 (arm_ashrdi3_1bit): Remove pattern.
2957 (lshrdi3): Remove shift by 1 expansion.
2958 (arm_lshrdi3_1bit): Remove pattern.
2959 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
2960 cost of ashldi3 by 1.
2961 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
2962 (<shift>di3_neon): Likewise.
2964 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2966 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
2967 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
2968 and (*aarch64_simd_mov<VQ:mode>).
2969 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
2970 pattern alternative.
2971 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
2972 attributes to match pattern alternative.
2974 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
2976 * config.gcc (powerpc*-*-*): Add emmintrin.h.
2977 * config/rs6000/emmintrin.h: New file.
2978 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
2980 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2982 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
2983 * (movdi_vfp_cortexa8): Remove pattern.
2985 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2987 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
2989 (Specific, alpha*-dec-osf5.1): Remove.
2990 (Specific, mips-sgi-irix5): Remove.
2991 (Specific, mips-sgi-irix6): Remove.
2993 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2996 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
2997 arguments to clear_bit_region_be.
2999 2017-10-30 Jim Wilson <wilson@tuliptree.org>
3001 * gimplify.c: Include memmodel.h.
3003 2017-10-30 Martin Jambor <mjambor@suse.cz>
3005 * omp-grid.c (grid_attempt_target_gridification): Also insert a
3006 condition whether loop should be executed at all.
3008 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3010 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3011 gimple folding of vec_madd() intrinsics.
3012 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
3013 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
3014 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
3016 2017-10-30 Richard Biener <rguenther@suse.de>
3018 PR tree-optimization/82762
3020 2017-10-23 Richard Biener <rguenther@suse.de>
3022 PR tree-optimization/82129
3024 2017-08-01 Richard Biener <rguenther@suse.de>
3026 PR tree-optimization/81181
3027 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3028 (compute_antic): ... end of iteration here.
3030 2017-10-30 Joseph Myers <joseph@codesourcery.com>
3032 * doc/invoke.texi (C Dialect Options): Document -std=c17,
3033 -std=iso9899:2017 and -std=gnu17.
3034 * doc/standards.texi (C Language): Document C17 support.
3035 * doc/cpp.texi (Overview): Mention -std=c17.
3036 (Standard Predefined Macros): Document C11 and C17 values of
3037 __STDC_VERSION__. Do not refer to C99 support as incomplete.
3038 * doc/extend.texi (Inline): Do not list individual options for
3039 standards newer than C99.
3040 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
3042 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
3045 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3047 * asan.c (asan_finish_file): Align asan globals array by shadow
3050 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3053 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
3054 (struct store_immediate_info): Add bitregion_start and bitregion_end
3056 (store_immediate_info::store_immediate_info): Add brs and bre
3057 arguments and initialize bitregion_{start,end} from those.
3058 (struct merged_store_group): Add bitregion_start, bitregion_end,
3059 align_base and mask fields. Drop unnecessary struct keyword from
3060 struct store_immediate_info. Add do_merge method.
3061 (clear_bit_region_be): Use memset instead of loop storing zeros.
3062 (merged_store_group::do_merge): New method.
3063 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
3064 stores as long as the surrounding bitregions have no gaps.
3065 (merged_store_group::merge_overlapping): Use do_merge.
3066 (merged_store_group::apply_stores): Test that bitregion_{start,end}
3067 is byte aligned, rather than requiring that start and width are
3068 byte aligned. Drop unnecessary struct keyword from
3069 struct store_immediate_info. Allocate and populate also mask array.
3070 Make start of the arrays relative to bitregion_start rather than
3071 start and size them according to bitregion_{end,start} difference.
3072 (struct imm_store_chain_info): Drop unnecessary struct keyword from
3073 struct store_immediate_info.
3074 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
3075 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
3076 struct keyword from struct store_immediate_info.
3077 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
3078 between stores as long as the surrounding bitregions have no gaps.
3080 (struct split_store): Add orig non-static data member.
3081 (split_store::split_store): Initialize orig to false.
3082 (find_constituent_stmts): Return store_immediate_info *, non-NULL
3083 if there is exactly a single original stmt. Change stmts argument
3084 to pointer from reference, if NULL, don't push anything to it. Add
3085 first argument, use it to optimize skipping over orig stmts that
3086 are known to be before bitpos already. Simplify.
3087 (split_group): Return unsigned int count how many stores are or
3088 would be needed rather than a bool. Add allow_unaligned argument.
3089 Change split_stores argument from reference to pointer, if NULL,
3090 only do a dry run computing how many stores would be produced.
3091 Rewritten algorithm to use both alignment and misalign if
3092 !allow_unaligned and handle bitfield stores with gaps.
3093 (imm_store_chain_info::output_merged_store): Set start_byte_pos
3094 from bitregion_start instead of start. Compute allow_unaligned
3095 here, if true, do 2 split_group dry runs to compute which one
3096 produces fewer stores and prefer aligned if equal. Punt if
3097 new count is bigger or equal than original before emitting any
3098 statements, rather than during that. Remove no longer needed
3099 new_ssa_names tracking. Replace num_stmts with
3100 split_stores.length (). Use 32-bit stack allocated entries
3101 in split_stores auto_vec. Try to reuse original store lhs/rhs1
3102 if possible. Handle bitfields with gaps.
3103 (pass_store_merging::execute): Ignore bitsize == 0 stores.
3104 Compute bitregion_{start,end} for the stores and construct
3105 store_immediate_info with that. Formatting fixes.
3107 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
3110 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
3111 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
3113 2017-10-29 Jim Wilson <wilson@tuliptree.org>
3115 * gimplify.c: Include tm_p.h.
3117 * common.opt (gcoff): Re-add as ignored option.
3118 (gcoff1, gcoff2, gcoff3): Likewise.
3120 * Makefile.in (OBJS): Delete sdbout.o.
3121 (GTFILES): Delete $(srcdir)/sdbout.c.
3122 * debug.h: Delete sdb_debug_hooks.
3123 * final.c: Delete sdbout.h include.
3124 (final_scan_insn): Delete SDB_DEBUG check.
3125 (rest_of_clean_state): Likewise.
3126 * output.h: Delete sdb_begin_function_line.
3129 * toplev.c: Delete sdbout.h include.
3130 (process_options): Delete SDB_DEBUG check.
3131 * tree-core.h (tree_type_common): Delete pointer field of
3133 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
3134 TYPE_SYMTAB_POINTER.
3135 * tree.h (TYPE_SYMTAB_POINTER): Delete.
3136 (TYPE_SYMTAB_IS_POINTER): Delete.
3137 (TYPE_SYMTAB_IS_DIE): Renumber.
3138 * xcoffout.c: Refer to former sdbout.c file.
3139 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
3141 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
3142 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
3143 (Debugging Options): Delete -gcoff.
3144 (-gstabs): Delete SDB reference.
3146 (-gcoff@var{level}): Delete.
3147 * doc/passes.texi (Debugging information output): Delete SDB and
3148 sdbout.c references.
3149 * doc/tm.texi: Regenerate.
3150 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
3151 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
3153 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
3154 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
3155 (SDB_DEBUGGING_INFO): Delete.
3156 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
3157 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
3158 * target.def (output_source_filename): Delete COFF reference.
3160 * common.opt (gcoff): Delete.
3161 (gxcoff+): Update Negative chain.
3162 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
3164 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
3165 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
3166 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
3168 (expand_function_start): Change sdb reference to past tense.
3169 (expand_function_end): Change sdb reference to past tense.
3170 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
3171 * opts.c (debug_type_names): Delete coff entry.
3172 (common_handle_option): Delete OPT_gcoff case.
3173 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
3175 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
3176 * config/cris/cris.h: Delete SDB reference in comment.
3177 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
3178 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
3179 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3180 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3182 (ix86_expand_prologue): Likewise.
3183 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3184 * config/ia64/ia64.h: Likewise.
3185 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3186 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3188 * config/mmix/mmix.h: Likewise.
3189 * config/nds32/nds32.c: Likewise.
3190 * config/stormy/storym16.h: Likewise.
3191 * config/visium/visium.h: Likewise.
3192 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3194 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3196 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3197 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3198 FRAME_POINTER_REGNUM point at high end of local var area.
3200 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3202 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3203 Move comment around. Do not reset best_edge for a copiable
3204 destination if the copy would cause a partition change.
3205 (better_edge_p): Remove redundant check.
3207 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3209 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3211 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3213 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3214 for math functions that have _Float<N> and _Float<N>X variants.
3215 (mathfn_built_in_2): Add support for math functions that have
3216 _Float<N> and _Float<N>X variants.
3217 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3218 (expand_builtin_mathfn_ternary): Add support for fma with
3219 _Float<N> and _Float<N>X variants.
3220 (expand_builtin): Likewise.
3221 (fold_builtin_3): Likewise.
3222 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3223 create math function _Float<N> and _Float<N>X variants as external
3225 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3226 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3227 the __builtin_ prefix and if not strict ansi, without the prefix.
3228 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3229 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3230 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3231 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3232 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3233 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3234 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3235 function signatures for fma _Float<N> and _Float<N>X variants.
3236 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3237 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3238 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3239 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3240 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3241 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3242 * gencfn-macros.c (print_case_cfn): Add support for math functions
3243 that have _Float<N> and _Float<N>X variants.
3244 (print_define_operator_list): Likewise.
3245 (fltfn_suffixes): Likewise.
3247 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3248 for math functions that have _Float<N> and _Float<N>X variants.
3249 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3250 and _Float<N>X variants.
3251 (COPYSIGN): Likewise.
3254 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3255 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3257 (integer_valued_read_call_p): Likewise.
3258 * fold-const-call.c (fold_const_call_ss): Likewise.
3259 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3260 _Float<N> and _Float<N>X variants.
3261 (fold_const_call_ssss): Add support for fma _Float<N> and
3262 _Float<N>X variants.
3263 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3264 support for copysign and fma _Float<N> and _Float<N>X variants.
3265 (backprop::process_builtin_call_use): Likewise.
3266 * tree-call-cdce.c (can_test_argument_range); Add support for
3267 sqrt _Float<N> and _Float<N>X variants.
3268 (edom_only_function): Likewise.
3269 (get_no_error_domain): Likewise.
3270 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3271 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3272 copysign _Float<N> and _Float<N>X variants.
3273 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3274 handled by machine independent code.
3275 (FMAF128): Likewise.
3276 * doc/cpp.texi (Common Predefined Macros): Document defining
3277 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3278 fma _Float<N> and _Float<N>X variants.
3280 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3283 * config/i386/i386-modes.def (CCFPU): Remove definition.
3284 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3285 (ix86_cc_modes_compatible): Ditto.
3286 (ix86_expand_carry_flag_compare): Ditto.
3287 (ix86_expand_int_movcc): Ditto.
3288 (ix86_expand_int_addcc): Ditto.
3289 (ix86_reverse_condition): Ditto.
3290 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3291 Return true/false for unordered/ordered fp comparisons.
3292 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3293 (ix86_prepare_fp_compare_args): Update for rename.
3294 (ix86_expand_fp_compare): Update for rename. Generate unordered
3295 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3296 (ix86_expand_sse_compare_and_jump): Ditto.
3297 * config/i386/predicates.md (fcmov_comparison_operator):
3298 Remove CCFPU mode handling.
3299 (ix86_comparison_operator): Ditto.
3300 (ix86_carry_flag_operator): Ditto.
3301 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3302 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3303 (*cmpu<mode>_cc_i387): Ditto.
3304 (FPCMP): Remove mode iterator.
3305 (unord): Remove mode attribute.
3306 (unord_subst): New define_subst transformation
3307 (unord): New define_subst attribute.
3309 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3310 (*cmpi<unord>xf_i387): Ditto.
3311 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3312 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3313 using unord_subst transformation.
3314 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3315 (round_saeonly): Also handle CCFP mode.
3316 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3317 Remove UNSPEC_SAHF unspec handling.
3319 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3321 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3323 2017-10-27 Jeff Law <law@redhat.com>
3325 * gimple-ssa-sprintf.c: Include domwalk.h.
3326 (class sprintf_dom_walker): New class, derived from dom_walker.
3327 (sprintf_dom_walker::before_dom_children): New function.
3328 (struct call_info): Moved into sprintf_dom_walker class
3329 (compute_formath_length, handle_gimple_call): Likewise.
3330 (sprintf_length::execute): Call the dominator walker rather
3331 than walking the statements.
3333 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3334 gimple statement locations.
3335 (check_array_bounds): Corresponding changes. Get the statement's
3336 location directly from wi->stmt.
3338 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3341 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3343 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3345 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3346 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3348 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3351 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3352 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3353 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3354 maybe_get_pool_constant.
3355 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3358 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3360 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3361 2.26 caveat. Update gas and gld versions.
3362 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3365 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3367 * cgraph.h (set_malloc_flag): Declare.
3368 * cgraph.c (set_malloc_flag_1): New function.
3369 (set_malloc_flag): Likewise.
3370 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3371 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3373 (read_ipa_call_summary): Add support for reading is_return_callee.
3374 (write_ipa_call_summary): Stream is_return_callee.
3375 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3376 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3377 ipa-prop.h, ipa-fnsummary.h.
3378 (pure_const_names): Change to static.
3379 (malloc_state_e): Define.
3380 (malloc_state_names): Define.
3381 (funct_state_d): Add field malloc_state.
3382 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3383 (check_retval_uses): New function.
3384 (malloc_candidate_p): Likewise.
3385 (analyze_function): Add support for malloc attribute.
3386 (pure_const_write_summary): Stream malloc_state.
3387 (pure_const_read_summary): Add support for reading malloc_state.
3388 (dump_malloc_lattice): New function.
3389 (propagate_malloc): New function.
3390 (warn_function_malloc): New function.
3391 (ipa_pure_const::execute): Call propagate_malloc and
3392 ipa_free_fn_summary.
3393 (pass_local_pure_const::execute): Add support for malloc attribute.
3394 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3395 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3397 2017-10-27 Martin Liska <mliska@suse.cz>
3399 PR gcov-profile/82457
3400 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3401 for fork-like functions to be properly instrumented.
3403 2017-10-27 Richard Biener <rguenther@suse.de>
3406 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3407 info when we redirected EH.
3409 2017-10-26 Michael Collison <michael.collison@arm.com>
3411 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3413 (<optab>_trunchf<GPI:mode>2: New pattern.
3414 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3415 * config/aarch64/iterators.md (wv): New mode attribute.
3416 (vf, VF): New mode attributes.
3417 (vgp, VGP): New mode attributes.
3418 (s): Update attribute with SImode and DImode prefixes.
3420 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3422 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3423 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3424 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3425 (nios2_option_overide): Initialize it. Don't allow R0-relative
3426 addressing with PIC.
3427 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3428 (nios2_symbolic_constant_p): Likewise.
3429 (nios2_legitimate_address_p): Likewise.
3430 (nios2_r0rel_section_name_p): New.
3431 (nios2_symbol_ref_in_r0rel_data_p): New.
3432 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3433 (r0rel_constant_p): New.
3434 (nios2_print_operand_address): Handle r0rel_constant_p.
3435 (nios2_cdx_narrow_form_p): Likewise.
3436 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3437 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3438 (Nios II Options): Document -mr0rel-sec.
3440 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3442 * config/nios2/nios2.c: Include xregex.h.
3443 (nios2_gprel_sec_regex): New.
3444 (nios2_option_overide): Initialize it. Don't allow GP-relative
3445 addressing with PIC.
3446 (nios2_small_section_name_p): Check for regex match.
3447 * config/nios2/nios2.opt (mgprel-sec=): New option.
3448 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3449 (Nios II Options): Document -mgprel-sec.
3451 2017-10-26 Jim Wilson <wilson@tuliptree.org>
3453 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
3455 2017-10-26 Tom de Vries <tom@codesourcery.com>
3457 PR tree-optimization/82707
3458 * gimple.c (gimple_copy): Fix unsharing of
3459 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3461 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3463 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3464 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3465 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3466 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3467 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3468 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3469 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3470 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
3471 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
3472 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
3473 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
3474 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
3475 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
3476 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
3477 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
3478 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
3479 _mm512_mask_cmpunord_ps_mask): New intrinsics.
3481 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3483 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
3485 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
3486 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
3487 warning to rs6000.c. Remove the Undocumented flag, since it has
3489 (-mabi=ibmlongdouble): Likewise.
3490 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
3491 already set, set the default format for long double.
3492 (rs6000_debug_reg_global): Print whether long double is IBM or
3494 (rs6000_option_override_internal): Rework setting long double
3495 format. Only warn if the user is changing the long double default
3496 and they did not use -Wno-psabi.
3497 * doc/invoke.texi (PowerPC options): Update the documentation for
3498 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
3500 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3501 Alan Hayward <alan.hayward@arm.com>
3502 David Sherwood <david.sherwood@arm.com>
3504 * rtl.h (wider_subreg_mode): New function.
3505 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
3506 rather than an unsigned int *.
3507 * ira-color.c (regno_max_ref_width): Replace with...
3508 (regno_max_ref_mode): ...this new variable.
3509 (coalesced_pseudo_reg_slot_compare): Update accordingly.
3510 Use wider_subreg_mode.
3511 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
3512 rather than an unsigned int *.
3513 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
3514 (process_alt_operands): Likewise.
3515 (invariant_p): Likewise.
3516 * lra-spills.c (assign_mem_slot): Likewise.
3517 (add_pseudo_to_slot): Likewise.
3518 * lra.c (collect_non_operand_hard_regs): Likewise.
3519 (add_regs_to_insn_regno_info): Likewise.
3520 * reload1.c (regno_max_ref_width): Replace with...
3521 (regno_max_ref_mode): ...this new variable.
3522 (reload): Update accordingly. Update call to
3523 ira_sort_regnos_for_alter_reg.
3524 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
3525 (init_eliminable_invariants): Update to use regno_max_ref_mode.
3526 (scan_paradoxical_subregs): Likewise.
3528 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3530 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
3531 (aarch64_frame): Add emit_frame_chain boolean.
3532 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3533 Move eh_return case to aarch64_layout_frame.
3534 (aarch64_layout_frame): Initialize emit_frame_chain.
3535 (aarch64_expand_prologue): Use emit_frame_chain.
3537 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3539 * config/aarch64/aarch64.c (aarch64_layout_frame):
3540 Ensure LR is always stored at the bottom of the callee-saves.
3541 Remove rarely used frame layout which saves callee-saves at top of
3542 frame, so the store of LR can be used as a valid probe in all cases.
3544 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3546 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
3547 Improve unaligned TImode/TFmode base/offset split.
3549 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3550 Alan Hayward <alan.hayward@arm.com>
3551 David Sherwood <david.sherwood@arm.com>
3553 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
3554 * combine.c (find_single_use_1): Likewise.
3555 (expand_field_assignment): Likewise.
3556 (move_deaths): Likewise.
3557 * lra-constraints.c (simplify_operand_subreg): Likewise.
3558 (curr_insn_transform): Likewise.
3559 * lra.c (collect_non_operand_hard_regs): Likewise.
3560 (add_regs_to_insn_regno_info): Likewise.
3561 * rtlanal.c (reg_referenced_p): Likewise.
3562 (covers_regno_no_parallel_p): Likewise.
3564 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3566 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
3567 Don't print any bits outside the precision of the value.
3568 * wide-int.cc (test_printing): Add some new tests.
3570 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3572 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
3573 supports -xbrace_comment option.
3574 * configure: Regenerate.
3575 * config.in: Regenerate.
3576 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
3577 (ASM_CPU_SPEC): Use it.
3579 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3581 * target.def (static_rtx_alignment): New hook.
3582 * targhooks.h (default_static_rtx_alignment): Declare.
3583 * targhooks.c (default_static_rtx_alignment): New function.
3584 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
3585 * doc/tm.texi: Regenerate.
3586 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
3587 instead of targetm.constant_alignment. Remove call to
3589 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3590 (cris_preferred_mininum_alignment): New function, split out from...
3591 (cris_constant_alignment): ...here.
3592 (cris_static_rtx_alignment): New function.
3593 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
3595 (ix86_constant_alignment): ...here.
3596 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3597 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3598 (mmix_static_rtx_alignment): New function.
3599 * config/spu/spu.c (spu_static_rtx_alignment): New function.
3600 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3602 2017-10-26 Tamar Christina <tamar.christina@arm.com>
3605 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
3606 Add flag_trapping_math and flag_fp_int_builtin_inexact.
3608 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
3610 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
3611 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
3612 mark as a sign-extending load.
3613 (local_pic_load_u): Define.
3615 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
3618 * fold-const.c (operand_equal_for_comparison_p): Also return true
3619 if ARG0 is a simple variant of ARG1 with narrower precision.
3620 (fold_ternary_loc): Always pass unstripped operands to the predicate.
3622 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
3624 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
3626 * i386.h (processor_costs): Add gather_static, gather_per_elt,
3627 scatter_static, scatter_per_elt.
3628 * x86-tune-costs.h: Add new cost entries.
3630 2017-10-25 Richard Biener <rguenther@suse.de>
3632 * tree-ssa-sccvn.h (vn_eliminate): Declare.
3633 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
3634 class pass_fre): Move to ...
3635 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
3636 class pass_fre): ... here and adjust for statistics.
3638 2017-10-25 Jakub Jelinek <jakub@redhat.com>
3641 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
3642 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
3643 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
3644 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
3647 2017-10-25 Richard Biener <rguenther@suse.de>
3649 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
3650 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
3651 eliminate_push_avail, eliminate_insert): Move inside...
3652 (class eliminate_dom_walker): ... this class in preparation
3654 (fini_eliminate): Remove by merging with ...
3655 (eliminate): ... this function. Adjust for class changes.
3656 (pass_pre::execute): Remove fini_eliminate call.
3657 (pass_fre::execute): Likewise.
3659 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3662 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
3663 (VPERMI2, VPERMI2I): New mode iterators.
3664 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
3665 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
3667 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
3668 mode iterator. Remove 3 old define_insn patterns.
3669 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
3670 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
3671 VPERMI2 mode iterator, remove the other two expanders.
3672 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
3673 to use VPERMI2 mode iterator, add another alternative for vpermi2*
3674 instructions, remove the other two patterns.
3675 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
3676 mode iterator, remove the other two patterns.
3677 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
3678 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
3679 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
3680 and adjust argument order accordingly.
3681 (ix86_expand_vec_perm): Adjust caller.
3682 (expand_vec_perm_1): Likewise.
3683 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
3684 (expand_vec_perm_vpermt2_vpshub2): ... this.
3685 (ix86_expand_vec_perm_const_1): Adjust caller.
3686 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
3689 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
3690 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
3691 (vec_shl_<mode>): Remove unused expander.
3692 (avx512bw_<shift_insn><mode>3): New define_insn.
3693 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
3694 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
3696 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
3699 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
3702 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3704 PR rtl-optimization/82396
3705 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
3706 (autopref_multipass_init): Simplify initialization.
3707 (autopref_rank_data): Simplify sort order.
3708 * gcc/sched-int.h (autopref_multipass_data_): Remove
3709 multi_mem_insn_p, min_offset and max_offset.
3711 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3714 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3715 Check special value of flag_omit_frame_pointer.
3716 (aarch64_can_eliminate): Likewise.
3717 (aarch64_override_options_after_change_1): Simplify handling of
3718 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
3720 2017-10-24 Richard Biener <rguenther@suse.de>
3722 PR tree-optimization/82697
3723 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
3724 zero for conditional load and unconditional store.
3726 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3728 * doc/install.texi: Document bootstrap-cet.
3730 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3733 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
3734 ENDBR instruction at function entrance if function is only
3737 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3740 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
3741 patterns to better describe from which operation the CF is computed.
3742 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
3743 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
3744 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
3745 is 0, use _0 suffixed expanders instead of emitting a comparison
3748 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
3750 * config/i386/i386.md(*movsf_internal, *movdf_internal):
3751 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
3753 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
3756 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
3757 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
3758 * loop-iv.c (iv_get_reaching_def): Likewise.
3759 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
3760 variable is promoted and the partition contains undefined values.
3762 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3764 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
3766 (nios2_address_cost): Define.
3767 (nios2_legitimize_address): Recognize (exp + constant) directly.
3768 (TARGET_ADDRESS_COST): Define.
3770 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3772 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
3773 (nios2_symbolic_memory_operand_p): Declare.
3774 (nios2_split_large_constant): Declare.
3775 (nios2_split_symbolic_memory_operand): Declare.
3776 * config/nios2/nios2.c: Adjust includes.
3777 (nios2_symbolic_constant_allowed): New.
3778 (nios2_symbolic_constant_p): New.
3779 (nios2_plus_symbolic_constant_p): New.
3780 (nios2_valid_addr_expr_p): Recognize addresses involving
3782 (nios2_legitimate_address_p): Likewise, also LO_SUM.
3783 (nios2_symbolic_memory_operand_p): New.
3784 (nios2_large_constant_p): New.
3785 (nios2_split_large_constant): New.
3786 (nios2_split_plus_large_constant): New.
3787 (nios2_split_symbolic_memory_operand): New.
3788 (nios2_legitimize_address): Code refactoring. Handle addresses
3789 involving symbolic constants.
3790 (nios2_emit_move_sequence): Likewise.
3791 (nios2_print_operand): Improve error output.
3792 (nios2_print_operand_address): Handle LO_SUM.
3793 (nios2_cdx_narrow_form_p): Likewise.
3794 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
3795 operands involving symbolic constants.
3796 (movhi_internal, movsi_internal): Likewise.
3797 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
3798 (extendhisi2, extendqi<mode>2): Likewise.
3800 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3802 * tree-pass.h (PROP_rtl_split_insns): Define.
3803 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
3805 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3807 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
3809 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3812 * target.def (const_not_ok_for_debug_p): Default to
3813 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
3814 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
3815 * targhooks.c (default_const_not_ok_for_debug_p): New function.
3816 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
3817 which targetm.const_not_ok_for_debug_p returned true.
3818 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
3820 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
3822 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
3823 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
3824 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
3826 (ix86_const_not_ok_for_debug_p): New function.
3827 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
3828 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
3830 2017-10-23 David Malcolm <dmalcolm@redhat.com>
3833 * system.h: Conditionally include "unique-ptr.h" if
3834 INCLUDE_UNIQUE_PTR is defined.
3835 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
3836 of defining INCLUDE_UNIQUE_PTR before including "system.h".
3838 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
3840 * config/rl78/rl78.md: New define_expand "subdi3".
3842 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
3845 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
3846 DF_REF_INSN if DF_REF_INSN_INFO is false.
3848 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
3850 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
3851 xmm_move instead of sse_move.
3852 (sse_store_index): New function.
3853 (ix86_register_move_cost): Be more sensible about mismatch stall;
3854 model AVX moves correctly; make difference between sse->integer and
3856 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
3857 moves; make difference between SSE and AVX.
3858 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
3859 and zmm_move. Increase size of sse load and store tables;
3860 add unaligned load and store tables; add ssemmx_to_integer.
3861 * x86-tune-costs.h: Update all entries according to real
3862 move latencies from Agner Fog's manual and chip documentation.
3864 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3867 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
3868 * config/i386/constraints.md (Wf): New constraint.
3869 * config/i386/i386.md (UNSPEC_SBB): New unspec.
3870 (cmp<dwi>_doubleword): Removed.
3871 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
3872 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
3873 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
3874 expand with cmp<dwi>_doubleword. For LTU and GEU use
3875 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
3877 * common.opt (gcolumn-info): Enable by default.
3878 * doc/invoke.texi (gcolumn-info): Document new default.
3880 2017-10-23 Richard Biener <rguenther@suse.de>
3882 PR tree-optimization/82672
3883 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
3884 Fold the stmt if we propagated into it.
3886 2017-10-23 Richard Biener <rguenther@suse.de>
3888 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
3889 (bitmap_remove_expr_from_set): ... this. All callers call this
3890 for non-constant values.
3891 (bitmap_set_subtract): Rename to...
3892 (bitmap_set_subtract_expressions): ... this. Adjust and
3894 (bitmap_set_contains_value): Remove superfluous check.
3895 (bitmap_set_replace_value): Inline into single caller ...
3896 (bitmap_value_replace_in_set): ... here and simplify.
3897 (dependent_clean): Merge into ...
3898 (clean): ... this using an overload. Adjust.
3899 (prune_clobbered_mems): Adjust.
3900 (compute_antic_aux): Likewise.
3901 (compute_partial_antic_aux): Likewise.
3903 2017-10-23 Richard Biener <rguenther@suse.de>
3905 PR tree-optimization/82129
3907 2017-08-01 Richard Biener <rguenther@suse.de>
3909 PR tree-optimization/81181
3910 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3911 (compute_antic): ... end of iteration here.
3913 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3915 * target.def (starting_frame_offset): New hook.
3916 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
3917 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
3918 * doc/tm.texi.in: Regenerate.
3919 * hooks.h (hook_hwi_void_0): Declare.
3920 * hooks.c (hook_hwi_void_0): New function.
3921 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
3922 STARTING_FRAME_OFFSET.
3923 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
3924 * reload1.c (reload): Likewise.
3925 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
3926 instead of STARTING_FRAME_OFFSET.
3927 * function.c (try_fit_stack_local): Likewise.
3928 (assign_stack_local_1): Likewise
3929 (instantiate_virtual_regs): Likewise.
3930 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
3931 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
3932 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
3933 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
3934 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
3935 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
3936 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
3937 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
3938 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
3939 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
3940 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
3941 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
3942 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
3943 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
3944 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
3945 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
3946 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
3947 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
3948 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
3949 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
3950 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
3951 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
3952 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
3953 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
3954 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
3955 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
3956 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
3957 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
3958 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
3959 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
3960 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
3961 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
3962 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
3963 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
3964 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
3965 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
3966 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
3967 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
3968 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
3969 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
3970 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
3971 * config/avr/avr.c (avr_starting_frame_offset): Make static and
3972 return a HOST_WIDE_INT.
3973 (avr_builtin_setjmp_frame_value): Use it instead of
3974 STARTING_FRAME_OFFSET.
3975 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3976 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
3977 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
3979 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3980 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
3981 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
3982 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3983 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
3984 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
3985 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3986 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
3987 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
3988 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3989 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
3990 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
3992 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3993 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
3994 * config/mips/mips.c (mips_compute_frame_info): Refer to
3995 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3996 (mips_starting_frame_offset): New function.
3997 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3998 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
3999 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
4000 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
4001 and return a HOST_WIDE_INT.
4002 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4003 (mmix_initial_elimination_offset): Refer to
4004 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4005 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
4006 * config/pa/pa.c (pa_starting_frame_offset): New function.
4007 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
4008 (pa_expand_prologue): Likewise.
4009 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4010 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
4011 !FRAME_GROWS_DOWNWARD handling to...
4012 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4013 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
4014 !FRAME_GROWS_DOWNWARD handling to...
4015 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4016 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
4017 !FRAME_GROWS_DOWNWARD handling to...
4018 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4019 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
4021 (rs6000_starting_frame_offset): New function.
4022 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
4023 !FRAME_GROWS_DOWNWARD handling to...
4024 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4025 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
4026 !FRAME_GROWS_DOWNWARD handling to...
4027 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4028 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
4029 !FRAME_GROWS_DOWNWARD handling to...
4030 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4031 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
4032 (rs6000_starting_frame_offset): New function.
4033 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
4034 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
4035 * config/vax/vax.c (vax_starting_frame_offset): New function.
4036 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
4037 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4038 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
4039 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
4040 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4041 * system.h (STARTING_FRAME_OFFSET): Poison.
4043 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4045 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4046 SCALAR_TYPE_MODE instead of TYPE_MODE.
4048 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4049 Alan Hayward <alan.hayward@arm.com>
4050 David Sherwood <david.sherwood@arm.com>
4052 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
4054 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4055 Alan Hayward <alan.hayward@arm.com>
4056 David Sherwood <david.sherwood@arm.com>
4058 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
4060 2017-10-23 Richard Biener <rguenther@suse.de>
4062 PR tree-optimization/82129
4063 * tree-ssa-pre.c (bitmap_set_and): Remove.
4064 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
4065 canonicalizing expressions in the set to those with lowest
4066 ID rather than taking that from the first edge.
4068 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4070 * combine.c (rtx_equal_for_field_assignment_p): Use
4071 byte_lowpart_offset.
4073 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4074 Alan Hayward <alan.hayward@arm.com>
4075 David Sherwood <david.sherwood@arm.com>
4077 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
4078 to a SUBREG_PROMOTED_VAR.
4080 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4081 Alan Hayward <alan.hayward@arm.com>
4082 David Sherwood <david.sherwood@arm.com>
4084 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
4085 (expand_debug_source_expr): Likewise.
4086 * combine.c (combine_simplify_rtx): Likewise.
4087 * cse.c (fold_rtx): Likewise.
4088 * optabs.c (expand_float): Likewise.
4089 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4090 (simplify_binary_operation_1): Likewise.
4092 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4093 Alan Hayward <alan.hayward@arm.com>
4094 David Sherwood <david.sherwood@arm.com>
4096 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
4097 (record_promoted_value): Likewise.
4098 * expr.c (expand_expr_real_2): Likewise.
4099 * ree.c (update_reg_equal_equiv_notes): Likewise.
4100 (combine_set_extension): Likewise.
4101 * rtlanal.c (low_bitmask_len): Likewise.
4102 * simplify-rtx.c (neg_const_int): Likewise.
4103 (simplify_binary_operation_1): Likewise.
4105 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4106 Alan Hayward <alan.hayward@arm.com>
4107 David Sherwood <david.sherwood@arm.com>
4109 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
4110 * regcprop.c (maybe_mode_change): Likewise.
4111 * reload1.c (alter_reg): Likewise.
4113 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4115 * inchash.h (inchash::hash::add_wide_int): New function.
4116 * lto-streamer-out.c (hash_tree): Use it.
4118 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4120 * inchash.h (inchash::hash::add_wide_int): Rename to...
4121 (inchash::hash::add_hwi): ...this.
4122 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
4123 (polymorphic_call_target_hasher::hash): Likewise.
4124 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
4125 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
4126 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
4127 * lto-streamer-out.c (hash_tree): Likewise.
4128 * optc-save-gen.awk: Likewise.
4129 * tree.c (add_expr): Likewise.
4131 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4134 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
4135 for ordered inequality comparisons even with TARGET_IEEE_FP.
4137 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4140 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
4141 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4142 Expand with cmp<dwi>_doubleword.
4144 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4146 * extend.texi: Add x86 specific to 'nocf_check' attribute.
4147 List CET intrinsics.
4148 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
4149 specific to -fcf-protection option.
4151 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4153 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
4154 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
4155 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
4156 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
4157 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
4158 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
4159 (extra_objs): Add cet.o for Linux/x86 targets.
4160 (tmake_file): Add i386/t-cet for Linux/x86 targets.
4161 * config/i386/cet.c: New file.
4162 * config/i386/cetintrin.h: Likewise.
4163 * config/i386/t-cet: Likewise.
4164 * config/i386/cpuid.h (bit_SHSTK): New.
4165 (bit_IBT): Likewise.
4166 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
4167 pass IBT and SHSTK bits.
4168 * config/i386/i386-builtin-types.def
4169 (VOID_FTYPE_UNSIGNED_PVOID): New.
4170 (VOID_FTYPE_UINT64_PVOID): Likewise.
4171 * config/i386/i386-builtin.def: Add CET intrinsics.
4172 * config/i386/i386-c.c (ix86_target_macros_internal): Add
4173 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
4174 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
4175 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
4177 * config/i386/i386.c (rest_of_insert_endbranch): New.
4178 (pass_data_insert_endbranch): Likewise.
4179 (pass_insert_endbranch): Likewise.
4180 (make_pass_insert_endbranch): Likewise.
4181 (ix86_notrack_prefixed_insn_p): Likewise.
4182 (ix86_target_string): Add -mibt, -mshstk flags.
4183 (ix86_option_override_internal): Add flag_cf_protection
4185 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4186 (ix86_print_operand): Add 'notrack' prefix output.
4187 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4188 (ix86_expand_builtin): Expand CET intrinsics.
4189 (x86_output_mi_thunk): Add 'endbranch' instruction.
4190 * config/i386/i386.h (TARGET_IBT): New.
4191 (TARGET_IBT_P): Likewise.
4192 (TARGET_SHSTK): Likewise.
4193 (TARGET_SHSTK_P): Likewise.
4194 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4195 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4196 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4197 (builtin_setjmp_setup): New pattern.
4198 (builtin_longjmp): Likewise.
4199 (rdssp<mode>): Likewise.
4200 (incssp<mode>): Likewise.
4201 (saveprevssp): Likewise.
4202 (rstorssp): Likewise.
4203 (wrss<mode>): Likewise.
4204 (wruss<mode>): Likewise.
4205 (setssbsy): Likewise.
4206 (clrssbsy): Likewise.
4207 (nop_endbr): Likewise.
4208 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4210 * config/i386/immintrin.h: Include <cetintrin.h>.
4211 * config/i386/linux-common.h
4212 (file_end_indicate_exec_stack_and_cet): New prototype.
4213 (TARGET_ASM_FILE_END): New.
4215 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4217 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4218 latencies instead of having separate table; make difference between
4219 integer and float costs.
4220 * i386.h (processor_costs): Remove scalar_stmt_cost,
4221 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4222 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4224 * x86-tune-costs.h: Remove entries which has been removed in
4225 procesor_costs from all tables; make cond_taken_branch_cost
4226 and cond_not_taken_branch_cost COST_N_INSNS based.
4228 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4230 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4232 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4234 * config/i386/i386.md (isa): Remove fma_avx512f.
4235 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4236 <avx512>_fmadd_<mode>_mask3<round_name>,
4237 <avx512>_fmsub_<mode>_mask<round_name>,
4238 <avx512>_fmsub_<mode>_mask3<round_name>,
4239 <avx512>_fnmadd_<mode>_mask<round_name>,
4240 <avx512>_fnmadd_<mode>_mask3<round_name>,
4241 <avx512>_fnmsub_<mode>_mask<round_name>,
4242 <avx512>_fnmsub_<mode>_mask3<round_name>,
4243 <avx512>_fmaddsub_<mode>_mask<round_name>,
4244 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4245 <avx512>_fmsubadd_<mode>_mask<round_name>,
4246 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4247 (*vec_widen_umult_even_v16si<mask_name>,
4248 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4249 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4251 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4253 * extend.texi: Add 'nocf_check' documentation.
4254 * gimple.texi: Add second parameter to
4255 gimple_build_call_from_tree.
4256 * invoke.texi: Add -fcf-protection documentation.
4257 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4259 2017-10-20 Richard Biener <rguenther@suse.de>
4261 PR tree-optimization/82473
4262 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4263 the largest input type.
4265 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4267 * c-attribs.c (handle_nocf_check_attribute): New function.
4268 (c_common_attribute_table): Add 'nocf_check' handling.
4269 * gimple-parser.c: Add second argument NULL to
4270 gimple_build_call_from_tree.
4271 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4272 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4274 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4275 * common.opt: Add fcf-protection flag.
4276 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4277 * flag-types.h: Add enum cf_protection_level.
4278 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4279 Add 'nocf_check' attribute propagation to gimple call.
4280 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4281 (gimple_build_call_from_tree): Update prototype.
4282 (gimple_call_nocf_check_p): New function.
4283 (gimple_call_set_nocf_check): Likewise.
4284 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4285 * ipa-icf.c: Add nocf_check attribute in statement hash.
4286 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4287 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4288 * toplev.c (process_options): Add flag_cf_protection handling.
4290 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4292 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4294 2017-10-20 Richard Biener <rguenther@suse.de>
4296 PR tree-optimization/82603
4297 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4298 remove false predicated stores.
4300 2017-10-20 Richard Biener <rguenther@suse.de>
4302 * graphite-isl-ast-to-gimple.c
4303 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4304 Remove return value and simplify, dump copied stmt after lhs
4306 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4307 Reduce dump verbosity.
4308 (gsi_insert_earliest): Likewise.
4309 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4310 * graphite.c (print_global_statistics): Adjust dumping.
4311 (print_graphite_scop_statistics): Likewise.
4312 (print_graphite_statistics): Do not dump loops here.
4313 (graphite_transform_loops): But here.
4315 2017-10-20 Nicolas Roche <roche@adacore.com>
4317 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4318 * configure: Regenerate.
4320 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4323 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4324 functions when optimizing replace GIMPLE_RETURN stmts with
4325 calls to __builtin_unreachable ().
4328 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4329 for -fsanitize=thread link of executables.
4330 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4331 link of executables.
4334 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4336 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4337 define_insns for logical vector shifts to use VI248_AVX512BW
4338 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4339 condition, useless isa and prefix attributes. Change the first
4340 2 of these define_insns to ...
4341 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4342 define_insn for avx512vl.
4343 (<shift_insn><mode>3): ... and this, new define_insn without
4344 masking for non-avx512vl.
4347 * config/i386/sse.md (*andnot<mode>3,
4348 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4349 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4350 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4351 not applied use empty suffix even for TARGET_AVX512VL.
4352 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4353 is applied, supply evex,evex or evex,evex,evex instead of just
4356 2017-10-20 Julia Koval <julia.koval@intel.com>
4358 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4359 (OPTION_MASK_ISA_GFNI_UNSET): New.
4360 (ix86_handle_option): Handle OPT_mgfni.
4361 * config/i386/cpuid.h (bit_GFNI): New.
4362 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4363 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4364 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4365 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4366 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4367 * config/i386/i386.opt: Add mgfni.
4369 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4371 * config/msp430/msp430.c (msp430_option_override): Disable
4372 -fdelete-null-pointer-checks.
4373 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4375 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4377 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4378 of x87 and SSE instructions.
4380 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4382 * asan.c (create_cond_insert_point): Do not update edge count.
4383 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4384 (afdo_propagate_circuit): Likewise.
4385 (afdo_calculate_branch_prob): Likewise.
4386 (afdo_annotate_cfg): Likewise.
4387 * basic-block.h (struct edge_def): Remove count.
4388 (edge_def::count): New accessor.
4389 * bb-reorder.c (rotate_loop): Update.
4390 (find_traces_1_round): Update.
4391 (connect_traces): Update.
4392 (sanitize_hot_paths): Update.
4393 * cfg.c (unchecked_make_edge): Update.
4394 (make_single_succ_edge): Update.
4395 (check_bb_profile): Update.
4396 (dump_edge_info): Update.
4397 (update_bb_profile_for_threading): Update.
4398 (scale_bbs_frequencies_int): Update.
4399 (scale_bbs_frequencies_gcov_type): Update.
4400 (scale_bbs_frequencies_profile_count): Update.
4401 (scale_bbs_frequencies): Update.
4402 * cfganal.c (connect_infinite_loops_to_exit): Update.
4403 * cfgbuild.c (compute_outgoing_frequencies): Update.
4404 (find_many_sub_basic_blocks): Update.
4405 * cfgcleanup.c (try_forward_edges): Update.
4406 (try_crossjump_to_edge): Update
4407 * cfgexpand.c (expand_gimple_cond): Update
4408 (expand_gimple_tailcall): Update
4409 (construct_exit_block): Update
4410 * cfghooks.c (verify_flow_info): Update
4411 (redirect_edge_succ_nodup): Update
4412 (split_edge): Update
4413 (make_forwarder_block): Update
4414 (duplicate_block): Update
4415 (account_profile_record): Update
4416 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4417 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4418 * cfgloopmanip.c (scale_loop_profile): Update.
4420 (lv_adjust_loop_entry_edge): Update.
4421 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4422 (force_nonfallthru_and_redirect): Update.
4423 (purge_dead_edges): Update.
4424 (rtl_flow_call_edges_add): Update.
4425 * cgraphunit.c (init_lowered_empty_function): Update.
4426 (cgraph_node::expand_thunk): Update.
4427 * gimple-pretty-print.c (dump_probability): Update.
4428 (dump_edge_probability): Update.
4429 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4430 * haifa-sched.c (sched_create_recovery_edges): Update.
4431 * hsa-gen.c (convert_switch_statements): Update.
4432 * ifcvt.c (dead_or_predicable): Update.
4433 * ipa-inline-transform.c (inline_transform): Update.
4434 * ipa-split.c (split_function): Update.
4435 * ipa-utils.c (ipa_merge_profiles): Update.
4436 * loop-doloop.c (add_test): Update.
4437 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4438 * lto-streamer-in.c (input_cfg): Update.
4439 (input_function): Update.
4440 * lto-streamer-out.c (output_cfg): Update.
4441 * modulo-sched.c (sms_schedule): Update.
4442 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4443 * predict.c (maybe_hot_edge_p): Update.
4444 (unlikely_executed_edge_p): Update.
4445 (probably_never_executed_edge_p): Update.
4446 (dump_prediction): Update.
4447 (drop_profile): Update.
4448 (propagate_unlikely_bbs_forward): Update.
4449 (determine_unlikely_bbs): Update.
4450 (force_edge_cold): Update.
4451 * profile.c (compute_branch_probabilities): Update.
4452 * reg-stack.c (better_edge): Update.
4453 * shrink-wrap.c (handle_simple_exit): Update.
4454 * tracer.c (better_p): Update.
4455 * trans-mem.c (expand_transaction): Update.
4456 (split_bb_make_tm_edge): Update.
4457 * tree-call-cdce.c: Update.
4458 * tree-cfg.c (gimple_find_sub_bbs): Update.
4459 (gimple_split_edge): Update.
4460 (gimple_duplicate_sese_region): Update.
4461 (gimple_duplicate_sese_tail): Update.
4462 (gimple_flow_call_edges_add): Update.
4463 (insert_cond_bb): Update.
4464 (execute_fixup_cfg): Update.
4465 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4466 * tree-complex.c (expand_complex_div_wide): Update.
4467 * tree-eh.c (lower_resx): Update.
4468 (unsplit_eh): Update.
4469 (cleanup_empty_eh_move_lp): Update.
4470 * tree-inline.c (copy_edges_for_bb): Update.
4471 (freqs_to_counts): Update.
4472 (copy_cfg_body): Update.
4473 * tree-ssa-dce.c (remove_dead_stmt): Update.
4474 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4475 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4476 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4477 (unloop_loops): Update.
4478 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4479 * tree-ssa-loop-split.c (connect_loops): Update.
4480 (split_loop): Update.
4481 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
4482 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4483 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4484 * tree-ssa-reassoc.c (branch_fixup): Update.
4485 * tree-ssa-tail-merge.c (replace_block_by): Update.
4486 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4487 (compute_path_counts): Update.
4488 (update_profile): Update.
4489 (recompute_probabilities): Update.
4490 (update_joiner_offpath_counts): Update.
4491 (estimated_freqs_path): Update.
4492 (freqs_to_counts_path): Update.
4493 (clear_counts_path): Update.
4494 (ssa_fix_duplicate_block_edges): Update.
4495 (duplicate_thread_path): Update.
4496 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4497 (case_bit_test_cmp): Update.
4498 (collect_switch_conv_info): Update.
4499 (gen_inbound_check): Update.
4500 (do_jump_if_equal): Update.
4501 (emit_cmp_and_jump_insns): Update.
4502 * tree-tailcall.c (decrease_profile): Update.
4503 (eliminate_tail_call): Update.
4504 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4505 (vect_do_peeling): Update.
4506 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4507 * ubsan.c (ubsan_expand_null_ifn): Update.
4508 (ubsan_expand_ptr_ifn): Update.
4509 * value-prof.c (gimple_divmod_fixed_value): Update.
4510 (gimple_mod_pow2): Update.
4511 (gimple_mod_subtract): Update.
4512 (gimple_ic): Update.
4513 (gimple_stringop_fixed_value): Update.
4515 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4518 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
4520 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
4522 PR rtl-optimization/82395
4523 * ira-color.c (allocno_priority_compare_func): Fix comparison step
4524 based on non_spilled_static_chain_regno_p.
4526 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4528 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
4529 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
4530 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
4532 2017-10-19 Martin Sebor <msebor@redhat.com>
4534 PR tree-optimization/82596
4535 * tree.c (array_at_struct_end_p): Handle STRING_CST.
4537 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4539 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
4540 (get_mem_refs_of_builtin_call): Likewise.
4541 * builtins.c (expand_builtin_apply): Adjust call to
4542 allocate_dynamic_stack_space.
4543 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
4544 the third argument to allocate_dynamic_stack_space, otherwise -1.
4545 (expand_builtin): Deal with all alloca variants.
4546 (is_inexpensive_builtin): Likewise.
4547 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
4548 * calls.c (special_function_p): Deal with all alloca variants.
4549 (initialize_argument_information): Adjust call to
4550 allocate_dynamic_stack_space.
4551 (expand_call): Likewise.
4552 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
4553 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
4554 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
4555 use it for the stack usage computation.
4556 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
4557 * function.c (gimplify_parameters): Call build_alloca_call_expr.
4558 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
4559 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
4560 (in_loop_p): Remove first argument and useless check.
4561 (pass_walloca::execute): Remove useless test and adjust call to above.
4562 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
4563 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
4564 (gimplify_call_expr): Deal with all alloca variants.
4565 * hsa-gen.c (gen_hsa_alloca): Likewise.
4566 (gen_hsa_insns_for_call): Likewise.
4567 * ipa-pure-const.c (special_builtin_state): Likewise.
4568 * tree-chkp.c (chkp_build_returned_bound): Likewise.
4569 * tree-object-size.c (alloc_object_size): Likewise.
4570 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4571 (call_may_clobber_ref_p_1): Likewise.
4572 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
4573 (ccp_fold_stmt): Likewise.
4574 (optimize_stack_restore): Likewise.
4575 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
4576 (mark_all_reaching_defs_necessary_1): Likewise.
4577 (propagate_necessity): Likewise.
4578 (eliminate_unnecessary_stmts): Likewise.
4579 * tree.c (build_common_builtin_nodes): Build
4580 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
4581 (build_alloca_call_expr): New function.
4582 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
4583 (CASE_BUILT_IN_ALLOCA): Likewise.
4584 (build_alloca_call_expr): Declare.
4585 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
4587 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4590 * dwarf2out.c (new_die_raw): New static inline function.
4591 (new_die): Use it to create the DIE.
4592 (add_AT_external_die_ref): Likewise.
4593 (clone_die): Likewise.
4594 (clone_as_declaration): Likewise.
4595 (dwarf2out_vms_debug_main_pointer): Likewise.
4596 (base_type_die): Likewise. Remove early return for corner cases.
4597 Do not call add_pubtype on the DIE here.
4598 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
4599 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
4600 typedefs for base types with DW_AT_endianity. Make sure a DIE with
4601 native order exists for base types, attach the DIE manually and call
4602 add_pubtype on it. Do not equate a reverse order DIE to the type.
4604 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
4606 * config/arm/arm.c (align_ok_ldrd_strd): New function.
4607 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
4609 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
4611 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4613 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
4614 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
4615 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
4616 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
4617 * opts.c (sanitizer_opts): Add builtin.
4618 * ubsan.c (instrument_builtin): New function.
4619 (pass_ubsan::execute): Call it.
4620 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
4621 * doc/invoke.texi: Document -fsanitize=builtin.
4623 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
4624 builtins, store max (log2 (align), 0) into uchar field instead of
4625 align into uptr field.
4626 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
4627 store uchar 0 field instead of uptr 0 field.
4628 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
4629 instead of passing one address of struct with 2 locations pass
4630 two addresses of structs with 1 location each.
4631 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
4632 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
4633 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
4634 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
4635 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
4636 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
4637 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
4638 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
4640 2017-10-19 Martin Liska <mliska@suse.cz>
4643 * file-find.c (remove_prefix): Remove.
4644 * file-find.h (remove_prefix): Likewise.
4645 * gcc-ar.c: Remove smartness of lookup.
4647 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
4649 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
4650 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
4651 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
4653 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4656 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
4657 (setcc + and to xor + setcc): New peephole2.
4659 2017-10-19 Tom de Vries <tom@codesourcery.com>
4661 * doc/sourcebuild.texi (Test Directives, Variants of
4662 dg-require-support): Add dg-require-stack-size.
4664 2017-10-19 Martin Liska <mliska@suse.cz>
4667 * gimplify.c (gimplify_decl_expr): Do not instrument variables
4668 that have a large alignment.
4669 (gimplify_target_expr): Likewise.
4671 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
4673 PR rtl-optimization/82602
4674 * ira.c (rtx_moveable_p): Return false for volatile asm.
4676 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
4679 * config/i386/i386-modes.def (CCGZ): New CC mode.
4680 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
4681 * config/i386/predicates.md (ix86_comparison_operator):
4683 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4684 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
4685 with double-word subtraction.
4686 (put_condition_code): Handle CCGZmode.
4688 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
4690 * wide-int.cc (debug (const wide_int &)): New.
4691 (debug (const wide_int *)): New.
4692 (debug (const widest_int &)): New.
4693 (debug (const widest_int *)): New.
4695 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
4698 * lra-constraints.c (curr_insn_transform): Use non-input operand
4699 instead of output one for matched reload.
4701 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4703 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
4704 (tree-ssa-loop-ivopts.h): New header file.
4705 (struct builtin_info): New fields.
4706 (classify_builtin_1): Compute and record base and offset parts for
4707 memset builtin partition by calling strip_offset.
4708 (offset_cmp, fuse_memset_builtins): New functions.
4709 (finalize_partitions): Fuse adjacent memset partitions by calling
4711 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
4712 Expose the interface.
4713 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
4715 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4717 PR tree-optimization/82574
4718 * tree-loop-distribution.c (find_single_drs): New parameter. Check
4719 that data reference must be executed exactly once per iteration
4720 against the outermost loop in nest.
4721 (classify_partition): Update call to above function.
4723 2017-10-18 Richard Biener <rguenther@suse.de>
4725 PR tree-optimization/82591
4726 * graphite.c (graphite_transform_loops): Move code gen message
4728 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4729 Here. Handle scop_to_isl_ast failing.
4730 (scop_to_isl_ast): Limit the number of ISL operations.
4732 2017-10-18 Richard Biener <rguenther@suse.de>
4734 * graphite-isl-ast-to-gimple.c
4735 (translate_isl_ast_to_gimple::set_rename): Simplify.
4736 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
4737 (graphite_copy_stmts_from_block): ... here.
4738 (copy_bb_and_scalar_dependences): Simplify.
4739 (add_parameters_to_ivs_params): Canonicalize.
4740 (generate_entry_out_of_ssa_copies): Simplify.
4741 * graphite-sese-to-poly.c (extract_affine_name): Simplify
4742 by passing in ISL dimension.
4743 (parameter_index_in_region_1): Rename to ...
4744 (parameter_index_in_region): ... this.
4745 (extract_affine): Adjust assert, pass down parameter index.
4746 (add_param_constraints): Use range-info when available.
4747 (build_scop_context): Adjust.
4748 * sese.c (new_sese_info): Adjust.
4749 (free_sese_info): Likewise.
4750 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
4751 Remove unused typedefs.
4752 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
4754 2017-10-18 Martin Liska <mliska@suse.cz>
4756 * combine.c (simplify_compare_const): Add gcc_fallthrough.
4758 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4760 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
4761 (s390_sched_init): Do not reset s390_sched_state if we entered the
4762 current basic block via a fallthru edge and all others are unlikely.
4764 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4766 * config/s390/s390.c (NUM_SIDES): New variable.
4767 (LONGRUNNING_THRESHOLD): New variable.
4768 (LATENCY_FACTOR): New variable.
4769 (s390_sched_score): Decrease score for long-running instructions on
4771 (s390_sched_variable_issue): Perform bookkeeping for long-running
4774 2017-10-18 Richard Biener <rguenther@suse.de>
4776 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
4777 Simplify with removal of the parameter rename map.
4778 (set_rename): Likewise.
4779 (should_copy_to_new_region): Likewise.
4780 (graphite_copy_stmts_from_block): Likewise.
4781 (copy_bb_and_scalar_dependences): Remove initialization of
4782 unused copied_bb_map.
4784 (copy_internal_parameters): Likewise.
4785 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
4786 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
4787 Use INTEGRAL_TYPE_P.
4788 (parameter_index_in_region_1): Rename to ...
4789 (assign_parameter_index_in_region): ... this. Assert we have
4790 a parameter we handle.
4791 (scan_tree_for_params): Adjust.
4792 * sese.h (parameter_rename_map_t): Remove.
4793 (struct sese_info_t): Remove unused parameter_rename_map and
4794 copied_bb_map members.
4795 * sese.c (new_sese_info): Adjust.
4796 (free_sese_info): Likewise.
4798 2017-10-18 Martin Liska <mliska@suse.cz>
4801 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
4802 on an abnormal edge.
4804 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
4806 * doc/invoke.texi (ffunction-sections and fdata-sections):
4809 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4811 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
4812 the use statement can throw internally.
4814 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4816 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
4817 any RTX present on the RHS of a SET.
4818 * compare-elim.c (try_eliminate_compare): Restore comment.
4820 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4822 * langhooks.h (struct lang_hooks): Document that tree_size langhook
4823 may be also called on tcc_type nodes.
4824 * langhooks.c (lhd_tree_size): Likewise.
4826 2017-10-17 David Malcolm <dmalcolm@redhat.com>
4828 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
4829 format_warning_at_substring.
4830 (maybe_warn): Convert source_range * param to a location_t. Pass
4831 UNKNOWN_LOCATION rather than NULL to fmtwarn.
4832 (format_directive): Remove code to extract source_ranges and
4833 source_range * in favor of just a location_t.
4834 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
4836 * substring-locations.c (format_warning_va): Convert
4837 source_range * param to a location_t.
4838 (format_warning_at_substring): Likewise.
4839 * substring-locations.h (format_warning_va): Likewise.
4840 (format_warning_at_substring): Likewise.
4842 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
4844 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
4846 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
4847 and scatter/gather ops.
4849 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
4850 vec_gather_load and vec_scatter_store.
4851 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
4852 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
4853 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
4854 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
4855 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
4856 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
4858 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
4860 * reg-stack.c (compare_for_stack_reg): Add bool argument.
4861 Detect FTST instruction and handle its register pops. Only pop
4862 second operand if can_pop_second_op is true.
4863 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
4864 set can_pop_second_op to false in the compare_for_stack_reg call.
4866 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
4867 output_fp_compare for stack register operands.
4868 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
4869 instructions here. Do not emit stack register pops here. Assert
4870 that FCOMPP pops next to top stack register. Rewrite function.
4872 2017-10-17 Nathan Sidwell <nathan@acm.org>
4875 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
4876 use DECL_ASSEMBLER_NAME_RAW.
4879 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
4882 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
4883 Wilco Dijkstra <wilco.dijkstra@arm.com>
4885 * builtins.c (expand_builtin_update_setjmp_buf): Add a
4886 converstion to Pmode from the buf_addr.
4888 2017-10-17 Richard Biener <rguenther@suse.de>
4890 * graphite-dependences.c (scop_get_reads_and_writes): Change
4891 output parameters to references.
4893 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
4895 PR 71026/tree-optimization
4896 * fold-const.c (distribute_real_division): Removed.
4897 (fold_binary_loc): Remove calls to distribute_real_divison.
4899 2017-10-17 Richard Biener <rguenther@suse.de>
4901 * graphite-scop-detection.c
4902 (scop_detection::stmt_has_simple_data_refs_p): Always use
4903 the full nest as region.
4904 (try_generate_gimple_bb): Likewise.
4905 * sese.c (scalar_evolution_in_region): Simplify now that
4906 SCEV can handle instantiation in regions.
4907 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
4908 in the non-loop part of a function if requested.
4910 2017-10-17 Richard Biener <rguenther@suse.de>
4912 PR tree-optimization/82563
4913 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
4915 (graphite_regenerate_ast_isl): Call it.
4916 * graphite-scop-detection.c (build_scops): Remove entry edge split.
4918 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4920 PR tree-optimization/82549
4921 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
4922 Formatting fixes. Instead of calling make_bit_field_ref with negative
4925 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
4927 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
4928 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
4929 _mm_maskz_reduce_ss): New.
4930 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
4931 __builtin_ia32_reducess_mask): Ditto..
4932 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
4933 * config/i386/sse.md (reduces<mode>): Renamed to ...
4934 (reduces<mode><mask_scalar_name>): ... this.
4935 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
4937 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
4938 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
4940 2017-10-16 David Malcolm <dmalcolm@redhat.com>
4942 * Makefile.in (OBJS): Add unique-ptr-tests.o.
4943 * selftest-run-tests.c (selftest::run_tests): Call
4944 selftest::unique_ptr_tests_cc_tests.
4945 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
4946 * unique-ptr-tests.cc: New file.
4948 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
4951 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4953 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4954 (make_hard_regno_born, make_hard_regno_dead): Update
4955 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
4957 2017-10-16 Jeff Law <law@redhat.com>
4959 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4961 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
4963 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
4965 2017-10-16 Olivier Hainque <hainque@adacore.com>
4967 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
4968 with_cpu if we were configured for an e500v2 target cpu name.
4970 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
4972 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
4973 * doc/invoke.texi: Document +nodsp as a valid extension for
4976 2017-10-16 Martin Liska <mliska@suse.cz>
4978 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
4979 (test_set_range): Likewise.
4980 (test_range_functions): Rename to ...
4981 (test_bit_in_range): ... this.
4982 (sbitmap_c_tests): Add new test.
4984 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4986 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
4988 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
4989 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
4991 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4993 * config/aarch64/aarch64-builtins.c
4994 (aarch64_types_quadopu_lane_qualifiers): New.
4995 (TYPES_QUADOPU_LANE): New.
4996 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
4997 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
4998 (aarch64_<sur>dot_laneq<vsi2qi>): New.
4999 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
5000 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
5001 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
5002 (Vdottype, DOTPROD): New.
5003 (sur): Add SDOT and UDOT.
5005 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5007 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
5008 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5009 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
5011 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
5012 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
5013 Enable TARGET_DOTPROD.
5014 (cortex-a75.cortex-a55): Likewise.
5015 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
5017 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5019 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
5020 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
5022 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
5024 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
5025 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
5026 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
5027 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
5028 * config/arm/types.md (neon_dot, neon_dot_q): New.
5029 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
5031 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5033 * config/arm/arm.h (TARGET_DOTPROD): New.
5034 * config/arm/arm.c (arm_arch_dotprod): New.
5035 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
5036 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
5037 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
5038 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
5039 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
5040 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
5041 * doc/invoke.texi (armv8.2-a): Document dotprod
5043 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5045 * i386.c (ix86_vec_cost): New function.
5046 (ix86_rtx_costs): Handle vector operations better.
5047 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
5048 * x86-tune-costs.h: Add new costs to all tables.
5050 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5052 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
5054 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
5055 divsd, sqrtss and sqrtsd
5056 * x86-tune-costs.h: Add new entries to all costs.
5057 (znver1_cost): Fix to match real instruction latencies.
5059 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5060 Michael Collison <michael.collison@arm.com>
5062 * compare-elim.c: Include emit-rtl.h.
5063 (can_merge_compare_into_arith): New function.
5064 (try_validate_parallel): Likewise.
5065 (try_merge_compare): Likewise.
5066 (try_eliminate_compare): Call the above when no previous clobber
5068 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
5071 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5075 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
5076 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
5080 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
5081 to be any operand_equal_p operands. For & (B - 1) require
5082 B to be power of 2. Recognize
5083 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
5085 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5088 * optabs.c (expand_memory_blockage): Fix call of
5089 targetm.have_memory_blockage.
5091 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5094 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
5095 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
5096 objects to extra_objs instead of overwriting it.
5098 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5100 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
5101 Use any_fp_register_operand as operand[3] predicate. Simplify
5102 equality test for operands[2] and operands[4] memory location.
5103 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
5104 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
5105 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
5106 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
5107 any_fp_register_operand as operand[1] predicate. Simplify
5108 equality test for operands[0] and operands[3] memory location.
5109 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
5110 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
5111 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
5113 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5115 * target-insns.def: Add memory_blockage.
5116 * optabs.c (expand_memory_blockage): New function.
5117 (expand_asm_memory_barrier): Rename ...
5118 (expand_asm_memory_blockage): ... to this.
5119 (expand_mem_thread_fence): Call expand_memory_blockage
5120 instead of expand_asm_memory_barrier.
5121 (expand_mem_singnal_fence): Ditto.
5122 (expand_atomic_load): Ditto.
5123 (expand_atomic_store): Ditto.
5124 * doc/md.texi (Standard Pattern Names For Generation):
5125 Document memory_blockage instruction pattern.
5127 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
5129 * config/rl78/rl78.c (rl78_emit_libcall): New function.
5130 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
5131 * config/rl78/rl78.md: New define_expand "adddi3".
5133 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5135 * cfghooks.c (verify_flow_info): Disable check that all probabilities
5138 2017-10-13 Jeff Law <law@redhat.com>
5140 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
5142 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5145 * internal-fn.c (expand_mul_overflow): If both operands have
5146 the same highpart of -1 or 0 and the topmost bit of lowpart
5147 is different, overflow is if res <= 0 rather than res < 0.
5149 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
5151 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
5152 TARGET_P9_VECTOR code for unaligned_load case.
5154 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5156 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
5158 2017-10-13 Nathan Sidwell <nathan@acm.org>
5160 * tree-core.h (tree_contains_struct): Make bool.
5161 * tree.c (tree_contains_struct): Likewise.
5162 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
5163 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
5164 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
5165 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
5166 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
5168 2017-10-13 Richard Biener <rguenther@suse.de>
5170 * graphite-isl-ast-to-gimple.c
5171 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
5172 parameters and dominance check.
5173 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
5174 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
5175 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
5176 Do not update SSA form here or do intermediate IL verification.
5177 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
5178 (graphite_initialize): Remove check on the number of loops in
5179 the function and inline into graphite_transform_loops.
5180 (graphite_finalize): Inline into graphite_transform_loops.
5181 (graphite_transform_loops): Perform SSA update and IL verification
5183 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5185 2017-10-13 Richard Biener <rguenther@suse.de>
5187 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5188 graphite_expression_type_precision): Avoid global constructor
5190 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5191 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5192 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5193 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5194 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5195 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5197 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5200 * config/i386/i386.h (ix86_red_zone_size): New.
5201 * config/i386/i386.md (push peephole2s): Replace
5202 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
5204 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5205 Alan Hayward <alan.hayward@arm.com>
5206 David Sherwood <david.sherwood@arm.com>
5208 * combine.c (can_change_dest_mode): Reject changes in
5209 REGMODE_NATURAL_SIZE.
5211 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5212 Alan Hayward <alan.hayward@arm.com>
5213 David Sherwood <david.sherwood@arm.com>
5215 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5216 (expand_debug_source_expr): Likewise.
5217 * combine.c (combine_simplify_rtx): Likewise.
5218 * cse.c (fold_rtx): Likewise.
5219 * fwprop.c (canonicalize_address): Likewise.
5220 * targhooks.c (default_shift_truncation_mask): Likewise.
5222 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5223 Alan Hayward <alan.hayward@arm.com>
5224 David Sherwood <david.sherwood@arm.com>
5226 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5227 (widened_mode): Likewise.
5228 (expand_unop): Likewise.
5229 * ree.c (transform_ifelse): Likewise.
5230 (merge_def_and_ext): Likewise.
5231 (combine_reaching_defs): Likewise.
5232 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5234 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5235 Alan Hayward <alan.hayward@arm.com>
5236 David Sherwood <david.sherwood@arm.com>
5238 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5239 * combine.c (gen_lowpart_for_combine): Likewise.
5240 * dwarf2out.c (rtl_for_decl_location): Likewise.
5241 * final.c (alter_subreg): Likewise.
5242 * rtlhooks.c (gen_lowpart_general): Likewise.
5243 (gen_lowpart_if_possible): Likewise.
5245 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5246 Alan Hayward <alan.hayward@arm.com>
5247 David Sherwood <david.sherwood@arm.com>
5249 * calls.c (expand_call): Use subreg_lowpart_offset.
5250 * cse.c (cse_insn): Likewise.
5251 * regcprop.c (copy_value): Likewise.
5252 (copyprop_hardreg_forward_1): Likewise.
5254 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5257 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5258 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5259 =Q constraints to +Q and into insn condition add check
5260 that operands[0] and operands[1] are equal.
5261 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5262 =Q constraints to +Q and into insn condition add check
5263 that operands[0] is equal to either operands[1] or operands[2].
5266 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5267 instead of handling MINUS_EXPR twice (once for each argument),
5268 canonicalize operand order and handle just once, use rtype where
5269 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5272 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5273 any values of __C while still being pattern recognizable as a simple
5276 2017-10-13 Richard Biener <rguenther@suse.de>
5278 PR tree-optimization/82451
5280 2017-10-02 Richard Biener <rguenther@suse.de>
5282 PR tree-optimization/82355
5283 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5284 a mapping for the enclosing loop but avoid generating one for
5286 (copy_bb_and_scalar_dependences): Remove premature codegen
5287 error on PHIs in blocks duplicated into multiple places.
5288 * graphite-scop-detection.c
5289 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5290 in the region use it as loop and nest to analyze the DR in.
5291 (try_generate_gimple_bb): Likewise.
5292 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5293 (add_loop_constraints): For blocks in a loop not in the region
5294 create a dimension with a single iteration.
5295 * sese.h (gbb_loop_at_index): Remove assert.
5297 * cfgloop.c (loop_preheader_edge): For the loop tree root
5298 return the single successor of the entry block.
5299 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5300 Reset the SCEV hashtable and niters.
5301 * graphite-scop-detection.c
5302 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5303 assert that we only have POLYNOMIAL_CHREC that vary in loops
5304 contained in the region.
5305 (scop_detection::graphite_can_represent_expr): Adjust.
5306 (scop_detection::stmt_has_simple_data_refs_p): For loops
5307 not in the region set loop to NULL. The nest is now the
5308 entry edge to the region.
5309 (try_generate_gimple_bb): Likewise.
5310 * sese.c (scalar_evolution_in_region): Adjust for
5311 instantiate_scev change.
5312 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5313 Make nest parameter the edge into the region.
5314 (create_data_ref): Likewise.
5315 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5316 entry edge into a region and adjust instantiate_scev calls.
5317 (create_data_ref): Likewise.
5318 (graphite_find_data_references_in_stmt): Likewise.
5319 (find_data_references_in_stmt): Pass the loop preheader edge
5320 from the nest argument.
5321 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5322 parameter the edge into the region.
5323 (instantiate_parameters): Use the loop preheader edge as entry.
5324 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5326 (get_instantiated_value_entry): Make instantiate_below parameter
5327 the edge into the region.
5328 (instantiate_scev_name): Likewise. Adjust dominance checks,
5329 when we cannot use loop-based instantiation instantiate by
5330 walking use-def chains.
5331 (instantiate_scev_poly): Adjust.
5332 (instantiate_scev_binary): Likewise.
5333 (instantiate_scev_convert): Likewise.
5334 (instantiate_scev_not): Likewise.
5335 (instantiate_array_ref): Remove.
5336 (instantiate_scev_3): Likewise.
5337 (instantiate_scev_2): Likewise.
5338 (instantiate_scev_1): Likewise.
5339 (instantiate_scev_r): Do not blindly handle N-operand trees.
5340 Do not instantiate array-refs. Handle all constants and invariants.
5341 (instantiate_scev): Make instantiate_below parameter
5342 the edge into the region.
5343 (resolve_mixers): Use the loop preheader edge for the region
5344 parameter to instantiate_scev_r.
5345 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5347 2017-10-13 Richard Biener <rguenther@suse.de>
5349 PR tree-optimization/82525
5350 * graphite-isl-ast-to-gimple.c
5351 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5353 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5354 Fail code generation when we cannot represent the isl integer.
5355 (binary_op_to_tree): Elide modulo operations that are no-ops
5356 in the type we code generate. Remove now superfluous code
5359 2017-10-13 Richard Biener <rguenther@suse.de>
5361 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5362 (scop_detection::harmful_loop_in_region): Remove premature
5363 IV type restriction.
5364 (scop_detection::graphite_can_represent_scev): We can handle
5365 pointer IVs just fine.
5367 2017-10-13 Alan Modra <amodra@gmail.com>
5369 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5370 "Clobbers and Scratch Registers". Add paragraph on
5371 alternative to clobbers for scratch registers and OpenBLAS
5374 2017-10-13 Alan Modra <amodra@gmail.com>
5376 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5377 example of a memory input for a string of known length. Move
5378 commentary out of table. Add a number of new examples
5379 covering array memory inputs.
5381 2017-10-12 Martin Liska <mliska@suse.cz>
5383 PR tree-optimization/82493
5384 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5385 (test_range_functions): New function.
5386 (sbitmap_c_tests): Likewise.
5387 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5388 * selftest.h (sbitmap_c_tests): New function.
5390 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5392 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5394 * config/rs6000/amo.h: Fix spacing issue.
5396 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5399 * config/i386/i386.md (*ashl<mode>3_mask_1,
5400 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5401 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5404 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5406 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5407 (profile_probability): Set max_probability
5408 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5410 * profile-count.c (profile_probability::differs_from_p): Do not
5411 rely on max_probaiblity == 10000
5413 2017-10-12 Jeff Law <law@redhat.com>
5415 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5418 2017-10-12 Martin Sebor <msebor@redhat.com>
5422 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5423 (handle_alias_pairs): Call it.
5424 * common.opt (-Wattribute-alias): New option.
5425 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5426 * doc/invoke.texi (-Wattribute-alias): Document.
5428 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5431 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5433 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5435 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5436 (make_hard_regno_born, make_hard_regno_dead): Update
5437 bb_killed_pseudos and bb_gen_pseudos.
5439 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5441 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5443 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5445 * config/alpha/alpha.c (alpha_split_conditional_move):
5446 Use std::swap instead of manually swapping.
5447 (alpha_stdarg_optimize_hook): Ditto.
5448 (alpha_canonicalize_comparison): Ditto.
5450 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5452 * tree-loop-distribution.c (struct builtin_info): New struct.
5453 (struct partition): Refactor fields into struct builtin_info.
5454 (partition_free): Free struct builtin_info.
5455 (build_size_arg_loc, build_addr_arg_loc): Delete.
5456 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5457 information from struct builtin_info.
5458 (find_single_drs): New function refactored from classify_partition.
5459 Also moved builtin validity checks to this function.
5460 (compute_access_range, alloc_builtin): New functions.
5461 (classify_builtin_st, classify_builtin_ldst): New functions.
5462 (classify_partition): Refactor code into functions find_single_drs,
5463 classify_builtin_st and classify_builtin_ldst.
5464 (distribute_loop): Don't do runtime alias check when distributing
5466 (find_seed_stmts_for_distribution): New function.
5467 (pass_loop_distribution::execute): Refactor code finding seed
5468 stmts into above function. Support distribution for the innermost
5469 two-level loop nest. Adjust dump information.
5471 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5473 * tree-loop-distribution.c: Adjust the general comment.
5474 (NUM_PARTITION_THRESHOLD): New macro.
5475 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
5476 (classify_partition): Skip builtin pattern of loop nest's inner loop.
5477 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
5478 in call to build_partition_graph.
5479 (finalize_partitions): New parameter. Make loop distribution more
5480 conservative by fusing more partitions.
5481 (distribute_loop): Don't do runtime alias check in case of loop nest
5483 (find_seed_stmts_for_distribution): New function.
5484 (prepare_perfect_loop_nest): New function.
5485 (pass_loop_distribution::execute): Refactor code finding seed stmts
5486 and loop nest into above functions. Support loop nest distribution.
5487 Adjust dump information accordingly.
5489 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5491 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
5492 and set PTYPE_SEQUENTIAL for merged partition.
5494 2017-10-12 Richard Biener <rguenther@suse.de>
5496 PR tree-optimization/69728
5498 2017-09-19 Richard Biener <rguenther@suse.de>
5500 PR tree-optimization/69728
5501 * graphite-sese-to-poly.c (schedule_error): New global.
5502 (add_loop_schedule): Handle empty domain by failing the
5504 (build_original_schedule): Handle schedule_error.
5506 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
5507 domain by returning an unchanged schedule.
5509 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5511 * genrecog.c (validate_pattern): For VEC_SELECT verify that
5512 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
5514 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
5516 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
5519 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5522 * expr.c (store_field): Don't optimize away bitsize == 0 store
5523 from CALL_EXPR with addressable return type.
5525 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5527 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
5528 * config/rs6000/rs6000.md (sel): Delete mode attribute.
5529 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
5530 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
5531 TARGET_ISEL instead of TARGET_ISEL<sel>.
5533 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
5535 * config/rs6000/rs6000.c
5536 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
5538 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5540 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
5542 (reg_or_cint_operand): Fix comment.
5543 (reg_or_zero_operand): New predicate.
5544 * config/rs6000/rs6000-protos.h (output_isel): Delete.
5545 * config/rs6000/rs6000.c (output_isel): Delete.
5546 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
5547 instead of reg_or_cint_operand. Output instruction directly (not via
5549 (isel_unsigned_<mode>): Ditto.
5550 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
5551 gpc_reg_operand. Add an instruction alternative for this. Output
5552 instruction directly.
5553 (*isel_reversed_unsigned_<mode>): Ditto.
5555 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5557 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
5558 (TARGET_CANONICALIZE_COMPARISON): Define.
5560 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
5563 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
5564 Check whether the dest is REG before adding REG_EQUIV note.
5566 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5569 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5571 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5572 (make_hard_regno_born, make_hard_regno_dead): Update
5573 bb_killed_pseudos and bb_gen_pseudos.
5575 2017-10-11 Nathan Sidwell <nathan@acm.org>
5577 * incpath.h (enum incpath_kind): Name enum, prefix values.
5578 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
5579 * incpath.c (heads, tails): Use INC_MAX.
5580 (add_env_var_paths, add_standard_paths): Use incpath_kind.
5581 (merge_include_chains, split_quote_chain,
5582 register_include_chains): Update incpath_kind names.
5583 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
5584 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
5586 (add_framework_path, darwin_register_objc_includes): Likewise.
5587 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
5589 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5591 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
5592 Do not use float_operator operator predicate.
5593 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5594 * config/i386/predicates.md (float_operator): Remove predicate.
5596 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5598 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
5599 (*jccxf_i387): Ditto.
5600 (*jcc<mode>_i387): Ditto.
5601 (*jccu<mode>_i387): Ditto.
5602 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
5603 (*jcc_*_i387 splitters): Remove.
5604 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
5605 * config/i386/i386.c (ix86_split_fp_branch): Remove.
5606 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
5609 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5611 * profile-count.h (slow_safe_scale_64bit): New function.
5612 (safe_scale_64bit): New inline.
5613 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
5614 * profile-count.c: Include wide-int.h
5615 (slow_safe_scale_64bit): New.
5617 2017-10-11 Nathan Sidwell <nathan@acm.org>
5619 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
5620 HAS_DECL_ASSEMBLER_NAME_P.
5621 * gimple-expr.c (gimple_decl_printable_name: Check
5622 HAS_DECL_ASSEMBLER_NAME_P too.
5623 * ipa-utils.h (type_in_anonymous_namespace_p): Check
5624 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
5625 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
5626 * passes.c (rest_of_decl_compilation): Check
5627 HAS_DECL_ASSEMBLER_NAME_P too.
5628 * recog.c (verify_changes): Likewise.
5629 * tree-pretty-print.c (dump_decl_name): Likewise.
5630 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
5632 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
5633 (SET_DECL_ASSEMBLER_NAME): Use it.
5634 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
5635 (COPY_DECL_ASSEMBLER_NAME): Likewise.
5636 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
5638 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5640 * config.gcc (i386, x86_64): Add extra objects.
5641 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
5642 (ix86_min_insn_size): Declare.
5643 (ix86_issue_rate): Declare.
5644 (ix86_adjust_cost): Declare.
5645 (ia32_multipass_dfa_lookahead): Declare.
5646 (ix86_macro_fusion_p): Declare.
5647 (ix86_macro_fusion_pair_p): Declare.
5648 (ix86_bd_has_dispatch): Declare.
5649 (ix86_bd_do_dispatch): Declare.
5650 (ix86_core2i7_init_hooks): Declare.
5651 (ix86_atom_sched_reorder): Declare.
5652 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
5653 (COSTS_N_BYTES): Move to x86-tune-costs.h.
5654 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
5655 (rip_relative_addr_p): Rename to ...
5656 (ix86_rip_relative_addr_p): ... this one; export.
5657 (memory_address_length): Update.
5658 (ix86_issue_rate): Move to x86-tune-sched.c.
5659 (ix86_flags_dependent): Move to x86-tune-sched.c.
5660 (ix86_agi_dependent): Move to x86-tune-sched.c.
5661 (exact_dependency_1): Move to x86-tune-sched.c.
5662 (exact_store_load_dependency): Move to x86-tune-sched.c.
5663 (ix86_adjust_cost): Move to x86-tune-sched.c.
5664 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
5665 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
5666 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
5667 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
5668 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
5669 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
5670 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
5671 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
5672 (min_insn_size): Rename to ...
5673 (ix86_min_insn_size): ... this one; export.
5674 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
5675 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
5676 (core2i7_first_cycle_multipass_backtrack): Move to
5677 x86-tune-sched-core.c.
5678 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
5679 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
5680 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
5681 (ix86_avoid_jump_mispredicts): Update.
5682 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
5683 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
5684 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
5685 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
5686 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
5687 (MAX_INSN): Move to ix86-tune-sched-bd.c.
5688 (MAX_IMM): Move to ix86-tune-sched-bd.c.
5689 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
5690 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
5691 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
5692 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
5693 (MAX_STORE): Move to ix86-tune-sched-bd.c.
5694 (BIG): Move to ix86-tune-sched-bd.c.
5695 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
5696 (enum insn_path): Move to ix86-tune-sched-bd.c.
5697 (get_mem_group): Move to ix86-tune-sched-bd.c.
5698 (is_cmp): Move to ix86-tune-sched-bd.c.
5699 (dispatch_violation): Move to ix86-tune-sched-bd.c.
5700 (is_branch): Move to ix86-tune-sched-bd.c.
5701 (is_prefetch): Move to ix86-tune-sched-bd.c.
5702 (init_window): Move to ix86-tune-sched-bd.c.
5703 (allocate_window): Move to ix86-tune-sched-bd.c.
5704 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
5705 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
5706 (process_end_window): Move to ix86-tune-sched-bd.c.
5707 (allocate_next_window): Move to ix86-tune-sched-bd.c.
5708 (find_constant): Move to ix86-tune-sched-bd.c.
5709 (get_num_immediates): Move to ix86-tune-sched-bd.c.
5710 (has_immediate): Move to ix86-tune-sched-bd.c.
5711 (get_insn_path): Move to ix86-tune-sched-bd.c.
5712 (get_insn_group): Move to ix86-tune-sched-bd.c.
5713 (count_num_restricted): Move to ix86-tune-sched-bd.c.
5714 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
5715 (add_insn_window): Move to ix86-tune-sched-bd.c.
5716 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
5717 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
5718 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
5719 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
5720 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
5721 (do_dispatch): Move to ix86-tune-sched-bd.c.
5722 (has_dispatch): Move to ix86-tune-sched-bd.c.
5723 * config/i386/t-i386: Add new object files.
5724 * config/i386/x86-tune-costs.h: New file.
5725 * config/i386/x86-tune-sched-atom.c: New file.
5726 * config/i386/x86-tune-sched-bd.c: New file.
5727 * config/i386/x86-tune-sched-core.c: New file.
5728 * config/i386/x86-tune-sched.c: New file.
5730 2017-10-11 Liu Hao <lh_mouse@126.com>
5732 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
5733 the generic version below instead.
5734 (should_colorize): Recognize Windows consoles as terminals
5736 * pretty-print.c [__MINGW32__] (write_all): New function.
5737 [__MINGW32__] (find_esc_head): Likewise.
5738 [__MINGW32__] (find_esc_terminator): Likewise.
5739 [__MINGW32__] (eat_esc_sequence): Likewise.
5740 [__MINGW32__] (mingw_ansi_fputs): New function that handles
5742 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
5745 2017-10-11 Richard Biener <rguenther@suse.de>
5747 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
5748 Properly call analyze_scalar_evolution with the loop of the stmt.
5750 2017-10-11 Richard Biener <rguenther@suse.de>
5752 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
5753 * tree-core.h (tree_base): Add chrec_var union member.
5754 * tree.h (CHREC_VAR): Remove.
5755 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
5756 * tree-chrec.h (build_polynomial_chrec): Adjust.
5757 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
5758 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
5760 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
5762 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
5763 * match.pd: ... here.
5764 ((T) X == (T) Y): Relax condition.
5766 2017-10-11 Bin Cheng <bin.cheng@arm.com>
5768 PR tree-optimization/82472
5769 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
5771 (break_alias_scc_partitions): Update postorder number.
5773 2017-10-11 Martin Liska <mliska@suse.cz>
5776 * opts.c (parse_no_sanitize_attribute): Do not use error_value
5778 * opts.h (parse_no_sanitize_attribute): Remove last argument.
5780 2017-10-11 Martin Liska <mliska@suse.cz>
5782 * print-rtl.c (print_insn): Move declaration of idbuf
5783 to same scope as name.
5785 2017-10-11 Martin Liska <mliska@suse.cz>
5790 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5793 2017-10-11 Martin Liska <mliska@suse.cz>
5796 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5799 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
5801 * config.gcc (default_gnu_indirect_function): Default to yes for
5802 arm*-*-linux* with glibc.
5804 2017-10-11 Richard Biener <rguenther@suse.de>
5806 * tree-scalar-evolution.c (get_scalar_evolution): Handle
5807 default-defs and types we do not want to analyze.
5808 (interpret_loop_phi): Replace unreachable code with an assert.
5809 (compute_scalar_evolution_in_loop): Remove and inline ...
5810 (analyze_scalar_evolution_1): ... here, replacing condition with
5811 what makes the intent clearer. Remove handling of cases
5812 get_scalar_evolution now handles.
5814 2017-10-10 Jim Wilson <wilson@tuliptree.org>
5816 PR rtl-optimization/81434
5817 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
5818 comment for main loop. In sched_group_found if, also add checks for
5819 pass and min_cost_group.
5821 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
5823 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
5824 (rs6000_insn_cost): New function.
5825 * config/rs6000/rs6000.md (cost): New attribute.
5827 2017-10-10 Jakub Jelinek <jakub@redhat.com>
5828 H.J. Lu <hongjiu.lu@intel.com>
5832 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
5833 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
5834 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
5835 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
5836 to OPTION_MASK_ISA_AVX512VL - builtins that have both
5837 OPTION_MASK_ISA_MMX and some other bit set require both
5838 mmx and the ISAs without the mmx bit.
5839 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
5840 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
5841 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
5842 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
5843 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
5844 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
5845 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
5846 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
5847 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
5848 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
5849 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
5850 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
5851 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
5852 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
5853 Add OPTION_MASK_ISA_MMX.
5855 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
5857 * config.gcc (armv7*-*-freebsd*): New target.
5858 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
5860 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
5862 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
5863 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
5866 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
5868 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
5869 a template parameter.
5870 (WIDE_INT_REF_FOR): Update accordingly.
5871 * tree.h (wi::int_traits <const_tree>): Delete.
5872 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
5873 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
5874 (wi::tree_to_wide_ref): New typedef.
5875 (wi::to_wide): New function.
5876 * calls.c (get_size_range): Use wi::to_wide when operating on
5878 * cgraph.c (cgraph_node::create_thunk): Likewise.
5879 * config/i386/i386.c (ix86_data_alignment): Likewise.
5880 (ix86_local_alignment): Likewise.
5881 * dbxout.c (stabstr_O): Likewise.
5882 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
5883 * expr.c (const_vector_from_tree): Likewise.
5884 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
5885 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
5886 (fold_negate_expr_1, int_const_binop_1, const_binop)
5887 (fold_convert_const_int_from_real, optimize_bit_field_compare)
5888 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
5889 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
5890 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
5891 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
5892 (fold_not_const, round_up_loc): Likewise.
5893 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
5894 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
5895 (alloca_call_type): Likewise.
5896 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
5897 * godump.c (go_output_typedef): Likewise.
5898 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
5899 * internal-fn.c (get_min_precision): Likewise.
5900 * ipa-cp.c (ipcp_store_vr_results): Likewise.
5901 * ipa-polymorphic-call.c
5902 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
5903 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
5904 (ipa_modify_call_arguments): Likewise.
5905 * match.pd: Likewise.
5906 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
5907 * print-tree.c (print_node_brief, print_node): Likewise.
5908 * stmt.c (expand_case): Likewise.
5909 * stor-layout.c (layout_type): Likewise.
5910 * tree-affine.c (tree_to_aff_combination): Likewise.
5911 * tree-cfg.c (group_case_labels_stmt): Likewise.
5912 * tree-data-ref.c (dr_analyze_indices): Likewise.
5913 (prune_runtime_alias_test_list): Likewise.
5914 * tree-dump.c (dequeue_and_dump): Likewise.
5915 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
5916 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
5917 * tree-pretty-print.c (dump_generic_node): Likewise.
5918 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
5919 (simple_iv_with_niters): Likewise.
5920 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
5921 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
5922 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
5923 * tree-ssa-loop-niter.c (split_to_var_and_offset)
5924 (refine_value_range_using_guard, number_of_iterations_ne_max)
5925 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
5926 (get_cst_init_from_scev, record_nonwrapping_iv)
5927 (scev_var_range_cant_overflow): Likewise.
5928 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
5929 * tree-ssa-pre.c (compute_avail): Likewise.
5930 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
5931 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
5932 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
5933 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
5934 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
5935 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
5936 (dump_case_nodes, try_switch_expansion): Likewise.
5937 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
5938 (vect_do_peeling): Likewise.
5939 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
5940 * tree-vect-stmts.c (vectorizable_load): Likewise.
5941 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
5942 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
5943 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
5944 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
5945 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
5946 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
5947 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
5948 (evrp_dom_walker::before_dom_children): Likewise.
5949 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
5950 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
5951 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
5952 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
5953 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
5954 (get_range_pos_neg): Likewise.
5955 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
5956 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
5957 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
5958 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
5959 * config/avr/avr.c (avr_fold_builtin): Likewise.
5960 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
5961 * config/msp430/msp430.c (msp430_attr): Likewise.
5962 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
5963 * config/powerpcspe/powerpcspe-c.c
5964 (altivec_resolve_overloaded_builtin): Likewise.
5965 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
5966 (rs6000_expand_ternop_builtin): Likewise.
5967 * config/rs6000/rs6000-c.c
5968 (altivec_resolve_overloaded_builtin): Likewise.
5969 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
5970 (rs6000_expand_ternop_builtin): Likewise.
5971 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
5973 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5975 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
5976 when copying loop nest with only one inner loop.
5978 2017-10-10 Richard Biener <rguenther@suse.de>
5980 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
5981 blocks if SCEV is active.
5982 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
5984 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
5985 (scev_initialize): Assert we are not yet initialized.
5987 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5989 * tree-loop-distribution.c (generate_loops_for_partition): Remove
5990 inner loop's exit stmt by making it always exit the loop, otherwise
5991 we would generate an infinite empty loop.
5993 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5995 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
5996 renaming variables in new preheader if it's deleted.
5998 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6000 * tree-loop-distribution.c (struct partition): Remove unused field
6001 loops of the structure.
6002 (partition_alloc, partition_free): Ditto.
6003 (build_rdg_partition_for_vertex): Ditto.
6005 2017-10-09 Jeff Law <law@redhat.com>
6007 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
6008 return type to match prototype and documentation.
6010 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6012 * config/rs6000/rs6000.c (processor_costs): Move to ...
6013 * config/rs6000/rs6000.h: ... here.
6014 (rs6000_cost): Declare.
6016 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
6018 * except.c (setjmp_fn): New global variable.
6019 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
6020 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6021 if DONT_USE_BUILTIN_SETJMP is defined.
6023 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6025 * target.def (insn_cost): New hook.
6026 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
6027 * doc/tm.texi: Regenerate.
6028 * rtlanal.c (insn_cost): Use the new hook.
6030 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6032 * combine.c (combine_validate_cost): Compute the new insn_cost,
6033 not just pattern_cost.
6034 (try_combine): Adjust comment.
6036 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6038 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
6040 * combine.c (uid_insn_cost): Adjust comment.
6041 (combine_validate_cost): Adjust comment. Use pattern_cost instead
6043 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
6044 * dse.c (find_shift_sequence): Ditto.
6045 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
6046 (bb_valid_for_noce_process_p): Use pattern_cost.
6047 * rtl.h (insn_rtx_cost): Delete.
6048 (pattern_cost): New prototype.
6049 (insn_cost): New prototype.
6050 * rtlanal.c (insn_rtx_cost): Rename to...
6051 (pattern_cost): ... this.
6054 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
6056 * config/i386/i386.md (*jcc_2): Remove insn pattern.
6057 (*jcc<mode>_0_r_i387): Ditto.
6058 (*jccxf_r_i387): Ditto.
6059 (*jcc<mode>_r_i387): Ditto.
6060 (*jccu<mode>_r_i387): Ditto.
6061 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
6062 (*jcc): Rename from *jcc_1.
6064 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6066 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
6067 deferred rescans after the lvx/stvx recombination pre-pass.
6069 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
6071 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
6072 memory operation instruction support.
6073 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
6074 (rs6000-ibm-aix[789]*): Likewise.
6075 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
6076 Document new functions.
6078 2017-10-09 Richard Biener <rguenther@suse.de>
6080 PR tree-optimization/82397
6081 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
6082 equality only for semantically equal trees.
6084 2017-10-09 Richard Biener <rguenther@suse.de>
6086 PR tree-optimization/82449
6087 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
6088 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
6089 allow constant addresses.
6090 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
6093 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6095 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
6098 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6101 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
6104 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6107 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
6109 2017-10-09 Jakub Jelinek <jakub@redhat.com>
6112 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
6113 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
6115 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
6117 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
6118 (WI_BINARY_PREDICATE_RESULT): Likewise.
6119 (wi::binary_traits::operator_result): New type.
6120 (wi::binary_traits::predicate_result): Likewise.
6121 (generic_wide_int::operator~, unary generic_wide_int::operator-)
6122 (generic_wide_int::operator==, generic_wide_int::operator!=)
6123 (generic_wide_int::operator&, generic_wide_int::and_not)
6124 (generic_wide_int::operator|, generic_wide_int::or_not)
6125 (generic_wide_int::operator^, generic_wide_int::operator+
6126 (binary generic_wide_int::operator-, generic_wide_int::operator*):
6128 (operator~, unary operator-, operator==, operator!=, operator&)
6129 (operator|, operator^, operator+, binary operator-, operator*): New
6131 * expr.c (get_inner_reference): Use wi::bit_and_not.
6132 * fold-const.c (fold_binary_loc): Likewise.
6133 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
6134 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
6135 (bit_value_binop): Likewise.
6136 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
6137 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
6138 (extract_range_from_binary_expr_1): Likewise.
6139 (masked_increment): Likewise.
6140 (simplify_bit_ops_using_ranges): Likewise.
6142 2017-10-09 Martin Jambor <mjambor@suse.cz>
6145 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
6146 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6147 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
6149 (hsa_fixup_mov_insn_type): New function.
6150 (hsa_op_with_type::get_in_type): Use it.
6151 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
6152 immediates in an assert.
6153 (hsa_op_with_type::extend_int_to_32bit): New method.
6154 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
6155 types. Convert to dest type if necessary.
6156 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
6157 (reg_for_gimple_ssa): Pass false as min32int to
6158 hsa_type_for_scalar_tree_type.
6159 (gen_hsa_addr): Fixup type when creating addresable temporary.
6160 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
6161 (gen_hsa_unary_operation): Extend operands and convert to dest type if
6162 necessary. Call hsa_fixup_mov_insn_type.
6163 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
6164 extend operands and convert to dest type if necessary.
6165 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
6166 to dest type if necessary.
6167 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
6168 if conversion nt necessary and size matches.
6169 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
6170 to dest type if necessary.
6171 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
6172 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
6174 (gen_hsa_clrsb): Likewise.
6175 (gen_hsa_ffs): Likewise.
6176 (gen_hsa_divmod): Extend operands and convert to dest type if
6178 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6180 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6182 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6183 Remove empty default arguments. Use a brace block as output
6185 (conditional return): Ditto.
6187 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6188 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6190 (group_ending_nop): Ditto.
6191 (doloop_end): Ditto.
6192 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6193 (splitters for those): Ditto.
6195 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6197 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6198 a conditional jump (and the compare for it) so that pc_rtx is the
6200 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6201 for the deleted and renamed ctr<mode>_internal[234] patterns.
6202 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6203 Delete second conditional return pattern.
6204 (ctr<mode>_internal2): Delete this second bdnz pattern.
6205 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6206 (ctr<mode>_internal4): Delete this second bdz pattern.
6208 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6210 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6211 (always_initialized_rtx_for_ssa_name_p): New predicate.
6212 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6213 (finish_out_of_ssa): Free new field of SA.
6214 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6215 * tree-ssa-coalesce.c: Include tree-ssa.h.
6216 (get_parm_default_def_partitions): Remove extern keyword.
6217 (get_undefined_value_partitions): New function.
6218 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6219 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6221 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6223 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6225 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6227 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6229 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6230 for targets that preffer 128bit.
6232 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6234 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6236 2017-10-08 Olivier Hainque <hainque@adacore.com>
6238 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6239 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6241 (thumb_set_return_address): Likewise.
6243 2017-10-08 Olivier Hainque <hainque@adacore.com>
6245 * common/config/arm/arm-common.c (arm_except_unwind_info):
6246 Handle DWARF2_UNWIND_INFO.
6248 2017-10-07 Michael Collison <michael.collison@arm.com>
6250 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6253 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6255 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6256 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6257 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6258 defined, force the creation of a new block for a dispatch label.
6260 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6262 * invoke.texi (Wsuggest-attribute=cold): Document.
6263 * common.opt (Wsuggest-attribute=cold): New
6264 * ipa-pure-const.c (warn_function_cold): New function.
6265 * predict.c (compute_function_frequency): Use it.
6266 * predict.h (warn_function_cold): Declare.
6268 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6270 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6273 2017-10-06 Martin Liska <mliska@suse.cz>
6275 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6276 keyword for member functions.
6277 (struct sanopt_tree_couple): New struct.
6278 (struct sanopt_tree_couple_hash): New function.
6279 (struct sanopt_ctx): Add new hash_map.
6280 (has_dominating_ubsan_ptr_check): New function.
6281 (record_ubsan_ptr_check_stmt): Likewise.
6282 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6283 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6284 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6286 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6289 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6290 aarch64_simd_valid_immediate on CONST_VECTORs.
6291 (aarch64_reg_or_bic_imm): Likewise.
6293 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6295 PR rtl-optimization/82396
6296 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6298 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6300 * graphite-dependences.c (scop_get_reads): Move code to...
6301 (scop_get_must_writes): Move code to...
6302 (scop_get_may_writes): Move code to...
6303 (scop_get_reads_and_writes): ... here.
6304 (scop_get_dependences): Call scop_get_reads_and_writes.
6306 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6308 PR tree-optimization/82434
6309 * fold-const.h (can_native_encode_type_p,
6310 can_native_encode_string_p): Remove.
6311 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6312 don't encode anything, just return what would be otherwise returned.
6313 (native_encode_fixed, native_encode_complex, native_encode_vector):
6315 (native_encode_string): Likewise. Inline by hand
6316 can_native_encode_string_p.
6317 (can_native_encode_type_p): Remove.
6318 (can_native_encode_string_p): Remove.
6319 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6320 STRING_CSTs using can_native_encode_string_p, test all
6321 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6322 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6323 argument from native_encode_expr.
6324 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6325 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6328 2017-10-06 Richard Biener <rguenther@suse.de>
6330 PR tree-optimization/82397
6331 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6332 operand_equal_p but rely on data_ref_compare_tree for detecting
6334 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6335 to match up with dr_group_sort_cmp.
6337 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6340 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6342 * config/s390/s390-builtin-types.def: Regenerate.
6344 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6347 * config/s390/s390-builtin-types.def: Regenerate.
6348 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6349 Change flag from B_VXE to B_VX.
6350 (s390_vec_min_dbl): Remove B_VXE flag.
6352 2017-10-06 Richard Biener <rguenther@suse.de>
6354 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6355 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6356 translate_isl_ast_to_gimple::is_valid_rename,
6357 translate_isl_ast_to_gimple::get_rename,
6358 translate_isl_ast_to_gimple::get_def_bb_for_const,
6359 translate_isl_ast_to_gimple::get_new_name,
6360 translate_isl_ast_to_gimple::collect_all_ssa_names,
6361 translate_isl_ast_to_gimple::copy_loop_phi_args,
6362 translate_isl_ast_to_gimple::collect_all_ssa_names,
6363 translate_isl_ast_to_gimple::copy_loop_phi_args,
6364 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6365 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6366 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6367 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6368 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6369 translate_isl_ast_to_gimple::copy_cond_phi_args,
6370 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6371 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6372 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6373 translate_isl_ast_to_gimple::rename_uses,
6374 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6375 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6376 (set_rename_for_each_def): Likewise.
6377 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6378 here. Handle rewriting SCEV analyzable uses here.
6379 (copy_bb_and_scalar_dependences): Generate code for PHI
6381 (graphite_regenerate_ast_isl): Adjust.
6382 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6383 (add_write, add_read): New functions.
6384 (build_cross_bb_scalars_def): Use it and simplify.
6385 (build_cross_bb_scalars_use): Likewise.
6386 (graphite_find_cross_bb_scalar_vars): Inline into...
6387 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6388 simulating out-of-SSA. Compute liveout and add dependencies.
6389 (build_scops): Force an empty entry block.
6390 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6392 (sese_build_liveouts): Declare.
6393 (sese_trivially_empty_bb_p): Likewise.
6394 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6395 compute liveout and debug_liveout.
6396 (sese_bad_liveouts_use): Remove.
6397 (sese_reset_debug_liveouts_bb): Likewise.
6398 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6399 (sese_build_liveouts): Build liveout and debug_liveout and store
6401 (new_sese_info): Adjust.
6402 (free_sese_info): Likewise.
6403 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6404 do not build liveout here.
6405 (move_sese_in_condition): Adjust region entry.
6406 (scev_analyzable_p): Match up with chrec_apply requirements.
6407 (sese_trivially_empty_bb_p): New.
6408 * tree-into-ssa.c (get_reaching_def): Properly support generating
6409 default-defs for incremental rewrite of anonymous names.
6411 2017-10-06 Richard Biener <rguenther@suse.de>
6413 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6414 precision do not perform modulo reduction.
6416 2017-10-06 Richard Biener <rguenther@suse.de>
6418 PR tree-optimization/82436
6419 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6420 conservatively choose the vectorization factor when checking
6421 whether we can perform the required load permutation.
6422 (vect_transform_slp_perm_load): Assert when we may not fail.
6424 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6426 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6427 message for incompatible -msdata=* and -mcall-* options.
6429 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6431 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6432 rate for post-reload scheduling.
6434 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6436 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6438 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6440 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6441 to improve monte carlo in scimark.
6443 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6445 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6446 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6447 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6448 pentium4_cost, nocona_cost): Set reassociation width to 1.
6449 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6450 width to 2 for fp operations and 1 otherwise.
6451 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
6453 (atom_cost): Set reassociation width to 2.
6454 (slm_cost, generic_cost): Set fp reassociation width
6455 to 2 and 1 otherwise.
6456 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6457 (core_cost): Set fp reassociation width to 4 and vector to 2.
6458 (ix86_reassociation_width): Rewrite using cost table; special case
6459 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6460 and TARGET_AVX128_OPTIMAL.
6461 * config/i386/i386.h (processor_costs): Add
6462 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6463 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6464 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6465 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6466 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6467 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6469 2017-10-05 Nathan Sidwell <nathan@acm.org>
6471 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
6473 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6475 * config/arm/arm.c (arm_test_fpu_data): New.
6476 (arm_run_selftests): Call arm_test_fpu_data.
6478 2017-10-04 Nathan Sidwell <nathan@acm.org>
6480 * toplev.c (toplev::main): Remove excess parens on pretty_printer
6482 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
6484 2017-10-04 Sudakshina Das <sudi.das@arm.com>
6486 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
6487 check type for aarch64_simd_valid_immediate.
6488 (aarch64_output_simd_mov_immediate): Update prototype.
6489 (aarch64_simd_valid_immediate): Update prototype.
6490 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
6491 support for ORR-immediate.
6492 (and<mode>3): modified pattern to add support for BIC-immediate.
6493 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
6494 now checks for valid immediate for BIC and ORR based on new enum
6496 (aarch64_output_simd_mov_immediate): Function now used to output
6497 BIC/ORR imm as well based on new enum argument.
6498 * config/aarch64/constraints.md (Do): New vector immediate constraint.
6500 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
6501 (aarch64_reg_or_bic_imm): Likewise.
6503 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6505 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
6506 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
6508 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6512 PR rtl-optimization/82396
6513 * haifa-sched.c (autopref_multipass_init): Simplify
6515 (autopref_rank_data): Simplify sort order.
6516 * sched-int.h (autopref_multipass_data_): Remove
6517 multi_mem_insn_p, min_offset and max_offset.
6519 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6521 * doc/sourcebuild.texi: Document vect_peeling_profitable.
6523 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6525 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
6528 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6530 * doc/sourcebuild.texi: Document vect_long_mult.
6532 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
6534 PR tree-optimization/82413
6535 * fold-const.c (build_range_check): Use widest_int when comparing
6536 the maximum ETYPE value with HIGH.
6538 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6540 PR rtl-optimization/82396
6541 * haifa-sched.c (autopref_multipass_init): Simplify
6543 (autopref_rank_data): Simplify sort order.
6544 * sched-int.h (autopref_multipass_data_): Remove
6545 multi_mem_insn_p, min_offset and max_offset.
6547 2017-10-04 Jakub Jelinek <jakub@redhat.com>
6549 PR tree-optimization/82381
6550 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
6551 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
6554 PR tree-optimization/82374
6555 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
6556 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
6557 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
6558 current_function_decl to the new decl.
6560 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6562 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
6563 helper macro for IEEE float128 hardware built-in functions.
6564 (SQRTF128_ODD): Add built-in functions with the round-to-odd
6566 (TRUNCF128_ODD): Likewise.
6567 (ADDF128_ODD): Likewise.
6568 (SUBF128_ODD): Likewise.
6569 (MULF128_ODD): Likewise.
6570 (DIVF128_ODD): Likewise.
6571 (FMAF128_ODD): Likewise.
6572 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
6573 UNSPEC_TRUNC_ROUND_TO_ODD.
6574 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
6575 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
6576 floating point round to odd instructions.
6577 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
6578 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
6579 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
6580 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
6581 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
6582 (trunc<mode>sf2_hw): Change the truncate with round to odd
6583 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
6584 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
6585 to odd hardware instructions.
6586 (sub<mode>3_odd): Likewise.
6587 (mul<mode>3_odd): Likewise.
6588 (div<mode>3_odd): Likewise.
6589 (sqrt<mode>2_odd): Likewise.
6590 (fma<mode>4_odd): Likewise.
6591 (fms<mode>4_odd): Likewise.
6592 (nfma<mode>4_odd): Likewise.
6593 (nfms<mode>4_odd): Likewise.
6594 (trunc<mode>df2_odd): Change the truncate with round to odd
6595 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
6597 * doc/extend.texi (PowerPC built-in functions): Update documentation
6598 for existing IEEE float128-bit built-in functions. Add built-in
6599 functions that generate the IEEE 128-bit floating point round to
6602 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
6604 PR rtl-optimization/77729
6605 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
6606 to (X&(C1&~C2))|C2 transformations.
6608 2017-10-03 Martin Jambor <mjambor@suse.cz>
6610 PR tree-optimization/82363
6611 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
6612 mismatch, mark lacc written regardless of racc.
6614 2017-10-03 Jakub Jelinek <jakub@redhat.com>
6616 PR tree-optimization/82381
6617 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
6618 stmt_to_insert nor wheather SSA_NAMEs are default defs.
6619 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
6620 fallthrough into reassoc_stmt_dominates_stmt_p.
6623 * combine.c (combine_instructions): Don't combine in unreachable
6626 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6629 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
6630 function to not use the have_cpu variable. Do not set cpu_index,
6631 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
6633 (rs6000_valid_attribute_p): Remove duplicate initializations of
6634 old_optimize and func_optimize.
6635 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
6636 (rs6000_activate_target_options): Make global.
6637 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
6640 2017-10-02 Jakub Jelinek <jakub@redhat.com>
6642 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
6643 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
6644 Set *poffset to 0 and *psize and *pmax_size to -1 if
6645 *poffset + *psize overflows in HOST_WIDE_INT.
6647 PR tree-optimization/82387
6648 PR tree-optimization/82388
6649 PR tree-optimization/82389
6650 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
6651 instead of live_bytes non-NULL.
6653 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
6656 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
6659 2017-10-02 Richard Biener <rguenther@suse.de>
6661 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
6662 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
6663 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
6665 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6667 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
6668 requested precision matches the type's.
6669 * calls.c (alloc_max_size): Calculate the new candidate size as
6670 a widest_int and use wi::to_widest when comparing it with the
6671 current candidate size.
6672 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
6673 zero rather than integer_zero_node.
6674 * match.pd: Check for a no-op conversion before using wi::add
6675 rather than after. Use tree_to_uhwi when summing small shift
6676 counts into an unsigned int.
6678 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6679 Alan Hayward <alan.hayward@arm.com>
6680 David Sherwood <david.sherwood@arm.com>
6683 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
6684 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
6685 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
6686 POINTER_AND_FP_REGS.
6688 2017-10-02 Richard Biener <rguenther@suse.de>
6690 PR tree-optimization/82355
6691 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6692 a mapping for the enclosing loop but avoid generating one for
6694 (copy_bb_and_scalar_dependences): Remove premature codegen
6695 error on PHIs in blocks duplicated into multiple places.
6696 * graphite-scop-detection.c
6697 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
6698 in the region use it as loop and nest to analyze the DR in.
6699 (try_generate_gimple_bb): Likewise.
6700 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
6701 (add_loop_constraints): For blocks in a loop not in the region
6702 create a dimension with a single iteration.
6703 * sese.h (gbb_loop_at_index): Remove assert.
6705 2017-10-01 Kevin Buettner <kevinb@redhat.com>
6707 * omp-expand.c (adjust_context_scope): New function.
6708 (expand_parallel_call): Call adjust_context_scope.
6710 2017-10-01 Jeff Law <law@redhat.com>
6712 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
6713 dom_opt_dom_walker class with direct access to private members.
6714 Add comments. Call test_for_singularity.
6715 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
6716 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
6717 m_dummy_cond anymore.
6718 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
6720 (pass_dominator:execute): Build the dummy_cond here and pass it
6721 to the dom_opt_dom_walker ctor.
6722 (test_for_singularity): New function.
6724 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
6725 Maya Rashish <coypu@sdf.org>
6727 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
6728 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
6729 (alpha*-*-netbsd*) Use nbsd_tm_file.
6730 (arm*-*-netbsdelf*) Likewise.
6731 (i[34567]86-*-netbsdelf*) Likewise.
6732 (x86_64-*-netbsd*) Likewise.
6733 (mips*-*-netbsd*) Likewise.
6734 (powerpc-*-netbsd*) Likewise.
6735 (sh*-*-netbsd*) Likewise.
6736 (sparc-*-netbsdelf*) Likewise.
6737 (sparc64-*-netbsd*) Likewise.
6738 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
6740 (vax-*-netbsdelf*) Likewise.
6741 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
6742 (UINT_FAST8_TYPE) Likewise.
6743 (INT_FAST16_TYPE) Check CHAR_FAST16.
6744 (UINT_FAST16_TYPE) Likewise.
6746 2017-09-30 Jakub Jelinek <jakub@redhat.com>
6749 * config/i386/i386.md
6750 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
6751 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
6752 *divmodsi4_zext_2): New define_insn_and_split.
6753 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
6754 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
6755 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
6756 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
6757 New define_insn_and_split.
6758 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
6759 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
6760 operands[1] having DImode when mode is SImode.
6762 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
6763 always SImode for DIV and MOD in REG_EQUAL notes.
6765 2017-09-29 Yury Gribov <tetra2005@gmail.com>
6768 * match.pd: Fix handling of NaNs in pattern.
6770 2017-09-29 Jeff Law <law@redhat.com>
6772 * sbitmap.c (bitmap_bit_in_range_p): New function.
6773 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
6774 * tree-ssa-dse.c (live_bytes_read): New function.
6775 (dse_classify_store): Ignore reads of dead bytes.
6777 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
6778 typos and whitespace errors.
6779 * config/i386/predicates.md (address_no_seg_operand): Likewise.
6780 * config/s390/s390.c (s390_emit_prologue): Likewise.
6782 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6785 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
6786 with a symbol for LRA.
6788 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6790 PR rtl-optimization/82338
6791 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
6793 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
6795 * genmodes.c (calc_wider_mode): Suppress qsort macro.
6796 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
6797 (qsort_chk): Declare.
6798 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
6799 (qsort_chk): New function.
6801 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6803 PR tree-optimization/82337
6804 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
6805 phi definition if the PHI result appears in an abnormal PHI.
6806 (find_basis_for_base_expr): Don't record a basis if the LHS of the
6807 basis appears in an abnormal PHI.
6809 2017-09-29 Richard Biener <rguenther@suse.de>
6811 * graphite-isl-ast-to-gimple.c
6812 (translate_isl_ast_to_gimple::set_codegen_error): New function.
6813 (binary_op_to_tree): Use it.
6814 (get_rename_from_scev): Likewise.
6815 (copy_loop_phi_nodes): Likewise.
6816 (copy_bb_and_scalar_dependences): Likewise.
6817 (translate_pending_phi_nodes): Likewise.
6819 2017-09-29 Jakub Jelinek <jakub@redhat.com>
6822 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
6823 for movabsq $(i32 << shift), r64.
6825 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
6827 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
6828 index when encoding %esp as %rsp to avoid 0x67 prefix.
6830 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
6832 * config/i386/i386.md (*movsf_internal, *movdf_internal):
6833 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
6835 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
6837 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
6838 Extensions with more than 16 double VFP registers.
6839 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
6840 to_clear_mask and all code related to it. Replace the remaining
6841 entry by a sbitmap and adapt code accordingly.
6843 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
6845 * brig-builtins.def: Change pure attributes to const.
6847 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6849 * config.gcc (default_gnu_indirect_function): Default to yes for
6850 sparc*-*-linux* with glibc.
6852 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6854 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
6855 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
6856 when creating .init_array and .fini_array sections with priority
6859 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
6862 * config/aarch64/aarch64.c
6863 (aarch64_builtin_support_vector_misalignment): Always return false
6864 when misalignment is unknown.
6866 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
6868 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
6869 this function to return false if the definition used by the swap
6870 instruction is artificial, or if the memory address from which the
6871 constant value is loaded is not represented by a base address held
6872 in a register or if the base address register is a frame or stack
6873 pointer. Additionally, return false if the base address of the
6874 loaded constant is a SYMBOL_REF but is not considered to be a
6876 (replace_swapped_load_constant): New function.
6877 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
6878 loaded constant vector with a load of a swapped constant vector.
6880 2017-09-27 Carl Love <cel@us.ibm.com>
6882 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
6883 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
6884 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
6887 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
6889 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
6890 first, always call autopref_rank_data otherwise.
6892 2017-09-27 Richard Biener <rguenther@suse.de>
6894 * graphite-scop-detection.c (find_scop_parameters): Move
6895 loop bound handling ...
6896 (gather_bbs::before_dom_children): ... here, avoiding the need
6897 to build scop_info->loop_nest.
6898 (record_loop_in_sese): Remove.
6899 * sese.h (sese_info_t::loop_nest): Remove.
6900 * sese.c (new_sese_info): Do not allocate loop_nest.
6901 (free_sese_info): Do not free loop_nest.
6903 2017-09-27 Jakub Jelinek <jakub@redhat.com>
6906 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
6907 lhs from calls if the lhs has addressable type.
6909 2017-09-27 Richard Biener <rguenther@suse.de>
6911 * graphite.h (scop::max_alias_set): New member.
6912 * graphite-scop-detection.c: Remove references to non-existing
6913 --param in comments.
6914 (build_alias_sets): Record the maximum alias set used for drs.
6915 (build_scops): Support zero as unlimited for
6916 --param graphite-max-arrays-per-scop.
6917 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
6919 (build_poly_sr_1): ... here. Compute alias set based on the
6920 maximum alias set used for drs rather than
6921 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
6923 2017-09-27 Richard Biener <rguenther@suse.de>
6925 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
6926 --param loop-block-tile-size=0 to disable tiling.
6928 2017-09-27 Richard Biener <rguenther@suse.de>
6930 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
6931 (graphite-max-nb-scop-params): Document special value zero.
6932 * domwalk.h (dom_walker::STOP): New symbolical constant.
6933 (dom_walker::dom_walker): Add optional parameter for bb to
6935 (dom_walker::~dom_walker): Declare.
6936 (dom_walker::before_dom_children): Document STOP return value.
6937 (dom_walker::m_user_bb_to_rpo): New member.
6938 (dom_walker::m_bb_to_rpo): Likewise.
6939 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
6940 mapping here if not provided by the user.
6941 (dom_walker::~dom_walker): Free bb to RPO mapping if not
6942 provided by the user.
6943 (dom_walker::STOP): Define.
6944 (dom_walker::walk): Do not compute bb to RPO mapping here.
6945 Support STOP return value from before_dom_children to stop
6947 * graphite-optimize-isl.c (optimize_isl): If the schedule
6948 is the same still generate code if -fgraphite-identity
6949 or -floop-parallelize-all are given.
6950 * graphite-scop-detection.c: Include cfganal.h.
6951 (gather_bbs::gather_bbs): Get and pass through bb to RPO
6953 (gather_bbs::before_dom_children): Return STOP for BBs
6955 (build_scops): Compute bb to RPO mapping and pass it to
6956 the domwalk. Treat --param graphite-max-nb-scop-params=0
6957 as not limiting the number of params.
6958 * graphite.c (graphite_initialize): Remove limit on the
6959 number of basic-blocks in a function.
6960 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
6961 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
6962 default value of 10.
6964 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6966 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
6967 Adjust code to eliminate needing to do the shift right 32-bits
6968 operation after XSCVDPSPN.
6970 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6972 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
6973 ((X / Y) != 0 -> X >= Y): Likewise.
6975 2017-09-26 Carl Love <cel@us.ibm.com>
6977 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
6978 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
6979 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
6980 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
6981 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
6982 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
6983 definitions and overloading.
6984 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
6985 statement for P9V_BUILTIN_XST_LEN_R.
6986 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
6987 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
6988 define_expand and define_insn for the instructions and builtins.
6989 * doc/extend.texi: Update the built-in documentation file for the new
6991 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
6992 define_insn for the instructions
6994 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
6997 * gcc/config/netbsd-protos.h: New file.
6998 * gcc/config/netbsd.c: New file.
6999 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
7000 * gcc/config/t-netbsd: New file.
7001 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
7002 (tmake_file) Add t-netbsd.
7003 (extra_objs) Add netbsd.o.
7005 2017-09-26 Janus Weil <janus@gcc.gnu.org>
7009 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
7011 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7013 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
7014 sign extension from a vector register to a GPR by doing a 32-bit
7015 direct move and then an EXTSW.
7016 (extendsi<mode>2 splitter): Likewise.
7017 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
7018 right or vector extract after doing XSCVDPSPN. Use
7019 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
7021 (movdi_from_sf_zero_ext): Likewise.
7022 (reload_gpr_from_vsxsf): Likewise.
7023 (p8_mfvsrd_4_disf): Delete, no longer used.
7024 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
7025 then needing to move the SFmode to a GPR to use the XSCVDPSP
7026 instruction instead of FRSP and XSCVDPSPN.
7027 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
7028 it is adjacent to the other XSCVSPDP insns.
7029 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
7030 SFmode to be in traditional Altivec registers.
7031 (vsx_xscvdpspn): Eliminate useless alternative constraint.
7032 (vsx_xscvspdpn): Likewise.
7033 (vsx_xscvspdpn_scalar): Likewise.
7035 2017-09-26 Martin Jambor <mjambor@suse.cz>
7037 * tree-sra.c (compare_access_positions): Put integral types first,
7038 stabilize sorting of integral types, remove conditions putting
7039 non-full-precision integers last.
7040 (sort_and_splice_var_accesses): Disable scalarization if a
7041 non-integert would be represented by a non-full-precision integer.
7043 2017-09-26 Joseph Myers <joseph@codesourcery.com>
7045 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
7046 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
7047 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
7048 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
7049 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
7050 conditionals inside the function instead of around it. Call
7051 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
7052 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
7054 2017-09-26 Richard Biener <rguenther@suse.de>
7056 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
7058 (scop_detection::build_scop_breadth): ... this. Removed.
7059 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
7060 (scop_detection::harmful_stmt_in_bb): Likewise.
7061 (scop_detection::graphite_can_represent_stmt): Likewise.
7062 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
7063 (scop_detection::can_represent_loop): Remove recursion, fold in ...
7064 (scop_detection::can_represent_loop_1): ... this. Removed.
7065 (scop_detection::harmful_loop_in_region): Simplify after inlining
7066 the above and remove more quadraticness.
7067 (build_scops): Adjust.
7068 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
7071 2017-09-26 Jakub Jelinek <jakub@redhat.com>
7074 * config/i386/i386.c (ix86_print_operand_address_as): Only test
7075 REGNO (base) == SP_REG if base is a REG.
7078 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
7079 if it is different SSA_NAME.
7080 (optimize_range_tests_cmp_bitwise): New function.
7081 (optimize_range_tests): Call it.
7083 2017-09-26 Richard Biener <rguenther@suse.de>
7085 PR tree-optimization/82321
7086 * graphite.c (canonicalize_loop_closed_ssa): Properly check
7087 for the def being inside the loop.
7089 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7091 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
7093 * config/s390/s390-builtins.def: Fix constraint on op4.
7095 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7097 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
7098 independent expanders.
7099 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
7100 ("vec_ordered", "vec_unordered"): New expanders.
7102 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7104 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
7107 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7109 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
7110 vec_unpacks_lo_v16qi.
7111 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
7113 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7115 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
7116 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
7117 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
7119 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7121 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
7123 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
7125 ("*vec_slb<mode>"): New insn pattern.
7126 ("vec_shr_<mode>"): New expander.
7127 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
7128 and force the shift count operand to V16QImode.
7129 ("vec_srb<mode>"): Set shift count mode to V16QI.
7131 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7133 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
7134 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
7135 ("vec_widen_smult_hi_<mode>"): New expander definitions.
7137 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
7140 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
7142 2017-09-26 Richard Biener <rguenther@suse.de>
7144 PR tree-optimization/82320
7145 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
7148 2017-09-25 Jeff Law <law@redhat.com>
7150 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
7151 prototype for new argument.
7152 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
7153 mostly extracted from rs6000_emit_allocate_stack.
7154 (rs6000_emit_probe_stack_range_stack_clash): New function.
7155 (rs6000_emit_allocate_stack): Call
7156 rs6000_emit_probe_stack_range_stack_clash as needed.
7157 (rs6000_emit_probe_stack_range): Add additional argument
7158 to call to gen_probe_stack_range{si,di}.
7159 (output_probe_stack_range): New.
7160 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
7161 (output_probe_stack_range_stack_clash): New.
7162 (rs6000_emit_prologue): Emit notes into dump file as requested.
7163 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
7164 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
7165 Add additional operand and pass it to output_probe_stack_range.
7167 2017-09-25 Bin Cheng <bin.cheng@arm.com>
7169 PR tree-optimization/82163
7170 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
7171 (checking_verify_loop_closed_ssa): New parameter.
7172 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
7173 (check_loop_closed_ssa_stmt): Delete.
7174 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
7175 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
7176 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
7179 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7181 * brig-builtins.def: Treat HSAIL barrier builtins as
7182 setjmp/longjump style functions.
7184 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7186 * target.def (constant_alignment): New hook.
7187 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7188 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7189 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7190 * doc/tm.texi: Regenerate.
7191 * targhooks.h (default_constant_alignment): Declare.
7192 (constant_alignment_word_strings): Likewise.
7193 * targhooks.c (default_constant_alignment): New function.
7194 (constant_alignment_word_strings): Likewise.
7195 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7196 instead of CONSTANT_ALIGNMENT.
7197 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7198 (force_const_mem): Likewise.
7199 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7200 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7201 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7202 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7203 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7205 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7206 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7207 constant_alignment_word_strings.
7208 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7209 (CONSTANT_ALIGNMENT): Likewise.
7210 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7211 (arm_constant_alignment): New function.
7212 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7213 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7214 constant_alignment_word_strings.
7215 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7216 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7217 constant_alignment_word_strings.
7218 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7219 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7220 (cris_constant_alignment): New function.
7221 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7222 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7223 (epiphany_constant_alignment): New function.
7224 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7225 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7226 constant_alignment_word_strings.
7227 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7228 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7229 constant_alignment_word_strings.
7230 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7231 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7232 constant_alignment_word_strings.
7233 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7234 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7235 * config/i386/i386.c (ix86_constant_alignment): Make static.
7236 Use the same interface as the target hook.
7237 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7238 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7239 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7240 constant_alignment_word_strings.
7241 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7242 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7243 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7244 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7245 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7246 constant_alignment_word_strings.
7247 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7248 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7249 constant_alignment_word_strings.
7250 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7251 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7252 constant_alignment_word_strings.
7253 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7254 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7256 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7257 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7258 * config/mips/mips.c (mips_constant_alignment): New function.
7259 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7260 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7261 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7262 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7263 (mmix_constant_alignment): Make static. Use the same interface
7265 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7266 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7267 constant_alignment_word_strings.
7268 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7269 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7270 constant_alignment_word_strings.
7271 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7272 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7273 constant_alignment_word_strings.
7274 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7275 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7276 (rs6000_constant_alignment): New function.
7277 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7278 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7279 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7280 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7281 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7282 (rs6000_constant_alignment): New function.
7283 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7284 * config/s390/s390.c (s390_constant_alignment): New function.
7285 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7286 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7287 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7288 constant_alignment_word_strings.
7289 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7290 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7291 (sparc_constant_alignment): New function.
7292 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7293 * config/spu/spu.c (spu_constant_alignment): New function.
7294 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7295 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7296 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7297 constant_alignment_word_strings.
7298 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7299 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7300 constant_alignment_word_strings.
7301 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7302 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7303 constant_alignment_word_strings.
7304 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7305 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7306 (visium_constant_alignment): New function.
7307 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7308 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7309 (xtensa_constant_alignment): New function.
7310 * system.h (CONSTANT_ALIGNMENT): Poison.
7312 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7314 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7315 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7316 (rs6000_builtin_valid_without_lhs): New helper function.
7317 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7318 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7320 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7322 * target.h (vec_perm_indices): Use unsigned short rather than
7324 (auto_vec_perm_indices): Likewise.
7325 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7326 Use unsigned int rather than unsigned char.
7327 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7329 2017-09-25 Richard Biener <rguenther@suse.de>
7331 * cfgloop.h (sort_sibling_loops): Declare.
7332 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7333 (sort_sibling_loops): New function sorting the sibling loop list
7335 * graphite.c (graphite_transform_loops): Sort sibling loops.
7337 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7339 * target.def (vec_perm_const_ok): Change sel parameter to
7341 * optabs-query.c (can_vec_perm_p): Update accordingly.
7342 * doc/tm.texi: Regenerate.
7343 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7344 auto_vec_perm_indices and remove separate nelt field.
7345 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7346 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7347 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7348 (aarch64_expand_vec_perm_const): Update accordingly.
7349 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7350 to vec_perm_indices.
7351 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7352 auto_vec_perm_indices and remove separate nelt field.
7353 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7354 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7355 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7357 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7358 to vec_perm_indices.
7359 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7360 sel to vec_perm_indices.
7361 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7362 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7363 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7365 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7368 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7371 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7372 on the FUNCTION_DECL function context if it has a DIE that is a
7375 2017-09-25 Richard Biener <rguenther@suse.de>
7377 PR tree-optimization/82285
7378 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7381 2017-09-25 Tom de Vries <tom@codesourcery.com>
7385 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7388 2017-09-25 Richard Biener <rguenther@suse.de>
7390 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7391 ISL errors other than isl_error_quota happen. Dump if the
7392 schedule is the same.
7393 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7394 errors instead of aborting inside ISL.
7396 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7399 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7400 of libgcc_eh for m64.
7401 * config/i386/darwin64.h: Likewise.
7403 2017-09-25 Richard Biener <rguenther@suse.de>
7406 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7407 attribute for incomplete types nor twice for complete ones.
7409 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7412 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7413 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7416 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7419 * config/i386/i386.opt (mprefer-avx256): Use
7420 ix86_target_flags variable.
7421 * config/i386/i386.c (ix86_target_string): Move
7422 -mprefer-avx256 to flag2_opts.
7424 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7427 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7428 and x != -1 | y != -1 into (x & y) != -1.
7430 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7432 * config.gcc: Add new case statement to set
7433 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7434 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7435 s390x-*-linux* case statements. Added aarch64 to the list of
7436 supported architectures.
7438 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7440 PR tree-optimization/82289
7441 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7442 STMT_VINFO_RELEVANT_P.
7444 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7445 Alan Hayward <alan.hayward@arm.com>
7446 David Sherwood <david.sherwood@arm.com>
7448 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7449 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7450 (extract_range_from_binary_expr_1): Don't call
7451 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
7453 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7454 Alan Hayward <alan.hayward@arm.com>
7455 David Sherwood <david.sherwood@arm.com>
7457 * target.def (preferred_vector_alignment): New hook.
7458 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7460 * doc/tm.texi: Regenerate.
7461 * targhooks.h (default_preferred_vector_alignment): Declare.
7462 * targhooks.c (default_preferred_vector_alignment): New function.
7463 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7465 (DR_TARGET_ALIGNMENT): New macro.
7466 (aligned_access_p): Update commentary.
7467 (vect_known_alignment_in_bytes): New function.
7468 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7470 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
7471 Calculate the misalignment based on the target alignment rather than
7473 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
7474 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
7475 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
7476 the target alignment, rather than masking the element misalignment
7477 with the number of elements in a vector. Also use the target
7478 alignment when calculating the maximum number of peels.
7479 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
7480 instead of TYPE_ALIGN_UNIT.
7481 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
7482 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
7483 (vect_create_addr_base_for_vector_ref): Update call accordingly.
7484 (vect_create_data_ref_ptr): Likewise.
7485 (vect_setup_realignment): Realign by ANDing with
7486 -DR_TARGET_MISALIGNMENT.
7487 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
7488 the number of peels based on DR_TARGET_ALIGNMENT.
7489 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
7490 with the guaranteed alignment boundary when deciding whether
7492 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
7493 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
7494 (ensure_base_align): Remove stmt_info parameter. Get the
7495 target base alignment from DR_TARGET_ALIGNMENT.
7496 (vectorizable_store): Update call accordingly. Interpret
7497 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
7499 (vectorizable_load): Likewise.
7501 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7502 Alan Hayward <alan.hayward@arm.com>
7503 David Sherwood <david.sherwood@arm.com>
7505 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
7506 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
7507 (vect_enhance_data_refs_alignment): Likewise.
7509 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7511 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7512 error. Only quit immediately if parsing is complete.
7513 (BEGIN): Initialize fatal_err and parse_done.
7514 (begin fpu, end fpu): Check number of arguments.
7515 (begin arch, end arch): Likewise.
7516 (begin cpu, end cpu): Likewise.
7517 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7518 (optalias): Likewise.
7520 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7522 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7523 * config/arm/arm-isa.h: Delete. Move definitions to ...
7524 * arm-cpus.in: ... here. Use new feature and fgroup values.
7525 * config/arm/arm.c (arm_option_override): Use lower case for feature
7527 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7528 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7529 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7531 (print_isa_bits_for): New function.
7532 (gen_isa): New function.
7533 (gen_comm_data): Use print_isa_bits_for.
7534 (define feature): New keyword.
7535 (define fgroup): New keyword.
7536 * config/arm/t-arm (TM_H): Remove.
7537 (GTM_H): Add arm-isa.h.
7538 (arm-isa.h): Add rule to generate file.
7539 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7540 case for feature bit names.
7542 2017-09-22 Richard Biener <rguenther@suse.de>
7544 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
7546 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
7547 print of no dependency loops ...
7548 * graphite.c (graphite_transform_loops): ... here.
7549 (canonicalize_loop_closed_ssa_form): Work from inner to outer
7551 (same_close_phi_node, remove_duplicate_close_phi,
7552 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
7553 (canonicalize_loop_closed_ssa): ... here and simplify.
7554 * graphite-optimize-isl.c: Include tree-vectorizer.h.
7555 (optimize_isl): Use dump_printf_loc to tell when we stopped
7556 optimizing because of an ISL timeout.
7558 2017-09-22 Richard Biener <rguenther@suse.de>
7560 PR tree-optimization/82291
7561 * tree-if-conv.c (predicate_mem_writes): Make sure to
7562 remove writes in blocks predicated with false.
7564 2017-09-22 Richard Biener <rguenther@suse.de>
7566 * sese.c: Include cfganal.h.
7567 (if_region_set_false_region): Remove.
7568 (create_if_region_on_edge): Likewise.
7569 (move_sese_in_condition): Re-implement without destroying
7572 2017-09-22 Richard Biener <rguenther@suse.de>
7574 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
7575 Verify both BBs contain loop PHI nodes before dispatching to
7577 (graphite_regenerate_ast_isl): Do not recompute dominators,
7578 do not verify three times. Restructure for clarity.
7579 * graphite-scop-detection.c (same_close_phi_node,
7580 remove_duplicate_close_phi, make_close_phi_nodes_unique,
7581 defined_in_loop_p, canonicalize_loop_closed_ssa,
7582 canonicalize_loop_closed_ssa_form): Simplify, remove excess
7583 checking and SSA rewrite, move to ...
7584 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
7585 (graphite_initialize): Do not pass in ctx, do not reset the
7586 SCEV cache, compute only dominators.
7587 (graphite_transform_loops): Allocate ISL ctx after
7588 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
7589 Maintain post-dominators only around build_scops.
7590 * sese.c (if_region_set_false_region): Make static. Free
7591 and recompute dominators.
7592 (move_sese_in_condition): Assert we don't get called with
7593 post-dominators computed.
7594 * sese.h (if_region_set_false_region): Remove.
7596 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
7598 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
7599 mode attribute for TARGET_AVX512VL.
7601 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
7603 * config/i386/i386.opt (mprefer-avx256): New option.
7604 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
7606 (ix86_preferred_simd_mode): Return 256-bit AVX modes
7607 for TARGET_PREFER_AVX256.
7608 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
7610 2017-09-21 Jeff Law <law@redhat.com>
7612 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
7613 Fix dump output if the only stack space is for pushed registers.
7615 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7617 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
7620 2017-09-21 Martin Sebor <msebor@redhat.com>
7623 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
7624 code (in C++) or code that triggers warnings.
7626 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
7628 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
7630 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
7632 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
7633 * sched-rgn.c: Ditto.
7634 * sel-sched-ir.c: Ditto.
7636 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
7638 * toplev.h (set_random_seed): Adjust return type.
7639 * toplev.c (init_local_tick): Move eager initialization of random_seed
7640 to get_random_seed. Adjust comment.
7641 (init_random_seed): Inline to get_random_seed, delete.
7642 (get_random_seed): Initialize random_seed lazily.
7643 (set_random_seed): Do not return previous value.
7644 (print_switch_value): Do not call get_random_seed.
7646 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
7648 * cgraph.c (delete_function_version): New, broken out from...
7649 (cgraph_node::delete_function_version): ...here. Rename to
7650 cgraph_node::delete_function_version_by_decl. Update all uses.
7651 (cgraph_node::remove): Call delete_function_version.
7653 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7656 * tree-inline.c (expand_call_inline): Emit clobber stmts for
7657 VAR_DECLs to which addressable non-volatile parameters are mapped
7658 and for id->retvar after the return value assignment. Clear
7659 id->retval and id->retbnd after inlining.
7661 2017-09-21 Richard Biener <rguenther@suse.de>
7663 PR tree-optimization/82276
7664 PR tree-optimization/82244
7665 * tree-vrp.c (build_assert_expr_for): Set
7666 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
7668 (remove_range_assertions): Revert earlier change.
7670 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
7673 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
7675 2017-09-21 Richard Biener <rguenther@suse.de>
7677 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7678 Restore valid IL after code generation errors.
7679 * graphite.c (graphite_transform_loops): Diagnose code
7680 generation issues as MSG_MISSED_OPTIMIZATION and continue
7681 with processing SCOPs.
7683 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7684 Alan Hayward <alan.hayward@arm.com>
7685 David Sherwood <david.sherwood@arm.com>
7687 * calls.c (compute_argument_addresses): Use simplify_gen_binary
7688 rather than choosing between plus_constant and gen_rtx_<CODE>.
7689 * expr.c (emit_push_insn): Likewise.
7690 (expand_expr_real_2): Likewise.
7692 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7693 Alan Hayward <alan.hayward@arm.com>
7694 David Sherwood <david.sherwood@arm.com>
7696 * loop-unroll.c (split_iv): Call copy_rtx on the step.
7698 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7699 Alan Hayward <alan.hayward@arm.com>
7700 David Sherwood <david.sherwood@arm.com>
7702 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
7703 calling tree_to_uhwi.
7705 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7706 Alan Hayward <alan.hayward@arm.com>
7707 David Sherwood <david.sherwood@arm.com>
7709 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
7710 INTEGER_CST rather than a negative test for ADDR_EXPR.
7712 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7713 Alan Hayward <alan.hayward@arm.com>
7714 David Sherwood <david.sherwood@arm.com>
7716 * tree-vrp.c (extract_range_from_binary_expr_1): Check
7717 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
7719 2017-09-21 Richard Biener <rguenther@suse.de>
7721 PR tree-optimization/71351
7722 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
7723 graphite_create_new_loop_guard): Remove, fold remaining parts
7725 (translate_isl_ast_node_for): ... here and simplify.
7727 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7730 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
7731 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
7732 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
7733 alternative always use QI mode, for -Os imov (=R,R) alternative
7734 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
7737 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7738 Jeff Law <law@redhat.com>
7740 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
7741 (allocate_stack_space): New function, partially extracted from
7743 (s390_emit_prologue): Track offset to most recent stack probe.
7744 Code to allocate space moved into allocate_stack_space.
7745 Dump actions when no stack is allocated.
7746 (s390_prologue_plus_offset): New function.
7747 (s390_emit_stack_probe): Likewise.
7749 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
7751 * common.opt (Wa, Wl, Wp, g, gz=): Add
7753 (gno-column-info): Remove.
7754 (gcolumn-info): Drop RejectNegative.
7756 (gno-record-gcc-switches): Remove.
7757 (grecord-gcc-switches): Drop RejectNegative.
7758 (gno-split-dwarf): Remove.
7759 (gsplit-dwarf): Drop RejectNegative.
7760 (gno-strict-dwarf): Remove.
7761 (gstrict-dwarf): Drop RejectNegative.
7762 * config/darwin.opt (gfull, gused): Add RejectNegative.
7763 * dwarf2out.c (gen_producer_string): Drop
7764 gno-record-gcc-switches handler.
7765 * optc-gen.awk: Add g to prefixes with negative forms.
7766 * opts-common.c (remapping_prefix_p): New.
7767 (find_opt): Check it.
7768 (generate_canonical_option): Test g prefix.
7769 (option_map): Add -gno- mapping.
7770 (add_misspelling_candidates): Check remapping_prefix_p.
7772 2017-09-20 Jeff Law <law@redhat.com>
7774 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
7775 thinko in stack clash protection support.
7777 * explow.c (compute_stack_clash_protection_loop_data): Use
7778 CONST_INT_P instead of explicit test. Verify object is a
7779 CONST_INT_P before looking at INTVAL.
7780 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
7781 instead of explicit test.
7783 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
7786 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
7787 address instead of to r1 and r11.
7789 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
7791 * config.gcc: Support "knm".
7792 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
7793 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7795 * config/i386/i386.c (m_KNM): Define.
7796 (processor_target_table): Add "knm".
7798 (ix86_option_override_internal): Add "knm".
7799 (ix86_issue_rate): Add PROCESSOR_KNM.
7800 (ix86_adjust_cost): Ditto.
7801 (ia32_multipass_dfa_lookahead): Ditto.
7802 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
7803 (fold_builtin_cpu): Add M_INTEL_KNM.
7804 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
7805 (processor_type): Add PROCESSOR_KNM.
7806 * config/i386/x86-tune.def: Add m_KNM.
7807 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
7809 2017-09-20 Richard Biener <rguenther@suse.de>
7811 PR tree-optimization/80213
7812 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
7813 are allowed in empty BBs as well.
7814 (canonicalize_loop_closed_ssa): Also look for other complex
7816 (scop_detection::get_sese): Include the loop-closed PHI block
7818 (scop_detection::merge_sese): Remove code adding extra blocks.
7819 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
7820 (build_scops): Assert the final returned scop is invalid.
7822 2017-09-20 Richard Biener <rguenther@suse.de>
7824 PR tree-optimization/82264
7825 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
7827 (vn_phi_lookup): Likewise.
7828 (vn_phi_insert): Likewise.
7830 2017-09-20 Jakub Jelinek <jakub@redhat.com>
7832 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
7833 that fits into uhwi or shwi, add DW_AT_const_value regardless
7834 of early_dwarf without going through RTL, using add_AT_unsigned
7837 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
7838 (DEBUG_LTO_ABBREV_SECTION): Likewise.
7839 (DEBUG_LTO_MACINFO_SECTION): Likewise.
7840 (DEBUG_MACRO_SECTION): Likewise.
7841 (DEBUG_LTO_MACRO_SECTION): Likewise.
7842 (DEBUG_STR_DWO_SECTION): Likewise.
7843 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
7844 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
7845 (DEBUG_LTO_DWO_LINE_SECTION): Define.
7846 (DEBUG_LTO_LINE_STR_SECTION): Define.
7847 (init_sections_and_labels): Initialize debug_line_str_section
7848 variable. Initialize debug_loc_section for -gdwarf-5 to
7849 DEBUG_LOCLISTS_SECTION. Formatting fixes.
7851 2017-09-20 Richard Biener <rguenther@suse.de>
7853 * graphite-sese-to-poly.c (extract_affine): Properly handle
7854 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
7856 2017-09-20 Richard Biener <rguenther@suse.de>
7858 PR tree-optimization/81373
7859 * graphite-scop-detection.c (build_cross_bb_scalars_def):
7860 Force SESE live-out defs to be handled even if they are
7863 2017-09-19 Jeff Law <law@redhat.com>
7865 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
7866 nothing for stack adjustments with REG_STACK_CHECK.
7867 * sched-deps.c (parse_add_or_inc): Reject insns with
7868 REG_STACK_CHECK from dependency breaking.
7869 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
7870 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
7871 * reg-notes.def (STACK_CHECK): New note.
7873 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
7874 (ix86_expand_prologue): Dump stack clash info as needed.
7875 Call ix86_adjust_stack_and_probe_stack_clash as needed.
7877 * function.c (dump_stack_clash_frame_info): New function.
7878 * function.h (dump_stack_clash_frame_info): Prototype.
7879 (enum stack_clash_probes): New enum.
7881 * config/alpha/alpha.c (alpha_expand_prologue): Also check
7882 flag_stack_clash_protection.
7883 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
7884 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
7885 (arm_frame_pointer_required): Likewise.
7886 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
7887 (ia64_expand_prologue): Likewise.
7888 * config/mips/mips.c (mips_expand_prologue): Likewise.
7889 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
7890 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7891 (sparc_flat_expand_prologue): Likewise.
7892 * config/spu/spu.c (spu_expand_prologue): Likewise.
7894 * explow.c: Include "params.h".
7895 (anti_adjust_stack_and_probe_stack_clash): New function.
7896 (get_stack_check_protect): Likewise.
7897 (compute_stack_clash_protection_loop_data): Likewise.
7898 (emit_stack_clash_protection_loop_start): Likewise.
7899 (emit_stack_clash_protection_loop_end): Likewise.
7900 (allocate_dynamic_stack_space): Use get_stack_check_protect.
7901 Use anti_adjust_stack_and_probe_stack_clash.
7902 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
7903 (emit_stack_clash_protection_loop_start): Likewise.
7904 (emit_stack_clash_protection_loop_end): Likewise.
7905 * rtl.h (get_stack_check_protect): Prototype.
7906 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
7907 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
7908 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
7910 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
7912 * doc/tm.texi: Rebuilt.
7913 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
7914 get_stack_check_protect.
7915 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
7916 * config/arm/arm.c (arm_expand_prologue): Likewise.
7917 (arm_frame_pointer_required): Likewise.
7918 * config/i386/i386.c (ix86_expand_prologue): Likewise.
7919 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
7920 * config/mips/mips.c (mips_expand_prologue): Likewise.
7921 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
7922 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
7923 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7924 (sparc_flat_expand_prologue): Likewise.
7926 * common.opt (-fstack-clash-protection): New option.
7927 * flag-types.h (enum stack_check_type): Note difference between
7928 -fstack-check= and -fstack-clash-protection.
7929 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
7930 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
7931 * toplev.c (process_options): Issue warnings/errors for cases
7932 not handled with -fstack-clash-protection.
7933 * doc/invoke.texi (-fstack-clash-protection): Document new option.
7934 (-fstack-check): Note additional problem with -fstack-check=generic.
7935 Note that -fstack-check is primarily for Ada and refer users
7936 to -fstack-clash-protection for stack-clash-protection.
7937 Document new params for stack clash protection.
7939 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
7941 * config/i386/i386.c (ix86_split_long_move): Do not handle
7942 address used for LEA in a special way.
7944 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
7946 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
7948 2017-09-19 Martin Sebor <msebor@redhat.com>
7951 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
7952 of incompatible types.
7954 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
7956 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7957 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
7958 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7959 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
7961 2017-09-19 Richard Biener <rguenther@suse.de>
7963 PR tree-optimization/82244
7964 * tree-vrp.c (remove_range_assertions): Do not propagate
7965 a constant to abnormals but replace the assert with a copy.
7967 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
7969 PR rtl-optimization/57878
7970 PR rtl-optimization/68988
7971 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
7972 avoidance test involving non_reload_pseudos. Move frequency test
7973 below the general fragmentation avoidance test.
7975 2017-09-19 Richard Biener <rguenther@suse.de>
7977 PR tree-optimization/69728
7978 * graphite-sese-to-poly.c (schedule_error): New global.
7979 (add_loop_schedule): Handle empty domain by failing the
7981 (build_original_schedule): Handle schedule_error.
7983 2017-09-19 Richard Biener <rguenther@suse.de>
7985 * graphite-scop-detection.c (scop_detection::can_represent_loop):
7986 Do not iterate to sibling loops but only to siblings of inner
7989 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
7992 * config/m68k/m68k.md (moveq feeding equality comparison): Check
7993 that the registers are different.
7995 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
7997 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
7998 to processor_model and "amdfam17h" to arch_names_table.
7999 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
8001 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8004 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
8006 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8007 Alan Hayward <alan.hayward@arm.com>
8008 David Sherwood <david.sherwood@arm.com>
8010 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
8012 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
8014 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
8015 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
8016 vect_schedule_slp_instance.
8017 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
8018 Update call to vect_slp_analyze_node_operations. Simplify return
8020 (vect_slp_analyze_bb_1): Update call accordingly.
8021 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
8022 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
8023 (vect_schedule_slp): Update call accordingly.
8025 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8026 Alan Hayward <alan.hayward@arm.com>
8027 David Sherwood <david.sherwood@arm.com>
8029 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
8030 with types that aren't in fact scalar.
8032 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8034 * tree-vect-slp.c (vect_record_max_nunits): New function,
8036 (vect_build_slp_tree_1): ...here.
8037 (vect_build_slp_tree_2): Call it for phis too.
8039 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8041 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
8042 to vect_get_vec_def_for_operand when getting the mask operand.
8044 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8045 Alan Hayward <alan.hayward@arm.com>
8046 David Sherwood <david.sherwood@arm.com>
8048 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
8051 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8052 Alan Hayward <alan.hayward@arm.com>
8053 David Sherwood <david.sherwood@arm.com>
8055 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
8056 calculation for vector booleans.
8058 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8059 Alan Hayward <alan.hayward@arm.com>
8060 David Sherwood <david.sherwood@arm.com>
8062 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
8064 (vect_transform_stmt): ...here.
8065 (vect_analyze_stmt): Use it instead of calling
8066 vectorizable_live_operation directly.
8068 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
8070 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
8071 non-SIMT targets in acc vector loops.
8073 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
8075 * configure.ac: Add arc and check if assembler supports gdwarf2.
8076 * configure: Regenerate.
8078 2017-09-18 Richard Biener <rguenther@suse.de>
8080 PR tree-optimization/82220
8081 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
8082 epilogue niters from the min_profitable_iters compute.
8084 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8087 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
8088 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
8089 (ix86_init_pic_reg): Revert 2017-09-01 changes.
8091 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
8094 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
8095 switching to a new text section.
8097 2017-09-18 Richard Biener <rguenther@suse.de>
8099 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
8101 (build_alias_set): Reject aliases with no access function.
8103 2017-09-18 Richard Biener <rguenther@suse.de>
8105 PR tree-optimization/79622
8106 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
8108 (build_cross_bb_scalars_use): Likewise.
8110 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
8112 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
8114 2017-09-18 Alan Modra <amodra@gmail.com>
8117 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
8118 stack_pointer_rtx for count 0. Update comments. Break up
8119 large rtl expression.
8121 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
8123 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
8124 Increase to 20 bytes.
8125 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
8126 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
8127 or avx version of the stub.
8129 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
8132 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
8133 compute the minimum stack alignment. Also update preferred stack
8134 boundary for leaf functions.
8136 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
8138 PR tree-optimization/82228
8139 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
8142 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8144 * common/config/nds32/nds32-common.c
8145 (nds32_option_optimization_table): Refine formatting.
8146 (nds32_option_optimization_table): Use -fsched-pressure and
8147 -fomit-frame-pointer for specific optimization level.
8149 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8151 * config/nds32/nds32.c: Refine formatting and comments.
8152 * config/nds32/nds32.h: Likewise.
8153 * config/nds32/nds32.md: Likewise.
8154 * config/nds32/nds32-cost.c: Likewise.
8155 * config/nds32/nds32-isr.c: Likewise.
8156 * config/nds32/nds32-md-auxiliary.c: Likewise.
8157 * config/nds32/nds32-multiple.md: Likewise.
8158 * config/nds32/nds32-predicates.c: Likewise.
8160 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
8161 Jakub Jelinek <jakub@redhat.com>
8163 Add support for -std=c++2a.
8164 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
8166 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
8168 2017-09-15 Steve Ellcey <sellcey@cavium.com>
8171 * doc/extend.texi (Common Function Attributes): Add
8172 references to ARM, AArch64, and S/390 specific attributes.
8173 (Function Specific Option Pragmas): Add AArch64 and S/390
8174 to list of back ends that support the target pragma.
8176 2017-09-15 Nathan Sidwell <nathan@acm.org>
8178 * doc/standards.texi: Fix C++17 description. Update URLs for
8181 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8183 * common.opt (Wcast-align=strict): New warning option.
8184 * doc/invoke.texi: Document -Wcast-align=strict.
8186 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8188 * cgraph.h (cgraph_thunk_info): Add comments.
8189 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8190 assert for VIRTUAL_* arguments stricter.
8192 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8194 PR tree-optimization/71026
8195 * match.pd: Move RDIV patterns from fold-const.c
8196 * fold-const.c (distribute_real_division): Removed.
8197 (fold_binary_loc): Remove calls to distribute_real_divison.
8199 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8201 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8202 c++1z and gnu++1z as deprecated. Change other references to
8203 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8204 Change -Wc++1z-compat to -Wc++17-compat.
8205 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8206 * dwarf2out.c (highest_c_language): Handle C++17.
8207 (gen_compile_unit_die): Likewise.
8209 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8211 PR rtl-optimization/82192
8212 * combine.c (make_extraction): Don't look through non-paradoxical
8213 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8216 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8217 Alan Hayward <alan.hayward@arm.com>
8218 David Sherwood <david.sherwood@arm.com>
8220 * target.def (function_arg_offset): New hook.
8221 * targhooks.h (default_function_arg_offset): Declare.
8222 * targhooks.c (default_function_arg_offset): New function.
8223 * function.c (locate_and_pad_parm): Use
8224 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8225 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8226 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8227 * doc/tm.texi: Regenerate.
8228 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8229 * config/spu/spu.c (spu_function_arg_offset): New function.
8230 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8231 * system.h (FUNCTION_ARG_OFFSET): Poison.
8233 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8234 Alan Hayard <alan.hayward@arm.com>
8235 David Sherwood <david.sherwood@arm.com>
8237 * target.def (truly_noop_truncation): New hook.
8238 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8239 than TRULY_NOOP_TRUNCATION.
8240 * hooks.h (hook_bool_uint_uint_true): Declare.
8241 * hooks.c (hook_bool_uint_uint_true): New function.
8242 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8243 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8244 * doc/tm.texi: Regenerate.
8245 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8246 rather than TRULY_NOOP_TRUNCATION in comments.
8247 (simplify_comparison): Likewise.
8248 (record_truncated_value): Likewise.
8249 * expmed.c (extract_bit_field_1): Likewise.
8250 (extract_split_bit_field): Likewise.
8251 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8252 instead of TRULY_NOOP_TRUNCATION.
8253 * function.c (assign_parm_setup_block): Likewise.
8254 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8255 * rtlhooks.c: Include target.h.
8256 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8257 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8258 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8259 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8260 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8261 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8262 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8263 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8264 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8265 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8266 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8267 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8268 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8269 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8270 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8271 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8272 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8273 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8274 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8275 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8276 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8277 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8278 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8279 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8280 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8281 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8282 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8283 rather than TRULY_NOOP_TRUNCATION in comments.
8284 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8285 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8286 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8287 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8288 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8289 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8290 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8291 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8292 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8293 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8294 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8295 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8296 rather than TRULY_NOOP_TRUNCATION in comments.
8297 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8298 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8299 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8300 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8301 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8302 TRULY_NOOP_TRUNCATION condition.
8303 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8304 (TRULY_NOOP_TRUNCATION): Delete.
8305 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8306 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8307 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8308 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8309 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8310 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8311 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8312 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8313 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8314 rather than TRULY_NOOP_TRUNCATION in comments.
8315 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8316 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8317 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8318 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8319 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8320 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8322 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8325 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8326 (*cmp_ior): Likewise.
8327 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8328 (*ior_scc_scc_cmp): Likewise.
8329 (*and_scc_scc): Likewise.
8330 (*and_scc_scc_cmp): Likewise.
8332 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8333 Alan Hayard <alan.hayward@arm.com>
8334 David Sherwood <david.sherwood@arm.com>
8336 * target.def (can_change_mode_class): New hook.
8337 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8338 (hard_regno_nregs): Likewise.
8339 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8340 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8341 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8342 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8343 (LOAD_EXTEND_OP): Update accordingly.
8344 * doc/tm.texi: Regenerate.
8345 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8346 CANNOT_CHANGE_MODE_CLASS.
8347 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8348 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8349 * combine.c (simplify_set): Update accordingly.
8350 * emit-rtl.c (validate_subreg): Likewise.
8351 * recog.c (general_operand): Likewise.
8352 * regcprop.c (mode_change_ok): Likewise.
8353 * reload1.c (choose_reload_regs): Likewise.
8354 (inherit_piecemeal_p): Likewise.
8355 * rtlanal.c (simplify_subreg_regno): Likewise.
8356 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8357 instead of CANNOT_CHANGE_MODE_CLASS.
8358 (reload_cse_simplify_operands): Likewise.
8359 * reload.c (push_reload): Use targetm.can_change_mode_class
8360 instead of CANNOT_CHANGE_MODE_CLASS.
8361 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8362 REG_CANNOT_CHANGE_MODE_P.
8363 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8364 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8365 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8366 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8367 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8368 (arm_can_change_mode_class): New function.
8369 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8370 than CANNOT_CHANGE_MODE_CLASS in comments.
8371 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8372 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8373 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8374 (ix86_can_change_mode_class): ...this new function, inverting the
8375 sense of the return value.
8376 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8377 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8378 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8379 (ia64_can_change_mode_class): New function.
8380 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8381 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8382 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8383 (m32c_can_change_mode_class): ...this new function, inverting the
8384 sense of the return value.
8385 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8386 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8387 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8388 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8389 (mips_can_change_mode_class): ...this new function, inverting the
8390 sense of the return value.
8391 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8392 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8393 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8394 (msp430_can_change_mode_class): New function.
8395 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8396 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8397 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8398 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8399 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8400 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8401 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8402 (pa_cannot_change_mode_class): Replace with...
8403 (pa_can_change_mode_class): ...this new function, inverting the
8404 sense of the return value.
8405 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8406 than CANNOT_CHANGE_MODE_CLASS in comments.
8407 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8408 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8409 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8410 (pdp11_cannot_change_mode_class): Replace with...
8411 (pdp11_can_change_mode_class): ...this new function, inverting the
8412 sense of the return value.
8413 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8414 * config/powerpcspe/powerpcspe-protos.h
8415 (rs6000_cannot_change_mode_class_ptr): Delete.
8416 * config/powerpcspe/powerpcspe.c
8417 (rs6000_cannot_change_mode_class_ptr): Delete.
8418 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8419 (rs6000_option_override_internal): Assign to
8420 targetm.can_change_mode_class instead of
8421 rs6000_cannot_change_mode_class_ptr.
8422 (rs6000_cannot_change_mode_class): Replace with...
8423 (rs6000_can_change_mode_class): ...this new function, inverting the
8424 sense of the return value.
8425 (rs6000_debug_cannot_change_mode_class): Replace with...
8426 (rs6000_debug_can_change_mode_class): ...this new function.
8427 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8428 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8429 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8430 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8431 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8433 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
8434 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8435 (rs6000_option_override_internal): Assign to
8436 targetm.can_change_mode_class instead of
8437 rs6000_cannot_change_mode_class_ptr.
8438 (rs6000_cannot_change_mode_class): Replace with...
8439 (rs6000_can_change_mode_class): ...this new function, inverting the
8440 sense of the return value.
8441 (rs6000_debug_cannot_change_mode_class): Replace with...
8442 (rs6000_debug_can_change_mode_class): ...this new function.
8443 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8444 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8445 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8446 (s390_can_change_mode_class): ...this new function, inverting the
8447 sense of the return value.
8448 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8449 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8450 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8451 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8452 (sh_cannot_change_mode_class): Replace with...
8453 (sh_can_change_mode_class): ...this new function, inverting the
8454 sense of the return value.
8455 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8456 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8457 (sparc_can_change_mode_class): New function.
8458 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8459 * config/spu/spu.c (spu_can_change_mode_class): New function.
8460 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8461 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8462 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8463 (visium_can_change_mode_class): New function.
8464 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8466 2017-09-15 Richard Biener <rguenther@suse.de>
8468 PR tree-optimization/82217
8469 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8470 but not undefined case.
8472 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8475 * postreload.c (reload_cse_simplify_operands): Skip
8476 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
8478 2017-09-15 Richard Biener <rguenther@suse.de>
8480 PR tree-optimization/68823
8481 * graphite-scop-detection.c (build_alias_set): If we have a
8482 possible dependence check whether we can handle them by just
8483 looking at the DRs DR_ACCESS_FNs.
8484 (build_scops): If build_alias_set fails, fail the SCOP.
8486 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
8488 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
8489 to support float128 built-in functions that require the ISA 3.0
8491 (BU_FLOAT128_3_HW): Likewise.
8492 (SQRTF128): Add support for the IEEE 128-bit square root and fma
8494 (FMAF128): Likewise.
8496 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8497 support for built-in functions that need the ISA 3.0 IEEE 128-bit
8498 floating point instructions.
8499 (rs6000_invalid_builtin): Likewise.
8500 (rs6000_builtin_mask_names): Likewise.
8501 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
8502 (RS6000_BTM_FLOAT128_HW): Likewise.
8503 (RS6000_BTM_COMMON): Likewise.
8504 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
8506 * doc/extend.texi (RS/6000 built-in functions): Document the
8507 IEEE 128-bit floating point square root and fused multiply-add
8510 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
8512 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
8513 reg (r2) isn't in the set of registers defined in the prologue.
8515 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8516 Alan Hayward <alan.hayward@arm.com>
8517 David Sherwood <david.sherwood@arm.com>
8519 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
8520 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
8521 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
8522 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
8523 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
8525 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
8526 max_vectorization_factor.
8527 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
8529 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8530 Alan Hayward <alan.hayward@arm.com>
8531 David Sherwood <david.sherwood@arm.com>
8533 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
8534 (vect_worthwhile_without_simd_p): Declare.
8535 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
8536 (vectorizable_reduction): Use it.
8537 * tree-vect-stmts.c (vectorizable_shift): Likewise.
8538 (vectorizable_operation): Likewise.
8540 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8541 Alan Hayward <alan.hayward@arm.com>
8542 David Sherwood <david.sherwood@arm.com>
8544 * tree-vectorizer.h (vect_get_num_copies): New function.
8545 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
8546 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8547 (vectorizable_induction): Likewise.
8548 (vectorizable_live_operation): Likewise.
8549 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8550 (vectorizable_bswap): Likewise.
8551 (vectorizable_call): Likewise.
8552 (vectorizable_conversion): Likewise.
8553 (vectorizable_assignment): Likewise.
8554 (vectorizable_shift): Likewise.
8555 (vectorizable_operation): Likewise.
8556 (vectorizable_store): Likewise.
8557 (vectorizable_load): Likewise.
8558 (vectorizable_condition): Likewise.
8559 (vectorizable_comparison): Likewise.
8560 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
8562 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8563 Alan Hayward <alan.hayward@arm.com>
8564 David Sherwood <david.sherwood@arm.com>
8566 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
8567 of vect_init_vector.
8569 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8570 Alan Hayward <alan.hayward@arm.com>
8571 David Sherwood <david.sherwood@arm.com>
8573 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
8574 an inline wrapper that provides a location.
8575 (gimple_build_vector): Likewise.
8576 * gimple-fold.c (gimple_build_vector_from_val): New function.
8577 (gimple_build_vector): Likewise.
8578 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
8579 functions to build the initial value. Always return a gimple value.
8580 (get_initial_defs_for_reduction): Likewise. Only compute
8582 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
8583 vect_init_vector on the results from get_initial_def(s)_for_reduction.
8584 (vectorizable_induction): Use gimple_build_vector rather than
8587 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8588 Alan Hayward <alan.hayward@arm.com>
8589 David Sherwood <david.sherwood@arm.com>
8591 * target.h (vec_perm_indices): New typedef.
8592 (auto_vec_perm_indices): Likewise.
8593 * optabs-query.h: Include target.h
8594 (can_vec_perm_p): Take a vec_perm_indices *.
8595 * optabs-query.c (can_vec_perm_p): Likewise.
8596 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
8597 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8598 * tree-vect-generic.c (lower_vec_perm): Likewise.
8599 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
8600 (vect_grouped_load_supported): Likewise.
8601 (vect_shift_permute_load_chain): Likewise.
8602 (vect_permute_store_chain): Use auto_vec_perm_indices.
8603 (vect_permute_load_chain): Likewise.
8604 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
8605 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
8606 Update uses of can_vec_perm_p.
8607 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
8608 mode with a number of elements. Take a vec_perm_indices *.
8609 (vect_create_epilog_for_reduction): Update accordingly.
8610 Use auto_vec_perm_indices.
8611 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
8612 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
8613 (vect_transform_slp_perm_load): Likewise.
8614 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
8615 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
8616 (vect_gen_perm_mask_checked): Likewise.
8617 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
8618 (vect_gen_perm_mask_checked): Likewise.
8619 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
8620 (vectorizable_store): Likewise.
8621 (vectorizable_load): Likewise.
8622 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
8623 (vectorizable_bswap): Likewise.
8625 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8626 Alan Hayward <alan.hayward@arm.com>
8627 David Sherwood <david.sherwood@arm.com>
8629 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
8630 * tree.c (build_vector): Likewise.
8631 (build_vector_from_ctor): Update accordingly.
8632 (build_vector_from_val): Likewise.
8633 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8634 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8635 * tree-vect-generic.c (add_rshift): Likewise.
8636 (expand_vector_divmod): Likewise.
8637 (optimize_vector_constructor): Likewise.
8638 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
8639 (vect_transform_slp_perm_load): Likewise.
8640 (vect_schedule_slp_instance): Likewise.
8641 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
8642 (vectorizable_call): Likewise.
8643 (vect_gen_perm_mask_any): Likewise. Add elements in order.
8644 * expmed.c (make_tree): Likewise.
8645 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
8646 a vector passed to build_vector.
8647 (fold_convert_const): Likewise.
8648 (exact_inverse): Likewise.
8649 (fold_ternary_loc): Likewise.
8650 (fold_relational_const): Likewise.
8651 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
8652 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
8653 (const_unop): Likewise. Store the reduction accumulator in a
8654 variable rather than an array.
8655 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
8656 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
8657 the new vector, rather than constructing it after the input arrays.
8658 (native_interpret_vector): Use auto_vec<tree> when building
8659 a vector passed to build_vector. Add elements in order.
8660 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
8661 auto_vec<tree> when building a vector passed to build_vector.
8662 (vect_create_epilog_for_reduction): Likewise.
8663 (vectorizable_induction): Likewise.
8664 (get_initial_def_for_reduction): Likewise. Fix indentation of
8666 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
8668 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
8669 passed to build_vector.
8671 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8672 Alan Hayward <alan.hayward@arm.com>
8673 David Sherwood <david.sherwood@arm.com>
8675 * tree-core.h (tree_base::u): Add an "nelts" field.
8676 (tree_vector): Use VECTOR_CST_NELTS as the length.
8677 * tree.c (tree_size): Likewise.
8678 (make_vector): Initialize VECTOR_CST_NELTS.
8679 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
8680 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
8681 TYPE_VECTOR_SUBPARTS.
8682 * expr.c (const_vector_mask_from_tree): Consistently use "units"
8683 as the number of units, setting it from VECTOR_CST_NELTS.
8684 (const_vector_from_tree): Likewise.
8685 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
8686 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
8687 (fold_negate_expr_1): Likewise.
8688 (fold_convert_const): Likewise.
8689 (const_binop): Likewise. Differentiate the number of output and
8691 (const_unop): Likewise.
8692 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
8693 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
8694 in cases that did the opposite.
8696 2017-09-14 Richard Biener <rguenther@suse.de>
8698 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
8701 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
8703 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
8705 2017-09-14 Jakub Jelinek <jakub@redhat.com>
8708 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
8709 if and where to split a bb, except for splitting before debug insn
8710 sequences followed by non-label real insn. Delete debug insns
8711 in between basic blocks.
8713 * combine.c (make_compound_operation_int): Formatting fixes.
8715 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
8716 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8717 * config/netbsd.h (LINK_EH_SPEC): Likewise.
8718 * config/sol2.h (LINK_EH_SPEC): Likewise.
8719 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8720 * config/s390/linux.h (LINK_SPEC): Likewise.
8721 * config/freebsd.h (LINK_EH_SPEC): Likewise.
8722 * config/openbsd.h (LINK_EH_SPEC): Likewise.
8723 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8724 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
8725 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
8726 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8727 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
8728 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
8730 2017-09-13 Jakub Jelinek <jakub@redhat.com>
8732 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
8734 (ENDFILE_LINUX_SPEC): Likewise.
8735 (LINK_EH_SPEC): Likewise.
8736 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
8737 (LINK_OS_LINUX_SPEC32): Likewise.
8738 (LINK_OS_LINUX_SPEC64): Likewise.
8739 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
8740 (LINK_OS_LINUX_SPEC): Likewise.
8742 2017-09-13 Martin Liska <mliska@suse.cz>
8745 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
8746 CASE_HIGH is NULL_TREE.
8748 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8749 Alan Hayward <alan.hayward@arm.com>
8750 David Sherwood <david.sherwood@arm.com>
8752 * target.def (secondary_memory_needed): New hook.
8753 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
8754 instead of SECONDARY_MEMORY_NEEDED.
8755 (secondary_memory_needed_mode): Likewise.
8756 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
8757 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
8758 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
8759 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
8760 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
8761 * doc/tm.texi: Regenerate.
8762 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
8763 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
8764 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8765 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
8766 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
8767 * config/i386/i386.c (inline_secondary_memory_needed): Put the
8768 mode argument first and change the reg_class arguments to reg_class_t.
8769 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
8770 Make static. Update the call to inline_secondary_memory_needed.
8771 (ix86_register_move_cost): Update the call to
8772 inline_secondary_memory_needed.
8773 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8774 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
8776 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
8777 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8779 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
8780 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
8781 * config/mips/mips.c (mips_secondary_memory_needed): Make static
8782 and match hook interface. Add comment from mips.h.
8783 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8784 * config/mmix/mmix.md (truncdfsf2): Refer to
8785 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8787 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
8788 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
8789 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8790 (pa_secondary_memory_needed): New function.
8791 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
8792 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
8793 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8794 (pdp11_secondary_memory_needed): Make static and match hook interface.
8795 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
8796 * config/powerpcspe/powerpcspe-protos.h
8797 (rs6000_secondary_memory_needed_ptr): Delete.
8798 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
8800 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8801 (rs6000_option_override_internal): Assign to
8802 targetm.secondary_memory_needed rather than
8803 rs6000_secondary_memory_needed_ptr.
8804 (rs6000_secondary_memory_needed): Match hook interface.
8805 (rs6000_debug_secondary_memory_needed): Likewise.
8806 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
8807 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
8808 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
8809 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8810 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
8811 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
8813 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
8814 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8815 (rs6000_option_override_internal): Assign to
8816 targetm.secondary_memory_needed rather than
8817 rs6000_secondary_memory_needed_ptr.
8818 (rs6000_secondary_memory_needed): Match hook interface.
8819 (rs6000_debug_secondary_memory_needed): Likewise.
8820 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
8821 * config/s390/s390.c (s390_secondary_memory_needed): New function.
8822 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8823 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
8824 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8825 (sparc_secondary_memory_needed): New function.
8826 * lra-constraints.c (check_and_process_move): Refer to
8827 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8829 (curr_insn_transform): Likewise.
8830 (process_alt_operands): Use targetm.secondary_memory_needed
8831 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8832 (check_secondary_memory_needed_p): Likewise.
8833 (choose_split_class): Likewise.
8834 * reload.c: Unconditionally include code that was previously
8835 conditional on SECONDARY_MEMORY_NEEDED.
8836 (push_secondary_reload): Use targetm.secondary_memory_needed
8837 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8838 (push_reload): Likewise.
8839 * reload1.c: Unconditionally include code that was previously
8840 conditional on SECONDARY_MEMORY_NEEDED.
8841 (choose_reload_regs): Use targetm.secondary_memory_needed
8842 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8843 (gen_reload): Likewise.
8844 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
8846 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8847 Alan Hayward <alan.hayward@arm.com>
8848 David Sherwood <david.sherwood@arm.com>
8850 * target.def (secondary_memory_needed_mode): New hook:
8851 * targhooks.c (default_secondary_memory_needed_mode): Declare.
8852 * targhooks.h (default_secondary_memory_needed_mode): New function.
8853 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
8854 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
8855 * doc/tm.texi: Regenerate.
8856 * lra-constraints.c (check_and_process_move): Use
8857 targetm.secondary_memory_needed_mode instead of
8858 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
8859 (curr_insn_transform): Likewise.
8860 * reload.c (get_secondary_mem): Likewise.
8861 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8862 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
8864 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8865 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8866 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
8867 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8868 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
8870 * config/powerpcspe/powerpcspe-protos.h
8871 (rs6000_secondary_memory_needed_mode): Delete.
8872 * config/powerpcspe/powerpcspe.c
8873 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8874 (rs6000_secondary_memory_needed_mode): Make static.
8875 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8876 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
8878 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8880 (rs6000_secondary_memory_needed_mode): Make static.
8881 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8882 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
8883 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8884 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8885 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8887 (sparc_secondary_memory_needed_mode): New function.
8888 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
8890 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
8892 * config/aarch64/constraints.md (Umq): New constraint.
8893 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
8895 (mov<mode>): Update condition.
8897 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8899 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
8900 when bitposition is the same.
8902 2017-09-13 Richard Biener <rguenther@suse.de>
8904 * dwarf2out.c (output_die_symbol): Remove.
8905 (output_die): Do not output a DIEs symbol.
8907 2017-09-13 Richard Biener <rguenther@suse.de>
8910 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
8911 default-def to avoid breaking iterator update with the weird
8912 interaction with cgraph_update_edges_for_call_stmt_node.
8914 2017-09-13 Richard Biener <rguenther@suse.de>
8916 * tree-cfg.c (verify_gimple_assign_binary): Add verification
8917 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
8918 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
8919 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
8921 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
8923 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
8924 Disable pc relative literal load irrespective of
8925 TARGET_FIX_ERR_A53_84341 for default.
8927 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
8929 * config/sparc/sparc.c (output_return): Output the source location of
8930 the insn in the delay slot, if any.
8931 (output_sibcall): Likewise.
8933 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
8936 * common.opt (-static-pie): New alias.
8937 (shared): Negate static-pie.
8938 (-no-pie): Update help text.
8940 (static-pie): New option.
8941 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
8942 -static-pie support.
8943 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
8944 (LINK_EH_SPEC): Likewise.
8945 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8946 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8947 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8948 * gcc.c (LINK_COMMAND_SPEC): Likewise.
8949 (init_gcc_specs): Likewise.
8950 (init_spec): Likewise.
8951 (display_help): Update help message for -pie.
8952 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
8955 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
8957 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
8958 (movdi_aarch64): Likewise.
8959 (movti_aarch64): Likewise.
8961 2017-09-12 Simon Wright <simon@pushface.org>
8964 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
8965 calculation of the minor version, always output as 0.
8967 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8970 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8971 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
8972 on it early, rather than manual conversion late. For
8973 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
8974 instead of performing manual conversion.
8976 2017-09-12 Carl Love <cel@us.ibm.com>
8978 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
8979 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
8981 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8982 VMULOSW): Add definitions.
8983 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8984 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8985 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8986 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8987 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8989 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8991 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
8993 (movti_aarch64): Likewise.
8994 (movdf_aarch64): Likewise.
8995 (movtf_aarch64): Likewise.
8996 (load_pairdi): Likewise.
8997 (store_pairdi): Likewise.
8998 (load_pairdf): Likewise.
8999 (store_pairdf): Likewise.
9000 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
9001 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
9002 (ldr_got_small_<mode>): Likewise.
9003 (ldr_got_small_28k_<mode>): Likewise.
9004 (ldr_got_tiny): Likewise.
9005 * config/aarch64/iterators.md (ldst_sz): New.
9006 (ldpstp_sz): Likewise.
9007 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
9009 (thunderx_load): Split load_8 to load_16.
9010 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
9012 (thunderx2t99_storepair_basic): Split store_8 to store_16.
9013 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
9014 (xgene1_store_pair): Split store_8 to store_16.
9015 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
9016 (falkor_st_0_st_sd): Split store_8 to store_16.
9018 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9020 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
9021 and store1/2/3/4 to store_4/8/12/16.
9022 * config/aarch64/aarch64.md: Update for rename.
9023 * config/arm/arm.md: Likewise.: Likewise.
9024 * config/arm/arm.c: Likewise.
9025 * config/arm/thumb1.md: Likewise.
9026 * config/arm/thumb2.md: Likewise.
9027 * config/arm/vfp.md: Likewise.
9028 * config/arm/arm-generic.md: Likewise.
9029 * config/arm/arm1020e.md: Likewise.
9030 * config/arm/arm1026ejs.md: Likewise.
9031 * config/arm/arm1136jfs.md: Likewise.
9032 * config/arm/arm926ejs.md: Likewise.
9033 * config/arm/cortex-a15.md: Likewise.
9034 * config/arm/cortex-a17.md: Likewise.
9035 * config/arm/cortex-a5.md: Likewise.
9036 * config/arm/cortex-a53.md: Likewise.
9037 * config/arm/cortex-a57.md: Likewise.
9038 * config/arm/cortex-a7.md: Likewise.
9039 * config/arm/cortex-a8.md: Likewise.
9040 * config/arm/cortex-a9.md: Likewise.
9041 * config/arm/cortex-m4.md: Likewise.
9042 * config/arm/cortex-m7.md: Likewise.
9043 * config/arm/cortex-r4.md: Likewise.
9044 * config/arm/exynos-m1.md: Likewise.
9045 * config/arm/fa526.md: Likewise.
9046 * config/arm/fa606te.md: Likewise.
9047 * config/arm/fa626te.md: Likewise.
9048 * config/arm/fa726te.md: Likewise.
9049 * config/arm/fmp626.md: Likewise.
9050 * config/arm/iwmmxt.md: Likewise.
9051 * config/arm/ldmstm.md: Likewise.
9052 * config/arm/marvell-pj4.md: Likewise.
9053 * config/arm/xgene1.md: Likewise.
9054 * config/aarch64/thunderx.md: Likewise.
9055 * config/aarch64/thunderx2t99.md: Likewise.
9056 * config/aarch64/falkor.md: Likewise.
9058 2017-09-12 Martin Liska <mliska@suse.cz>
9060 * attribs.c (private_lookup_attribute): New function.
9061 * attribs.h (private_lookup_attribute): Declared here.
9062 (lookup_attribute): Called from this place.
9064 2017-09-12 Richard Biener <rguenther@suse.de>
9066 PR tree-optimization/82157
9067 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
9068 stmts with side-effects.
9070 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9071 Alan Hayward <alan.hayward@arm.com>
9072 David Sherwood <david.sherwood@arm.com>
9074 * target.def (hard_regno_nregs): New hook.
9075 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
9076 * targhooks.h (default_hard_regno_nregs): Declare.
9077 * targhooks.c (default_hard_regno_nregs): New function.
9078 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
9079 (TARGET_HARD_REGNO_NREGS): ...this hook.
9080 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
9081 (CLASS_MAX_NREGS): Likewise.
9082 * doc/tm.texi: Regenerate.
9083 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
9084 instead of HARD_REGNO_NREGS.
9085 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
9086 HARD_REGNO_NREGS in the comment.
9087 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
9088 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
9089 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
9090 Return an unsigned int.
9091 (TARGET_HARD_REGNO_NREGS): Redefine.
9092 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
9093 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
9094 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9095 (arc_hard_regno_nregs): New function.
9096 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
9097 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
9098 (arm_hard_regno_nregs): New function.
9099 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
9100 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
9101 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
9102 (TARGET_HARD_REGNO_NREGS): Redefine.
9103 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
9104 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
9105 (HARD_REGNO_NREGS): Delete.
9106 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
9107 (cr16_hard_regno_nregs): New function.
9108 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
9109 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
9110 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
9111 (cris_hard_regno_nregs): New function.
9112 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
9113 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
9114 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9115 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
9116 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
9117 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
9118 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
9119 (frv_hard_regno_nregs): Make static. Take and return an
9121 (frv_class_max_nregs): Remove outdated copy of documentation.
9122 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
9123 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
9124 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
9125 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
9126 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
9127 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
9128 (TARGET_HARD_REGNO_NREGS): Redefine.
9129 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
9130 (CLASS_MAX_NREGS): Update comment.
9131 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
9132 (ia64_hard_regno_nregs): New function.
9133 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
9134 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
9135 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
9136 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
9137 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
9139 (m32c_hard_regno_nregs): Likewise. Make static.
9140 (TARGET_HARD_REGNO_NREGS): Redefine.
9141 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
9142 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
9143 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
9144 (m68k_hard_regno_nregs): New function.
9145 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
9146 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
9147 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
9148 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
9149 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
9150 Take and return an unsigned int.
9151 (TARGET_HARD_REGNO_NREGS): Redefine.
9152 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
9153 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9154 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
9155 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
9156 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
9157 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
9158 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
9159 (msp430_hard_regno_nregs): Make static. Take and return an
9161 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
9162 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
9163 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
9164 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
9165 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
9166 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
9167 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
9168 (TARGET_HARD_REGNO_NREGS): Redefine.
9169 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
9170 (PA_HARD_REGNO_NREGS): ...this.
9171 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
9172 (PA_HARD_REGNO_NREGS): ...this.
9173 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9174 (pa_hard_regno_nregs): New function.
9175 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
9176 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
9177 (pdp11_hard_regno_nregs): New function.
9178 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
9179 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9180 (rs6000_hard_regno_nregs_hook): New function.
9181 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9182 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9183 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9184 Take and return an unsigned int. Move earlier in file.
9185 (TARGET_HARD_REGNO_NREGS): Redefine.
9186 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9187 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9188 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9189 (rl78_hard_regno_nregs): Make static. Take and return an
9191 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9192 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9193 (rs6000_hard_regno_nregs_hook): New function.
9194 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9195 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9196 (TARGET_HARD_REGNO_NREGS): Redefine.
9197 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9198 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9199 instead of HARD_REGNO_NREGS.
9200 (s390_hard_regno_nregs): New function.
9201 (s390_hard_regno_mode_ok): Add comment from s390.h.
9202 (TARGET_HARD_REGNO_NREGS): Redefine.
9203 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9204 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9205 (sh_hard_regno_nregs): New function.
9206 (sh_pass_in_reg_p): Use it.
9207 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9208 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9209 (sparc_hard_regno_nregs): New function.
9210 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9211 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9212 (spu_function_arg_advance): Use it, supplying a valid register number.
9213 (TARGET_HARD_REGNO_NREGS): Redefine.
9214 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9215 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9216 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9217 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9218 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9219 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9220 (CLASS_MAX_NREGS): Remove copy of old documentation.
9221 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9222 (visium_hard_regno_nregs): New function.
9223 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9224 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9225 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9226 xtensa_hard_regno_nregs): New function.
9227 * system.h (HARD_REGNO_NREGS): Poison.
9229 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9231 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9232 hard_regno_nregs instead of HARD_REGNO_NREGS.
9233 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9234 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9235 (c6x_expand_epilogue): Likewise.
9236 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9237 (frv_read_iacc_argument): Likewise.
9238 * config/sh/sh.c: Include regs.h.
9239 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9240 (regs_used): Likewise.
9241 (output_stack_adjust): Likewise.
9242 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9243 * expmed.c: Include regs.h.
9244 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9245 * ree.c: Include regs.h.
9246 (combine_reaching_defs): Use hard_regno_nregs instead of
9248 (add_removable_extension): Likewise.
9250 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9252 * regs.h (hard_regno_nregs): Turn into a function.
9253 (end_hard_regno): Update accordingly.
9254 * caller-save.c (setup_save_areas): Likewise.
9255 (save_call_clobbered_regs): Likewise.
9256 (replace_reg_with_saved_mem): Likewise.
9257 (insert_restore): Likewise.
9258 (insert_save): Likewise.
9259 * combine.c (can_change_dest_mode): Likewise.
9260 (move_deaths): Likewise.
9261 (distribute_notes): Likewise.
9262 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9263 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9264 (rs6000_split_multireg_move): Likewise.
9265 (rs6000_register_move_cost): Likewise.
9266 (rs6000_memory_move_cost): Likewise.
9267 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9268 (rs6000_split_multireg_move): Likewise.
9269 (rs6000_register_move_cost): Likewise.
9270 (rs6000_memory_move_cost): Likewise.
9271 * cselib.c (cselib_reset_table): Likewise.
9272 (cselib_lookup_1): Likewise.
9273 * emit-rtl.c (set_mode_and_regno): Likewise.
9274 * function.c (aggregate_value_p): Likewise.
9275 * ira-color.c (setup_profitable_hard_regs): Likewise.
9276 (check_hard_reg_p): Likewise.
9277 (calculate_saved_nregs): Likewise.
9278 (assign_hard_reg): Likewise.
9279 (improve_allocation): Likewise.
9280 (calculate_spill_cost): Likewise.
9281 * ira-emit.c (modify_move_list): Likewise.
9282 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9283 (ira_hard_reg_in_set_p): Likewise.
9284 * ira.c (setup_reg_mode_hard_regset): Likewise.
9285 (clarify_prohibited_class_mode_regs): Likewise.
9286 (check_allocation): Likewise.
9287 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9288 (lra_setup_reg_renumber): Likewise.
9289 (setup_try_hard_regno_pseudos): Likewise.
9290 (spill_for): Likewise.
9291 (assign_hard_regno): Likewise.
9292 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9293 * lra-constraints.c (in_class_p): Likewise.
9294 (lra_constraint_offset): Likewise.
9295 (simplify_operand_subreg): Likewise.
9296 (lra_constraints): Likewise.
9297 (split_reg): Likewise.
9298 (split_if_necessary): Likewise.
9299 (invariant_p): Likewise.
9300 (inherit_in_ebb): Likewise.
9301 * lra-lives.c (process_bb_lives): Likewise.
9302 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9303 (get_hard_regs): Likewise.
9304 (do_remat): Likewise.
9305 * lra-spills.c (assign_spill_hard_regs): Likewise.
9306 * mode-switching.c (create_pre_exit): Likewise.
9307 * postreload.c (reload_combine_recognize_pattern): Likewise.
9308 * recog.c (peep2_find_free_register): Likewise.
9309 * regcprop.c (kill_value_regno): Likewise.
9310 (set_value_regno): Likewise.
9311 (copy_value): Likewise.
9312 (maybe_mode_change): Likewise.
9313 (find_oldest_value_reg): Likewise.
9314 (copyprop_hardreg_forward_1): Likewise.
9315 * regrename.c (check_new_reg_p): Likewise.
9316 (regrename_do_replace): Likewise.
9317 * reload.c (push_reload): Likewise.
9318 (combine_reloads): Likewise.
9319 (find_dummy_reload): Likewise.
9320 (operands_match_p): Likewise.
9321 (find_reloads): Likewise.
9322 (find_equiv_reg): Likewise.
9323 (reload_adjust_reg_for_mode): Likewise.
9324 * reload1.c (count_pseudo): Likewise.
9325 (count_spilled_pseudo): Likewise.
9326 (find_reg): Likewise.
9327 (clear_reload_reg_in_use): Likewise.
9328 (free_for_value_p): Likewise.
9329 (allocate_reload_reg): Likewise.
9330 (choose_reload_regs): Likewise.
9331 (reload_adjust_reg_for_temp): Likewise.
9332 (emit_reload_insns): Likewise.
9333 (delete_output_reload): Likewise.
9334 * rtlanal.c (subreg_get_info): Likewise.
9335 * sched-deps.c (sched_analyze_reg): Likewise.
9336 * sel-sched.c (init_regs_for_mode): Likewise.
9337 (mark_unavailable_hard_regs): Likewise.
9338 (choose_best_reg_1): Likewise.
9339 (verify_target_availability): Likewise.
9340 * valtrack.c (dead_debug_insert_temp): Likewise.
9341 * var-tracking.c (track_loc_p): Likewise.
9342 (emit_note_insn_var_location): Likewise.
9343 * varasm.c (make_decl_rtl): Likewise.
9344 * reginfo.c (choose_hard_reg_mode): Likewise.
9345 (init_reg_modes_target): Refer directly to
9346 this_target_regs->x_hard_regno_nregs.
9348 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9350 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9351 instead of hard_regno_nregs.
9353 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9355 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9356 end_hard_regno instead of hard_regno_nregs.
9357 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9358 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9359 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9360 * ira-color.c (improve_allocation): Likewise.
9361 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9362 * lra-lives.c (mark_regno_live): Likewise.
9363 (mark_regno_dead): Likewise.
9364 * lra-remat.c (operand_to_remat): Likewise.
9365 * lra.c (collect_non_operand_hard_regs): Likewise.
9366 * postreload.c (reload_combine_note_store): Likewise.
9367 (move2add_valid_value_p): Likewise.
9368 * reload.c (regno_clobbered_p): Likewise.
9370 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9372 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9374 * config/v850/v850.c (v850_reorg): Likewise.
9375 * reload.c (refers_to_regno_for_reload_p): Likewise.
9376 (find_equiv_reg): Likewise.
9377 * reload1.c (reload_reg_reaches_end_p): Likewise.
9379 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9381 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9383 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9384 * config/arm/arm.c (output_move_neon): Likewise.
9385 (arm_attr_length_move_neon): Likewise.
9386 (neon_split_vcombine): Likewise.
9387 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9388 (c6x_mark_reg_written): Likewise.
9389 (c6x_dwarf_register_span): Likewise.
9390 * config/i386/i386.c (ix86_save_reg): Likewise.
9391 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9392 (rws_access_reg): Likewise.
9393 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9394 * mode-switching.c (create_pre_exit): Likewise.
9395 * ree.c (combine_reaching_defs): Likewise.
9396 (add_removable_extension): Likewise.
9397 * regcprop.c (find_oldest_value_reg): Likewise.
9398 (copyprop_hardreg_forward_1): Likewise.
9399 * reload.c (reload_inner_reg_of_subreg): Likewise.
9400 (push_reload): Likewise.
9401 (combine_reloads): Likewise.
9402 (find_dummy_reload): Likewise.
9403 (reload_adjust_reg_for_mode): Likewise.
9404 * reload1.c (find_reload_regs): Likewise.
9405 (forget_old_reloads_1): Likewise.
9406 (reload_reg_free_for_value_p): Likewise.
9407 (reload_adjust_reg_for_temp): Likewise.
9408 (emit_reload_insns): Likewise.
9409 (delete_output_reload): Likewise.
9410 * sel-sched.c (choose_best_reg_1): Likewise.
9411 (choose_best_pseudo_reg): Likewise.
9413 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9414 Alan Hayward <alan.hayward@arm.com>
9415 David Sherwood <david.sherwood@arm.com>
9417 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9418 * target.def (slow_unaligned_access): New hook.
9419 * targhooks.h (default_slow_unaligned_access): Declare.
9420 * targhooks.c (default_slow_unaligned_access): New function.
9421 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9422 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9423 * doc/tm.texi: Regenerate.
9424 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9425 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9426 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9428 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9429 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9431 (rs6000_slow_unaligned_access): New function.
9432 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9433 (expand_block_compare): Likewise.
9434 (expand_strn_compare): Likewise.
9435 (rs6000_rtx_costs): Likewise.
9436 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9437 (riscv_slow_unaligned_access): Likewise.
9438 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9439 (riscv_slow_unaligned_access_p): ...this and make static.
9440 (riscv_option_override): Update accordingly.
9441 (riscv_slow_unaligned_access): New function.
9442 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9443 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9444 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9445 (rs6000_slow_unaligned_access): New function.
9446 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9447 (rs6000_rtx_costs): Likewise.
9448 * config/rs6000/rs6000-string.c (expand_block_compare)
9449 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9450 of SLOW_UNALIGNED_ACCESS.
9451 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9452 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
9453 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
9454 of SLOW_UNALIGNED_ACCESS.
9455 * expmed.c (simple_mem_bitfield_p): Likewise.
9456 * expr.c (alignment_for_piecewise_move): Likewise.
9457 (emit_group_load_1): Likewise.
9458 (emit_group_store): Likewise.
9459 (copy_blkmode_from_reg): Likewise.
9460 (emit_push_insn): Likewise.
9461 (expand_assignment): Likewise.
9462 (store_field): Likewise.
9463 (expand_expr_real_1): Likewise.
9464 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9465 * lra-constraints.c (simplify_operand_subreg): Likewise.
9466 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9467 * gimple-ssa-store-merging.c: Likewise in block comment at start
9469 * tree-ssa-strlen.c: Include target.h.
9470 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
9471 of SLOW_UNALIGNED_ACCESS.
9472 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
9474 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9476 PR rtl-optimization/82185
9477 * expmed.c (emit_store_flag_int): Only test tem if it has been
9480 2017-09-12 Richard Biener <rguenther@suse.de>
9483 * match.pd ((FTYPE) N CMP CST): Fix typo.
9485 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
9487 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
9489 (mips_near_type_p): Add 'short_call' attribute as a synonym
9491 * doc/extend.texi (short_call): Document new function attribute.
9493 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9496 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
9497 assertion check that in the condition.
9498 (get_atomic_generic_size): Likewise. Before testing if parameter
9499 has pointer type, if it has array type, call for C++
9500 default_conversion to perform array-to-pointer conversion.
9502 2017-09-12 Richard Biener <rguenther@suse.de>
9504 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
9505 for operations we cannot scalarize.
9507 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
9509 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
9510 vectors heap vectors. Clean up comments.
9511 Make visited_bbs a reference.
9512 (profitable_jump_thread_path): Make GC
9513 vectors heap vectors. Clean up comments.
9515 (convert_and_register_jump_thread_path): Make GC vectors heap
9517 (check_subpath_and_update_thread_path): Same. Clean up comments.
9518 Make visited_bbs a reference.
9519 (handle_phi): Abstract common code to to
9520 register_jump_thread_path_if_profitable.
9521 Rename VAR_BB to DEF_BB.
9523 Make GC vectors heap vectors.
9524 Make visited_bbs a reference.
9525 (handle_assignment): Same.
9526 (register_jump_thread_path_if_profitable): New.
9527 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
9529 Make GC vectors heap vectors. Clean up comments.
9530 Make visited_bbs a reference.
9531 (find_jump_threads_backwards): Make visited_bbs live in the stack.
9532 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
9535 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
9538 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
9539 words of E_DImode object are reachable by xtensa_uimm8x4 access.
9541 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
9543 Revert r251800 and r251799.
9545 2017-09-11 Martin Jambor <mjambor@suse.cz>
9548 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
9549 arguments in advance.
9550 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
9551 use it to find predecessor edges.
9552 (naive_outof_ssa): Collect vector of predecessors.
9554 2017-09-08 Jason Merrill <jason@redhat.com>
9556 PR c++/70029 - ICE with ref-qualifier and -flto
9557 * langhooks.h (struct lang_hooks_for_types): Add
9558 copy_lang_qualifiers.
9559 * attribs.c (build_type_attribute_qual_variant): Use it.
9560 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
9562 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
9563 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
9565 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
9568 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
9569 (*mulsi3_sp64): New instruction.
9570 (mulsi3): New expander.
9572 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
9574 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
9576 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9578 * sancov.c: Include memmodel.h.
9580 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
9583 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
9586 2017-09-07 Carl Love <cel@us.ibm.com>
9588 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
9589 the sldi instruction.
9591 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
9593 * sancov.c: Include tm_p.h.
9595 2017-09-07 Jakub Jelinek <jakub@redhat.com>
9598 * output.h (switch_to_other_text_partition): New declaration.
9599 * varasm.c (switch_to_other_text_partition): New function.
9600 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
9601 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
9602 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
9603 to the other text partition before emitting LCL label and switch back
9604 after emitting the word after it.
9606 2017-09-07 Richard Biener <rguenther@suse.de>
9608 * passes.def (pass_split_crit_edges): Remove instance before PRE.
9609 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
9610 critical edges here, after loop init.
9611 (pass_data_pre): Remove PROP_no_crit_edges flags.
9612 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
9613 for valueization of call args to avoid leaking VN_TOP.
9614 (visit_use): Assert we do not visit default defs.
9615 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
9616 Use error_mark_node to more easily detect leaking VN_TOP.
9617 All default-defs are varying, not VN_TOP. Mark them visited.
9618 (run_scc_vn): Make code match comment.
9620 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9622 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
9623 OPTION_MASK_FLOAT128_KEYWORD.
9624 (POWERPC_MASKS): Likewise.
9625 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
9626 support for the -mfloat128-type option, and make -mfloat128
9627 default on PowerPC Linux systems. Define or undefine
9628 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
9629 Define __float128 to be __ieee128 if IEEE 128-bit support is
9630 enabled, or undefine it.
9631 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
9632 Delete defining __FLOAT128_TYPE__.
9633 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
9634 -mfloat128-type option and make -mfloat128 default on PowerPC
9636 (TARGET_FLOAT128_TYPE): Likewise.
9637 (-mfloat128-type): Likewise.
9638 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9639 Delete the -mfloat128-type option and make -mfloat128 default on
9640 PowerPC Linux systems. Always use __ieee128 to be the keyword for
9641 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
9642 128-bit floating point is enabled. Change tests from using
9643 -mfloat128-type to -mfloat128.
9644 (rs6000_mangle_type): Use the correct mangling for the __float128
9645 type even if normal long double is restricted to 64-bits.
9646 (floatn_mode): Enable the _Float128 type by default on VSX Linux
9648 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
9649 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
9650 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
9652 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
9653 documentation for -mfloat128.
9655 2017-09-06 Olivier Hainque <hainque@adacore.com>
9657 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
9659 2017-09-06 Wish Wu <wishwu007@gmail.com>
9660 Jakub Jelinek <jakub@redhat.com>
9662 * asan.c (initialize_sanitizer_builtins): Add
9663 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
9664 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
9665 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
9666 BT_FN_VOID_UINT64_PTR variables.
9667 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
9668 (BT_FN_VOID_UINT16_UINT16): Likewise.
9669 (BT_FN_VOID_UINT32_UINT32): Likewise.
9670 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
9671 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
9672 (BT_FN_VOID_UINT64_PTR): Likewise.
9673 * common.opt (flag_sanitize_coverage): New variable.
9674 (fsanitize-coverage=trace-pc): Remove.
9675 (fsanitize-coverage=): Add.
9676 * flag-types.h (enum sanitize_coverage_code): New enum.
9677 * fold-const.c (fold_range_test): Disable non-short-circuit
9678 optimization if flag_sanitize_coverage.
9679 (fold_truth_andor): Likewise.
9680 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
9681 * opts.c (COVERAGE_SANITIZER_OPT): Define.
9682 (coverage_sanitizer_opts): New array.
9683 (get_closest_sanitizer_option): Add OPTS argument, handle also
9684 OPT_fsanitize_coverage_.
9685 (parse_sanitizer_options): Adjusted to also handle
9686 OPT_fsanitize_coverage_.
9687 (common_handle_option): Add OPT_fsanitize_coverage_.
9688 * sancov.c (instrument_comparison, instrument_switch): New function.
9689 (sancov_pass): Add trace-cmp support.
9690 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
9691 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
9692 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
9693 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
9694 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
9695 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
9696 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
9697 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
9698 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
9699 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
9701 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9703 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9704 error. Only quit immediately if parsing is complete.
9705 (BEGIN): Initialize fatal_err and parse_done.
9706 (begin fpu, end fpu): Check number of arguments.
9707 (begin arch, end arch): Likewise.
9708 (begin cpu, end cpu): Likewise.
9709 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9710 (optalias): Likewise.
9712 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9714 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9715 * config/arm/arm-isa.h: Delete. Move definitions to ...
9716 * arm-cpus.in: ... here. Use new feature and fgroup values.
9717 * config/arm/arm.c (arm_option_override): Use lower case for feature
9719 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9720 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9721 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9723 (print_isa_bits_for): New function.
9724 (gen_isa): New function.
9725 (gen_comm_data): Use print_isa_bits_for.
9726 (define feature): New keyword.
9727 (define fgroup): New keyword.
9728 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
9729 (arm-isa.h): Add rule to generate file.
9730 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9731 case for feature bit names.
9733 2017-09-06 Richard Biener <rguenther@suse.de>
9735 * tree-ssa-pre.c (NECESSARY): Remove.
9736 (create_expression_by_pieces): Do not touch pass-local flags.
9737 (insert_into_preds_of_block): Likewise.
9738 (do_pre_regular_insertion): Likewise.
9739 (eliminate_insert): Likewise.
9740 (eliminate_dom_walker::before_dom_children): Likewise.
9741 (fini_eliminate): Do not look at inserted_exprs.
9742 (mark_operand_necessary): Remove.
9743 (remove_dead_inserted_code): Replace with simple work-list
9744 algorithm based on inserted_exprs and SSA uses.
9745 (pass_pre::execute): Re-order fini_eliminate and
9746 remove_dead_inserted_code.
9748 2017-09-06 Olivier Hainque <hainque@adacore.com>
9750 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
9751 for VxWorks 7. Adjust surrounding comments.
9753 2017-09-06 Richard Biener <rguenther@suse.de>
9755 * gimple-ssa-strength-reduction.c
9756 (find_candidates_dom_walker::before_dom_children): Also allow
9759 2017-09-06 Richard Biener <rguenther@suse.de>
9761 PR tree-optimization/82108
9762 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
9763 for gap in the non-permutation SLP case.
9765 2017-09-06 Martin Jambor <mjambor@suse.cz>
9767 PR tree-optimization/82078
9768 * tree-sra.c (sort_and_splice_var_accesses): Move call to
9769 add_access_to_work_queue...
9770 (build_accesses_from_assign): ...here.
9771 (propagate_all_subaccesses): Make sure racc is the group
9772 representative, if there is one.
9774 2017-09-06 Jakub Jelinek <jakub@redhat.com>
9777 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
9780 2017-09-06 Richard Biener <rguenther@suse.de>
9782 * gimple-ssa-strength-reduction.c
9783 (find_candidates_dom_walker::before_doom_children): Use a
9784 type and not a mode check.
9786 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
9789 * config/arm/predicates.md (arm_general_adddi_operand): Create new
9791 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
9793 2017-09-05 Jeff Law <law@redhat.com>
9795 PR tree-optimization/64910
9796 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
9797 cases where we have 3 or more operands.
9799 2017-09-05 Jakub Jelinek <jakub@redhat.com>
9802 * omp-low.c (lower_omp_for): Recompute tree invariant if
9803 gimple_omp_for_initial/final is ADDR_EXPR.
9806 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
9807 into gimple val before gimplification fo the COND_EXPR.
9809 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
9811 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
9812 REGION_COPY argument.
9813 (thread_through_all_blocks): Remove unused argument to
9814 duplicate_thread_path.
9816 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9817 Alan Hayward <alan.hayward@arm.com>
9818 David Sherwood <david.sherwood@arm.com>
9820 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
9821 Take a scalar_mode rather than a machine_mode.
9822 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9823 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
9824 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9825 (aarch64_gen_adjusted_ldpstp): Likewise.
9826 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
9828 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9829 Alan Hayward <alan.hayward@arm.com>
9830 David Sherwood <david.sherwood@arm.com>
9832 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
9833 Take a scalar_int_mode instead of a machine_mode.
9834 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9835 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9836 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9837 (aarch64_simd_attr_length_rglist): Delete.
9838 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
9839 a scalar_int_mode instead of a machine_mode.
9840 (aarch64_add_offset): Likewise.
9841 (aarch64_internal_mov_immediate): Likewise
9842 (aarch64_add_constant_internal): Likewise.
9843 (aarch64_add_constant): Likewise.
9844 (aarch64_movw_imm): Likewise.
9845 (aarch64_rtx_arith_op_extract_p): Likewise.
9846 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9847 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9848 Remove assert that the mode isn't a vector.
9849 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9850 (aarch64_expand_mov_immediate): Update calls after above changes.
9851 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
9852 (aarch64_and_bitmask_imm): Check for scalar integer modes.
9853 (aarch64_move_imm): Likewise.
9854 (aarch64_can_const_movi_rtx_p): Likewise.
9855 (aarch64_strip_extend): Likewise.
9856 (aarch64_extr_rtx_p): Likewise.
9857 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
9858 a CONST_INT when the mode parameter is VOIDmode.
9859 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
9861 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9863 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
9864 * stor-layout.c (bitwise_mode_for_mode): Likewise.
9865 (bitwise_type_for_mode): Update accordingly.
9867 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9869 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
9870 * stor-layout.c (mode_for_size_tree): Likewise.
9871 (mode_for_array): Update accordingly.
9872 (layout_decl): Likewise.
9873 (compute_record_mode): Likewise. Only set the mode once.
9875 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9877 * target.def (get_mask_mode): Change return type to opt_mode.
9879 * doc/tm.texi: Regenerate.
9880 * targhooks.h (default_get_mask_mode): Return an opt_mode.
9881 * targhooks.c (default_get_mask_mode): Likewise.
9882 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
9883 * optabs-query.c (can_vec_mask_load_store_p): Update use of
9884 targetm.get_mask_mode.
9885 * tree.c (build_truth_vector_type): Likewise.
9887 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9889 * machmode.h (mode_for_vector): Return an opt_mode.
9890 * stor-layout.c (mode_for_vector): Likewise.
9891 (mode_for_int_vector): Update accordingly.
9892 (layout_type): Likewise.
9893 * config/i386/i386.c (emit_memmov): Likewise.
9894 (ix86_expand_set_or_movmem): Likewise.
9895 (ix86_expand_vector_init): Likewise.
9896 (ix86_get_mask_mode): Likewise.
9897 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
9899 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
9900 * expmed.c (extract_bit_field_1): Likewise.
9901 * expr.c (expand_expr_real_2): Likewise.
9902 * optabs-query.c (can_vec_perm_p): Likewise.
9903 (can_vec_mask_load_store_p): Likewise.
9904 * optabs.c (expand_vec_perm): Likewise.
9905 * targhooks.c (default_get_mask_mode): Likewise.
9906 * tree-vect-stmts.c (vectorizable_store): Likewise.
9907 (vectorizable_load): Likewise.
9908 (get_vectype_for_scalar_type_and_size): Likewise.
9910 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9912 * machmode.h (mode_for_int_vector): New function.
9913 * stor-layout.c (mode_for_int_vector): Likewise.
9914 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
9915 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
9916 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
9917 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
9918 (s390_expand_vcond): Likewise.
9920 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9922 * machmode.h (opt_machine_mode): New type.
9923 (opt_mode<T>): Allow construction from anything that can be
9925 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
9926 (mode_for_size): Return an opt_machine_mode.
9927 * stor-layout.c (mode_for_size): Likewise.
9928 (mode_for_size_tree): Update call accordingly.
9929 (bitwise_mode_for_mode): Likewise.
9930 (make_fract_type): Likewise.
9931 (make_accum_type): Likewise.
9932 * caller-save.c (replace_reg_with_saved_mem): Update call
9934 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9935 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9936 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9937 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9938 * expmed.c (extract_bit_field_1): Likewise.
9939 * reload.c (get_secondary_mem): Likewise.
9940 * varasm.c (assemble_integer): Likewise.
9941 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
9944 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9946 * machmode.h (decimal_float_mode_for_size): New function.
9947 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
9948 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
9949 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
9950 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
9951 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
9952 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
9954 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9956 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
9957 (get_builtin_sync_mode): Likewise.
9958 (expand_ifn_atomic_compare_exchange): Likewise.
9959 (expand_builtin_atomic_clear): Likewise.
9960 (expand_builtin_atomic_test_and_set): Likewise.
9961 (fold_builtin_atomic_always_lock_free): Likewise.
9962 * calls.c (compute_argument_addresses): Likewise.
9963 (emit_library_call_value_1): Likewise.
9964 (store_one_arg): Likewise.
9965 * combine.c (combine_instructions): Likewise.
9966 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
9967 * config/arm/arm.c (arm_function_value): Likewise.
9968 (aapcs_allocate_return_reg): Likewise.
9969 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
9970 * config/i386/i386.c (construct_container): Likewise.
9971 (ix86_gimplify_va_arg): Likewise.
9972 (ix86_expand_sse_cmp): Likewise.
9973 (emit_memmov): Likewise.
9974 (emit_memset): Likewise.
9975 (expand_small_movmem_or_setmem): Likewise.
9976 (ix86_expand_pextr): Likewise.
9977 (ix86_expand_pinsr): Likewise.
9978 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
9979 * config/microblaze/microblaze.c (microblaze_block_move_straight):
9981 * config/mips/mips.c (mips_function_value_1) Likewise.
9982 (mips_block_move_straight): Likewise.
9983 (mips_expand_ins_as_unaligned_store): Likewise.
9984 * config/powerpcspe/powerpcspe.c
9985 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9986 (rs6000_darwin64_record_arg_flush): Likewise.
9987 * config/rs6000/rs6000.c
9988 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9989 (rs6000_darwin64_record_arg_flush): Likewise.
9990 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
9991 (sparc_function_value_1): Likewise.
9992 * config/spu/spu.c (adjust_operand): Likewise.
9993 (spu_emit_branch_or_set): Likewise.
9994 (arith_immediate_p): Likewise.
9995 * emit-rtl.c (gen_lowpart_common): Likewise.
9996 * expr.c (expand_expr_real_1): Likewise.
9997 * function.c (assign_parm_setup_block): Likewise.
9998 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9999 * reload1.c (alter_reg): Likewise.
10000 * stor-layout.c (mode_for_vector): Likewise.
10001 (layout_type): Likewise.
10003 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10005 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
10006 (spu_convert_move): Likewise.
10007 * lower-subreg.c (resolve_simple_move): Likewise.
10009 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10012 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
10013 define_insn to a define_expand.
10014 (altivec_vsum2sws_direct): New define_insn.
10015 (altivec_vsumsws): Convert from a define_insn to a define_expand.
10017 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10019 * config/arm/arm.c (arm_option_params_internal): Improve setting of
10022 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
10027 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
10028 realign stack if stack alignment needed is less than incoming
10031 2017-09-05 Marek Polacek <polacek@redhat.com>
10034 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
10037 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10039 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
10041 2017-09-05 Richard Biener <rguenther@suse.de>
10043 PR tree-optimization/82084
10044 * fold-const.c (can_native_encode_string_p): Handle wide characters.
10046 2017-09-05 Richard Biener <rguenther@suse.de>
10048 PR tree-optimization/82102
10049 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
10051 2017-09-05 Martin Liska <mliska@suse.cz>
10053 PR tree-optimization/82032
10054 * tree-cfg.c (generate_range_test): New function.
10055 * tree-cfg.h (generate_range_test): Declared here.
10056 * tree-cfgcleanup.c (convert_single_case_switch): New function.
10057 (cleanup_control_expr_graph): Use it.
10058 * tree-switch-conversion.c (try_switch_expansion): Remove
10060 (emit_case_nodes): Use generate_range_test.
10062 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10065 * config/i386/i386.md (*<btsc><mode>_mask): Add
10066 TARGET_USE_BT to insn constraint.
10067 (*btr<mode>_mask): Ditto.
10069 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
10071 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
10072 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
10074 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
10077 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
10078 TARGET_NEON and TARGET_IWMMXT.
10079 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
10080 TARGET_NEON and TARGET_IWMMXT.
10081 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
10083 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10085 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
10086 (ix86_rewrite_tls_address): Ditto.
10087 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
10088 (ix86_rewrite_tls_address_1): Ditto.
10089 (ix86_rewrite_tls_address): Ditto.
10090 * config/i386/predicates.md (tls_address_pattern): New predicate.
10091 * config/i386/i386.md (TLS address splitter): New splitter.
10093 2017-09-04 Richard Biener <rguenther@suse.de>
10095 PR tree-optimization/82084
10096 * fold-const.h (can_native_encode_string_p): Declare.
10097 * fold-const.c (can_native_encode_string_p): Factor out from ...
10098 (native_encode_string): ... here.
10099 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
10100 vectorizing stores from constants we later cannot handle.
10102 2017-09-04 Marek Polacek <polacek@redhat.com>
10105 * doc/invoke.texi: Update -Wtautological-compare documentation.
10107 2017-09-04 Jeff Law <law@redhat.com>
10109 PR tree-optimization/64910
10110 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
10111 swap the first and last operand if the last is a constant.
10113 2017-09-04 Marek Polacek <polacek@redhat.com>
10116 * convert.c (do_narrow): When sanitizing signed integer overflows,
10117 bail out for signed types.
10118 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
10120 2017-09-04 Richard Biener <rguenther@suse.de>
10122 PR tree-optimization/82060
10123 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
10124 Move devirtualization after stmt folding and before EH/AB/noreturn
10125 cleanup to get the stmt refs canonicalized. Use a bool instead
10126 of gimple_modified_p since that doesn't work for NOPs. Schedule
10127 NOPs generated by folding for removal.
10129 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10130 Alan Hayward <alan.hayward@arm.com>
10131 David Sherwood <david.sherwood@arm.com>
10133 * coretypes.h (pad_direction): New enum.
10134 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
10135 (FUNCTION_ARG_PADDING): Likewise.
10136 * target.def (function_arg_padding): New hook.
10137 * targhooks.h (default_function_arg_padding): Declare.
10138 * targhooks.c (default_function_arg_padding): New function.
10139 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
10140 (TARGET_FUNCTION_ARG_PADDING): ...this.
10141 * doc/tm.texi: Regenerate.
10142 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
10143 instead of direction.
10144 (compute_argument_addresses): Likewise.
10145 (load_register_parameters): Likewise.
10146 (emit_library_call_value_1): Likewise.
10147 (store_one_arg): Use targetm.calls.function_arg_padding instead
10148 of FUNCTION_ARG_PADDING.
10149 (must_pass_in_stack_var_size_or_pad): Likewise.
10150 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
10151 (emit_group_store): Likewise.
10152 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
10153 instead of FUNCTION_ARG_PADDING.
10154 (emit_push_insn): Likewise, and propagate enum change throughout
10156 * function.h (direction): Delete.
10157 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
10159 * function.c (assign_parm_find_stack_rtl): Likewise.
10160 (assign_parm_setup_block_p): Likewise.
10161 (assign_parm_setup_block): Likewise.
10162 (gimplify_parameters): Likewise.
10163 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
10164 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
10166 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
10167 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10168 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
10169 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
10170 (aarch64_function_arg_padding): ...this new function.
10171 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
10172 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10173 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
10174 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10175 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
10176 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10177 (arm_pad_arg_upward): Replace with...
10178 (arm_function_arg_padding): ...this new function.
10179 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10181 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10182 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10183 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10184 (ia64_hpux_function_arg_padding): Replace with...
10185 (ia64_function_arg_padding): ...this new function. Use pad_direction
10186 instead of direction. Check for TARGET_HPUX.
10187 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10188 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10189 (iq2000_function_arg_padding): New function.
10190 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10191 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10192 (mips_function_arg_padding): ...this new function.
10193 (mips_pad_reg_upward): Update accordingly.
10194 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10195 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10196 targetm.calls.function_arg_padding.
10197 (FUNCTION_ARG_PADDING): Delete.
10198 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10199 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10200 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10201 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10202 (nios2_block_reg_padding): Return pad_direction instead of direction.
10203 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10204 instead of direction.
10205 (nios2_function_arg_padding): Likewise. Make static.
10206 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10207 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10208 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10209 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10210 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10211 (pa_function_arg_padding): Make static. Return pad_direction instead
10213 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10214 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10215 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10216 instead of direction. Use targetm.calls.function_arg_padding.
10217 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10218 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10219 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10220 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10221 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10223 (function_arg_padding): Rename to...
10224 (rs6000_function_arg_padding): ...this. Make static. Return
10225 pad_direction instead of direction.
10226 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10227 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10228 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10229 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10230 instead of direction. Use targetm.calls.function_arg_padding.
10231 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10232 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10233 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10234 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10235 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10236 (function_arg_padding): Rename to...
10237 (rs6000_function_arg_padding): ...this. Make static. Return
10238 pad_direction instead of direction.
10239 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10240 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10241 * config/s390/s390.c (s390_function_arg_padding): New function.
10242 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10243 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10244 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10245 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10246 (function_arg_padding): Rename to...
10247 (sparc_function_arg_padding): ...this. Make static. Return
10248 pad_direction instead of direction.
10249 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10250 * config/spu/spu.c (spu_function_arg_padding): New function.
10251 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10252 * system.h (FUNCTION_ARG_PADDING): Poison.
10254 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10255 Alan Hayward <alan.hayward@arm.com>
10256 David Sherwood <david.sherwood@arm.com>
10258 * target.def (modes_tieable_p): New hook.
10259 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10260 (TARGET_MODES_TIEABLE_P): ...this.
10261 * doc/tm.texi.in: Regenerate.
10262 * hooks.h (hook_bool_mode_mode_true): Declare.
10263 * hooks.c (hook_bool_mode_mode_true): New function.
10264 * combine.c (subst): Use targetm.modes_tieable_p instead of
10266 * dse.c (find_shift_sequence): Likewise.
10267 * expmed.c (extract_low_bits): Likewise.
10268 * lower-subreg.c: Include target.h.
10269 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10271 * rtlanal.c (rtx_cost): Likewise.
10272 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10273 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10274 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10275 (TARGET_MODES_TIEABLE_P): Redefine.
10276 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10277 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10278 (TARGET_MODES_TIEABLE_P): Redefine.
10279 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10280 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10281 (arc_modes_tieable_p): New function.
10282 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10283 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10284 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10285 (arm_modes_tieable_p): Make static.
10286 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10287 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10288 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10289 (TARGET_MODES_TIEABLE_P): Redefine.
10290 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10291 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10292 (TARGET_MODES_TIEABLE_P): Redefine.
10293 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10294 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10295 (cr16_modes_tieable_p): New function.
10296 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10297 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10298 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10299 (TRULY_NOOP_TRUNCATION): Update comment.
10300 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10301 (TRULY_NOOP_TRUNCATION): Update comment.
10302 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10303 (frv_modes_tieable_p): New function.
10304 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10305 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10306 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10307 (TARGET_MODES_TIEABLE_P): Redefine.
10308 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10309 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10310 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10311 (TARGET_MODES_TIEABLE_P): Redefine.
10312 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10313 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10314 (ia64_modes_tieable_p): New function.
10315 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10316 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10317 (iq2000_modes_tieable_p): New function.
10318 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10319 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10320 (lm32_modes_tieable_p): New function.
10321 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10322 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10323 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10324 (TARGET_MODES_TIEABLE_P): Redefine.
10325 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10326 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10327 (m32r_modes_tieable_p): New function.
10328 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10329 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10330 (m68k_modes_tieable_p): New function.
10331 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10332 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10333 (mcore_modes_tieable_p): New function.
10334 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10335 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10337 (TARGET_MODES_TIEABLE_P): Redefine.
10338 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10339 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10340 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10341 (TARGET_MODES_TIEABLE_P): Redefine.
10342 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10343 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10344 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10345 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10346 (mn10300_modes_tieable_p): ...this and make static.
10347 (TARGET_MODES_TIEABLE_P): Redefine.
10348 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10349 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10350 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10351 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10352 (msp430_modes_tieable_p): Make static.
10353 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10354 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10355 (TARGET_MODES_TIEABLE_P): Redefine.
10356 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10357 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10358 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10359 (TARGET_MODES_TIEABLE_P): Redefine.
10360 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10361 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10362 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10363 (TARGET_MODES_TIEABLE_P): Redefine.
10364 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10365 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10366 (pdp11_modes_tieable_p): New function.
10367 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10368 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10369 (rs6000_modes_tieable_p): New function.
10370 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10371 * config/powerpcspe/powerpcspe.md: Update comment.
10372 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10373 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10374 (TARGET_MODES_TIEABLE_P): Redefine.
10375 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10376 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10377 (rl78_modes_tieable_p): New function.
10378 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10379 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10380 (rs6000_modes_tieable_p): New function.
10381 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10382 * config/rs6000/rs6000.md: Update comment.
10383 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10384 * config/rx/rx.c (rx_modes_tieable_p): New function.
10385 (TARGET_MODES_TIEABLE_P): Redefine.
10386 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10387 * config/s390/s390.c (s390_modes_tieable_p): New function.
10388 (TARGET_MODES_TIEABLE_P): Redefine.
10389 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10390 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10391 (sh_modes_tieable_p): New function.
10392 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10393 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10394 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10395 (sparc_modes_tieable_p): Make static.
10396 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10397 * config/spu/spu.c (spu_modes_tieable_p): New function.
10398 (TARGET_MODES_TIEABLE_P): Redefine.
10399 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10400 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10401 (TARGET_MODES_TIEABLE_P): Redefine.
10402 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10403 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10404 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10405 * config/v850/v850.c (v850_modes_tieable_p): New function.
10406 (TARGET_MODES_TIEABLE_P): Redefine.
10407 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10408 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10409 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10410 (visium_modes_tieable_p): New function.
10411 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10412 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10413 (xtensa_modes_tieable_p): New function.
10414 * system.h (MODES_TIEABLE_P): Poison.
10416 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10417 Alan Hayward <alan.hayward@arm.com>
10418 David Sherwood <david.sherwood@arm.com>
10420 * target.def (hard_regno_mode_ok): New hook.
10421 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10422 (TARGET_HARD_REGNO_MODE_OK): ...this.
10423 * doc/tm.texi.in: Regenerate.
10424 * hooks.h (hook_bool_uint_mode_true): Declare.
10425 * hooks.c (hook_bool_uint_mode_true): New function.
10426 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10427 HARD_REGNO_MODE_OK.
10428 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10429 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10430 instead of HARD_REGNO_MODE_OK.
10431 * caller-save.c: Include target.h.
10432 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10433 HARD_REGNO_MODE_OK.
10434 * combine.c (can_combine_p): Likewise.
10435 (combinable_i3pat): Likewise.
10436 (can_change_dest_mode): Likewise.
10437 * expr.c (init_expr_target): Likewise.
10438 (convert_move): Likewise.
10439 (convert_modes): Likewise.
10440 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10441 (setup_prohibited_mode_move_regs): Likewise.
10442 * ira.h (target_ira): Likewise.
10443 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10444 * lra-constraints.c (process_alt_operands): Likewise.
10445 (split_reg): Likewise.
10446 * recog.c (peep2_find_free_register): Likewise.
10447 * ree.c (combine_reaching_defs): Likewise.
10448 * regcprop.c (maybe_mode_change): Likewise.
10449 * reginfo.c (init_reg_sets_1): Likewise.
10450 (choose_hard_reg_mode): Likewise.
10451 (simplifiable_subregs): Likewise.
10452 * regrename.c (check_new_reg_p): Likewise.
10453 * reload.c (find_valid_class): Likewise.
10454 (find_valid_class_1): Likewise.
10455 (reload_inner_reg_of_subreg): Likewise.
10456 (push_reload): Likewise.
10457 (combine_reloads): Likewise.
10458 (find_dummy_reload): Likewise.
10459 (find_reloads): Likewise.
10460 * reload1.c (find_reg): Likewise.
10461 (set_reload_reg): Likewise.
10462 (allocate_reload_reg): Likewise.
10463 (choose_reload_regs): Likewise.
10464 (reload_adjust_reg_for_temp): Likewise.
10465 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10466 (simplify_subreg_regno): Likewise.
10467 * sel-sched.c (init_regs_for_mode): Likewise.
10468 * varasm.c (make_decl_rtl): Likewise.
10469 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10470 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
10471 HARD_REGNO_MODE_OK.
10472 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
10473 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
10474 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10475 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
10476 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
10477 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10478 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
10479 (arc_mode_class): Delete.
10480 (HARD_REGNO_MODE_OK): Delete.
10481 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10482 (arc_hard_regno_mode_ok): Rename old array to...
10483 (arc_hard_regno_mode_ok_modes): ...this.
10484 (arc_conditional_register_usage): Update accordingly.
10485 (arc_mode_class): Make static.
10486 (arc_hard_regno_mode_ok): New function.
10487 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
10488 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
10489 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10490 (arm_hard_regno_mode_ok): Make static.
10491 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
10492 HARD_REGNO_MODE_OK.
10493 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
10494 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
10495 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
10497 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10498 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
10499 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
10500 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
10501 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
10502 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10503 * config/bfin/predicates.md (valid_reg_operand): Use
10504 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
10505 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
10506 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
10507 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10508 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
10509 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
10510 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10511 (cr16_hard_regno_mode_ok): Make static and return a bool.
10512 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
10513 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10514 (cris_hard_regno_mode_ok): New function.
10515 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
10516 (epiphany_mode_class): Delete.
10517 (HARD_REGNO_MODE_OK): Delete.
10518 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
10519 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10520 (hard_regno_mode_ok): Rename to...
10521 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
10522 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
10523 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
10524 HARD_REGNO_MODE_OK.
10525 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
10526 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
10527 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10528 (frv_hard_regno_mode_ok): Make static and return a bool.
10529 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
10530 HARD_REGNO_MODE_OK.
10531 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
10532 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
10533 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
10534 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
10536 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10537 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
10538 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
10539 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
10541 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10542 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
10543 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10544 (ia64_hard_regno_mode_ok): New function.
10545 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
10546 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10547 (iq2000_hard_regno_mode_ok): New function.
10548 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
10549 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10550 (lm32_hard_regno_mode_ok): New function.
10551 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
10552 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
10553 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
10554 instead of HARD_REGNO_MODE_OK.
10555 (m32c_hard_regno_ok): Rename to...
10556 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
10557 (m32c_cannot_change_mode_class): Update accordingly.
10558 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10559 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
10560 (m32r_mode_class): Delete.
10561 (HARD_REGNO_MODE_OK): Delete.
10562 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10563 (m32r_hard_regno_mode_ok): Rename to...
10564 (m32r_hard_regno_modes): ...this.
10565 (m32r_mode_class): Make static.
10566 (m32r_hard_regno_mode_ok): New function.
10567 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
10568 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
10569 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10570 (m68k_hard_regno_mode_ok): Make static.
10571 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
10572 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10573 (mcore_hard_regno_mode_ok): New function.
10574 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
10575 (HARD_REGNO_MODE_OK): Delete.
10576 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
10578 (microblaze_hard_regno_mode_ok_p): ...this and make static.
10579 (microblaze_hard_regno_mode_ok): New function.
10580 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10581 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
10582 (mips_hard_regno_mode_ok): Delete.
10583 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
10584 (mips_hard_regno_mode_ok_p): ...this and make static.
10585 (mips_hard_regno_mode_ok_p): Rename to...
10586 (mips_hard_regno_mode_ok_uncached): ...this.
10587 (mips_hard_regno_mode_ok): New function.
10588 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
10589 of HARD_REGNO_MODE_OK.
10590 (mips_option_override): Update after above name changes.
10591 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10592 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
10593 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
10594 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
10595 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
10596 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10597 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
10598 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
10599 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
10600 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10601 (msp430_hard_regno_mode_ok): Make static and return a bool.
10602 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
10603 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
10604 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
10606 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10607 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
10608 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
10609 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
10610 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
10611 (PA_HARD_REGNO_MODE_OK): ...this
10612 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
10613 (PA_HARD_REGNO_MODE_OK): ...this.
10614 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10615 (pa_hard_regno_mode_ok): New function.
10616 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
10617 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10618 (pdp11_hard_regno_mode_ok): New function.
10619 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
10620 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
10622 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
10624 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10625 (rs6000_hard_regno_mode_ok): Rename to...
10626 (rs6000_hard_regno_mode_ok_uncached): ...this.
10627 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10628 (rs6000_hard_regno_mode_ok): New function.
10629 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
10630 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
10631 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
10632 (riscv_hard_regno_mode_ok): ...this and make static.
10633 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10634 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
10635 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
10636 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10637 (rl78_hard_regno_mode_ok): Make static and return bool.
10638 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
10639 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
10641 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
10642 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10643 (rs6000_hard_regno_mode_ok): Rename to...
10644 (rs6000_hard_regno_mode_ok_uncached): ...this.
10645 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10646 (rs6000_hard_regno_mode_ok): New function.
10647 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
10648 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
10649 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10650 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
10651 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
10652 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
10653 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10654 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
10655 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
10656 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10657 (sh_hard_regno_mode_ok): Make static.
10658 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
10659 instead of HARD_REGNO_MODE_OK.
10660 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
10661 (sparc_mode_class): Delete.
10662 (HARD_REGNO_MODE_OK): Delete.
10663 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10664 (hard_regno_mode_classes): Make static.
10665 (sparc_mode_class): Likewise.
10666 (sparc_hard_regno_mode_ok): New function.
10667 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
10668 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
10669 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
10671 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10672 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
10673 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
10674 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
10675 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
10676 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10677 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
10678 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
10679 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10680 (visium_hard_regno_mode_ok): New function.
10681 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
10682 instead of HARD_REGNO_MODE_OK.
10683 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
10684 (HARD_REGNO_MODE_OK): Delete.
10685 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
10686 (xtensa_hard_regno_mode_ok_p): ...this and make static.
10687 (xtensa_option_override): Update accordingly.
10688 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10689 (xtensa_hard_regno_mode_ok): New function.
10690 * system.h (HARD_REGNO_MODE_OK): Poison.
10692 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10693 Alan Hayward <alan.hayward@arm.com>
10694 David Sherwood <david.sherwood@arm.com>
10696 * target.def (hard_regno_call_part_clobbered): New hook.
10697 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
10698 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
10699 * doc/tm.texi: Regenerate.
10700 * hooks.h (hook_bool_uint_mode_false): Declare.
10701 * hooks.c (hook_bool_uint_mode_false): New function.
10702 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10703 * cselib.c (cselib_process_insn): Use
10704 targetm.hard_regno_call_part_clobbered instead of
10705 HARD_REGNO_CALL_PART_CLOBBERED.
10706 * ira-conflicts.c (ira_build_conflicts): Likewise.
10707 * ira-costs.c (ira_tune_allocno_costs): Likewise.
10708 * lra-constraints.c (need_for_call_save_p): Likewise.
10709 * lra-lives.c: Include target.h.
10710 (check_pseudos_live_through_calls): Use
10711 targetm.hard_regno_call_part_clobbered instead of
10712 HARD_REGNO_CALL_PART_CLOBBERED.
10713 * regcprop.c: Include target.h.
10714 (copyprop_hardreg_forward_1): Use
10715 targetm.hard_regno_call_part_clobbered instead of
10716 HARD_REGNO_CALL_PART_CLOBBERED.
10717 * reginfo.c (choose_hard_reg_mode): Likewise.
10718 * regrename.c (check_new_reg_p): Likewise.
10719 * reload.c (find_equiv_reg): Likewise.
10720 * reload1.c (emit_reload_insns): Likewise.
10721 * sched-deps.c (deps_analyze_insn): Likewise.
10722 * sel-sched.c (init_regs_for_mode): Likewise.
10723 (mark_unavailable_hard_regs): Likewise.
10724 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
10725 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10726 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
10728 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10729 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10730 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
10732 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
10734 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10735 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10736 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
10738 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10739 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10740 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
10742 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10743 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
10745 * config/powerpcspe/powerpcspe.c
10746 (rs6000_hard_regno_call_part_clobbered): New function.
10747 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10748 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10749 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
10751 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10752 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10753 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
10755 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10756 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10757 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
10759 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10760 Alan Hayward <alan.hayward@arm.com>
10761 David Sherwood <david.sherwood@arm.com>
10763 * rtl.h (subreg_memory_offset): Declare.
10764 * emit-rtl.c (subreg_memory_offset): New function.
10765 * expmed.c (store_bit_field_1): Use it.
10766 * expr.c (undefined_operand_subword_p): Likewise.
10767 * simplify-rtx.c (simplify_subreg): Likewise.
10769 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
10771 PR rtl-optimization/57448
10774 * optabs.c (expand_atomic_load): Place compiler memory barriers if
10775 using atomic_load pattern.
10776 (expand_atomic_store): Likewise.
10778 2017-09-04 Jakub Jelinek <jakub@redhat.com>
10781 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
10782 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
10783 handling. Use replace_call_with_value with NULL instead of
10784 gsi_replace, unlink_stmt_vdef and release_defs.
10786 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
10789 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
10791 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10794 * rtl.h (emit_library_call_value_1): Declare.
10795 (emit_library_call): Replace declaration with a series of overloads.
10796 Remove the parameter count argument.
10797 (emit_library_call_value): Likewise.
10798 * calls.c (emit_library_call_value_1): Make global. Replace varargs
10799 with an "rtx_mode_t *".
10800 (emit_library_call_value): Delete.
10801 (emit_library_call): Likewise.
10802 * asan.c (asan_emit_stack_protection): Update calls accordingly.
10803 (asan_emit_allocas_unpoison): Likewise.
10804 * builtins.c (expand_builtin_powi): Likewise.
10805 (expand_asan_emit_allocas_unpoison): Likewise.
10806 * cfgexpand.c (expand_main_function): Likewise.
10807 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
10808 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
10809 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
10810 * config/arm/arm.c (arm_trampoline_init): Likewise.
10811 (arm_call_tls_get_addr): Likewise.
10812 (arm_expand_divmod_libfunc): Likewise.
10813 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
10814 (smulsi3_highpart): Likewise.
10815 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
10816 (c6x_expand_compare): Likewise.
10817 (c6x_expand_movmem): Likewise.
10818 * config/frv/frv.c (frv_trampoline_init): Likewise.
10819 * config/i386/i386.c (ix86_trampoline_init): Likewise.
10820 (ix86_expand_divmod_libfunc): Likewise.
10821 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
10822 (ia64_expand_compare): Likewise.
10823 (ia64_profile_hook): Likewise.
10824 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
10825 (nonlocal_goto): Likewise.
10826 (restore_stack_nonlocal): Likewise.
10827 * config/m32r/m32r.c (block_move_call): Likewise.
10828 (m32r_trampoline_init): Likewise.
10829 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
10830 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
10831 (m68k_call_m68k_read_tp): Likewise.
10832 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
10833 (microblaze_expand_divide): Likewise.
10834 * config/mips/mips.h (mips_args): Likewise.
10835 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
10836 (MIPS_ICACHE_SYNC): Likewise.
10837 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
10838 (nios2_trampoline_init): Likewise.
10839 * config/pa/pa.c (hppa_tls_call): Likewise.
10840 (pa_trampoline_init): Likewise.
10841 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
10842 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
10843 (expand_strn_compare): Likewise.
10844 (rs6000_generate_compare): Likewise.
10845 (rs6000_expand_float128_convert): Likewise.
10846 (output_profile_hook): Likewise.
10847 (rs6000_trampoline_init): Likewise.
10848 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
10849 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
10850 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
10851 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
10852 (rs6000_generate_compare): Likewise.
10853 (rs6000_expand_float128_convert): Likewise.
10854 (output_profile_hook): Likewise.
10855 (rs6000_trampoline_init): Likewise.
10856 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
10857 * config/sh/sh.c (sh_trampoline_init): Likewise.
10858 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
10859 (sparc_emit_float_lib_cmp): Likewise.
10860 (sparc32_initialize_trampoline): Likewise.
10861 (sparc64_initialize_trampoline): Likewise.
10862 (sparc_profile_hook): Likewise.
10863 * config/spu/spu.c (ea_load_store): Likewise.
10864 * config/spu/spu.md (floatunssidf2): Likewise.
10865 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
10866 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
10867 * config/visium/visium.c (expand_block_move_4): Likewise.
10868 (expand_block_move_2): Likewise.
10869 (expand_block_move_1): Likewise.
10870 (expand_block_set_4): Likewise.
10871 (expand_block_set_2): Likewise.
10872 (expand_block_set_1): Likewise.
10873 (visium_trampoline_init): Likewise.
10874 (visium_profile_hook): Likewise.
10875 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
10876 (xtensa_setup_frame_addresses): Likewise.
10877 (xtensa_trampoline_init): Likewise.
10878 * except.c (sjlj_emit_function_enter): Likewise.
10879 (sjlj_emit_function_exit): Likewise.
10880 * explow.c (allocate_dynamic_stack_space): Likewise.
10881 (probe_stack_range): Likewise.
10882 * expr.c (convert_mode_scalar): Likewise.
10883 * optabs.c (expand_binop): Likewise.
10884 (expand_twoval_binop_libfunc): Likewise.
10885 (expand_unop): Likewise.
10886 (prepare_cmp_insn): Likewise.
10887 (prepare_float_lib_cmp): Likewise.
10888 (expand_float): Likewise.
10889 (expand_fix): Likewise.
10890 (expand_fixed_convert): Likewise.
10891 (maybe_emit_sync_lock_test_and_set): Likewise.
10892 (expand_atomic_compare_and_swap): Likewise.
10893 (expand_mem_thread_fence): Likewise.
10894 (expand_atomic_fetch_op): Likewise.
10896 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
10898 * doc/generic.texi (OpenACC): Adjust URL.
10899 * doc/invoke.texi (C Dialect Options): Ditto.
10901 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
10903 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
10904 predicate for operand 1. Add (m,<S>) constraint.
10905 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
10906 Prevent memory operand 1 with register operand 2.
10908 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
10910 PR rtl-optimization/82024
10911 * combine.c (try_combine): If the combination result is a PARALLEL,
10912 and we only need to retain the SET in there that would be placed
10913 at I2, check that we can place that at I3 instead, before doing so.
10915 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10918 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
10920 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
10921 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
10924 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
10925 Jeff Law <law@redhat.com>
10927 * varasm.c (bss_initializer_p): Do not put constants into .bss
10928 (categorize_decl_for_section): Handle bss_initializer_p returning
10929 false when DECL_INITIAL is NULL.
10931 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10934 * config/s390/s390.c (s390_can_inline_p): New function.
10936 2017-09-01 Jeff Law <law@redhat.com>
10938 PR tree-optimization/82052
10939 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
10940 Always initialize the returned slot after a hash table miss
10941 when INSERT is true.
10943 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
10945 * config/s390/s390.md (mem_signal_fence): Remove.
10946 * doc/md.texi (mem_signal_fence): Remove.
10947 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
10949 * target-insns.def (mem_signal_fence): Remove.
10951 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10954 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
10957 * asan.c (create_odr_indicator): Strip name encoding from assembler
10958 name before appending it after __odr_asan_.
10960 2017-09-01 Martin Liska <mliska@suse.cz>
10962 PR tree-optimization/82059
10963 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
10964 frequency only when an edge is redirected.
10966 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10968 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
10969 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
10970 (arc_conditional_register_usage): Remove ARC600 lp_count
10972 (arc_file_start): Emit Tag_ARC_CPU_variation.
10973 (arc_can_use_doloop_p): New conditions to use ZOLs.
10974 (hwloop_fail): New function.
10975 (hwloop_optimize): Likewise.
10976 (hwloop_pattern_reg): Likewise.
10977 (arc_doloop_hooks): New struct, to be used with reorg_loops.
10978 (arc_reorg_loops): New function, calls reorg_loops.
10979 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
10980 (arc600_corereg_hazard): Remove ZOL checking, case handled by
10982 (arc_loop_hazard): Remove function, functionality moved into
10984 (arc_hazard): Remove arc_loop_hazard call.
10985 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
10986 into hwloop_optimize.
10987 (arc_label_align): Remove ZOL handling.
10988 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
10989 * config/arc/arc.md (doloop_begin): Remove pattern.
10990 (doloop_begin_i): Likewise.
10991 (doloop_end_i): Likewise.
10992 (doloop_fallback): Likewise.
10993 (doloop_fallback_m): Likewise.
10994 (doloop_end): Reimplement expand.
10995 (arc_lp): New pattern for LP instruction.
10996 (loop_end): New pattern.
10997 (loop_fail): Likewise.
10998 (decrement_and_branch_until_zero): Likewise.
10999 * config/arc/arc.opt (mlpc-width): New option.
11000 * doc/invoke.texi (mlpc-width): Document option.
11002 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11004 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
11005 (arc_ccfsm_advance): Fix checking for delay slots.
11006 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
11008 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11010 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11012 (movhi_insn): Update store instruction constraint which are saving
11013 6-bit short immediates.
11014 (movsi_insn): Consider also short scaled load operations.
11015 (zero_extendhisi2_i): Use Usd constraint instead of T.
11016 (extendhisi2_i): Add q constraint.
11017 (arc_clzsi2): Add type and length attributes.
11018 (arc_ctzsi2): Likewise.
11019 * config/arc/constraints.md (Usc): Update constraint, the
11020 assembler can parse two relocations for a single instruction.
11022 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11024 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
11025 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
11027 2017-08-31 Olivier Hainque <hainque@adacore.com>
11029 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
11030 match as powerpc-wrs-vxworks*.
11032 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
11034 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
11035 register constraint for by-element operand.
11036 (aarch64_mls_elt_merge<mode>): Likewise.
11038 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11040 * config/arc/arc.c (arc_can_follow_jump): Check for short
11043 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11045 * config.gcc: Use g.opt for arc.
11046 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
11047 functionality moved to ...
11048 (legitimate_scaled_address_p): New function, ...here.
11049 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
11050 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
11051 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
11053 (arc_override_options): Handle G option.
11054 (arc_output_pic_addr_const): Correct function definition.
11055 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
11056 (arc_decl_anon_ns_mem_p): Delete.
11057 (arc_in_small_data_p): Overhaul this function to take into
11058 consideration the value given via G option.
11059 (arc_rewrite_small_data_1): Renamed and corrected old
11060 arc_rewrite_small_data function.
11061 (arc_rewrite_small_data): New function.
11062 (small_data_pattern): Don't use pic_offset_table_rtx.
11063 * config/arc/arc.h (CC1_SPEC): Recognize G option.
11064 * config/arc/simdext.md (movmisalignv2hi): Use
11065 prepare_move_operands function.
11067 (movmisalign*): Likewise.
11068 * doc/invoke.texi (ARC options): Document -G option.
11070 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11072 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
11074 * config/arc/arc.c (arc_print_operand): Output scalled address for
11075 sdata whenever is possible.
11076 (arc_in_small_data_p): Allow sdata for 64bit datum when double
11077 load/stores are available.
11078 (compact_sda_memory_operand): Check for the alignment required by
11079 code density instructions.
11080 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
11082 * config/arc/constraints.md (Usd): Update constraint.
11083 (Us0): New constraint.
11084 (Usc): Update constraint.
11086 2017-08-31 Richard Biener <rguenther@suse.de>
11088 PR middle-end/82054
11089 * dwarf2out.c (dwarf2out_early_global_decl): Process each
11090 function only once.
11092 2017-08-31 Tamar Christina <tamar.christina@arm.com>
11094 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
11095 Resize type_signature.
11097 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11098 Alan Hayward <alan.hayward@arm.com>
11099 David Sherwood <david.sherwood@arm.com>
11101 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
11102 subregs whose inner modes can be stored in GPRs.
11103 (aarch64_classify_index): Likewise.
11105 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11106 Alan Hayward <alan.hayward@arm.com>
11107 David Sherwood <david.sherwood@arm.com>
11109 * config/aarch64/iterators.md (V_cmp_result): Rename to...
11110 (V_INT_EQUIV): ...this.
11111 (v_cmp_result): Rename to...
11112 (v_int_equiv): ...this.
11113 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
11114 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
11115 (copysign<mode>3): Likewise.
11116 (aarch64_simd_bsl<mode>_internal): Likewise.
11117 (aarch64_simd_bsl<mode>): Likewise.
11118 (vec_cmp<mode><mode>): Likewise.
11119 (vcond<mode><mode>): Likewise.
11120 (vcond<v_cmp_mixed><mode>): Likewise.
11121 (vcondu<mode><v_cmp_mixed>): Likewise.
11122 (aarch64_cm<optab><mode>): Likewise.
11123 (aarch64_cmtst<mode>): Likewise.
11124 (aarch64_fac<optab><mode>): Likewise.
11125 (vec_perm_const<mode>): Likewise.
11126 (vcond_mask_<mode><v_cmp_result>): Rename to...
11127 (vcond_mask_<mode><v_int_equiv>): ...this.
11128 (vec_cmp<mode><v_cmp_result>): Rename to...
11129 (vec_cmp<mode><v_int_equiv>): ...this.
11131 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11132 Alan Hayward <alan.hayward@arm.com>
11133 David Sherwood <david.sherwood@arm.com>
11135 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
11137 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
11138 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
11139 (UNSPEC_LD4_DREG): New unspecs.
11140 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
11141 (aarch64_ld2<mode>_dreg_be): Replace with...
11142 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
11144 (aarch64_ld3<mode>_dreg_le)
11145 (aarch64_ld3<mode>_dreg_be): Replace with...
11146 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
11148 (aarch64_ld4<mode>_dreg_le)
11149 (aarch64_ld4<mode>_dreg_be): Replace with...
11150 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
11153 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11155 PR tree-optimization/81987
11156 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
11157 insert an initializer in a location not dominated by the stride
11160 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11162 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
11163 on the entire header of the finally block in the fallthru case.
11165 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11167 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
11169 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
11171 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
11172 rs6000_emit_move_from_cr and call renamed function.
11173 (rs6000_emit_prologue): Call renamed functions.
11174 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
11175 movesi_from_cr, remove volatile CRs.
11177 2017-08-30 Jon Beniston <jon@beniston.com>
11178 Richard Biener <rguenther@suse.de>
11180 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11181 of VECTOR_MODE_P check.
11182 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11183 element vector types.
11185 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11187 * df.h (df_read_modify_subreg_p): Remove in favor of...
11188 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11189 const_rtx instead of an rtx.
11190 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11191 * df-problems.c (df_word_lr_mark_ref): Likewise.
11192 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11193 (mark_pseudo_reg_dead): Likewise.
11194 (mark_ref_dead): Likewise.
11195 * reginfo.c (init_subregs_of_mode): Likewise.
11196 * sched-deps.c (sched_analyze_1): Likewise.
11197 * df-scan.c (df_def_record_1): Likewise.
11198 (df_uses_record): Likewise.
11199 (df_read_modify_subreg_p): Remove in favor of...
11200 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11201 const_rtx instead of an rtx.
11203 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11204 Alan Hayward <alan.hayward@arm.com>
11205 David Sherwood <david.sherwood@arm.com>
11207 * rtl.h (partial_subreg_p): New function.
11208 * caller-save.c (save_call_clobbered_regs): Use it.
11209 * calls.c (expand_call): Likewise.
11210 * combine.c (combinable_i3pat): Likewise.
11211 (simplify_set): Likewise.
11212 (make_extraction): Likewise.
11213 (make_compound_operation_int): Likewise.
11214 (gen_lowpart_or_truncate): Likewise.
11215 (force_to_mode): Likewise.
11216 (make_field_assignment): Likewise.
11217 (reg_truncated_to_mode): Likewise.
11218 (record_truncated_value): Likewise.
11219 (move_deaths): Likewise.
11220 * cse.c (record_jump_cond): Likewise.
11221 (cse_insn): Likewise.
11222 * cselib.c (cselib_lookup_1): Likewise.
11223 * expmed.c (extract_bit_field_using_extv): Likewise.
11224 * function.c (assign_parm_setup_reg): Likewise.
11225 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11226 * ira-build.c (create_insn_allocnos): Likewise.
11227 * lra-coalesce.c (merge_pseudos): Likewise.
11228 * lra-constraints.c (match_reload): Likewise.
11229 (simplify_operand_subreg): Likewise.
11230 (curr_insn_transform): Likewise.
11231 * lra-lives.c (process_bb_lives): Likewise.
11232 * lra.c (new_insn_reg): Likewise.
11233 (lra_substitute_pseudo): Likewise.
11234 * regcprop.c (mode_change_ok): Likewise.
11235 (maybe_mode_change): Likewise.
11236 (copyprop_hardreg_forward_1): Likewise.
11237 * reload.c (push_reload): Likewise.
11238 (find_reloads): Likewise.
11239 (find_reloads_subreg_address): Likewise.
11240 * reload1.c (alter_reg): Likewise.
11241 (eliminate_regs_1): Likewise.
11242 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11244 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11246 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11247 back to if statements, including unpack.
11249 2017-08-30 Martin Liska <mliska@suse.cz>
11251 PR inline-asm/82001
11252 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11254 (func_checker::compare_asm_inputs_outputs): ... this function.
11255 (func_checker::compare_gimple_asm): Use the function to compare
11256 also ASM constrains.
11257 * ipa-icf-gimple.h: Rename the function.
11259 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11260 Alan Hayward <alan.hayward@arm.com>
11261 David Sherwood <david.sherwood@arm.com>
11263 * coretypes.h (complex_mode): New type.
11264 * gdbhooks.py (build_pretty_printer): Handle it.
11265 * machmode.h (complex_mode): New class.
11266 (complex_mode::includes_p): New function.
11267 (is_complex_int_mode): Likewise.
11268 (is_complex_float_mode): Likewise.
11269 * genmodes.c (get_mode_class): Handle complex mode classes.
11270 * function.c (expand_function_end): Use is_complex_int_mode.
11272 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11273 Alan Hayward <alan.hayward@arm.com>
11274 David Sherwood <david.sherwood@arm.com>
11276 * coretypes.h (scalar_mode_pod): New typedef.
11277 * gdbhooks.py (build_pretty_printer): Handle it.
11278 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11279 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11280 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11281 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11282 as_a <scalar_mode>.
11284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11285 Alan Hayward <alan.hayward@arm.com>
11286 David Sherwood <david.sherwood@arm.com>
11288 * machmode.h (mode_for_vector): Take a scalar_mode instead
11290 * stor-layout.c (mode_for_vector): Likewise.
11291 * explow.c (promote_mode): Use as_a <scalar_mode>.
11292 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11294 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11295 Alan Hayward <alan.hayward@arm.com>
11296 David Sherwood <david.sherwood@arm.com>
11298 * target.def (preferred_simd_mode): Take a scalar_mode
11299 instead of a machine_mode.
11300 * targhooks.h (default_preferred_simd_mode): Likewise.
11301 * targhooks.c (default_preferred_simd_mode): Likewise.
11302 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11303 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11304 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11305 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11306 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11307 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11308 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11309 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11311 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11312 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11313 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11314 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11315 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11316 * doc/tm.texi: Regenerate.
11317 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11320 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11321 Alan Hayward <alan.hayward@arm.com>
11322 David Sherwood <david.sherwood@arm.com>
11324 * target.def (scalar_mode_supported_p): Take a scalar_mode
11325 instead of a machine_mode.
11326 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11327 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11328 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11329 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11330 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11331 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11332 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11333 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11334 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11335 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11336 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11337 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11338 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11339 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11341 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11342 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11343 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11344 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11345 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11347 * doc/tm.texi: Regenerate.
11349 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11350 Alan Hayward <alan.hayward@arm.com>
11351 David Sherwood <david.sherwood@arm.com>
11353 * coretypes.h (opt_scalar_mode): New typedef.
11354 * gdbhooks.py (build_pretty_printers): Handle it.
11355 * machmode.h (mode_iterator::get_2xwider): Add overload for
11357 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11359 * expr.c (convert_mode_scalar): Likewise.
11360 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11361 * optabs.c (expand_float): Likewise.
11362 (expand_fix): Likewise.
11363 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11365 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11366 Alan Hayward <alan.hayward@arm.com>
11367 David Sherwood <david.sherwood@arm.com>
11369 * optabs.c (expand_float): Explicitly check for scalars before
11370 using a branching expansion.
11371 (expand_fix): Likewise.
11373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11374 Alan Hayward <alan.hayward@arm.com>
11375 David Sherwood <david.sherwood@arm.com>
11377 * expr.c (convert_mode): Split scalar handling out into...
11378 (convert_mode_scalar): ...this new function. Treat the modes
11381 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11382 Alan Hayward <alan.hayward@arm.com>
11383 David Sherwood <david.sherwood@arm.com>
11385 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11387 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
11389 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11390 Alan Hayward <alan.hayward@arm.com>
11391 David Sherwood <david.sherwood@arm.com>
11393 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11394 rather than a machine_mode.
11395 (fixed_from_string): Likewise.
11396 (fixed_convert): Likewise.
11397 (fixed_convert_from_int): Likewise.
11398 (fixed_convert_from_real): Likewise.
11399 (real_convert_from_fixed): Likewise.
11400 * fixed-value.c (fixed_from_double_int): Likewise.
11401 (fixed_from_string): Likewise.
11402 (fixed_convert): Likewise.
11403 (fixed_convert_from_int): Likewise.
11404 (fixed_convert_from_real): Likewise.
11405 (real_convert_from_fixed): Likewise.
11406 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
11408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11409 Alan Hayward <alan.hayward@arm.com>
11410 David Sherwood <david.sherwood@arm.com>
11412 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11413 of separate mode class checks. Do not allow vector modes here.
11414 (immed_wide_int_const): Use as_a <scalar_mode>.
11415 * explow.c (trunc_int_for_mode): Likewise.
11416 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11417 (wi::shwi): Likewise.
11418 (wi::min_value): Likewise.
11419 (wi::max_value): Likewise.
11420 * dwarf2out.c (loc_descriptor): Likewise.
11421 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11422 for CONST_WIDE_INT.
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 * tree.h (SCALAR_TYPE_MODE): New macro.
11429 * expr.c (expand_expr_addr_expr_1): Use it.
11430 (expand_expr_real_2): Likewise.
11431 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11432 (fold_convert_const_fixed_from_int): Likewise.
11433 (fold_convert_const_fixed_from_real): Likewise.
11434 (native_encode_fixed): Likewise
11435 (native_encode_complex): Likewise
11436 (native_encode_vector): Likewise.
11437 (native_interpret_fixed): Likewise.
11438 (native_interpret_real): Likewise.
11439 (native_interpret_complex): Likewise.
11440 (native_interpret_vector): Likewise.
11441 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11442 (simd_clone_adjust_argument_types): Likewise.
11443 (simd_clone_init_simd_arrays): Likewise.
11444 (simd_clone_adjust): Likewise.
11445 * stor-layout.c (layout_type): Likewise.
11446 * tree.c (build_minus_one_cst): Likewise.
11447 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11448 * tree-inline.c (estimate_move_cost): Likewise.
11449 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11450 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11451 (vectorizable_reduction): Likewise.
11452 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
11453 (vect_recog_mixed_size_cond_pattern): Likewise.
11454 (check_bool_pattern): Likewise.
11455 (adjust_bool_pattern): Likewise.
11456 (search_type_for_mask_1): Likewise.
11457 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11458 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11459 (vectorizable_load): Likewise.
11460 (vectorizable_store): Likewise.
11461 * ubsan.c (ubsan_encode_value): Likewise.
11462 * varasm.c (output_constant): Likewise.
11464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11465 Alan Hayward <alan.hayward@arm.com>
11466 David Sherwood <david.sherwood@arm.com>
11468 * coretypes.h (scalar_mode): New class.
11469 * machmode.h (scalar_mode): Likewise.
11470 (scalar_mode::includes_p): New function.
11471 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
11472 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
11473 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11474 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
11475 * expmed.c (store_bit_field_1): Likewise.
11476 (extract_bit_field_1): Likewise.
11477 * expr.c (write_complex_part): Likewise.
11478 (read_complex_part): Likewise.
11479 (emit_move_complex_push): Likewise.
11480 (expand_expr_real_2): Likewise.
11481 * function.c (assign_parm_setup_reg): Likewise.
11482 (assign_parms_unsplit_complex): Likewise.
11483 * optabs.c (expand_binop): Likewise.
11484 * rtlanal.c (subreg_get_info): Likewise.
11485 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11486 * varasm.c (output_constant_pool_2): Likewise.
11488 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11489 Alan Hayward <alan.hayward@arm.com>
11490 David Sherwood <david.sherwood@arm.com>
11492 * expmed.c (extract_high_half): Use scalar_int_mode and remove
11494 (expmed_mult_highpart_optab): Likewise.
11495 (expmed_mult_highpart): Likewise.
11497 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11498 Alan Hayward <alan.hayward@arm.com>
11499 David Sherwood <david.sherwood@arm.com>
11501 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
11502 instead of a machine_mode.
11503 (builtin_memset_read_str): Likewise.
11504 * builtins.c (c_readstr): Likewise.
11505 (builtin_memcpy_read_str): Likewise.
11506 (builtin_strncpy_read_str): Likewise.
11507 (builtin_memset_read_str): Likewise.
11508 (builtin_memset_gen_str): Likewise.
11509 (expand_builtin_signbit): Use scalar_int_mode for local variables.
11510 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
11511 instead of a machine_mode.
11512 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
11514 (make_extraction): Likewise.
11515 (try_widen_shift_mode): Take and return scalar_int_modes instead
11517 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
11518 a scalar_int_mode instead of a machine_mode.
11519 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
11520 (avr_addr_space_pointer_mode): Likewise.
11521 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
11522 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
11523 (msp430_unwind_word_mode): Likewise.
11524 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
11525 (spu_addr_space_pointer_mode): Likewise.
11526 (spu_addr_space_address_mode): Likewise.
11527 (spu_libgcc_cmp_return_mode): Likewise.
11528 (spu_libgcc_shift_count_mode): Likewise.
11529 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
11530 (rl78_addr_space_pointer_mode): Likewise.
11531 (fl78_unwind_word_mode): Likewise.
11532 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
11534 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
11535 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
11536 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
11537 (mips_valid_pointer_mode): Likewise.
11538 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
11539 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
11540 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
11542 (ft32_addr_space_address_mode): Likewise.
11543 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
11544 scalar_int_mode instead of a machine_mode.
11545 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
11547 (m32c_addr_space_address_mode): Likewise.
11548 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
11549 (rs6000_eh_return_filter_mode): Likewise.
11550 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
11551 (rs6000_eh_return_filter_mode): Likewise.
11552 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
11553 (s390_libgcc_shift_count_mode): Likewise.
11554 (s390_unwind_word_mode): Likewise.
11555 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
11557 * target.def (mode_rep_extended): Likewise.
11558 (valid_pointer_mode): Likewise.
11559 (addr_space.valid_pointer_mode): Likewise.
11560 (eh_return_filter_mode): Return a scalar_int_mode rather than
11562 (libgcc_cmp_return_mode): Likewise.
11563 (libgcc_shift_count_mode): Likewise.
11564 (unwind_word_mode): Likewise.
11565 (addr_space.pointer_mode): Likewise.
11566 (addr_space.address_mode): Likewise.
11567 * doc/tm.texi: Regenerate.
11568 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
11570 (do_jump): Use scalar_int_mode for local variables.
11571 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
11572 rather than a machine_mode.
11573 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
11574 (scompare_loc_descriptor_wide): Likewise.
11575 (scompare_loc_descriptor_narrow): Likewise.
11576 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
11578 * except.c (sjlj_emit_dispatch_table): Likewise.
11579 (expand_builtin_eh_copy_values): Likewise.
11580 * explow.c (convert_memory_address_addr_space_1): Likewise.
11581 Take a scalar_int_mode rather than a machine_mode.
11582 (convert_memory_address_addr_space): Take a scalar_int_mode rather
11583 than a machine_mode.
11584 (memory_address_addr_space): Use scalar_int_mode for local variables.
11585 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
11586 rather than a machine_mode.
11587 * expmed.c (mask_rtx): Likewise.
11588 (init_expmed_one_conv): Likewise.
11589 (expand_mult_highpart_adjust): Likewise.
11590 (extract_high_half): Likewise.
11591 (expmed_mult_highpart_optab): Likewise.
11592 (expmed_mult_highpart): Likewise.
11593 (expand_smod_pow2): Likewise.
11594 (expand_sdiv_pow2): Likewise.
11595 (emit_store_flag_int): Likewise.
11596 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
11598 (extract_low_bits): Likewise.
11599 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
11601 * expr.c (pieces_addr::adjust): Likewise.
11602 (can_store_by_pieces): Likewise.
11603 (store_by_pieces): Likewise.
11604 (clear_by_pieces_1): Likewise.
11605 (expand_expr_addr_expr_1): Likewise.
11606 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
11607 (expand_expr_real_1): Likewise.
11608 (try_casesi): Likewise.
11609 * final.c (shorten_branches): Likewise.
11610 * fold-const.c (fold_convert_const_int_from_fixed): Change the
11611 type of "mode" to machine_mode.
11612 * internal-fn.c (expand_arith_overflow_result_store): Take a
11613 scalar_int_mode rather than a machine_mode.
11614 (expand_mul_overflow): Use scalar_int_mode for local variables.
11615 * loop-doloop.c (doloop_modify): Likewise.
11616 (doloop_optimize): Likewise.
11617 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
11618 than a machine_mode.
11619 (expand_doubleword_shift_condmove): Likewise.
11620 (expand_doubleword_shift): Likewise.
11621 (expand_doubleword_clz): Likewise.
11622 (expand_doubleword_popcount): Likewise.
11623 (expand_doubleword_parity): Likewise.
11624 (expand_absneg_bit): Use scalar_int_mode for local variables.
11625 (prepare_float_lib_cmp): Likewise.
11626 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
11627 rather than a machine_mode.
11628 (convert_memory_address_addr_space): Likewise.
11629 (get_mode_bounds): Likewise.
11630 (get_address_mode): Return a scalar_int_mode rather than a
11632 * rtlanal.c (get_address_mode): Likewise.
11633 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
11634 than a machine_mode.
11635 * targhooks.c (default_mode_rep_extended): Likewise.
11636 (default_valid_pointer_mode): Likewise.
11637 (default_addr_space_valid_pointer_mode): Likewise.
11638 (default_eh_return_filter_mode): Return a scalar_int_mode rather
11639 than a machine_mode.
11640 (default_libgcc_cmp_return_mode): Likewise.
11641 (default_libgcc_shift_count_mode): Likewise.
11642 (default_unwind_word_mode): Likewise.
11643 (default_addr_space_pointer_mode): Likewise.
11644 (default_addr_space_address_mode): Likewise.
11645 * targhooks.h (default_eh_return_filter_mode): Likewise.
11646 (default_libgcc_cmp_return_mode): Likewise.
11647 (default_libgcc_shift_count_mode): Likewise.
11648 (default_unwind_word_mode): Likewise.
11649 (default_addr_space_pointer_mode): Likewise.
11650 (default_addr_space_address_mode): Likewise.
11651 (default_mode_rep_extended): Take a scalar_int_mode rather than
11653 (default_valid_pointer_mode): Likewise.
11654 (default_addr_space_valid_pointer_mode): Likewise.
11655 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
11657 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
11658 rather than a machine_mode.
11659 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
11660 for local variables.
11661 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11662 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
11663 than a machine_mode.
11665 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11666 Alan Hayward <alan.hayward@arm.com>
11667 David Sherwood <david.sherwood@arm.com>
11669 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
11670 the mode argument to scalar_int_mode.
11671 (do_jump_by_parts_zero_rtx): Likewise.
11672 (do_jump_by_parts_equality_rtx): Likewise.
11673 (do_jump_by_parts_greater): Take a mode argument.
11674 (do_jump_by_parts_equality): Likewise.
11675 (do_jump_1): Update calls accordingly.
11677 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11678 Alan Hayward <alan.hayward@arm.com>
11679 David Sherwood <david.sherwood@arm.com>
11681 * is-a.h (safe_dyn_cast): New function.
11682 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
11683 (jump_table_for_label): Likewise.
11684 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
11685 instead of an rtx_insn *.
11686 (shorten_branches): Use dyn_cast instead of LABEL_P and
11687 JUMP_TABLE_DATA_P. Use jump_table_for_label and
11688 rtx_jump_table_data::get_data_mode.
11689 (final_scan_insn): Likewise.
11691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11692 Alan Hayward <alan.hayward@arm.com>
11693 David Sherwood <david.sherwood@arm.com>
11695 * combine.c (try_combine): Use is_a <scalar_int_mode> when
11696 trying to combine a full-register integer set with a subreg
11699 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11700 Alan Hayward <alan.hayward@arm.com>
11701 David Sherwood <david.sherwood@arm.com>
11703 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
11704 that is always either address_mode or pointer_mode.
11706 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11707 Alan Hayward <alan.hayward@arm.com>
11708 David Sherwood <david.sherwood@arm.com>
11710 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
11711 when the two are known to be equal.
11713 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11714 Alan Hayward <alan.hayward@arm.com>
11715 David Sherwood <david.sherwood@arm.com>
11717 * simplify-rtx.c (simplify_const_unary_operation): Use
11718 is_a <scalar_int_mode> instead of checking for a nonzero
11719 precision. Forcibly convert op_mode to a scalar_int_mode
11720 in that case. More clearly differentiate the operand and
11721 result modes and use the former when deciding what the value
11722 of a count-bits operation should be. Use is_int_mode instead
11723 of checking for a MODE_INT. Remove redundant check for whether
11724 this mode has a zero precision.
11726 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11727 Alan Hayward <alan.hayward@arm.com>
11728 David Sherwood <david.sherwood@arm.com>
11730 * optabs.c (widen_leading): Change the type of the mode argument
11731 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
11732 (widen_bswap): Likewise.
11733 (expand_parity): Likewise.
11734 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
11735 (expand_ffs): Likewise.
11736 (epand_unop): Check for scalar integer modes before calling the
11739 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11740 Alan Hayward <alan.hayward@arm.com>
11741 David Sherwood <david.sherwood@arm.com>
11743 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
11745 (expand_expr_real_1): Update call accordingly.
11747 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11748 Alan Hayward <alan.hayward@arm.com>
11749 David Sherwood <david.sherwood@arm.com>
11751 * expmed.c (store_bit_field_using_insv): Add op0_mode and
11752 value_mode arguments. Use scalar_int_mode internally.
11753 (store_bit_field_1): Rename the new integer mode from imode
11754 to op0_mode and use it instead of GET_MODE (op0). Update calls
11755 to store_split_bit_field, store_bit_field_using_insv and
11756 store_fixed_bit_field.
11757 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
11758 Use scalar_int_mode internally. Use a bit count rather than a mode
11759 when calculating the largest bit size for get_best_mode.
11760 Update calls to store_split_bit_field and store_fixed_bit_field_1.
11761 (store_fixed_bit_field_1): Add mode and value_mode arguments.
11762 Remove assertion that OP0 has a scalar integer mode.
11763 (store_split_bit_field): Add op0_mode and value_mode arguments.
11764 Update calls to extract_fixed_bit_field.
11765 (extract_bit_field_using_extv): Add an op0_mode argument.
11766 Use scalar_int_mode internally.
11767 (extract_bit_field_1): Rename the new integer mode from imode to
11768 op0_mode and use it instead of GET_MODE (op0). Update calls to
11769 extract_split_bit_field, extract_bit_field_using_extv and
11770 extract_fixed_bit_field.
11771 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
11772 to extract_split_bit_field and extract_fixed_bit_field_1.
11773 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
11774 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
11775 on the target mode.
11776 (extract_split_bit_field): Add an op0_mode argument. Update call
11777 to extract_fixed_bit_field.
11779 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11780 Alan Hayward <alan.hayward@arm.com>
11781 David Sherwood <david.sherwood@arm.com>
11783 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
11784 * explow.c (hard_function_value): Likewise.
11785 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
11786 convert_to_mode call outside the loop.
11787 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
11788 for the mode iterator. Require the mode specified by max_pieces
11790 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
11792 (copy_blkmode_to_reg): Likewise.
11793 (set_storage_via_setmem): Likewise.
11794 * optabs.c (prepare_cmp_insn): Likewise.
11795 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11796 * stor-layout.c (finish_bitfield_representative): Likewise.
11798 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11799 Alan Hayward <alan.hayward@arm.com>
11800 David Sherwood <david.sherwood@arm.com>
11802 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
11803 * expr.c (convert_move): Use them.
11804 (convert_modes): Likewise.
11805 (store_expr_with_bounds): Likewise.
11807 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11808 Alan Hayward <alan.hayward@arm.com>
11809 David Sherwood <david.sherwood@arm.com>
11811 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
11812 parameter for the mode of "x". Remove the "known_x", "known_mode"
11813 and "known_ret" arguments. Change the type of the mode argument
11814 to scalar_int_mode.
11815 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
11816 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
11817 (reg_num_sign_bit_copies_for_combine): Likewise.
11818 * rtlanal.c (nonzero_bits1): Likewise.
11819 (num_sign_bit_copies1): Likewise.
11820 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
11821 (reg_num_sign_bit_copies_general): Likewise.
11822 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
11823 (reg_nonzero_bits_general): Likewise.
11825 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11826 Alan Hayward <alan.hayward@arm.com>
11827 David Sherwood <david.sherwood@arm.com>
11829 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
11830 than in subroutines. Return 1 for non-integer modes.
11831 (cached_num_sign_bit_copies): Change the type of the mode parameter
11832 to scalar_int_mode.
11833 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
11834 classes. Handle CONST_INT_P first and then check whether X also
11835 has a scalar integer mode. Check the same thing for inner registers
11836 of a SUBREG and for values that are being extended or truncated.
11838 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11839 Alan Hayward <alan.hayward@arm.com>
11840 David Sherwood <david.sherwood@arm.com>
11842 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
11843 in subroutines. Return the mode mask for non-integer modes.
11844 (cached_nonzero_bits): Change the type of the mode parameter
11845 to scalar_int_mode.
11846 (nonzero_bits1): Likewise. Remove early exit for other mode
11847 classes. Handle CONST_INT_P first and then check whether X
11848 also has a scalar integer mode.
11850 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11851 Alan Hayward <alan.hayward@arm.com>
11852 David Sherwood <david.sherwood@arm.com>
11854 * expr.c (widest_int_mode_for_size): Make the comment match the code.
11855 Return a scalar_int_mode and assert that the size is greater than
11857 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
11858 (op_by_pieces_d::op_by_pieces_d): Likewise.
11859 (op_by_pieces_d::run): Likewise.
11860 (can_store_by_pieces): Likewise.
11862 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11863 Alan Hayward <alan.hayward@arm.com>
11864 David Sherwood <david.sherwood@arm.com>
11866 * combine.c (extract_left_shift): Add a mode argument and update
11868 (make_compound_operation_int): Change the type of the mode parameter
11869 to scalar_int_mode and update the call to extract_left_shift.
11871 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11872 Alan Hayward <alan.hayward@arm.com>
11873 David Sherwood <david.sherwood@arm.com>
11875 * combine.c (simplify_and_const_int): Change the type of the mode
11876 parameter to scalar_int_mode.
11877 (simplify_and_const_int_1): Likewise. Update recursive call.
11879 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11880 Alan Hayward <alan.hayward@arm.com>
11881 David Sherwood <david.sherwood@arm.com>
11883 * combine.c (simplify_compare_const): Check that the mode is a
11884 scalar_int_mode (rather than VOIDmode) before testing its
11886 (simplify_comparison): Move COMPARISON_P handling out of the
11887 loop and restrict the latter part of the loop to scalar_int_modes.
11888 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
11889 and when considering SUBREG_REGs. Use is_int_mode instead of
11890 checking GET_MODE_CLASS against MODE_INT.
11892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11893 Alan Hayward <alan.hayward@arm.com>
11894 David Sherwood <david.sherwood@arm.com>
11896 * combine.c (try_widen_shift_mode): Move check for equal modes to...
11897 (simplify_shift_const_1): ...here. Use scalar_int_mode for
11898 shift_unit_mode and for modes involved in scalar shifts.
11900 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11901 Alan Hayward <alan.hayward@arm.com>
11902 David Sherwood <david.sherwood@arm.com>
11904 * combine.c (force_int_to_mode): New function, split out from...
11905 (force_to_mode): ...here. Keep xmode up-to-date and use it
11906 instead of GET_MODE (x).
11908 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11909 Alan Hayward <alan.hayward@arm.com>
11910 David Sherwood <david.sherwood@arm.com>
11912 * optabs-query.h (extraction_insn::struct_mode): Change type to
11913 opt_scalar_int_mode and update comment.
11914 (extraction_insn::field_mode): Change type to scalar_int_mode.
11915 (extraction_insn::pos_mode): Likewise.
11916 * combine.c (make_extraction): Update accordingly.
11917 * optabs-query.c (get_traditional_extraction_insn): Likewise.
11918 (get_optab_extraction_insn): Likewise.
11919 * recog.c (simplify_while_replacing): Likewise.
11920 * expmed.c (narrow_bit_field_mem): Change the type of the mode
11921 parameter to opt_scalar_int_mode.
11923 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11924 Alan Hayward <alan.hayward@arm.com>
11925 David Sherwood <david.sherwood@arm.com>
11927 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
11928 to a scalar_int_mode instead of a machine_mode.
11929 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
11930 (get_best_mode): Return a boolean and use a pointer argument to store
11931 the selected mode. Replace the limit mode parameter with a bit limit.
11932 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
11933 for the values returned by bit_field_mode_iterator::next_mode.
11934 (store_bit_field): Update call to get_best_mode.
11935 (store_fixed_bit_field): Likewise.
11936 (extract_fixed_bit_field): Likewise.
11937 * expr.c (optimize_bitfield_assignment_op): Likewise.
11938 * fold-const.c (optimize_bit_field_compare): Likewise.
11939 (fold_truth_andor_1): Likewise.
11940 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
11941 Update for new type of m_mode.
11942 (get_best_mode): As above.
11944 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11945 Alan Hayward <alan.hayward@arm.com>
11946 David Sherwood <david.sherwood@arm.com>
11948 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
11949 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
11950 (store_bit_field): Check is_a <scalar_int_mode> before calling
11951 strict_volatile_bitfield_p.
11952 (extract_bit_field): Likewise.
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 * target.def (cstore_mode): Return a scalar_int_mode.
11959 * doc/tm.texi: Regenerate.
11960 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
11961 * targhooks.h (default_cstore_mode): Likewise.
11962 * targhooks.c (default_cstore_mode): Likewise, using a forced
11964 * expmed.c (emit_cstore): Expect the target of the cstore to be
11967 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11968 Alan Hayward <alan.hayward@arm.com>
11969 David Sherwood <david.sherwood@arm.com>
11971 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
11973 (niter_desc): Likewise mode.
11974 (iv_analyze): Add a mode parameter.
11976 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
11977 and change its type to scalar_int_mode.
11978 * loop-iv.c: Update commentary at head of file.
11979 (iv_constant): Pass the mode paraeter before the rtx it describes
11980 and change its type to scalar_int_mode. Remove VOIDmode handling.
11981 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
11982 (iv_extend): Likewise.
11983 (shorten_into_mode): Likewise.
11984 (iv_add): Use scalar_int_mode.
11985 (iv_mult): Likewise.
11986 (iv_shift): Likewise.
11987 (canonicalize_iv_subregs): Likewise.
11988 (get_biv_step_1): Pass the outer_mode parameter before the rtx
11989 it describes and change its mode to scalar_int_mode. Also change
11990 the type of the returned inner_mode to scalar_int_mode.
11991 (get_biv_step): Likewise, turning outer_mode from a pointer
11992 into a direct parameter. Update call to get_biv_step_1.
11993 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
11994 iv_constant and get_biv_step.
11995 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
11996 and change its type to scalar_int_mode. Don't initialise iv->mode
11997 to VOIDmode and remove later checks for its still being VOIDmode.
11998 Update calls to iv_analyze_op and iv_analyze_expr. Check
11999 is_a <scalar_int_mode> when changing the mode under consideration.
12000 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
12001 Update call to iv_analyze_expr.
12002 (iv_analyze_op): Add a mode parameter. Reject subregs whose
12003 inner register is not also a scalar_int_mode. Update call to
12005 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
12006 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
12007 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
12008 separate mode class checks. Update calls to iv_analyze. Remove
12009 fix-up of VOIDmodes after iv_analyze_biv.
12010 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
12011 don't have a scalar_int_mode. Update call to biv_p.
12013 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12014 Alan Hayward <alan.hayward@arm.com>
12015 David Sherwood <david.sherwood@arm.com>
12017 * cfgexpand.c (convert_debug_memory_address): Use
12018 as_a <scalar_int_mode>.
12019 * combine.c (expand_compound_operation): Likewise.
12020 (make_extraction): Likewise.
12021 (change_zero_ext): Likewise.
12022 (simplify_comparison): Likewise.
12023 * cse.c (cse_insn): Likewise.
12024 * dwarf2out.c (minmax_loc_descriptor): Likewise.
12025 (mem_loc_descriptor): Likewise.
12026 (loc_descriptor): Likewise.
12027 * expmed.c (init_expmed_one_mode): Likewise.
12028 (synth_mult): Likewise.
12029 (emit_store_flag_1): Likewise.
12030 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
12031 of a comparison with size.
12032 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
12033 (reduce_to_bit_field_precision): Likewise.
12034 * function.c (expand_function_end): Likewise.
12035 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
12036 * loop-doloop.c (doloop_modify): Likewise.
12037 * optabs.c (expand_binop): Likewise.
12038 (expand_unop): Likewise.
12039 (expand_copysign_absneg): Likewise.
12040 (prepare_cmp_insn): Likewise.
12041 (maybe_legitimize_operand): Likewise.
12042 * recog.c (const_scalar_int_operand): Likewise.
12043 * rtlanal.c (get_address_mode): Likewise.
12044 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12045 (simplify_cond_clz_ctz): Likewise.
12046 * tree-nested.c (get_nl_goto_field): Likewise.
12047 * tree.c (build_vector_type_for_mode): Likewise.
12048 * var-tracking.c (use_narrower_mode): Likewise.
12050 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12051 Alan Hayward <alan.hayward@arm.com>
12052 David Sherwood <david.sherwood@arm.com>
12054 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
12055 * builtins.c (expand_builtin_signbit): Use it.
12056 * cfgexpand.c (expand_debug_expr): Likewise.
12057 * dojump.c (do_jump): Likewise.
12058 (do_compare_and_jump): Likewise.
12059 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
12060 * expmed.c (make_tree): Likewise.
12061 * expr.c (expand_expr_real_2): Likewise.
12062 (expand_expr_real_1): Likewise.
12063 (try_casesi): Likewise.
12064 * fold-const-call.c (fold_const_call_ss): Likewise.
12065 * fold-const.c (unextend): Likewise.
12066 (extract_muldiv_1): Likewise.
12067 (fold_single_bit_test): Likewise.
12068 (native_encode_int): Likewise.
12069 (native_encode_string): Likewise.
12070 (native_interpret_int): Likewise.
12071 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
12072 * internal-fn.c (expand_addsub_overflow): Likewise.
12073 (expand_neg_overflow): Likewise.
12074 (expand_mul_overflow): Likewise.
12075 (expand_arith_overflow): Likewise.
12076 * match.pd: Likewise.
12077 * stor-layout.c (layout_type): Likewise.
12078 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12079 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
12080 * tree-ssanames.c (get_range_info): Likewise.
12081 * tree-switch-conversion.c (array_value_type) Likewise.
12082 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
12083 (vect_recog_divmod_pattern): Likewise.
12084 (vect_recog_mixed_size_cond_pattern): Likewise.
12085 * tree-vrp.c (extract_range_basic): Likewise.
12086 (simplify_float_conversion_using_ranges): Likewise.
12087 * tree.c (int_fits_type_p): Likewise.
12088 * ubsan.c (instrument_bool_enum_load): Likewise.
12089 * varasm.c (mergeable_string_section): Likewise.
12090 (narrowing_initializer_constant_valid_p): Likewise.
12091 (output_constant): Likewise.
12093 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12094 Alan Hayward <alan.hayward@arm.com>
12095 David Sherwood <david.sherwood@arm.com>
12097 * machmode.h (NARROWEST_INT_MODE): New macro.
12098 * expr.c (alignment_for_piecewise_move): Use it instead of
12099 GET_CLASS_NARROWEST_MODE (MODE_INT).
12100 (push_block): Likewise.
12101 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
12103 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12105 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12106 Alan Hayward <alan.hayward@arm.com>
12107 David Sherwood <david.sherwood@arm.com>
12109 * postreload.c (move2add_valid_value_p): Change the type of the
12110 mode parameter to scalar_int_mode.
12111 (move2add_use_add2_insn): Add a mode parameter and use it instead
12113 (move2add_use_add3_insn): Likewise.
12114 (reload_cse_move2add): Update accordingly.
12116 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12117 Alan Hayward <alan.hayward@arm.com>
12118 David Sherwood <david.sherwood@arm.com>
12120 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
12122 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
12123 * optabs.c (expand_unop): Likewise.
12125 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12126 Alan Hayward <alan.hayward@arm.com>
12127 David Sherwood <david.sherwood@arm.com>
12129 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
12130 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
12131 (popcount_loc_descriptor): Likewise.
12132 (bswap_loc_descriptor): Likewise.
12133 (rotate_loc_descriptor): Likewise.
12134 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
12135 calling the functions above.
12137 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12138 Alan Hayward <alan.hayward@arm.com>
12139 David Sherwood <david.sherwood@arm.com>
12141 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
12143 (try_combine): Likewise.
12144 (simplify_if_then_else): Likewise.
12145 * cse.c (cse_insn): Likewise.
12146 * dwarf2out.c (mem_loc_descriptor): Likewise.
12147 * emit-rtl.c (gen_lowpart_common): Likewise.
12148 * simplify-rtx.c (simplify_truncation): Likewise.
12149 (simplify_binary_operation_1): Likewise.
12150 (simplify_const_relational_operation): Likewise.
12151 (simplify_ternary_operation): Likewise.
12152 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): 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 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
12159 * reload.c (push_reload): Likewise.
12160 (find_reloads): Likewise.
12162 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12163 Alan Hayward <alan.hayward@arm.com>
12164 David Sherwood <david.sherwood@arm.com>
12166 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
12167 (make_compound_operation_int): Likewise.
12168 (change_zero_ext): Likewise.
12169 * expr.c (convert_move): Likewise.
12170 (convert_modes): Likewise.
12171 * fwprop.c (forward_propagate_subreg): Likewise.
12172 * loop-iv.c (get_biv_step_1): Likewise.
12173 * optabs.c (widen_operand): Likewise.
12174 * postreload.c (move2add_valid_value_p): Likewise.
12175 * recog.c (simplify_while_replacing): Likewise.
12176 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12177 (simplify_binary_operation_1): Likewise. Remove redundant
12178 mode equality check.
12180 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12181 Alan Hayward <alan.hayward@arm.com>
12182 David Sherwood <david.sherwood@arm.com>
12184 * combine.c (combine_simplify_rtx): Add checks for
12185 is_a <scalar_int_mode>.
12186 (simplify_if_then_else): Likewise.
12187 (make_field_assignment): Likewise.
12188 (simplify_comparison): Likewise.
12189 * ifcvt.c (noce_try_bitop): Likewise.
12190 * loop-invariant.c (canonicalize_address_mult): Likewise.
12191 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12193 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12194 Alan Hayward <alan.hayward@arm.com>
12195 David Sherwood <david.sherwood@arm.com>
12197 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12198 is_a <scalar_int_mode> instead of != BLKmode.
12200 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12201 Alan Hayward <alan.hayward@arm.com>
12202 David Sherwood <david.sherwood@arm.com>
12204 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12205 instead of != VOIDmode.
12206 * combine.c (if_then_else_cond): Likewise.
12207 (change_zero_ext): Likewise.
12208 * dwarf2out.c (mem_loc_descriptor): Likewise.
12209 (loc_descriptor): Likewise.
12210 * rtlanal.c (canonicalize_condition): Likewise.
12211 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
12213 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12214 Alan Hayward <alan.hayward@arm.com>
12215 David Sherwood <david.sherwood@arm.com>
12217 * simplify-rtx.c (simplify_binary_operation_1): Use
12218 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
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 * wide-int.h (int_traits<unsigned char>) New class.
12225 (int_traits<unsigned short>) Likewise.
12226 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12227 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12229 * combine.c (set_nonzero_bits_and_sign_copies): Use
12230 is_a <scalar_int_mode>.
12231 (find_split_point): Likewise.
12232 (combine_simplify_rtx): Likewise.
12233 (simplify_logical): Likewise.
12234 (expand_compound_operation): Likewise.
12235 (expand_field_assignment): Likewise.
12236 (make_compound_operation): Likewise.
12237 (extended_count): Likewise.
12238 (change_zero_ext): Likewise.
12239 (simplify_comparison): Likewise.
12240 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12241 (ucompare_loc_descriptor): Likewise.
12242 (minmax_loc_descriptor): Likewise.
12243 (mem_loc_descriptor): Likewise.
12244 (loc_descriptor): Likewise.
12245 * expmed.c (init_expmed_one_mode): Likewise.
12246 * lra-constraints.c (lra_constraint_offset): Likewise.
12247 * optabs.c (prepare_libcall_arg): Likewise.
12248 * postreload.c (move2add_note_store): Likewise.
12249 * reload.c (operands_match_p): Likewise.
12250 * rtl.h (load_extend_op): Likewise.
12251 * rtlhooks.c (gen_lowpart_general): Likewise.
12252 * simplify-rtx.c (simplify_truncation): Likewise.
12253 (simplify_unary_operation_1): Likewise.
12254 (simplify_binary_operation_1): Likewise.
12255 (simplify_const_binary_operation): Likewise.
12256 (simplify_const_relational_operation): Likewise.
12257 (simplify_subreg): Likewise.
12258 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12259 * var-tracking.c (adjust_mems): Likewise.
12260 (prepare_call_arguments): Likewise.
12262 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12263 Alan Hayward <alan.hayward@arm.com>
12264 David Sherwood <david.sherwood@arm.com>
12266 * machmode.h (is_int_mode): New fuction.
12267 * combine.c (find_split_point): Use it.
12268 (combine_simplify_rtx): Likewise.
12269 (simplify_if_then_else): Likewise.
12270 (simplify_set): Likewise.
12271 (simplify_shift_const_1): Likewise.
12272 (simplify_comparison): Likewise.
12273 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12274 * cse.c (notreg_cost): Likewise.
12275 (cse_insn): Likewise.
12276 * cselib.c (cselib_lookup_1): Likewise.
12277 * dojump.c (do_jump_1): Likewise.
12278 (do_compare_rtx_and_jump): Likewise.
12279 * dse.c (get_call_args): Likewise.
12280 * dwarf2out.c (rtl_for_decl_init): Likewise.
12281 (native_encode_initializer): Likewise.
12282 * expmed.c (emit_store_flag_1): Likewise.
12283 (emit_store_flag): Likewise.
12284 * expr.c (convert_modes): Likewise.
12285 (store_field): Likewise.
12286 (expand_expr_real_1): Likewise.
12287 * fold-const.c (fold_read_from_constant_string): Likewise.
12288 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12289 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12290 * optabs.c (expand_binop): Likewise.
12291 (expand_unop): Likewise.
12292 (expand_abs_nojump): Likewise.
12293 (expand_one_cmpl_abs_nojump): Likewise.
12294 * simplify-rtx.c (mode_signbit_p): Likewise.
12295 (val_signbit_p): Likewise.
12296 (val_signbit_known_set_p): Likewise.
12297 (val_signbit_known_clear_p): Likewise.
12298 (simplify_relational_operation_1): Likewise.
12299 * tree.c (vector_type_mode): Likewise.
12301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12302 Alan Hayward <alan.hayward@arm.com>
12303 David Sherwood <david.sherwood@arm.com>
12305 * machmode.h (smallest_mode_for_size): Fix formatting.
12306 (smallest_int_mode_for_size): New function.
12307 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12308 instead of smallest_mode_for_size.
12309 * combine.c (make_extraction): Likewise.
12310 * config/arc/arc.c (arc_expand_movmem): Likewise.
12311 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12312 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12313 * config/s390/s390.c (s390_expand_insv): Likewise.
12314 * config/sparc/sparc.c (assign_int_registers): Likewise.
12315 * config/spu/spu.c (spu_function_value): Likewise.
12316 (spu_function_arg): Likewise.
12317 * coverage.c (get_gcov_type): Likewise.
12318 (get_gcov_unsigned_t): Likewise.
12319 * dse.c (find_shift_sequence): Likewise.
12320 * expmed.c (store_bit_field_1): Likewise.
12321 * expr.c (convert_move): Likewise.
12322 (store_field): Likewise.
12323 * internal-fn.c (expand_arith_overflow): Likewise.
12324 * optabs-query.c (get_best_extraction_insn): Likewise.
12325 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12326 * stor-layout.c (layout_type): Likewise.
12327 (initialize_sizetypes): Likewise.
12328 * targhooks.c (default_get_mask_mode): Likewise.
12329 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12331 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12332 Alan Hayward <alan.hayward@arm.com>
12333 David Sherwood <david.sherwood@arm.com>
12335 * machmode.h (opt_mode::else_blk): New function.
12336 (int_mode_for_mode): Declare.
12337 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12338 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12340 * cfgexpand.c (expand_debug_expr): Likewise.
12341 * combine.c (gen_lowpart_or_truncate): Likewise.
12342 (gen_lowpart_for_combine): Likewise.
12343 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12344 * config/avr/avr.c (avr_to_int_mode): Likewise.
12345 (avr_out_plus_1): Likewise.
12346 (avr_out_plus): Likewise.
12347 (avr_out_round): Likewise.
12348 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12349 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12350 (s390_expand_vcond): Likewise.
12351 * config/spu/spu.c (spu_split_immediate): Likewise.
12352 (spu_expand_mov): Likewise.
12353 * dse.c (get_stored_val): Likewise.
12354 * expmed.c (store_bit_field_1): Likewise.
12355 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12357 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12358 (extract_low_bits): Likewise.
12359 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12360 handling rather than repeating the check.
12361 (emit_group_store): Likewise.
12362 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12363 * optabs.c (expand_absneg_bit): Likewise.
12364 (expand_copysign_absneg): Likewise.
12365 (expand_copysign_bit): Likewise.
12366 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12367 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12368 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12369 * var-tracking.c (prepare_call_arguments): Likewise.
12370 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12371 int_mode_for_mode instead of mode_for_size.
12372 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12374 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12375 Alan Hayward <alan.hayward@arm.com>
12376 David Sherwood <david.sherwood@arm.com>
12378 * machmode.h (int_mode_for_size): New function.
12379 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12380 instead of mode_for_size.
12381 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12383 * combine.c (expand_field_assignment): Use int_mode_for_size
12384 instead of mode_for_size.
12385 (make_extraction): Likewise.
12386 (simplify_shift_const_1): Likewise.
12387 (simplify_comparison): Likewise.
12388 * dojump.c (do_jump): Likewise.
12389 * dwarf2out.c (mem_loc_descriptor): Likewise.
12390 * emit-rtl.c (init_derived_machine_modes): Likewise.
12391 * expmed.c (flip_storage_order): Likewise.
12392 (convert_extracted_bit_field): Likewise.
12393 * expr.c (copy_blkmode_from_reg): Likewise.
12394 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12395 * internal-fn.c (expand_mul_overflow): Likewise.
12396 * lower-subreg.c (simple_move): Likewise.
12397 * optabs-libfuncs.c (init_optabs): Likewise.
12398 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12399 * tree.c (vector_type_mode): Likewise.
12400 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12401 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12402 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12403 * tree-vect-stmts.c (vectorizable_load): Likewise.
12404 (vectorizable_store): Likewise.
12406 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12407 Alan Hayward <alan.hayward@arm.com>
12408 David Sherwood <david.sherwood@arm.com>
12410 * coretypes.h (pod_mode): New type.
12411 (scalar_int_mode_pod): New typedef.
12412 * machmode.h (pod_mode): New class.
12413 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12414 * genmodes.c (emit_mode_int_n): Update accordingly.
12415 * lower-subreg.h (target_lower_subreg): Change type to
12416 scalar_int_mode_pod.
12417 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12418 scalar_int_mode_pod.
12420 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12421 Alan Hayward <alan.hayward@arm.com>
12422 David Sherwood <david.sherwood@arm.com>
12424 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12425 machine_mode to scalar_int_mode.
12426 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12427 (rs6000_option_override_internal): Remove cast to int.
12428 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12429 machine_mode to scalar_int_mode.
12430 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12431 (rs6000_option_override_internal): Remove cast to int.
12432 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12433 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12435 * config/s390/s390.c (s390_expand_builtin): Likewise.
12436 * coretypes.h (scalar_int_mode): New type.
12437 (opt_scalar_int_mode): New typedef.
12438 * machmode.h (scalar_int_mode): New class.
12439 (scalar_int_mode::includes_p): New function.
12440 (byte_mode): Change type to scalar_int_mode.
12441 (word_mode): Likewise.
12442 (ptr_mode): Likewise.
12443 * emit-rtl.c (byte_mode): Likewise.
12444 (word_mode): Likewise.
12445 (ptr_mode): Likewise.
12446 (init_derived_machine_modes): Update accordingly.
12447 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12448 and MODE_PARTIAL_INT.
12449 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12450 opt_scalar_int_mode.
12452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12453 Alan Hayward <alan.hayward@arm.com>
12454 David Sherwood <david.sherwood@arm.com>
12456 * target.def (libgcc_floating_mode_supported_p): Take a
12458 * doc/tm.texi: Regenerate.
12459 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12461 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12462 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12465 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12466 Alan Hayward <alan.hayward@arm.com>
12467 David Sherwood <david.sherwood@arm.com>
12469 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12470 * doc/tm.texi: Regenerate.
12471 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
12472 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
12473 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
12474 * targhooks.h (default_floatn_mode): Likewise.
12475 * targhooks.c (default_floatn_mode): Likewise.
12476 * tree.c (build_common_tree_nodes): Update accordingly.
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 (mode_iterator::start): Provide overload for opt_modes.
12483 (mode_iterator::iterate_p): Likewise.
12484 (mode_iterator::get_wider): Likewise.
12485 * expr.c (init_expr_target): Use opt_scalar_float_mode.
12487 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12488 Alan Hayward <alan.hayward@arm.com>
12489 David Sherwood <david.sherwood@arm.com>
12491 * coretypes.h (opt_scalar_float_mode): New typedef.
12492 * machmode.h (float_mode_for_size): New function.
12493 * emit-rtl.c (double_mode): Delete.
12494 (init_emit_once): Use float_mode_for_size.
12495 * stor-layout.c (layout_type): Likewise.
12496 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
12498 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12499 Alan Hayward <alan.hayward@arm.com>
12500 David Sherwood <david.sherwood@arm.com>
12502 * output.h (assemble_real): Take a scalar_float_mode.
12503 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
12504 * config/arm/arm.md (consttable_4): Likewise.
12505 (consttable_8): Likewise.
12506 (consttable_16): Likewise.
12507 * config/mips/mips.md (consttable_float): Likewise.
12508 * config/s390/s390.c (s390_output_pool_entry): Likewise.
12509 * varasm.c (assemble_real): Take a scalar_float_mode.
12510 (output_constant_pool_2): Update accordingly.
12511 (output_constant): Likewise.
12513 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12514 Alan Hayward <alan.hayward@arm.com>
12515 David Sherwood <david.sherwood@arm.com>
12517 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
12518 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
12519 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
12520 (native_encode_real): Likewise.
12521 (native_interpret_real): Likewise.
12522 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
12523 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12525 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12526 Alan Hayward <alan.hayward@arm.com>
12527 David Sherwood <david.sherwood@arm.com>
12529 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
12530 <scalar_float_mode>. Simplify.
12531 (gen_extend_conv_libfunc): Likewise.
12533 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12534 Alan Hayward <alan.hayward@arm.com>
12535 David Sherwood <david.sherwood@arm.com>
12537 * coretypes.h (scalar_float_mode): New type.
12538 * machmode.h (mode_traits::from_int): Use machine_mode if
12539 USE_ENUM_MODES is defined.
12540 (is_a): New function.
12542 (dyn_cast): Likewise.
12543 (scalar_float_mode): New class.
12544 (scalar_float_mode::includes_p): New function.
12545 (is_float_mode): Likewise.
12546 * gdbhooks.py (MachineModePrinter): New class.
12547 (build_pretty_printer): Use it for scalar_float_mode.
12548 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
12549 (format_helper::format_helper): Turn into a template.
12550 * genmodes.c (get_mode_class): New function.
12551 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
12552 or machine_mode if none.
12553 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
12554 as_a <scalar_float_mode>.
12555 * dwarf2out.c (mem_loc_descriptor): Likewise.
12556 (insert_float): Likewise.
12557 (add_const_value_attribute): Likewise.
12558 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12559 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
12560 (expand_unop): Update accordingly.
12561 (expand_abs_nojump): Likewise.
12562 (expand_copysign_absneg): Take a scalar_float_mode.
12563 (expand_copysign_bit): Likewise.
12564 (expand_copysign): Update accordingly.
12566 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12567 Alan Hayward <alan.hayward@arm.com>
12568 David Sherwood <david.sherwood@arm.com>
12570 * coretypes.h (opt_mode): New class.
12571 * machmode.h (opt_mode): Likewise.
12572 (opt_mode::else_void): New function.
12573 (opt_mode::require): Likewise.
12574 (opt_mode::exists): Likewise.
12575 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
12576 (GET_MODE_2XWIDER_MODE): Likewise.
12577 (mode_iterator::get_wider): Update accordingly.
12578 (mode_iterator::get_2xwider): Likewise.
12579 (mode_iterator::get_known_wider): Likewise, turning into a template.
12580 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
12581 forcing a wider mode to exist.
12582 * config/cr16/cr16.h (LONG_REG_P): Likewise.
12583 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12584 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
12585 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
12586 * lower-subreg.c (init_lower_subreg): Likewise.
12587 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
12588 on the final iteration.
12589 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
12590 a wider mode exists before asking for a move pattern.
12591 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
12592 forcing a wider mode to exist.
12593 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
12594 returning false if no such mode exists.
12595 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
12596 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
12597 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
12598 Avoid checking for a MODE_INT if we already know the mode is not a
12600 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
12601 forcing a wider mode to exist.
12602 (expmed_mult_highpart_optab): Likewise.
12603 (expmed_mult_highpart): Likewise.
12604 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
12606 * lto-streamer-in.c (lto_input_mode_table): Likewise.
12607 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
12608 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12609 * internal-fn.c (expand_mul_overflow): Update use of
12610 GET_MODE_2XWIDER_MODE.
12611 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12612 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
12613 GET_MODE_WIDER_MODE.
12614 (convert_plusminus_to_widen): Likewise.
12615 * tree-switch-conversion.c (array_value_type): Likewise.
12616 * var-tracking.c (emit_note_insn_var_location): Likewise.
12617 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12618 Return false inside rather than outside the loop if no wider mode
12620 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
12621 and GET_MODE_2XWIDER_MODE
12622 (can_compare_p): Use else_void.
12623 * gdbhooks.py (OptMachineModePrinter): New class.
12624 (build_pretty_printer): Use it for opt_mode.
12626 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12627 Alan Hayward <alan.hayward@arm.com>
12628 David Sherwood <david.sherwood@arm.com>
12630 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
12631 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
12633 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12634 Alan Hayward <alan.hayward@arm.com>
12635 David Sherwood <david.sherwood@arm.com>
12637 * machmode.h (mode_traits): New structure.
12638 (get_narrowest_mode): New function.
12639 (mode_iterator::start): Likewise.
12640 (mode_iterator::iterate_p): Likewise.
12641 (mode_iterator::get_wider): Likewise.
12642 (mode_iterator::get_known_wider): Likewise.
12643 (mode_iterator::get_2xwider): Likewise.
12644 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
12645 (FOR_EACH_MODE): Likewise.
12646 (FOR_EACH_MODE_FROM): Likewise.
12647 (FOR_EACH_MODE_UNTIL): Likewise.
12648 (FOR_EACH_WIDER_MODE): Likewise.
12649 (FOR_EACH_2XWIDER_MODE): Likewise.
12650 * builtins.c (expand_builtin_strlen): Use new mode iterators.
12651 * combine.c (simplify_comparison): Likewise
12652 * config/i386/i386.c (type_natural_mode): Likewise.
12653 * cse.c (cse_insn): Likewise.
12654 * dse.c (find_shift_sequence): Likewise.
12655 * emit-rtl.c (init_derived_machine_modes): Likewise.
12656 (init_emit_once): Likewise.
12657 * explow.c (hard_function_value): Likewise.
12658 * expmed.c (extract_fixed_bit_field_1): Likewise.
12659 (extract_bit_field_1): Likewise.
12660 (expand_divmod): Likewise.
12661 (emit_store_flag_1): Likewise.
12662 * expr.c (init_expr_target): Likewise.
12663 (convert_move): Likewise.
12664 (alignment_for_piecewise_move): Likewise.
12665 (widest_int_mode_for_size): Likewise.
12666 (emit_block_move_via_movmem): Likewise.
12667 (copy_blkmode_to_reg): Likewise.
12668 (set_storage_via_setmem): Likewise.
12669 (compress_float_constant): Likewise.
12670 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12671 * optabs-query.c (get_best_extraction_insn): Likewise.
12672 * optabs.c (expand_binop): Likewise.
12673 (expand_twoval_unop): Likewise.
12674 (expand_twoval_binop): Likewise.
12675 (widen_leading): Likewise.
12676 (widen_bswap): Likewise.
12677 (expand_parity): Likewise.
12678 (expand_unop): Likewise.
12679 (prepare_cmp_insn): Likewise.
12680 (prepare_float_lib_cmp): Likewise.
12681 (expand_float): Likewise.
12682 (expand_fix): Likewise.
12683 (expand_sfix_optab): Likewise.
12684 * postreload.c (move2add_use_add2_insn): Likewise.
12685 * reg-stack.c (reg_to_stack): Likewise.
12686 * reginfo.c (choose_hard_reg_mode): Likewise.
12687 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12688 * stor-layout.c (mode_for_size): Likewise.
12689 (smallest_mode_for_size): Likewise.
12690 (mode_for_vector): Likewise.
12691 (finish_bitfield_representative): Likewise.
12692 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
12693 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
12694 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12695 * var-tracking.c (prepare_call_arguments): Likewise.
12697 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12698 Alan Hayward <alan.hayward@arm.com>
12699 David Sherwood <david.sherwood@arm.com>
12701 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
12702 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
12703 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
12704 * machmode.h (mode_size): Move earlier in file.
12705 (mode_precision): Likewise.
12706 (mode_inner): Likewise.
12707 (mode_nunits): Likewise.
12708 (mode_unit_size): Likewise.
12709 (unit_unit_precision): Likewise.
12710 (mode_wider): Likewise.
12711 (mode_2xwider): Likewise.
12712 (machine_mode): New class.
12713 (mode_to_bytes): New function.
12714 (mode_to_bits): Likewise.
12715 (mode_to_precision): Likewise.
12716 (mode_to_inner): Likewise.
12717 (mode_to_unit_size): Likewise.
12718 (mode_to_unit_precision): Likewise.
12719 (mode_to_nunits): Likewise.
12720 (GET_MODE_SIZE): Use mode_to_bytes.
12721 (GET_MODE_BITSIZE): Use mode_to_bits.
12722 (GET_MODE_PRECISION): Use mode_to_precision.
12723 (GET_MODE_INNER): Use mode_to_inner.
12724 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
12725 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
12726 (GET_MODE_NUNITS): Use mode_to_nunits.
12727 * system.h (ALWAYS_INLINE): New macro.
12728 * config/powerpcspe/powerpcspe-c.c
12729 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
12730 int for arg1_mode and arg2_mode.
12732 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12733 Alan Hayward <alan.hayward@arm.com>
12734 David Sherwood <david.sherwood@arm.com>
12736 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12737 Prefix mode names with E_ in case statements.
12738 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12739 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
12740 (aarch64_split_simd_move): Likewise.
12741 (aarch64_gen_storewb_pair): Likewise.
12742 (aarch64_gen_loadwb_pair): Likewise.
12743 (aarch64_gen_store_pair): Likewise.
12744 (aarch64_gen_load_pair): Likewise.
12745 (aarch64_get_condition_code_1): Likewise.
12746 (aarch64_constant_pool_reload_icode): Likewise.
12747 (get_rsqrte_type): Likewise.
12748 (get_rsqrts_type): Likewise.
12749 (get_recpe_type): Likewise.
12750 (get_recps_type): Likewise.
12751 (aarch64_gimplify_va_arg_expr): Likewise.
12752 (aarch64_simd_container_mode): Likewise.
12753 (aarch64_emit_load_exclusive): Likewise.
12754 (aarch64_emit_store_exclusive): Likewise.
12755 (aarch64_expand_compare_and_swap): Likewise.
12756 (aarch64_gen_atomic_cas): Likewise.
12757 (aarch64_emit_bic): Likewise.
12758 (aarch64_emit_atomic_swap): Likewise.
12759 (aarch64_emit_atomic_load_op): Likewise.
12760 (aarch64_evpc_trn): Likewise.
12761 (aarch64_evpc_uzp): Likewise.
12762 (aarch64_evpc_zip): Likewise.
12763 (aarch64_evpc_ext): Likewise.
12764 (aarch64_evpc_rev): Likewise.
12765 (aarch64_evpc_dup): Likewise.
12766 (aarch64_gen_ccmp_first): Likewise.
12767 (aarch64_gen_ccmp_next): Likewise.
12768 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
12769 (alpha_emit_xfloating_libcall): Likewise.
12770 (emit_insxl): Likewise.
12771 (alpha_arg_type): Likewise.
12772 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
12773 (arc_preferred_simd_mode): Likewise.
12774 (arc_secondary_reload): Likewise.
12775 (get_arc_condition_code): Likewise.
12776 (arc_print_operand): Likewise.
12777 (arc_legitimate_constant_p): Likewise.
12778 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12779 * config/arc/arc.md (casesi_load): Likewise.
12780 (casesi_compact_jump): Likewise.
12781 * config/arc/predicates.md (proper_comparison_operator): Likewise.
12782 (cc_use_register): Likewise.
12783 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12784 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12785 (arm_init_iwmmxt_builtins): Likewise.
12786 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
12787 (neon_expand_vector_init): Likewise.
12788 (arm_attr_length_move_neon): Likewise.
12789 (maybe_get_arm_condition_code): Likewise.
12790 (arm_emit_vector_const): Likewise.
12791 (arm_preferred_simd_mode): Likewise.
12792 (arm_output_iwmmxt_tinsr): Likewise.
12793 (thumb1_output_casesi): Likewise.
12794 (thumb2_output_casesi): Likewise.
12795 (arm_emit_load_exclusive): Likewise.
12796 (arm_emit_store_exclusive): Likewise.
12797 (arm_expand_compare_and_swap): Likewise.
12798 (arm_evpc_neon_vuzp): Likewise.
12799 (arm_evpc_neon_vzip): Likewise.
12800 (arm_evpc_neon_vrev): Likewise.
12801 (arm_evpc_neon_vtrn): Likewise.
12802 (arm_evpc_neon_vext): Likewise.
12803 (arm_validize_comparison): Likewise.
12804 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
12805 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
12806 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
12807 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
12808 (c6x_preferred_simd_mode): Likewise.
12809 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
12810 (epiphany_rtx_costs): Likewise.
12811 * config/epiphany/predicates.md (proper_comparison_operator):
12813 * config/frv/frv.c (condexec_memory_operand): Likewise.
12814 (frv_emit_move): Likewise.
12815 (output_move_single): Likewise.
12816 (output_condmove_single): Likewise.
12817 (frv_hard_regno_mode_ok): Likewise.
12818 (frv_matching_accg_mode): Likewise.
12819 * config/h8300/h8300.c (split_adds_subs): Likewise.
12820 (h8300_rtx_costs): Likewise.
12821 (h8300_print_operand): Likewise.
12822 (compute_mov_length): Likewise.
12823 (output_logical_op): Likewise.
12824 (compute_logical_op_length): Likewise.
12825 (compute_logical_op_cc): Likewise.
12826 (h8300_shift_needs_scratch_p): Likewise.
12827 (output_a_shift): Likewise.
12828 (compute_a_shift_length): Likewise.
12829 (compute_a_shift_cc): Likewise.
12830 (expand_a_rotate): Likewise.
12831 (output_a_rotate): Likewise.
12832 * config/i386/i386.c (classify_argument): Likewise.
12833 (function_arg_advance_32): Likewise.
12834 (function_arg_32): Likewise.
12835 (function_arg_64): Likewise.
12836 (function_value_64): Likewise.
12837 (ix86_gimplify_va_arg): Likewise.
12838 (ix86_legitimate_constant_p): Likewise.
12839 (put_condition_code): Likewise.
12840 (split_double_mode): Likewise.
12841 (ix86_avx256_split_vector_move_misalign): Likewise.
12842 (ix86_expand_vector_logical_operator): Likewise.
12843 (ix86_split_idivmod): Likewise.
12844 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
12845 (ix86_build_const_vector): Likewise.
12846 (ix86_build_signbit_mask): Likewise.
12847 (ix86_match_ccmode): Likewise.
12848 (ix86_cc_modes_compatible): Likewise.
12849 (ix86_expand_branch): Likewise.
12850 (ix86_expand_sse_cmp): Likewise.
12851 (ix86_expand_sse_movcc): Likewise.
12852 (ix86_expand_int_sse_cmp): Likewise.
12853 (ix86_expand_vec_perm_vpermi2): Likewise.
12854 (ix86_expand_vec_perm): Likewise.
12855 (ix86_expand_sse_unpack): Likewise.
12856 (ix86_expand_int_addcc): Likewise.
12857 (ix86_split_to_parts): Likewise.
12858 (ix86_vectorize_builtin_gather): Likewise.
12859 (ix86_vectorize_builtin_scatter): Likewise.
12860 (avx_vpermilp_parallel): Likewise.
12861 (inline_memory_move_cost): Likewise.
12862 (ix86_tieable_integer_mode_p): Likewise.
12863 (x86_maybe_negate_const_int): Likewise.
12864 (ix86_expand_vector_init_duplicate): Likewise.
12865 (ix86_expand_vector_init_one_nonzero): Likewise.
12866 (ix86_expand_vector_init_one_var): Likewise.
12867 (ix86_expand_vector_init_concat): Likewise.
12868 (ix86_expand_vector_init_interleave): Likewise.
12869 (ix86_expand_vector_init_general): Likewise.
12870 (ix86_expand_vector_set): Likewise.
12871 (ix86_expand_vector_extract): Likewise.
12872 (emit_reduc_half): Likewise.
12873 (ix86_emit_i387_round): Likewise.
12874 (ix86_mangle_type): Likewise.
12875 (ix86_expand_round_sse4): Likewise.
12876 (expand_vec_perm_blend): Likewise.
12877 (canonicalize_vector_int_perm): Likewise.
12878 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
12879 (expand_vec_perm_1): Likewise.
12880 (expand_vec_perm_interleave3): Likewise.
12881 (expand_vec_perm_even_odd_pack): Likewise.
12882 (expand_vec_perm_even_odd_1): Likewise.
12883 (expand_vec_perm_broadcast_1): Likewise.
12884 (ix86_vectorize_vec_perm_const_ok): Likewise.
12885 (ix86_expand_vecop_qihi): Likewise.
12886 (ix86_expand_mul_widen_hilo): Likewise.
12887 (ix86_expand_sse2_abs): Likewise.
12888 (ix86_expand_pextr): Likewise.
12889 (ix86_expand_pinsr): Likewise.
12890 (ix86_preferred_simd_mode): Likewise.
12891 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
12892 * config/i386/sse.md (*andnot<mode>3): Likewise.
12893 (<mask_codefor><code><mode>3<mask_name>): Likewise.
12894 (*<code><mode>3): Likewise.
12895 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
12896 (ia64_expand_atomic_op): Likewise.
12897 (ia64_arg_type): Likewise.
12898 (ia64_mode_to_int): Likewise.
12899 (ia64_scalar_mode_supported_p): Likewise.
12900 (ia64_vector_mode_supported_p): Likewise.
12901 (expand_vec_perm_broadcast): Likewise.
12902 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
12903 (iq2000_function_arg_advance): Likewise.
12904 (iq2000_function_arg): Likewise.
12905 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
12906 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
12907 (m68k_libcall_value): Likewise.
12908 (m68k_function_value): Likewise.
12909 (sched_attr_op_type): Likewise.
12910 * config/mcore/mcore.c (mcore_output_move): Likewise.
12911 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
12913 (microblaze_function_arg): Likewise.
12914 * config/mips/mips.c (mips16_build_call_stub): Likewise.
12915 (mips_print_operand): Likewise.
12916 (mips_mode_ok_for_mov_fmt_p): Likewise.
12917 (mips_vector_mode_supported_p): Likewise.
12918 (mips_preferred_simd_mode): Likewise.
12919 (mips_expand_vpc_loongson_even_odd): Likewise.
12920 (mips_expand_vec_unpack): Likewise.
12921 (mips_expand_vi_broadcast): Likewise.
12922 (mips_expand_vector_init): Likewise.
12923 (mips_expand_vec_reduc): Likewise.
12924 (mips_expand_msa_cmp): Likewise.
12925 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
12926 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
12927 (cc_flags_for_mode): Likewise.
12928 * config/msp430/msp430.c (msp430_print_operand): Likewise.
12929 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
12930 (nds32_output_casesi_pc_relative): Likewise.
12931 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12932 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
12933 (nvptx_gen_unpack): Likewise.
12934 (nvptx_gen_pack): Likewise.
12935 (nvptx_gen_shuffle): Likewise.
12936 (nvptx_gen_wcast): Likewise.
12937 (nvptx_preferred_simd_mode): Likewise.
12938 * config/pa/pa.c (pa_secondary_reload): Likewise.
12939 * config/pa/predicates.md (base14_operand): Likewise.
12940 * config/powerpcspe/powerpcspe-c.c
12941 (altivec_resolve_overloaded_builtin): Likewise.
12942 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
12944 (rs6000_preferred_simd_mode): Likewise.
12945 (output_vec_const_move): Likewise.
12946 (rs6000_expand_vector_extract): Likewise.
12947 (rs6000_split_vec_extract_var): Likewise.
12948 (reg_offset_addressing_ok_p): Likewise.
12949 (rs6000_legitimate_offset_address_p): Likewise.
12950 (rs6000_legitimize_address): Likewise.
12951 (rs6000_emit_set_const): Likewise.
12952 (rs6000_const_vec): Likewise.
12953 (rs6000_emit_move): Likewise.
12954 (spe_build_register_parallel): Likewise.
12955 (rs6000_darwin64_record_arg_recurse): Likewise.
12956 (swap_selector_for_mode): Likewise.
12957 (spe_init_builtins): Likewise.
12958 (paired_init_builtins): Likewise.
12959 (altivec_init_builtins): Likewise.
12960 (do_load_for_compare): Likewise.
12961 (rs6000_generate_compare): Likewise.
12962 (rs6000_expand_float128_convert): Likewise.
12963 (emit_load_locked): Likewise.
12964 (emit_store_conditional): Likewise.
12965 (rs6000_output_function_epilogue): Likewise.
12966 (rs6000_handle_altivec_attribute): Likewise.
12967 (rs6000_function_value): Likewise.
12968 (emit_fusion_gpr_load): Likewise.
12969 (emit_fusion_p9_load): Likewise.
12970 (emit_fusion_p9_store): Likewise.
12971 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
12972 (fusion_gpr_mem_load): Likewise.
12973 (fusion_addis_mem_combo_load): Likewise.
12974 (fusion_addis_mem_combo_store): Likewise.
12975 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
12976 (fusion_gpr_mem_load): Likewise.
12977 (fusion_addis_mem_combo_load): Likewise.
12978 (fusion_addis_mem_combo_store): Likewise.
12979 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12981 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
12982 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12983 (rs6000_preferred_simd_mode): Likewise.
12984 (output_vec_const_move): Likewise.
12985 (rs6000_expand_vector_extract): Likewise.
12986 (rs6000_split_vec_extract_var): Likewise.
12987 (reg_offset_addressing_ok_p): Likewise.
12988 (rs6000_legitimate_offset_address_p): Likewise.
12989 (rs6000_legitimize_address): Likewise.
12990 (rs6000_emit_set_const): Likewise.
12991 (rs6000_const_vec): Likewise.
12992 (rs6000_emit_move): Likewise.
12993 (rs6000_darwin64_record_arg_recurse): Likewise.
12994 (swap_selector_for_mode): Likewise.
12995 (paired_init_builtins): Likewise.
12996 (altivec_init_builtins): Likewise.
12997 (rs6000_expand_float128_convert): Likewise.
12998 (emit_load_locked): Likewise.
12999 (emit_store_conditional): Likewise.
13000 (rs6000_output_function_epilogue): Likewise.
13001 (rs6000_handle_altivec_attribute): Likewise.
13002 (rs6000_function_value): Likewise.
13003 (emit_fusion_gpr_load): Likewise.
13004 (emit_fusion_p9_load): Likewise.
13005 (emit_fusion_p9_store): Likewise.
13006 * config/rx/rx.c (rx_gen_move_template): Likewise.
13007 (flags_from_mode): Likewise.
13008 * config/s390/predicates.md (s390_alc_comparison): Likewise.
13009 (s390_slb_comparison): Likewise.
13010 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
13011 (s390_vector_mode_supported_p): Likewise.
13012 (s390_cc_modes_compatible): Likewise.
13013 (s390_match_ccmode_set): Likewise.
13014 (s390_canonicalize_comparison): Likewise.
13015 (s390_emit_compare_and_swap): Likewise.
13016 (s390_branch_condition_mask): Likewise.
13017 (s390_rtx_costs): Likewise.
13018 (s390_secondary_reload): Likewise.
13019 (__SECONDARY_RELOAD_CASE): Likewise.
13020 (s390_expand_cs): Likewise.
13021 (s390_preferred_simd_mode): Likewise.
13022 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
13023 * config/sh/sh.c (sh_print_operand): Likewise.
13024 (dump_table): Likewise.
13025 (sh_secondary_reload): Likewise.
13026 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13027 * config/sh/sh.md (casesi_worker_1): Likewise.
13028 (casesi_worker_2): Likewise.
13029 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
13030 (fcc_comparison_operator): Likewise.
13031 * config/sparc/sparc.c (sparc_expand_move): Likewise.
13032 (emit_soft_tfmode_cvt): Likewise.
13033 (sparc_preferred_simd_mode): Likewise.
13034 (output_cbranch): Likewise.
13035 (sparc_print_operand): Likewise.
13036 (sparc_expand_vec_perm_bmask): Likewise.
13037 (vector_init_bshuffle): Likewise.
13038 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13039 (spu_vector_mode_supported_p): Likewise.
13040 (spu_expand_insv): Likewise.
13041 (spu_emit_branch_or_set): Likewise.
13042 (spu_handle_vector_attribute): Likewise.
13043 (spu_builtin_splats): Likewise.
13044 (spu_builtin_extract): Likewise.
13045 (spu_builtin_promote): Likewise.
13046 (spu_expand_sign_extend): Likewise.
13047 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13048 (tilegx_simd_int): Likewise.
13049 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
13050 (tilepro_simd_int): Likewise.
13051 * config/v850/v850.c (const_double_split): Likewise.
13052 (v850_print_operand): Likewise.
13053 (ep_memory_offset): Likewise.
13054 * config/vax/vax.c (vax_rtx_costs): Likewise.
13055 (vax_output_int_move): Likewise.
13056 (vax_output_int_add): Likewise.
13057 (vax_output_int_subtract): Likewise.
13058 * config/visium/predicates.md (visium_branch_operator): Likewise.
13059 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
13060 (visium_print_operand_address): Likewise.
13061 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13062 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
13063 (xtensa_expand_conditional_branch): Likewise.
13064 (xtensa_copy_incoming_a7): Likewise.
13065 (xtensa_output_literal): Likewise.
13066 * dfp.c (decimal_real_maxval): Likewise.
13067 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13069 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13070 Alan Hayward <alan.hayward@arm.com>
13071 David Sherwood <david.sherwood@arm.com>
13073 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
13074 (mode_nunits_inline): Likewise.
13075 (mode_inner_inline): Likewise.
13076 (mode_unit_size_inline): Likewise.
13077 (mode_unit_precision_inline): Likewise.
13078 (emit_insn_modes_h): Likewise. Also emit a #define of the
13080 (emit_mode_wider): Add an E_ prefix to mode names.
13081 (emit_mode_complex): Likewise.
13082 (emit_mode_inner): Likewise.
13083 (emit_mode_adjustments): Likewise.
13084 (emit_mode_int_n): Likewise.
13085 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
13086 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
13087 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
13088 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
13089 (CRC32_BUILTIN, ENTRY): Likewise.
13090 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
13091 (aarch64_pop_regs): Likewise.
13092 (aarch64_process_components): Likewise.
13093 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
13094 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
13095 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
13096 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
13097 * config/arm/arm.c (arm_init_libfuncs): Likewise.
13098 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
13100 * config/i386/i386-builtin.def (pcmpestr): Likewise.
13101 (pcmpistr): Likewise.
13102 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
13103 * config/mmix/mmix.c (mmix_output_condition): Likewise.
13104 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13106 * config/rl78/rl78.c (mduc_regs): Likewise.
13107 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13108 (htm_expand_builtin): Likewise.
13109 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
13110 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
13111 * config/xtensa/xtensa.c (print_operand): Likewise.
13112 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
13113 (NUM_MODE_VECTOR_INT): Likewise.
13114 * genoutput.c (null_operand): Likewise.
13115 (output_operand_data): Likewise.
13116 * genrecog.c (print_parameter_value): Likewise.
13117 * lra.c (debug_operand_data): Likewise.
13119 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13121 * dwarf2out.c (scompare_loc_descriptor_wide)
13122 (scompare_loc_descriptor_narrow): New functions, split out from...
13123 (scompare_loc_descriptor): ...here.
13124 * expmed.c (emit_store_flag_int): New function, split out from...
13125 (emit_store_flag): ...here.
13127 2017-08-30 Richard Biener <rguenther@suse.de>
13129 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
13130 (dwarf2out_early_finish): Move setting of AT_pubnames from
13131 early debug output to early finish.
13133 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13135 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
13136 and -mdata-region to the assembler.
13138 2017-08-30 Richard Biener <rguenther@suse.de>
13140 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
13142 (gen_subprogram_die): Add DW_AT_object_pointer only early.
13143 (dwarf2out_early_global_decl): Only generate a DIE for the
13144 abstract origin if it doesn't already exist or is a declaration DIE.
13145 (resolve_addr): Do not add the linkage name twice when
13146 generating a stub DIE for the DW_TAG_GNU_call_site target.
13148 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13150 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13151 Use machine_mode rather than int for arg1_mode.
13153 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13156 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
13157 that the second argument of the built-in functions to unpack
13158 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
13159 switch statement instead a lot of if statements.
13160 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
13161 Allow 64-bit values to be in Altivec registers as well as
13162 traditional floating point registers.
13163 (pack<mode>, FMOVE128_VSX iterator): Likewise.
13165 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
13167 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
13168 MAX_REGS_PER_ADDRESS == 1.
13170 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
13172 * config/i386/i386.opt (flag_fentry): Do not init to -1.
13173 * config/i386/i386.c (ix86_option_override_internal): Simplify
13174 setting of opts->x_flag_entry.
13176 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13177 Jakub Jelinek <jakub@redhat.com>
13178 Richard Biener <rguenther@suse.de>
13180 PR tree-optimization/81503
13181 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13182 folded constant fits in the target type; reorder tests for clarity.
13184 2017-08-29 Martin Liska <mliska@suse.cz>
13186 * passes.def: Include pass_lower_switch.
13187 * stmt.c (dump_case_nodes): Remove and move to
13188 tree-switch-conversion.
13189 (case_values_threshold): Likewise.
13190 (expand_switch_as_decision_tree_p): Likewise.
13191 (emit_case_decision_tree): Likewise.
13192 (expand_case): Likewise.
13193 (balance_case_nodes): Likewise.
13194 (node_has_low_bound): Likewise.
13195 (node_has_high_bound): Likewise.
13196 (node_is_bounded): Likewise.
13197 (emit_case_nodes): Likewise.
13198 (struct simple_case_node): New struct.
13199 (add_case_node): Remove.
13200 (emit_case_dispatch_table): Use vector instead of case_list.
13201 (reset_out_edges_aux): Remove.
13202 (compute_cases_per_edge): Likewise.
13203 (expand_case): Build list of simple_case_node.
13204 (expand_sjlj_dispatch_table): Use it.
13205 * tree-switch-conversion.c (struct case_node): Moved from
13206 stmt.c and adjusted.
13207 (emit_case_nodes): Likewise.
13208 (node_has_low_bound): Likewise.
13209 (node_has_high_bound): Likewise.
13210 (node_is_bounded): Likewise.
13211 (case_values_threshold): Likewise.
13212 (reset_out_edges_aux): Likewise.
13213 (compute_cases_per_edge): Likewise.
13214 (add_case_node): Likewise.
13215 (dump_case_nodes): Likewise.
13216 (balance_case_nodes): Likewise.
13217 (expand_switch_as_decision_tree_p): Likewise.
13218 (emit_jump): Likewise.
13219 (emit_case_decision_tree): Likewise.
13220 (try_switch_expansion): Likewise.
13221 (do_jump_if_equal): Likewise.
13222 (emit_cmp_and_jump_insns): Likewise.
13223 (fix_phi_operands_for_edge): New function.
13224 (record_phi_operand_mapping): Likewise.
13225 (class pass_lower_switch): New pass.
13226 (pass_lower_switch::execute): New function.
13227 (make_pass_lower_switch): Likewise.
13228 (conditional_probability):
13229 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13230 * tree-pass.h: Add make_pass_lower_switch.
13232 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13235 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13238 2017-08-29 Richard Biener <rguenther@suse.de>
13240 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13241 we do not add a DW_AT_inline attribute twice.
13242 (gen_subprogram_die): Remove code setting DW_AT_inline on
13243 DECL_ABSTRACT_P nodes.
13245 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13247 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13248 calls to internal functions.
13249 (gimplify_modify_expr): Likewise.
13250 * tree-call-cdce.c (use_internal_fn): Likewise.
13251 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13252 (convert_to_divmod): Set the nothrow flag.
13253 * tree-if-conv.c (predicate_mem_writes): Likewise.
13254 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13255 (vectorizable_call): Likewise.
13256 (vectorizable_store): Likewise.
13257 (vectorizable_load): Likewise.
13258 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13259 (vect_recog_mask_conversion_pattern): Likewise.
13261 2017-08-29 Martin Liska <mliska@suse.cz>
13264 * gcc.c (driver_handle_option): Add new argument.
13265 * opts-common.c (handle_option): Pass
13266 target_option_override_hook.
13267 * opts-global.c (lang_handle_option): Add new option.
13268 (set_default_handlers): Add new argument.
13269 (decode_options): Likewise.
13270 * opts.c (target_handle_option): Likewise.
13271 (common_handle_option): Call target_option_override_hook.
13272 * opts.h (struct cl_option_handler_func): Add hook for
13273 target option override.
13274 (struct cl_option_handlers): Likewise.
13275 (set_default_handlers): Add new argument.
13276 (decode_options): Likewise.
13277 (common_handle_option): Likewise.
13278 (target_handle_option): Likewise.
13279 * toplev.c (toplev::main): Pass targetm.target_option.override
13282 2017-08-29 Richard Biener <rguenther@suse.de>
13283 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13285 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13286 life to the active subtree.
13288 2017-08-28 Jeff Law <law@redhat.com>
13290 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13291 derive_equivalences.
13292 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13293 Code moved into....
13294 (edge_info::derive_equivalences): New private member function
13296 * tree-ssa-dom.c (class edge_info): Changed from a struct
13297 to a class. Add ctor/dtor, methods and data members.
13298 (edge_info::edge_info): Renamed from allocate_edge_info.
13299 Initialize additional members.
13300 (edge_info::~edge_info): New.
13301 (free_dom_edge_info): Delete the edge info.
13302 (record_edge_info): Use new class & associated member functions.
13303 Tighten forms for testing for edge equivalences.
13304 (record_temporary_equivalences): Iterate over the simple
13305 equivalences rather than assuming there's only one per edge.
13306 (cprop_into_successor_phis): Iterate over the simple
13307 equivalences rather than assuming there's only one per edge.
13308 (optimize_stmt): Use operand_equal_p rather than pointer
13309 equality for mini-DSE code.
13311 2017-08-28 Nathan Sidwell <nathan@acm.org>
13313 * gcc.c (execute): Fold SIGPIPE handling into switch
13314 statement. Adjust internal error message.
13316 2017-08-28 Richard Biener <rguenther@suse.de>
13319 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13320 Do nothing for removed DIEs.
13322 2017-08-28 Richard Biener <rguenther@suse.de>
13324 PR tree-optimization/81977
13325 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13328 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13331 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13332 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13333 using targetm.gen_mem_thread_fence.
13335 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13338 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13339 predicate to register_operand. Reorder operands.
13340 (*btr<mode>): Ditto.
13341 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13342 (*btr<mode>_mask): Ditto.
13344 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13346 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13347 * config/rs6000/xmmintrin.h: New file.
13348 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13350 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13353 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13354 parameter and_insn and return it.
13355 (recombine_lvx_pattern): Insert a copy to ensure availability of
13356 the base register of the copied masking operation at the point of
13357 the instruction replacement.
13358 (recombine_stvx_pattern): Likewise.
13360 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13362 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13363 undocumented switches.
13364 (-mpower9-dform-vector): Likewise.
13365 (-mpower9-dform): Likewise.
13366 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13367 comments to delete references to -mpower9-dform* switches.
13368 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13369 Delete reference to -mpower9-dform* switches, test for
13370 -mpower9-vector instead.
13371 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13372 (OTHER_P9_VECTOR_MASKS): Likewise.
13373 (POWERPC_MASKS): Likewise.
13374 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13375 tests against -mpower9-dform* to -mpower9-vector. Delete code
13376 that checked for -mpower9-dform* consistancy with other options.
13377 Add test for -mpower9-misc to enable other power9 switches.
13378 (rs6000_init_hard_regno_mode_ok): Likewise.
13379 (rs6000_option_override_internal): Likewise.
13380 (rs6000_emit_prologue): Likewise.
13381 (rs6000_emit_epilogue): Likewise.
13382 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13383 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13384 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13386 (emit_fusion_p9_store): Likewise.
13387 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13388 resetting these macros if the assembler does not support ISA 3.0
13390 (TARGET_P9_DFORM_VECTOR): Likewise.
13391 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13392 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13394 2017-08-25 Alan Modra <amodra@gmail.com>
13396 PR rtl-optimization/81747
13397 * cse.c (cse_extended_basic_block): Don't attempt to record
13398 equivalences for degenerate conditional branches that branch
13399 to their fall-through.
13401 2017-08-24 Martin Sebor <msebor@redhat.com>
13403 PR middle-end/81908
13404 * gimple-fold.c (size_must_be_zero_p): New function.
13405 (gimple_fold_builtin_memory_op): Call it.
13407 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13409 * config/rs6000/mm_malloc.h: New file.
13411 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13413 PR tree-optimization/81913
13414 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13415 analysis when either IVs in condition can wrap.
13417 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13419 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13420 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13422 2017-08-24 Richard Biener <rguenther@suse.de>
13425 * targhooks.c (default_target_can_inline_p): Properly
13426 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13427 is present and always compare.
13428 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13429 infer -mfpmath=sse from TARGET_SSE_P.
13430 (ix86_can_inline_p): Properly use target_option_default_node when
13431 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13433 2017-08-24 Richard Biener <rguenther@suse.de>
13436 * dwarf2out.c (output_die): Handle flag_generate_offload like
13438 (output_comp_unit): Likewise.
13439 (gen_array_type_die): Likewise.
13440 (dwarf2out_early_finish): Likewise.
13441 (note_variable_value_in_expr): Likewise.
13442 (dwarf2out_finish): Likewise. Adjust assert.
13443 * cgraphunit.c (symbol_table::compile): Move setting of
13444 flag_generate_offload earlier ...
13445 (symbol_table::finalize_compilation_unit): ... here, before
13446 early debug finalization.
13448 2017-08-24 Richard Biener <rguenther@suse.de>
13450 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13451 and ipa-fnsummary.h.
13452 (ix86_can_inline_p): When ix86_fpmath flags do not match
13453 check whether the callee uses FP math at all.
13455 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13457 PR middle-end/81931
13458 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13459 instead of TYPE_PRECISION.
13461 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13462 Alan Hayward <alan.hayward@arm.com>
13463 David Sherwood <david.sherwood@arm.com>
13465 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13467 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13469 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
13471 (ix86_frame::stack_realign_allocate): New field.
13472 (struct machine_frame_state): Modify comments.
13473 (machine_frame_state::sp_realigned_fp_end): New field.
13474 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
13475 layout calculation.
13476 (sp_valid_at): Add assertion to assure no attempt to access invalid
13477 offset of a realigned stack.
13478 (fp_valid_at): Likewise.
13479 (choose_baseaddr): Modify comments.
13480 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
13481 ix86_expand_prologue.
13482 (ix86_expand_prologue): Modify stack realignment and allocation.
13483 (ix86_expand_epilogue): Modify comments.
13484 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
13485 avx2_runtime, avx512f, and avx512f_runtime.
13487 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
13489 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
13490 (mstackrealign): Do not init to -1.
13491 * config/i386/i386.c (ix86_option_override_internal):
13492 Check opts_set, not opts when setting default value of
13493 opts->x_ix86_force_align_arg_pointer.
13495 2017-08-23 Richard Biener <rguenther@suse.de>
13497 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
13498 lang_hooks.decl_printable_name.
13499 * print-rtl-function.c (print_rtx_function): Likewise.
13500 * tree-pretty-print.c (dump_function_header): Likewise.
13502 2017-08-23 Richard Biener <rguenther@suse.de>
13505 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
13506 -g0 at compile-time.
13508 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13510 PR middle-end/19706
13511 * doc/sourcebuild.texi (Other hardware attributes):
13514 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13516 PR middle-end/19706
13517 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
13518 Fix single-use check.
13520 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
13522 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
13524 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13526 * doc/install.texi: Modify to add more details on running selected
13529 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13531 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
13532 is combined with -mabi=ms.
13533 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
13536 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13538 PR tree-optimization/81488
13539 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
13540 and cached_basis fields.
13541 (MAX_SPREAD): New constant.
13542 (alloc_cand_and_find_basis): Initialize new fields.
13543 (clear_visited): New function.
13544 (create_phi_basis_1): Rename from create_phi_basis, set visited
13545 and cached_basis fields.
13546 (create_phi_basis): New wrapper function.
13547 (phi_add_costs_1): Rename from phi_add_costs, add spread
13548 parameter, set visited field, short-circuit when limits reached.
13549 (phi_add_costs): New wrapper function.
13550 (record_phi_increments_1): Rename from record_phi_increments, set
13552 (record_phi_increments): New wrapper function.
13553 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
13554 (phi_incr_cost): New wrapper function.
13555 (all_phi_incrs_profitable_1): Rename from
13556 all_phi_incrs_profitable, set visited field.
13557 (all_phi_incrs_profitable): New wrapper function.
13559 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13560 Alan Hayward <alan.hayward@arm.com>
13561 David Sherwood <david.sherwood@arm.com>
13563 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
13564 that takes the outer and inner modes.
13565 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
13566 comparison as the canonical test for a paradoxical subreg.
13567 * combine.c (simplify_set): Use paradoxical_subreg_p.
13568 (make_extraction): Likewise.
13569 (force_to_mode): Likewise.
13570 (rtx_equal_for_field_assignment_p): Likewise.
13571 (gen_lowpart_for_combine): Likewise.
13572 (simplify_comparison): Likewise.
13573 * cse.c (equiv_constant): Likewise.
13574 * expmed.c (store_bit_field_1): Likewise.
13575 * final.c (alter_subreg): Likewise.
13576 * fwprop.c (propagate_rtx): Likewise.
13577 (forward_propagate_subreg): Likewise.
13578 * ira-conflicts.c (ira_build_conflicts): Likewise.
13579 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
13580 * lra-constraints.c (curr_insn_transform): Likewise.
13581 (split_reg): Likewise.
13582 * lra-eliminations.c (move_plus_up): Likewise.
13583 (lra_eliminate_regs_1): Likewise.
13584 * recog.c (general_operand): Likewise.
13585 * ree.c (combine_reaching_defs): Likewise.
13586 * reload.c (push_reload): Likewise.
13587 (find_reloads): Likewise.
13588 * reload1.c (elimination_effects): Likewise.
13589 (compute_reload_subreg_offset): Likewise.
13590 (choose_reload_regs): Likewise.
13591 * rtlanal.c (subreg_lsb_1): Likewise.
13592 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13593 (simplify_subreg): Likewise.
13594 * var-tracking.c (track_loc_p): Likewise.
13595 * emit-rtl.c (byte_lowpart_offset): Likewise.
13596 (paradoxical_subreg_p): Delete out-of-line definition.
13598 2017-08-22 Jeff Law <law@redhat.com>
13600 PR tree-optimization/81741
13601 PR tree-optimization/71947
13602 * tree-ssa-dom.c: Include tree-inline.h.
13603 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
13604 equivalences if one is more expensive to compute than the other.
13605 * tree-ssa-scopedtables.h (class const_or_copies): Make
13606 record_const_or_copy_raw method private.
13607 (class avail_exprs_stack): New method simplify_binary_operation.
13608 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
13609 avail_exprs_stack::simplify_binary_operation as needed.
13610 (avail_exprs_stack::simplify_binary_operation): New function.
13612 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
13614 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
13615 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
13616 (DOT_SYMBOLS): Likewise.
13617 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
13618 (RELOCATABLE_NEEDS_FIXUP): Likewise.
13619 (RS6000_ABI_NAME): Likewise.
13620 (TARGET_CMODEL): Likewise.
13621 (TOC_SECTION_ASM_OP): Likewise.
13622 (SET_CMODEL): New macro.
13623 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
13625 2017-08-22 Richard Biener <rguenther@suse.de>
13627 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
13628 to allow for free-lang-data replacements similar to verify_type_variant.
13630 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13631 Alan Hayward <alan.hayward@arm.com>
13632 David Sherwood <david.sherwood@arm.com>
13634 * config/aarch64/aarch64.md (casesi): Use DImode rather than
13635 VOIDmode for the LABEL_REF.
13637 2017-08-22 Richard Biener <rguenther@suse.de>
13639 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
13641 2017-08-22 Richard Biener <rguenther@suse.de>
13643 * common.opt (feliminate-dwarf2-dups): Ignore.
13644 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
13645 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
13646 same_die_p_wrap, compute_section_prefix,
13647 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
13648 (comdat_symbol_id, comdat_symbol_number): Likewise.
13649 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
13651 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
13652 (output_die): Mark unreachable path unreachable.
13653 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
13654 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
13655 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
13656 (dwarf2out_early_finish): Likewise.
13658 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
13660 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
13662 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
13665 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
13666 not VAR_P. Filter attribute warnings with OPT_Wattributes.
13667 (avr_attribute_table) <io, io_low, address>: Initialize
13668 .decl_required with true.
13670 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
13672 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
13673 undocumented debugging options.
13674 (-mvsx-scalar-double): Likewise.
13675 (-mallow-df-permute): Likewise.
13676 (-mvectorize-builtins): Likewise.
13677 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13678 (rs6000_builtin_vectorized_function): Likewise.
13679 (rs6000_builtin_md_vectorized_function): Likewise.
13680 (rs6000_opt_vars): Likewise.
13682 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
13685 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
13686 (*btrq_imm): Rename from *btrq.
13687 (*btcq_imm): Rename from *btcq.
13688 (btsc): New code attribute.
13689 (*<btsc><mode>): New insn pattern.
13690 (*btr<mode>): Ditto.
13691 (*<btsc><mode>_mask): New insn_and_split pattern.
13692 (*btr<mode>_mask): Ditto.
13694 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13695 Alan Hayward <alan.hayward@arm.com>
13696 David Sherwood <david.sherwood@arm.com>
13698 * function.c (pad_below): Simplify padding calculation.
13700 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13701 Alan Hayward <alan.hayward@arm.com>
13702 David Sherwood <david.sherwood@arm.com>
13704 * target.def (function_prologue): Remove frame size argument.
13705 (function_epilogue): Likewise.
13706 * doc/tm.texi: Regenerate.
13707 * final.c (final_start_function): Update call to function_prologue.
13708 (final_end_function): Update call to function_epilogue.
13709 (default_function_pro_epilogue): Remove frame size argument.
13710 * output.h (default_function_pro_epilogue): Likewise.
13711 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
13712 (arm_output_function_prologue): Likewise.
13713 * config/frv/frv.c (frv_function_prologue): Likewise.
13714 (frv_function_epilogue): Likewise.
13715 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
13716 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
13717 (ia64_output_function_epilogue): Likewise.
13718 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
13719 (m32r_output_function_epilogue): Likewise.
13720 * config/microblaze/microblaze.c (microblaze_function_prologue)
13721 (microblaze_function_epilogue): Likewise.
13722 * config/mips/mips.c (mips_output_function_prologue): Likewise.
13723 (mips_output_function_epilogue): Likewise.
13724 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
13725 (mmix_target_asm_function_epilogue): Likewise.
13726 * config/msp430/msp430.c (msp430_start_function): Likewise.
13727 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
13728 (nds32_asm_function_epilogue): Likewise.
13729 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
13730 * config/pa/pa.c (pa_output_function_prologue): Likewise.
13731 (pa_output_function_epilogue): Likewise.
13732 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
13733 (rs6000_output_function_epilogue): Likewise.
13734 * config/rl78/rl78.c (rl78_start_function): Likewise.
13735 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
13736 (rs6000_output_function_epilogue): Likewise.
13737 * config/rx/rx.c (rx_output_function_prologue): Likewise.
13738 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
13739 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
13740 (sparc_asm_function_epilogue): Likewise.
13742 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13744 * tree.h (type_has_mode_precision_p): New function.
13745 * convert.c (convert_to_integer_1): Use it.
13746 * expr.c (expand_expr_real_2): Likewise.
13747 (expand_expr_real_1): Likewise.
13748 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
13749 * match.pd: Likewise.
13750 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
13751 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
13752 * tree-tailcall.c (process_assignment): Likewise.
13753 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13754 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
13755 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
13756 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13757 (vectorizable_assignment): Likewise.
13758 (vectorizable_shift): Likewise.
13759 (vectorizable_operation): Likewise.
13760 * tree-vrp.c (register_edge_assert_for_2): Likewise.
13762 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
13764 * match.pd: Add pow (C, x) simplification.
13766 2017-08-21 Richard Biener <rguenther@suse.de>
13768 PR tree-optimization/81900
13769 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
13770 for blocks with abnormal predecessors.
13771 (compute_antic): Do not set visited flag prematurely.
13773 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
13776 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
13778 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13780 * stor-layout.h (vector_type_mode): Move to...
13781 * tree.h (vector_type_mode): ...here.
13782 * stor-layout.c (vector_type_mode): Move to...
13783 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
13785 2017-08-21 Richard Biener <rguenther@suse.de>
13787 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
13788 register_external_die hooks.
13789 (debug_false_tree_charstarstar_uhwistar): Declare.
13790 (debug_nothing_tree_charstar_uhwi): Likewise.
13791 * debug.c (do_nothing_debug_hooks): Adjust.
13792 (debug_false_tree_charstarstar_uhwistar): New do nothing.
13793 (debug_nothing_tree_charstar_uhwi): Likewise.
13794 * dbxout.c (dbx_debug_hooks): Adjust.
13795 (xcoff_debug_hooks): Likewise.
13796 * sdbout.c (sdb_debug_hooks): Likewise.
13797 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13798 * dwarf2out.c (macinfo_label_base): New global.
13799 (dwarf2out_register_external_die): New function for the
13800 register_external_die hook.
13801 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
13802 (dwarf2_debug_hooks): Use them.
13803 (dwarf2_lineno_debug_hooks): Adjust.
13804 (struct die_struct): Add with_offset flag.
13805 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
13806 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
13807 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
13808 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
13809 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
13810 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
13811 defining section names for the early LTO debug variants.
13812 (reset_indirect_string): New helper.
13813 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
13814 (print_dw_val): Add support for offsetted symbol references.
13815 (get_ultimate_context): Split out from is_cxx.
13816 (is_cxx): Use get_ultimate_context.
13817 (is_fortran): Add decl overload.
13818 (compute_comp_unit_symbol): Split out worker from
13819 compute_section_prefix.
13820 (compute_section_prefix): Call compute_comp_unit_symbol and
13821 set comdat_type_p here.
13822 (output_die): Skip DIE symbol output for the LTO added one.
13823 Handle DIE symbol references with offset.
13824 (output_comp_unit): Guard section name mangling properly.
13825 For LTO debug sections emit a symbol at the section beginning
13826 which we use to refer to its DIEs.
13827 (add_abstract_origin_attribute): For DIEs registered via
13828 dwarf2out_register_external_die directly refer to the early
13829 DIE rather than indirectly through the shadow one we created.
13830 Remove obsolete call to dwarf2out_abstract_function for
13831 non-function/block origins.
13832 (gen_array_type_die): When generating early LTO debug do
13833 not emit DW_AT_string_length.
13834 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
13835 late when in LTO. As suggested place a gcc_unreachable for
13836 the DECL_ABSTRACT_P case.
13837 (gen_subprogram_die): Avoid another specification DIE
13838 for early built declarations/definitions for the late LTO case.
13839 (gen_variable_die): Add type references for late duplicated VLA dies
13841 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
13842 we have the abstract instance already.
13843 (process_scope_var): Adjust decl DIE contexts in LTO which
13844 first puts them in limbo.
13845 (gen_decl_die): Do not generate type DIEs late apart from
13846 types for VLAs or for decls we do not yet have a DIE. Do not
13847 call dwarf2out_abstract_function late.
13848 (dwarf2out_early_global_decl): Make sure to create DIEs
13849 for abstract instances of a decl first.
13850 (dwarf2out_late_global_decl): Adjust comment.
13851 (output_macinfo_op): With multiple macro sections use
13852 macinfo_label_base to distinguish labels.
13853 (output_macinfo): Likewise. Update macinfo_label_base.
13854 Pass in the line info label.
13855 (note_variable_value_in_expr): When generating LTO resolve
13856 all variable values here by generating DIEs as needed.
13857 (init_sections_and_labels): Add early LTO debug flag parameter
13858 and generate different sections and names if set. Add generation
13859 counter for the labels so we can have multiple of them.
13860 (reset_dies): Helper to allow DIEs to be output multiple times.
13861 (dwarf2out_finish): When outputting DIEs to the fat part of an
13862 LTO object first reset DIEs.
13863 (dwarf2out_early_finish): Output early DIEs when generating LTO.
13864 (modified_type_die): Check for decl_ultimate_origin being self
13866 (gen_type_die_with_usage): Likewise.
13867 (gen_typedef_die): Allow decl_ultimate_origin being self.
13868 (set_decl_abstract_flags): Remove.
13869 (set_block_abstract_flags): Likewise.
13870 (dwarf2out_abstract_function): Treat the early generated DIEs
13871 as the abstract copy and only add DW_AT_inline and
13872 DW_AT_artificial here and call set_decl_origin_self.
13873 If the DIE has an abstract origin don't do anything.
13874 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
13875 if we have none yet (Go fails to build one, PR78628).
13876 (variably_modified_type_p): Prevent endless recursion for Ada
13877 cyclic pointer types.
13878 * lto-streamer-in.c: Include debug.h.
13879 (dref_queue): New global.
13880 (lto_read_tree_1): Stream in DIE references.
13881 (lto_input_tree): Register DIE references.
13882 (input_function): Stream DECL_DEBUG_ARGS.
13883 * lto-streamer-out.c: Include debug.h.
13884 (lto_write_tree_1): Output DIE references.
13885 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
13886 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
13887 (output_function): Stream DECL_DEBUG_ARGS.
13888 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
13889 Stream DECL_ABSTRACT_ORIGIN.
13890 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
13891 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
13892 DECL_CONTEXT for file-scope decls.
13893 * lto-streamer.h (struct dref_entry): Declare.
13894 (dref_queue): Likewise.
13895 * cfgexpand.c (pass_expand::execute): Do not call the
13896 outlining_inline_function hook here.
13897 * lto-wrapper.c (debug_obj): New global.
13898 (tool_cleanup): Unlink it if required.
13899 (debug_objcopy): New function.
13900 (run_gcc): Handle early debug sections in the IL files by
13901 extracting them to separate files, partially linkin them and
13902 feeding the result back as result to the linker.
13903 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
13904 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
13905 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
13906 sections into a separate segment.
13907 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
13909 (darwin_asm_dwarf_section): Likewise.
13910 (darwin_asm_output_dwarf_offset): Likewise.
13911 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
13913 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13914 Alan Hayward <alan.hayward@arm.com>
13915 David Sherwood <david.sherwood@arm.com>
13917 * read-md.h (md_reader::record_potential_iterator_use): Replace
13918 pointer argument with an rtx and an index.
13919 * read-rtl.c (iterator_group::apply_iterator): Likewise.
13920 (apply_mode_iterator): Likewise.
13921 (apply_code_iterator): Likewise.
13922 (apply_int_iterator): Likewise.
13923 (apply_subst_iterator): Likewise.
13924 (record_iterator_use): Likewise.
13925 (record_attribute_use): Likewise.
13926 (md_reader::record_potential_iterator_use): Likewise. Update calls
13927 to record_iterator_use and apply_iterator.
13928 (iterator_use): Replace ptr with x and index.
13929 (attribute_use): Likewise.
13930 (apply_attribute_uses): Update calls to apply_iterator.
13931 (apply_iterators): Likewise. Update initialization of iterator_use.
13932 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
13933 and record_potential_iterator_use.
13934 (rtx_reader::read_rtx_operand): Likewise.
13936 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13937 Alan Hayward <alan.hayward@arm.com>
13938 David Sherwood <david.sherwood@arm.com>
13940 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
13943 2017-08-21 Richard Biener <rguenther@suse.de>
13945 PR middle-end/81884
13946 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
13947 at struct end conservatively when comparing common bases.
13949 2017-08-21 Richard Biener <rguenther@suse.de>
13951 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
13952 (mem_ref_in_stmt): Remove.
13953 (determine_max_movement): Use ref index to get at the reference.
13954 (invariantness_dom_walker::before_dom_children): Deal with
13955 lim data already initialized.
13956 (gather_mem_refs_stmt): Initialize lim data and record ref index.
13958 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13960 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
13961 (TARGET_ISA_ROUND): Ditto.
13962 (TARGET_ROUND): Ditto.
13963 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
13964 * config/i386/i386.md: Ditto.
13965 * config/i386/sse.md: Ditto.
13966 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
13967 with OPTION_MASK_ISA_SSE4_1.
13969 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13972 * doc/extend.texi (x86 Built-in Functions): Correct the name of
13973 __builtin_ia32_lzcnt_u16.
13975 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
13978 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
13979 (rs6000_set_current_function): Rewrite function to use it.
13981 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
13984 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
13985 and TYPE_WARN_IF_NOT_ALIGN.
13986 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
13987 (handle_warn_if_not_align): New.
13988 (place_union_field): Call handle_warn_if_not_align.
13989 (place_field): Call handle_warn_if_not_align.
13990 Copy TYPE_WARN_IF_NOT_ALIGN.
13991 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
13992 (layout_type): Likewise.
13993 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
13995 (tree_decl_common): Add warn_if_not_align.
13996 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
13997 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
13998 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
13999 (DECL_WARN_IF_NOT_ALIGN): Likewise.
14000 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
14001 * doc/extend.texi: Document warn_if_not_aligned attribute.
14002 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
14004 2017-08-17 Martin Liska <mliska@suse.cz>
14007 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
14008 (get_data_dependence): Use it as pointer type.
14009 (distribute_loop): Likewise.
14011 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14013 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
14014 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
14015 (p8_vmrgow_<mode>_direct): New define_insn.
14016 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
14017 handle endianness for vmrgew and vmrgow permute patterns.
14019 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14021 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
14022 * config/rs6000/rs6000-cpus.def: Remove comment.
14023 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
14024 (POWERPC_MASKS): Likewise.
14025 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
14026 use of TARGET_VSX_TIMODE.
14027 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
14028 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
14029 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
14030 (rs6000_option_override_internal): Remove dead code.
14031 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
14032 (rs6000_legitimize_reload_address): Likewise.
14033 (rs6000_legitimate_address_p): Likewise.
14034 (rs6000_opt_masks): Delete "vsx-timode".
14035 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
14036 from function comment.
14037 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
14038 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
14039 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
14041 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
14042 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
14043 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
14044 (VSX_TI): Likewise.
14046 (define_peephole2): Likewise.
14048 2017-08-17 Martin Sebor <msebor@redhat.com>
14051 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
14052 past the end of an array.
14053 (test_pp_format): Add test cases.
14055 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
14057 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
14058 missing ECF_NOTHROW flags.
14060 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14063 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
14064 operands residing in integer registers.
14065 (*vsx_le_perm_load_<mode>): Likewise.
14066 (*vsx_le_perm_store_<mode>): Likewise.
14067 (define_peephole2): Add peepholes to optimize the above.
14069 2017-08-17 Marek Polacek <polacek@redhat.com>
14071 PR middle-end/81814
14072 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
14073 to mimic what shorten_compare did. Change the return type to bool.
14074 (fold_cond_expr_with_comparison): Update call to
14075 operand_equal_for_comparison_p.
14076 (fold_ternary_loc): Likewise.
14078 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
14080 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
14082 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
14084 2017-08-17 Richard Biener <rguenther@suse.de>
14086 * tree-ssa-structalias.c (solve_graph): When propagating
14087 to successors update the graphs succ edges and avoid duplicate work.
14089 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
14092 * config/i386/i386.c (ix86_option_override_internal): Save target
14093 specific options after ix86_stack_protector_guard_reg was changed.
14095 2017-08-17 Richard Biener <rguenther@suse.de>
14097 PR tree-optimization/81827
14098 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
14099 (new_var_info): Initialize it conservatively.
14100 (get_call_vi): Mark register vars.
14101 (new_scalar_tmp_constraint_exp): Likewise.
14102 (handle_rhs_call): Likewise.
14103 (handle_const_call): Likewise.
14104 (create_function_info_for): Likewise.
14105 (solve_constraints): Sort varinfos to separate register from
14106 non-register vars to pack points-to solution bitmaps during
14109 2017-08-17 Marek Polacek <polacek@redhat.com>
14111 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
14113 2017-08-17 Richard Biener <rguenther@suse.de>
14115 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
14116 to true when overflow is undefined and we saturated the result.
14118 2017-08-17 Alan Modra <amodra@gmail.com>
14121 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
14122 Don't use store multiple if only one reg needs saving.
14123 (interesting_frame_related_regno): New function.
14124 (rs6000_frame_related): Don't emit frame info for regs that
14126 (rs6000_emit_epilogue): Likewise.
14128 2017-08-16 Nathan Sidwell <nathan@acm.org>
14130 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
14131 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
14132 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
14133 * tree.c (free_lang_data_in_type): Use else-if chain. Always
14134 clear TYPE_LANG_1. Remove obsolete member-function stripping.
14135 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
14136 (verify_type): Adjust for TYPE_BINFO move.
14137 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
14138 process TYPE_BINFO directly.
14139 (hash_tree): Likewise.
14140 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14142 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14145 2017-08-16 David Malcolm <dmalcolm@redhat.com>
14147 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
14149 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
14152 * config/i386/i386.md (*anddi_1_btr): Change predicates of
14153 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
14154 Add ix86_binary_operator_ok to insn constraint.
14155 (*iordi_1_bts): Ditto.
14156 (*xordi_1_btc): Ditto.
14157 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
14158 Update corresponding peephole2 pattern.
14162 2017-08-16 Bin Cheng <bin.cheng@arm.com>
14164 PR tree-optimization/81832
14165 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
14166 copy loop header which has IFN_LOOP_DIST_ALIAS call.
14168 2017-08-16 Marek Polacek <polacek@redhat.com>
14171 * fold-const.c (fold_indirect_ref_1): Restore original behavior
14172 regarding size_zero_node.
14174 2017-08-16 Martin Liska <mliska@suse.cz>
14177 * config.gcc: Respect previously set extra_objs in case
14180 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14182 PR tree-optimization/81835
14183 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14184 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14185 not depend on the phi.
14187 2017-08-16 Alan Modra <amodra@gmail.com>
14189 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14192 2017-08-16 Alan Modra <amodra@gmail.com>
14194 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14195 (save_reg_p): ..into this. Update all callers.
14196 (first_reg_to_save): Simplify.
14198 2017-08-16 Alan Modra <amodra@gmail.com>
14200 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14203 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14207 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14208 constant count if that count is less than 32.
14210 2017-08-15 Nathan Sidwell <nathan@acm.org>
14212 * gcc.c (execute): Emit friendlier message if inferior is killed
14213 by an external cause.
14215 2017-08-15 Richard Biener <rguenther@suse.de>
14217 PR tree-optimization/81790
14218 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14219 CONSTRUCTORs from simplifying and VN.
14221 2017-08-14 Martin Sebor <msebor@redhat.com>
14223 * builtin-attrs.def: Add comments.
14225 2017-08-14 Martin Sebor <msebor@redhat.com>
14228 * doc/extend.texi (attribute nonstring): Document new attribute.
14230 2017-08-14 Martin Sebor <msebor@redhat.com>
14233 * tree-diagnostic.c (default_tree_printer): Handle %G.
14234 * gimple-pretty-print.h (percent_G_format): Declare new function.
14235 * gimple-pretty-print.c (percent_G_format): Define.
14236 * tree-pretty-print.c (percent_K_format): Add argument.
14238 2017-08-14 Martin Sebor <msebor@redhat.com>
14240 PR translation/79998
14241 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14242 Remove a stray space.
14244 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14247 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14248 (*iordi_1_bts): Ditto.
14249 (*xordi_1_btc): Ditto.
14251 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14254 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14255 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14257 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14258 quoted strings, and make more translator-friendly.
14259 (darwin_rs6000_override_options): Likewise.
14260 (rs6000_option_override_internal): Likewise.
14261 (rs6000_return_in_memory): Fix overlong line.
14262 (init_cmulative_args): Use quoted strings, and make more
14263 translator-friendly.
14264 (rs6000_pass_by_reference): Fix overlong line.
14265 (def_builtin): Use quoted strings.
14266 (altivec_expand_predicate_builtin): Use quoted strings, and make
14267 more translator-friendly.
14268 (htm_expand_builtin): Use quoted strings.
14269 (cpu_expand_builtin): Use quoted strings, and make more
14270 translator-friendly.
14271 (altivec_expand_builtin): Likewise.
14272 (paired_expand_predicate_builtin): Likewise.
14273 (rs6000_invalid_builtin): Likewise.
14274 (builtin_function_type): Use quoted strings.
14275 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14276 more translator-friendly.
14277 (rs6000_trampoline_init): Likewise.
14278 (rs6000_handle_altivec_attribute): Likewise.
14279 (rs6000_inner_target_options): Use quoted strings.
14280 (rs6000_disable_incompatible_switches): Likewise.
14281 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14282 strings, and make more translator-friendly.
14283 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14285 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14287 PR tree-optimization/81799
14288 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14289 cond_expr to simple gimple operand.
14291 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14293 PR middle-end/46932
14294 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14296 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14300 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14301 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14302 TARGET_GASISR_PROLOGUES.
14303 * config/avr/avr.c (avr_option_override): Same.
14304 (avr_pass_pre_proep::execute): Same.
14306 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14309 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14310 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14312 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14314 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14315 define_insn_and_split. Split to a memory load from 0 in
14316 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14317 using PTR mode iterator.
14318 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14319 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14320 (*add_tp_<mode>): Redefine as define_insn_and_split.
14321 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14322 address space. Merge with *add_tp_x32 using PTR mode iterator.
14323 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14324 Split to an add with a memory load from 0 in
14325 DEFAULT_TLS_SEG_REG address space.
14327 2017-08-12 Andrew Pinski <apinski@cavium.com>
14329 * config/aarch64/aarch64-option-extensions.def (rdma):
14330 Fix feature string to what Linux prints out in /proc/cpuinfo.
14332 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14335 * dwarf2out.c (modified_type_die): For C typedef types that have
14336 an ultimate origin, process the ultimate origin instead of the
14338 (gen_typedef_die): Assert that input DECLs have no ultimate
14340 (gen_type_die_with_usage): For typedef variants that have an
14341 ultimate origin, just call gen_decl_die on the original DECL.
14342 (process_scope_var): Avoid creating DIEs for local typedefs and
14343 concrete static variables.
14345 2017-08-12 Alan Modra <amodra@gmail.com>
14349 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14350 match gnu-user.h startfile.
14351 (ENDFILE_LINUX_SPEC): Similarly.
14353 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14356 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14358 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14360 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14361 * config/aarch64/aarch64.md (mov<mode>): Change.
14362 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14363 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14364 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14366 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14368 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14369 for storage order barriers.
14371 2017-08-11 Martin Liska <mliska@suse.cz>
14374 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14375 variables of void type.
14377 2017-08-11 Martin Liska <mliska@suse.cz>
14379 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14380 TARGET_SUPPORTS_ALIASES.
14381 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14382 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14383 (optimize_weakref): Likewise.
14384 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14385 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14386 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14388 2017-08-11 Martin Liska <mliska@suse.cz>
14391 * config/i386/i386.c (make_resolver_func): Do complete
14392 refactoring of the function.
14394 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14397 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14398 * config/i386/i386.c (ix86_stack_protect_guard): Use
14399 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14400 * doc/invoke.texi (x86 Options): Document
14401 -mstack-protector-guard-symbol= option.
14403 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14405 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14406 * config/i386/i386.c (ix86_split_stack_guard): New function.
14407 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14408 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14409 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14410 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14411 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14412 (split_stack_space_check): Call ix86_split_stack_guard.
14414 2017-08-10 Martin Sebor <msebor@redhat.com>
14416 * print-tree.c (print_node): Print location using the established
14418 Replace spaces with colons.
14419 (debug_raw, debug): Ditto.
14421 2017-08-10 Martin Sebor <msebor@redhat.com>
14424 * pretty-print.c (pp_format): Correct the handling of %s precision.
14426 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14429 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14431 (ix86_finalize_stack_frame_flags): This. Also clear
14432 frame_pointer_needed if -fno-omit-frame-pointer is used without
14434 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14435 with ix86_finalize_stack_frame_flags.
14436 (ix86_expand_epilogue): Likewise.
14437 (ix86_expand_split_stack_prologue): Likewise.
14438 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14440 2017-08-10 Martin Liska <mliska@suse.cz>
14443 * c-attribs.c (handle_target_attribute):
14444 Report warning for an empty string argument of target attribute.
14446 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14449 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14451 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14452 or DECL_NONLOCAL labels.
14453 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
14454 or DECL_NONLOCAL labels here.
14456 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14458 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14459 to indicate when early gimple folding has been disabled.
14460 (rs6000_gimple_fold_builtin): Add debug content.
14461 (rs6000_invalid_builtin): Fix whitespace.
14462 (rs6000_expand_builtin): Fix whitespace.
14463 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14465 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14468 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14469 SAVE_MULTIPLE if not all the registers that saves, should be saved.
14471 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
14473 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
14474 (qdf24xx): Likewise.
14475 * config/aarch64/aarch64.md: Include falkor.md.
14476 * config/aarch64/falkor.md: New.
14478 2017-08-09 Marek Polacek <polacek@redhat.com>
14481 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
14482 * diagnostic.c (emit_diagnostic): Add a comment.
14483 (emit_diagnostic_valist): New function.
14485 2017-08-09 Marek Polacek <polacek@redhat.com>
14488 * input.c (make_location): New overload.
14489 * input.h (make_location): Declare.
14491 2017-08-08 Alan Modra <amodra@gmail.com>
14492 H.J. Lu <hongjiu.lu@intel.com>
14495 * gcc.c (NO_PIE_SPEC): Delete.
14496 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
14498 (LINK_PIE_SPEC): ..to here.
14499 (LINK_COMMAND_SPEC): Support -no-pie.
14500 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
14501 chain of crtbegin*.o selection, update for PIE_SPEC changes and
14503 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
14504 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
14505 (ENDFILE_CRTEND_SPEC): Similarly.
14507 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
14510 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
14511 (mstack-protector-guard-offset=): Ditto.
14512 * config/i386/i386.c (ix86_option_override): Handle
14513 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
14515 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
14516 ix86_stack_protect_guard_offset variables.
14517 (TARGET_STACK_PROTECT_GUARD): Always define.
14518 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
14519 and -mstack-protector-guard-offset= options.
14521 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14523 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
14524 boundary case for the last candidate.
14526 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14528 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
14529 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
14531 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14533 PR middle-end/19706
14534 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
14535 * config/aarch64/aarch64-builtins.c
14536 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
14537 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
14538 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
14540 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14541 Andrew Pinski <pinskia@gmail.com>
14543 PR middle-end/19706
14544 * internal-fn.def (XORSIGN): New.
14545 * optabs.def (xorsign_optab): New.
14546 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
14547 (convert_expand_mult_copysign): New.
14548 (pass_optimize_widening_mul::execute): Call
14549 convert_expand_mult_copysign.
14551 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14553 PR tree-optimization/81354
14554 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
14555 Insert on edges rather than explicitly creating landing pads.
14556 (analyze_candidates_and_replace): Commit edge inserts.
14558 2017-08-08 Richard Biener <rguenther@suse.de>
14560 PR middle-end/81719
14561 * tree-ssa-loop-niter.c: Include tree-dfa.h.
14562 (expand_simple_operations): Also look through ADDR_EXPRs with
14563 MEM_REF bases treating them as POINTER_PLUS_EXPR.
14565 2017-08-08 Richard Biener <rguenther@suse.de>
14567 PR tree-optimization/81723
14568 * tree-vect-slp.c (struct bst_traits): New hash traits.
14569 (bst_fail): New global.
14570 (vect_build_slp_tree_2): New worker, split out from ...
14571 (vect_build_slp_tree): ... this now wrapping it with using
14572 bst_fail set to cache SLP tree build fails. Properly handle
14574 (vect_analyze_slp_instance): Allocate and free bst_fail.
14576 2017-08-08 Martin Liska <mliska@suse.cz>
14579 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
14580 LABEL_DECLs that can be from a different function.
14582 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14584 PR tree-optimization/81744
14585 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
14586 loop's number of iterations.
14588 2017-08-08 Martin Liska <mliska@suse.cz>
14590 * asan.c: Include header files.
14591 * attribs.c (build_decl_attribute_variant): New function moved
14593 (build_type_attribute_qual_variant): Likewise.
14594 (cmp_attrib_identifiers): Likewise.
14595 (simple_cst_list_equal): Likewise.
14596 (omp_declare_simd_clauses_equal): Likewise.
14597 (attribute_value_equal): Likewise.
14598 (comp_type_attributes): Likewise.
14599 (build_type_attribute_variant): Likewise.
14600 (lookup_ident_attribute): Likewise.
14601 (remove_attribute): Likewise.
14602 (merge_attributes): Likewise.
14603 (merge_type_attributes): Likewise.
14604 (merge_decl_attributes): Likewise.
14605 (merge_dllimport_decl_attributes): Likewise.
14606 (handle_dll_attribute): Likewise.
14607 (attribute_list_equal): Likewise.
14608 (attribute_list_contained): Likewise.
14609 * attribs.h (lookup_attribute): New function moved from tree.[ch].
14610 (lookup_attribute_by_prefix): Likewise.
14611 * bb-reorder.c: Include header files.
14612 * builtins.c: Likewise.
14613 * calls.c: Likewise.
14614 * cfgexpand.c: Likewise.
14615 * cgraph.c: Likewise.
14616 * cgraphunit.c: Likewise.
14617 * convert.c: Likewise.
14618 * dwarf2out.c: Likewise.
14619 * final.c: Likewise.
14620 * fold-const.c: Likewise.
14621 * function.c: Likewise.
14622 * gimple-expr.c: Likewise.
14623 * gimple-fold.c: Likewise.
14624 * gimple-pretty-print.c: Likewise.
14625 * gimple.c: Likewise.
14626 * gimplify.c: Likewise.
14627 * hsa-common.c: Likewise.
14628 * hsa-gen.c: Likewise.
14629 * internal-fn.c: Likewise.
14630 * ipa-chkp.c: Likewise.
14631 * ipa-cp.c: Likewise.
14632 * ipa-devirt.c: Likewise.
14633 * ipa-fnsummary.c: Likewise.
14634 * ipa-inline.c: Likewise.
14635 * ipa-visibility.c: Likewise.
14637 * lto-cgraph.c: Likewise.
14638 * omp-expand.c: Likewise.
14639 * omp-general.c: Likewise.
14640 * omp-low.c: Likewise.
14641 * omp-offload.c: Likewise.
14642 * omp-simd-clone.c: Likewise.
14643 * opts-global.c: Likewise.
14644 * passes.c: Likewise.
14645 * predict.c: Likewise.
14646 * sancov.c: Likewise.
14647 * sanopt.c: Likewise.
14648 * symtab.c: Likewise.
14649 * toplev.c: Likewise.
14650 * trans-mem.c: Likewise.
14651 * tree-chkp.c: Likewise.
14652 * tree-eh.c: Likewise.
14653 * tree-into-ssa.c: Likewise.
14654 * tree-object-size.c: Likewise.
14655 * tree-parloops.c: Likewise.
14656 * tree-profile.c: Likewise.
14657 * tree-ssa-ccp.c: Likewise.
14658 * tree-ssa-live.c: Likewise.
14659 * tree-ssa-loop.c: Likewise.
14660 * tree-ssa-sccvn.c: Likewise.
14661 * tree-ssa-structalias.c: Likewise.
14662 * tree-ssa.c: Likewise.
14663 * tree-streamer-in.c: Likewise.
14664 * tree-vectorizer.c: Likewise.
14665 * tree-vrp.c: Likewise.
14666 * tsan.c: Likewise.
14667 * ubsan.c: Likewise.
14668 * varasm.c: Likewise.
14669 * varpool.c: Likewise.
14670 * tree.c: Remove functions moved to attribs.[ch].
14671 * tree.h: Likewise.
14672 * config/aarch64/aarch64.c: Add attrs.h header file.
14673 * config/alpha/alpha.c: Likewise.
14674 * config/arc/arc.c: Likewise.
14675 * config/arm/arm.c: Likewise.
14676 * config/avr/avr.c: Likewise.
14677 * config/bfin/bfin.c: Likewise.
14678 * config/c6x/c6x.c: Likewise.
14679 * config/cr16/cr16.c: Likewise.
14680 * config/cris/cris.c: Likewise.
14681 * config/darwin.c: Likewise.
14682 * config/epiphany/epiphany.c: Likewise.
14683 * config/fr30/fr30.c: Likewise.
14684 * config/frv/frv.c: Likewise.
14685 * config/ft32/ft32.c: Likewise.
14686 * config/h8300/h8300.c: Likewise.
14687 * config/i386/winnt.c: Likewise.
14688 * config/ia64/ia64.c: Likewise.
14689 * config/iq2000/iq2000.c: Likewise.
14690 * config/lm32/lm32.c: Likewise.
14691 * config/m32c/m32c.c: Likewise.
14692 * config/m32r/m32r.c: Likewise.
14693 * config/m68k/m68k.c: Likewise.
14694 * config/mcore/mcore.c: Likewise.
14695 * config/microblaze/microblaze.c: Likewise.
14696 * config/mips/mips.c: Likewise.
14697 * config/mmix/mmix.c: Likewise.
14698 * config/mn10300/mn10300.c: Likewise.
14699 * config/moxie/moxie.c: Likewise.
14700 * config/msp430/msp430.c: Likewise.
14701 * config/nds32/nds32-isr.c: Likewise.
14702 * config/nds32/nds32.c: Likewise.
14703 * config/nios2/nios2.c: Likewise.
14704 * config/nvptx/nvptx.c: Likewise.
14705 * config/pa/pa.c: Likewise.
14706 * config/pdp11/pdp11.c: Likewise.
14707 * config/powerpcspe/powerpcspe.c: Likewise.
14708 * config/riscv/riscv.c: Likewise.
14709 * config/rl78/rl78.c: Likewise.
14710 * config/rx/rx.c: Likewise.
14711 * config/s390/s390.c: Likewise.
14712 * config/sh/sh.c: Likewise.
14713 * config/sol2.c: Likewise.
14714 * config/sparc/sparc.c: Likewise.
14715 * config/spu/spu.c: Likewise.
14716 * config/stormy16/stormy16.c: Likewise.
14717 * config/tilegx/tilegx.c: Likewise.
14718 * config/tilepro/tilepro.c: Likewise.
14719 * config/v850/v850.c: Likewise.
14720 * config/vax/vax.c: Likewise.
14721 * config/visium/visium.c: Likewise.
14722 * config/xtensa/xtensa.c: Likewise.
14724 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14727 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
14728 constraints since the -mupper-regs-* switches have been
14730 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
14731 into a vector from a double word element that was extracted from
14732 another vector, and eliminate extra XXPERMDI instructions.
14733 (vsx_concat_<mode>_2): Likewise.
14734 (vsx_concat_<mode>_3): Likewise.
14735 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
14736 concat to allow optimizing inserts from previous extracts.
14738 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
14740 * config/i386/i386.c (ix86_stack_protect_guard): Generate
14741 memory reference to a SSP offset in TLS address space.
14742 (ix86_print_operand) <case '@'>: Remove.
14743 (ix86_print_operand_punct_valid_p): Remove '@' code.
14744 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
14745 UNSPEC_SP_TLS_TEST.
14746 (stack_tls_protect_set_<mode>): Remove.
14747 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
14748 (stack_tls_protect_test_<mode>): Remove.
14749 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
14751 2017-08-07 Olivier Hainque <hainque@adacore.com>
14754 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
14756 2017-08-07 Douglas Rupp <rupp@adacore.com>
14758 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
14759 variable was referenced as multidir in command.
14761 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14764 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
14767 2017-08-07 Martin Liska <mliska@suse.cz>
14769 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
14770 * config/rl78/rl78.c: Add include of attribs.h.
14771 * config/sh/sh.c: Likewise.
14772 * config/v850/v850.c: Likewise.
14774 2017-08-07 Tom de Vries <tom@codesourcery.com>
14776 PR middle-end/78266
14777 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
14779 2017-08-07 Martin Liska <mliska@suse.cz>
14781 * config/mips/mips.c: Include attribs.h.
14783 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
14786 * doc/invoke.texi: Document change in behvaior for -Ofast for
14789 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
14791 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
14793 (aarch64_pop_regs): Likewise.
14794 (aarch64_gen_load_pair): Likewise.
14795 (aarch64_save_callee_saves): Likewise.
14796 (aarch64_restore_callee_saves): Likewise.
14798 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14800 * config/i386/i386.c: Revert the last change.
14802 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14805 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14807 (ix86_finalize_stack_frame_flags): This. Also clear
14808 frame_pointer_needed if -fno-omit-frame-pointer is used without
14810 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14811 with ix86_finalize_stack_frame_flags.
14812 (ix86_expand_epilogue): Likewise.
14813 (ix86_expand_split_stack_prologue): Likewise.
14815 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14818 * config/i386/i386.c (get_builtin_code_for_version): Set priority
14819 to P_AES for Westmere.
14821 2017-08-07 Jonathan Yong <10walls@gmail.com>
14823 * config/i386/mingw.opt (fset-stack-executable): Removed.
14824 * config/i386/cygming.opt (fset-stack-executable): Moved
14826 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
14828 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
14830 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
14832 2017-08-07 Marek Polacek <polacek@redhat.com>
14834 PR middle-end/81737
14835 * fold-const.c (fold_indirect_ref_1): Check type_domain.
14837 2017-08-07 Martin Liska <mliska@suse.cz>
14839 * attribs.h (canonicalize_attr_name): New function.
14840 (cmp_attribs): Move from c-format.c and adjusted.
14841 (is_attribute_p): Moved from tree.h.
14842 * tree-inline.c: Add new includes.
14843 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
14844 (private_is_attribute_p): Remove.
14845 (private_lookup_attribute): Likewise.
14846 (private_lookup_attribute_by_prefix): Simplify.
14847 (remove_attribute): Use is_attribute_p.
14848 * tree.h: Remove removed declarations.
14850 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14852 PR middle-end/81698
14853 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
14854 instead of computing it in the function. Formatting fix.
14855 (expand_case): Don't rely on default_edge being the first edge,
14856 clear it if removing it, pass default_edge to
14857 emit_case_dispatch_table.
14858 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
14861 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
14863 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
14864 insn in the function, emit NOP after the insn.
14866 2017-08-06 Tom de Vries <tom@codesourcery.com>
14868 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
14871 2017-08-06 Tom de Vries <tom@codesourcery.com>
14873 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
14876 2017-08-04 Yury Gribov <tetra2005@gmail.com>
14878 PR tree-optimization/57371
14879 * match.pd: New pattern.
14881 2017-08-04 Marek Polacek <polacek@redhat.com>
14883 PR middle-end/81695
14884 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
14885 perform the computation in offset_int.
14887 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14889 PR tree-optimization/81136
14890 * tree-vectorizer.h: Include tree-hash-traits.h.
14891 (vec_base_alignments): New typedef.
14892 (vec_info): Add a base_alignments field.
14893 (vect_record_base_alignments): Declare.
14894 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
14896 (DR_IS_CONDITIONAL_IN_STMT): New macro.
14897 (create_data_ref): Add an is_conditional_in_stmt argument.
14898 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
14899 the is_conditional_in_stmt field.
14900 (data_ref_loc): Add an is_conditional_in_stmt field.
14901 (get_references_in_stmt): Set the is_conditional_in_stmt field.
14902 (find_data_references_in_stmt): Update call to create_data_ref.
14903 (graphite_find_data_references_in_stmt): Likewise.
14904 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
14905 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
14906 (vect_record_base_alignment): New function.
14907 (vect_record_base_alignments): Likewise.
14908 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
14909 for nested statements even if we fail to compute a misalignment.
14910 Use pooled base alignments for unconditional references.
14911 (vect_find_same_alignment_drs): Compare base addresses instead
14913 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
14914 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
14916 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14918 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
14919 Add an explicit name for the enum. Use auto_vec for slp_instances
14920 and grouped_stores.
14921 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
14923 (_bb_vec_info): Add a constructor and destructor.
14924 (vinfo_for_stmt): Return NULL for uids of -1 as well.
14925 (destroy_loop_vec_info): Delete.
14926 (vect_destroy_datarefs): Likewise.
14927 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
14928 (vec_info::vec_info): New function.
14929 (vec_info::~vec_info): Likewise.
14930 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
14931 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
14932 destroy_loop_vec_info.
14933 * tree-vect-loop.c (new_loop_vec_info): Replace with...
14934 (_loop_vec_info::_loop_vec_info): ...this.
14935 (destroy_loop_vec_info): Replace with...
14936 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
14937 the stmt_vec_infos. Leave handling of vec_info information to its
14938 destructor. Remove explicit vector releases.
14939 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
14940 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
14941 * tree-vect-slp.c (new_bb_vec_info): Replace with...
14942 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
14943 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
14944 (destroy_bb_vec_info): Replace with...
14945 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
14946 information to its destructor.
14947 (vect_slp_analyze_bb_1): Use new and delete instead of
14948 new_bb_vec_info and destroy_bb_vec_info.
14949 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
14952 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14954 * tree-data-ref.h (subscript): Add access_fn field.
14955 (data_dependence_relation): Add could_be_independent_p.
14956 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
14957 (same_access_functions): Move to tree-data-ref.c.
14958 * tree-data-ref.c (ref_contains_union_access_p): New function.
14959 (access_fn_component_p): Likewise.
14960 (access_fn_components_comparable_p): Likewise.
14961 (dr_analyze_indices): Add a reference to access_fn_component_p.
14962 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
14964 (constant_access_functions): Likewise.
14965 (add_other_self_distances): Likewise.
14966 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
14967 (initialize_data_dependence_relation): Use XCNEW and remove
14968 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
14969 of access functions that have the same type. Allow the
14970 subsequence to end with different bases in some circumstances.
14971 Record the chosen access functions in SUB_ACCESS_FN.
14972 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
14973 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
14974 (subscript_dependence_tester_1): Likewise dra and drb.
14975 (build_classic_dist_vector): Update calls accordingly.
14976 (subscript_dependence_tester): Likewise.
14977 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
14978 DDR_COULD_BE_INDEPENDENT_P.
14979 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
14980 comp_alias_ddrs instead of may_alias_ddrs.
14981 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
14983 (vect_analyze_data_ref_dependence): Use it if
14984 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
14985 distance vectors if that fails.
14986 (dependence_distance_ge_vf): New function.
14987 (vect_prune_runtime_alias_test_list): Use it. Don't clear
14988 LOOP_VINFO_MAY_ALIAS_DDRS.
14990 2017-08-04 Richard Biener <rguenther@suse.de>
14992 PR middle-end/81705
14993 * fold-const.c (fold_binary_loc): Properly restrict
14994 minus_var0 && minus_var1 case when associating undefined overflow
14997 2017-08-04 Tom de Vries <tom@codesourcery.com>
14999 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
15001 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15003 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15004 Don't start diagnostic messages with a capital letter.
15005 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15007 (rs6000_invalid_builtin): Likewise.
15008 (rs6000_trampoline_init): Likewise.
15010 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15013 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
15014 after setting changeable df flags.
15016 2017-08-03 Richard Biener <rguenther@suse.de>
15018 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
15019 up if the use is in USE - X.
15021 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
15023 * toplev.c (dumpfile.h): New include.
15024 (internal_error_reentered): New static function. Use it...
15025 (internal_error_function): ...here to handle reentered internal_error.
15027 2017-08-03 Richard Biener <rguenther@suse.de>
15029 PR middle-end/81148
15030 * fold-const.c (split_tree): Add minus_var and minus_con
15031 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
15032 here but always use minus_*.
15033 (associate_trees): Assert we never associate with MINUS_EXPR
15034 and NULL first operand. Do not recurse for PLUS_EXPR operands
15035 when associating as MINUS_EXPR either.
15036 (fold_binary_loc): Track minus_var and minus_con.
15038 2017-08-03 Tom de Vries <tom@codesourcery.com>
15041 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
15042 ACCEL_COMPILER, apply finish_options on
15043 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
15045 2017-08-03 Tom de Vries <tom@codesourcery.com>
15048 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
15049 function_entry_patch_area_size > 0.
15051 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15054 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
15055 instead of 10??LU, perform unit multiplication in wide_int,
15056 don't change alloc_object_size_limit if the limit is larger
15059 PR tree-optimization/81655
15060 PR tree-optimization/81588
15061 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
15062 the case when ranges[i].low and high are 1 for unsigned type with
15065 PR middle-end/81052
15066 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
15067 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
15069 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15071 * tree-vrp.h: Add include guard.
15073 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
15076 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
15077 (ud2): New insn pattern.
15078 * config/i386/i386.c (ix86_expand_epilogue):
15079 For naked functions, generate ud2 instead of trap insn.
15081 2017-08-02 Marek Polacek <polacek@redhat.com>
15084 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
15086 2017-08-02 Tom de Vries <tom@codesourcery.com>
15087 Cesar Philippidis <cesar@codesourcery.com>
15089 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
15090 Add missing edge probabilities.
15092 2017-08-02 Tamar Christina <tamar.christina@arm.com>
15094 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
15095 Correct endianness.
15097 2017-08-02 Jakub Jelinek <jakub@redhat.com>
15099 PR middle-end/79499
15100 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
15101 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
15102 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
15103 prologue_seq sequences - if any.
15105 2017-08-02 Richard Biener <rguenther@suse.de>
15107 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
15108 via vectors if supported, integer extracts via punning if supported
15109 or otherwise vector extracts.
15111 2017-08-02 Richard Biener <rguenther@suse.de>
15113 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
15115 (bitmap_insert_into_set): ... this.
15117 2017-08-02 Richard Biener <rguenther@suse.de>
15119 PR tree-optimization/81633
15121 2015-08-17 Alan Hayward <alan.hayward@arm.com>
15123 PR tree-optimization/71752
15124 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
15126 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
15128 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
15129 (machine_function::call_ms2sysv_pad_out): Remove field.
15130 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
15131 (ix86_compute_frame_layout): Likewise.
15133 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
15136 * config/i386/i386.c (ix86_set_func_type): Disallow naked
15137 attribute with interrupt attribute.
15139 2017-08-01 Andrew Pinski <apinski@cavium.com>
15141 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
15142 BIT_INSERT_EXPR's operand 1
15143 to see if the types precision matches.
15145 2017-08-01 Martin Liska <mliska@suse.cz>
15147 PR middle-end/70140
15148 * builtins.c (expand_builtin_memcpy_args): Remove.
15149 (expand_builtin_memcpy): Call newly added function
15150 expand_builtin_memory_copy_args.
15151 (expand_builtin_memcpy_with_bounds): Likewise.
15152 (expand_builtin_mempcpy): Remove last argument.
15153 (expand_builtin_mempcpy_with_bounds): Likewise.
15154 (expand_builtin_memory_copy_args): New function created from
15155 expand_builtin_mempcpy_args with small modifications.
15156 (expand_builtin_mempcpy_args): Remove.
15157 (expand_builtin_stpcpy): Remove unused argument.
15158 (expand_builtin): Likewise.
15159 (expand_builtin_with_bounds): Likewise.
15161 2017-08-01 Martin Liska <mliska@suse.cz>
15164 Make mempcpy more optimal (PR middle-end/70140).
15166 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15169 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15170 __builtin_vec_cmpne verify both arguments are compatible vectors
15171 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
15172 verify arg1_type is a pointer or array type. For __builtin_vec_st,
15173 move computation of aligned to after checking the argument types.
15177 * config/rs6000/vsx.md (vextract_fp_from_shorth,
15178 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15181 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15182 Doug Rupp <rupp@adacore.com>
15183 Olivier Hainque <hainque@adacore.com>
15185 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15186 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15188 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15189 for TARGET_OS_CPP_BUILTIN.
15190 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15191 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15193 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15194 passing required abi options to the assembler for EABI configurations.
15195 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15196 of .text.hot and .text.unlikely sections for kernel modules when
15197 using ARM style exceptions.
15198 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15199 options. Add EXTRA_CC1_SPEC.
15200 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15202 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15204 (ARM_TARGET2_DWARF_FORMAT): Define.
15205 * config/arm/t-vxworks: Adjust multilib control to removal of the
15206 Diab command line options.
15208 2017-08-01 Martin Liska <mliska@suse.cz>
15210 PR gcov-profile/81561
15211 * gcov.c (unblock): Make unblocking safe as we need to preserve
15212 index correspondence of blocks and block_lists.
15214 2017-08-01 Richard Biener <rguenther@suse.de>
15216 PR tree-optimization/81181
15217 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15218 (compute_antic): ... end of iteration here.
15220 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15222 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15223 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15224 (ftree-slp-vectorize): Likewise.
15225 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15226 can no longer be set independent of flag_tree_loop_vectorize.
15227 * omp-general.c (emp_max_vf): Likewise.
15228 * opts.c (enable_fdo_optimizations): Remove references to
15229 flag_tree_vectorize, these are now implicit.
15230 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15231 and leave it for the options machinery.
15233 2017-08-01 Martin Liska <mliska@suse.cz>
15235 PR middle-end/70140
15236 * builtins.c (expand_builtin_memcpy_args): Remove.
15237 (expand_builtin_memcpy): Call newly added function
15238 expand_builtin_memory_copy_args.
15239 (expand_builtin_memcpy_with_bounds): Likewise.
15240 (expand_builtin_mempcpy): Remove last argument.
15241 (expand_builtin_mempcpy_with_bounds): Likewise.
15242 (expand_builtin_memory_copy_args): New function created from
15243 expand_builtin_mempcpy_args with small modifications.
15244 (expand_builtin_mempcpy_args): Remove.
15245 (expand_builtin_stpcpy): Remove unused argument.
15246 (expand_builtin): Likewise.
15247 (expand_builtin_with_bounds): Likewise.
15249 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15252 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15253 print "ds:" only for immediates in generic address space.
15255 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15258 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15259 (ix86_function_ok_for_sibcall): Return false for naked functions.
15261 2017-08-01 Richard Biener <rguenther@suse.de>
15263 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15264 (compute_antic): Seed worklist with exit block predecessors.
15265 * cfganal.c (dfs_find_deadend): For a cycle return the source
15266 of the edge closing it.
15268 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15270 * config/aarch64/aarch64.c
15271 (aarch64_can_const_movi_rtx_p): Move 0 check.
15273 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15275 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15276 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15278 * match.pd: Ditto in address comparison pattern.
15280 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15282 PR tree-optimization/81627
15283 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15284 closed ssa form for store-store chain.
15286 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15288 PR tree-optimization/81620
15289 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15290 for store-store chain.
15292 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15294 PR tree-optimization/81588
15295 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15296 ranges[i].in_p, invert comparison code ccode. For >/>=,
15297 swap rhs1 and rhs2 and comparison code unconditionally,
15298 for </<= don't do that. Don't swap rhs1/rhs2 again if
15299 ranges[i].in_p, instead invert comparison code ccode if
15300 opcode or oe->rank is BIT_IOR_EXPR.
15303 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15304 a direct optab to conversion optab.
15305 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15306 with GET_MODE_INNER as last argument instead of optab_handler.
15307 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15308 vector extraction if possible and optab is available.
15309 * expr.c (store_constructor): Use convert_optab_handler instead
15310 of optab_handler. Use vector initialization from smaller
15311 vectors if possible and optab is available.
15312 * tree-vect-stmts.c (vectorizable_load): Likewise.
15313 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15314 now have two modes.
15315 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15316 of vec_init from half-sized vectors with the same element mode.
15317 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15318 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15319 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15320 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15321 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15322 after mode in gen_vec_extract* calls.
15323 (vec_extract<mode>): Renamed to ...
15324 (vec_extract<mode><ssescalarmodelower>): ... this.
15325 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15326 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15327 element mode after mode in gen_vec_init* calls.
15328 (VEC_INIT_HALF_MODE): New mode iterator.
15329 (vec_init<mode>): Renamed to ...
15330 (vec_init<mode><ssescalarmodelower>): ... this.
15331 (vec_init<mode><ssehalfvecmodelower>): New expander.
15332 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15333 (vec_extractv2sfsf): ... this.
15334 (vec_initv2sf): Renamed to ...
15335 (vec_initv2sfsf): ... this.
15336 (vec_extractv2si): Renamed to ...
15337 (vec_extractv2sisi): ... this.
15338 (vec_initv2si): Renamed to ...
15339 (vec_initv2sisi): ... this.
15340 (vec_extractv4hi): Renamed to ...
15341 (vec_extractv4hihi): ... this.
15342 (vec_initv4hi): Renamed to ...
15343 (vec_initv4hihi): ... this.
15344 (vec_extractv8qi): Renamed to ...
15345 (vec_extractv8qiqi): ... this.
15346 (vec_initv8qi): Renamed to ...
15347 (vec_initv8qiqi): ... this.
15348 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15349 (vec_init<mode>): Renamed to ...
15350 (vec_init<mode><VEC_base_l>): ... this.
15351 (vec_extract<mode>): Renamed to ...
15352 (vec_extract<mode><VEC_base_l>): ... this.
15353 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15354 (vec_initv2sfsf): ... this.
15355 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15356 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15357 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15358 element mode after mode in gen_vec_init* calls.
15359 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15360 (vec_init<mode><Vel>): ... this.
15361 (vec_extract<mode>): Renamed to ...
15362 (vec_extract<mode><Vel>): ... this.
15363 * config/aarch64/iterators.md (Vel): New mode attribute.
15364 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15365 Add element mode after mode in gen_vec_extract* calls.
15366 * config/s390/vector.md (non_vec_l): New mode attribute.
15367 (vec_extract<mode>): Renamed to ...
15368 (vec_extract<mode><non_vec_l>): ... this.
15369 (vec_init<mode>): Renamed to ...
15370 (vec_init<mode><non_vec_l>): ... this.
15371 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15372 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15374 * config/arm/iterators.md (V_elem_l): New mode attribute.
15375 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15376 (vec_extract<mode><V_elem_l>): ... this.
15377 (vec_extractv2di): Renamed to ...
15378 (vec_extractv2didi): ... this.
15379 (vec_init<mode>): Renamed to ...
15380 (vec_init<mode><V_elem_l>): ... this.
15381 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15382 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15383 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15384 Add element mode after gen_vec_extract* calls.
15385 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15386 (vec_init<mode><unitmode>): ... this.
15387 (vec_extract<mode>): Renamed to ...
15388 (vec_extract<mode><unitmode>): ... this.
15389 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15390 (vec_init<mode><unitmode>): ... this.
15391 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15392 (vec_initv2sfsf): ... this.
15393 (vec_extractv2sf): Renamed to ...
15394 (vec_extractv2sfsf): ... this.
15395 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15396 Add element mode after gen_vec_extract* calls.
15397 * config/mips/mips.md (unitmode): New mode iterator.
15398 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15399 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15400 * config/spu/spu.md (inner_l): New mode attribute.
15401 (vec_init<mode>): Renamed to ...
15402 (vec_init<mode><inner_l>): ... this.
15403 (vec_extract<mode>): Renamed to ...
15404 (vec_extract<mode><inner_l>): ... this.
15405 * config/sparc/sparc.md (veltmode): New mode iterator.
15406 (vec_init<VMALL:mode>): Renamed to ...
15407 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15408 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15409 (vec_initv2sisi): ... this.
15410 (vec_initv2sf): Renamed to ...
15411 (vec_initv2sfsf): ... this.
15412 (vec_extractv2sf): Renamed to ...
15413 (vec_extractv2sfsf): ... this.
15414 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15415 (vec_init<mode>): Renamed to ...
15416 (vec_init<mode><VEC_base_l>): ... this.
15417 (vec_extract<mode>): Renamed to ...
15418 (vec_extract<mode><VEC_base_l>): ... this.
15419 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15420 (vec_initv2sfsf): ... this.
15421 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15422 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15423 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15424 gen_vec_init* calls.
15426 2017-08-01 Richard Biener <rguenther@suse.de>
15428 PR tree-optimization/81297
15429 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15430 TREE_OVERFLOW from INTEGER_CSTs.
15432 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15434 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15436 2017-07-31 Carl Love <cel@us.ibm.com>
15438 * config/rs6000/rs6000-c: Add support for built-in functions
15439 vector signed char vec_xl_be (signed long long, signed char *);
15440 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15441 vector signed int vec_xl_be (signed long long, signed int *);
15442 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15443 vector signed long long vec_xl_be (signed long long, signed long long *);
15444 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15445 vector signed short vec_xl_be (signed long long, signed short *);
15446 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15447 vector double vec_xl_be (signed long long, double *);
15448 vector float vec_xl_be (signed long long, float *);
15449 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15450 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15451 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
15453 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
15454 (altivec_expand_builtin): Add switch statement to call
15455 altivec_expand_xl_be for each builtin.
15456 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15457 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15458 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15459 __builtin_vsx_le_be_v16qi.
15460 * doc/extend.texi: Update the built-in documentation file for the
15461 new built-in functions.
15463 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15466 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15468 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15470 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15472 * config.gcc: Add z14.
15473 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
15474 CPU model numbers for z13s and z14.
15475 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
15477 * config/s390/s390-opts.h (enum processor_type): Rename
15478 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15479 * config/s390/s390.c (processor_table): Add field for CPU name to
15480 be passed to Binutils.
15481 (s390_asm_output_machine_for_arch): Use the new field in
15482 processor_table for Binutils.
15483 (s390_expand_builtin): Replace arch12 with z14.
15484 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15485 (s390_get_sched_attrmask): Likewise.
15486 (s390_get_unit_mask): Likewise.
15487 * config/s390/s390.opt: Add z14 to processor_type enum.
15489 2017-07-31 Martin Jambor <mjambor@suse.cz>
15492 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
15493 regardless of optimization level.
15495 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
15496 Martin Liska <mliska@suse.cz>
15498 * predict.def: Remove old comment and adjust probability.
15499 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
15500 PREDICT statements.
15502 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15505 * config/i386/i386.c (ix86_function_naked): New function.
15506 (ix86_can_use_return_insn_p): Return false for naked functions.
15507 (ix86_expand_prologue): Skip prologue for naked functions.
15508 (ix86_expand_epilogue): Skip epilogue for naked functions
15509 and emit trap instruction.
15510 (ix86_warn_func_return): New function.
15511 (ix86_attribute_table): Add "naked" attribute specification.
15512 (TARGET_WARN_FUNC_RETURN): Define.
15513 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
15515 2017-07-31 Martin Liska <mliska@suse.cz>
15517 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
15518 (dump_gimple_bb_header): Always dump BB info.
15519 (pp_cfg_jump): Do not append info about BB when dumping a jump.
15521 2017-07-31 Martin Liska <mliska@suse.cz>
15524 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
15525 also with current_function_decl non-null equality.
15527 2017-07-31 Jakub Jelinek <jakub@redhat.com>
15530 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
15531 change type to the element type, instead add eltype variable and
15532 use it where we are interested in the element type.
15534 PR tree-optimization/81603
15535 * ipa-polymorphic-call.c
15536 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
15537 offset arithmetic in offset_int, bail out if the resulting bit offset
15538 doesn't fit into shwi.
15540 2017-07-31 Martin Liska <mliska@suse.cz>
15542 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
15543 (gimplify_save_expr): Fix comment.
15545 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
15548 * config/i386/i386.c (ix86_function_arg): Update arguments for
15550 (ix86_compute_frame_layout): Set the initial stack offset to
15551 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
15552 INCOMING_FRAME_SP_OFFSET.
15553 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
15554 stack before exception handler returns.
15555 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
15556 the 'ERROR_CODE' for exception handler.
15558 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
15560 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
15561 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
15562 (ASM_OUTPUT_REG_POP): Ditto.
15563 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
15564 instead of asm_fprintf to output pure string.
15566 2017-07-29 Jakub Jelinek <jakub@redhat.com>
15568 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
15569 to imported_module_or_decl hook.
15570 (debug_nothing_tree_tree_tree_bool): Remove.
15571 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
15572 * debug.c (do_nothing_debug_hooks): Use
15573 debug_nothing_tree_tree_tree_bool_bool instead of
15574 debug_nothing_tree_tree_tree_bool.
15575 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15576 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
15577 * sdbout.c (sdb_debug_hooks): Likewise.
15578 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
15579 (gen_namespace_die): Add DW_AT_export_symbols attribute if
15581 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
15582 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
15583 attribute, don't add anything.
15585 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15587 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
15588 (fold_build2_stat_loc): Likewise.
15589 (fold_build3_stat_loc): Likewise.
15590 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
15591 (fold_build1_loc): Remove macro.
15592 (fold_build2_loc): Likewise.
15593 (fold_build3_loc): Likewise.
15595 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15597 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
15598 (gimple_build_debug_bind_source_stat): Likewise.
15599 * gimple.h (gimple_build_debug_bind): Remove macro.
15600 (gimple_build_debug_bind_source): Likewise.
15602 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15604 * bitmap.c (bitmap_alloc): Adjust.
15605 (bitmap_gc_alloc): Likewise.
15606 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
15608 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15610 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
15611 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
15612 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
15613 (bitmap_gc_alloc_stat): Likewise.
15614 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
15616 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15618 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
15619 * rtl.h (shallow_copy_rtx): Remove macro.
15621 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15623 * emit-rtl.c (gen_raw_REG): Adjust.
15624 * gengenrtl.c (gendef): Likewise.
15625 * rtl.c (rtx_alloc_stat): Remove _stat from name.
15626 * rtl.h (rtx_alloc): Remove macro.
15628 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15630 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
15631 (build_tree_list_stat): Likewise.
15632 * tree.h (build_tree_list): Remove macro.
15633 (build_tree_list_vec): Likewise.
15635 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15637 * tree.c (make_vector_stat): Remove _stat from name.
15638 (build_vector_stat): Likewise.
15639 * tree.h (make_vector_stat): Remove macro.
15640 (build_vector_stat): Likewise.
15642 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15644 * tree.h (build_var_debug_value): Remove prototype.
15646 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15648 * tree.c (tree_cons_stat): Remove _stat from name.
15649 * tree.h (tree_cons): Remove macro.
15651 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15653 * tree.c (build_vl_exp_stat): Remove _stat from name.
15654 * tree.h (build_vl_exp): Remove macro.
15656 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15658 * tree.c (build_decl_stat): Remove _stat from name.
15659 * tree.h (build_decl): Remove macro.
15661 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15663 * gimple.c (gimple_build_with_ops_stat): Adjust.
15664 (gimple_alloc_stat): Remove _stat from name.
15665 * gimple.h (gimple_alloc): Remove macro.
15667 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15669 * tree.c (make_tree_vec_stat): Remove _stat from name.
15670 (grow_tree_vec_stat): Likewise.
15671 * tree.h (make_tree_vec_stat): Adjust prototype.
15672 (grow_tree_vec_stat): Likewise.
15673 (make_tree_vec): Remove macro.
15674 (grow_tree_vec): Likewise.
15676 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15678 * fold-const.c (fold_build1_stat_loc): Adjust.
15679 (fold_build2_stat_loc): Likewise.
15680 (fold_build3_stat_loc): Likewise.
15681 * tree.c (build0_stat): Remove _stat from name.
15682 (build1_stat): Likewise.
15683 (build2_stat): Likewise.
15684 (build3_stat): Likewise.
15685 (build4_stat): Likewise.
15686 (build5_stat): Likewise.
15687 * tree.h (build1_loc): Remove macro, and rename _stat function
15689 (build2_loc): Likewise.
15690 (build3_loc): Likewise.
15691 (build4_loc): Likewise.
15692 (build5_loc): Likewise.
15694 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15696 * tree.c (make_int_cst_stat): Remove _stat from name.
15697 * tree.h (make_int_cst_stat): Adjust prototype.
15698 (make_int_cst): Remove macro.
15700 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15702 * tree.c (make_tre_binfo_stat): Remove _stat from name.
15703 * tree.h (make_tree_binfo_stat): Adjust prototype.
15704 (make_tree_binfo): Remove.
15706 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15708 * tree.c (copy_node_stat): Rename to copy_node.
15709 (build_distinct_type_copy): Adjust.
15710 * tree.h (copy_node_stat): Adjust prototype.
15711 (copy_node): Remove macro.
15713 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15715 * tree.c (make_node_stat): rename to make_node.
15716 (build_tree_list_stat): Adjust.
15717 (build0_stat): Likewise.
15718 (build2_stat): Likewise.
15719 (build3_stat): Likewise.
15720 (build4_stat): Likewise.
15721 (build5_stat): Likewise.
15722 (build_decl_stat): Likewise.
15723 * tree.h (make_node_stat): Adjust prototype.
15724 (make_node): remove macro.
15726 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
15728 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
15729 (PPC_FEATURE2_SCV): Likewise.
15730 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
15732 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15734 * config/aarch64/aarch64.c
15735 (aarch64_internal_mov_immediate): Add new special pattern.
15736 * config/aarch64/aarch64.md (*movdi_aarch64):
15737 Add reg/32bit const mov case.
15739 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15740 Richard Sandiford <richard.sandiford@linaro.org>
15742 * config/aarch64/aarch64.md (mov<mode>): Generalize.
15743 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15744 Add integer and movi cases.
15745 (movi-split-hf-df-sf split, fp16): New.
15746 (enabled): Added TARGET_FP_F16INST.
15747 * config/aarch64/iterators.md (GPF_HF): New.
15748 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
15750 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15752 * config/aarch64/aarch64.c
15753 (aarch64_simd_container_mode): Add prototype.
15754 (aarch64_expand_mov_immediate): Add HI support.
15755 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
15756 (aarch64_can_const_movi_rtx_p): New.
15757 (aarch64_preferred_reload_class):
15758 Remove restrictions of using FP registers for certain SIMD operations.
15759 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
15760 (aarch64_valid_floating_const): Add integer move validation.
15761 (aarch64_simd_imm_scalar_p): Remove.
15762 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
15763 (aarch64_legitimate_constant_p): Expand list of supported cases.
15764 * config/aarch64/aarch64-protos.h
15765 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
15766 (aarch64_reinterpret_float_as_int): New.
15767 (aarch64_simd_imm_scalar_p): Remove.
15768 * config/aarch64/constraints.md (Uvi): New.
15769 (Dd): Split into Ds and new Dd.
15770 * config/aarch64/aarch64.md (*movsi_aarch64):
15772 (*movdi_aarch64): Add SIMD mov case.
15774 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15776 * tree-predcom.c: (struct chain): Handle store-store chain in which
15777 stores for elimination only store loop invariant values.
15778 (execute_pred_commoning_chain): Ditto.
15779 (prepare_initializers_chain_store_elim): Ditto.
15780 (prepare_finalizers): Ditto.
15781 (is_inv_store_elimination_chain): New function.
15782 (initialize_root_vars_store_elim_1): New function.
15784 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15786 * tree-predcom.c: Revise general description of the pass.
15787 (enum chain_type): New enum type for store elimination.
15788 (struct chain): New field supporting store elimination.
15789 (struct component): Ditto.
15790 (dump_chain): Dump store-stores chain.
15791 (release_chain): Release resources.
15792 (split_data_refs_to_components): Compute and create component
15793 contains only stores for elimination.
15794 (get_chain_last_ref_at): New function.
15795 (make_invariant_chain): Initialization.
15796 (make_rooted_chain): Specify chain type in parameter and record it.
15797 (add_looparound_copies): Skip for store-stores chain.
15798 (determine_roots_comp): Compute type of chain and pass it to
15800 (initialize_root_vars_store_elim_2): New function.
15801 (finalize_eliminated_stores): New function.
15802 (remove_stmt): Handle store for elimination.
15803 (execute_pred_commoning_chain): Execute predictive commoning on
15804 store-store chains.
15805 (determine_unroll_factor): Skip unroll for store-stores chain.
15806 (prepare_initializers_chain_store_elim): New function.
15807 (prepare_initializers_chain): Hanlde store-store chain.
15808 (prepare_finalizers_chain, prepare_finalizers): New function.
15809 (tree_predictive_commoning_loop): Return integer value indicating
15810 if loop is unrolled or lcssa form is corrupted.
15811 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
15813 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15815 * tree-predcom.c (initialize_root): Delete.
15816 (execute_pred_commoning_chain): Initialize root vars and replace
15817 reference of non-combined chain directly, rather than call above
15820 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15822 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
15823 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
15825 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15827 * tree-predcom.c (struct chain): New field init_seq.
15828 (release_chain): Release init_seq.
15829 (prepare_initializers_chain): Record intialization stmts in above
15831 (insert_init_seqs): New function.
15832 (tree_predictive_commoning_loop): Call insert_init_seqs.
15834 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15836 * tree-predcom.c (determine_roots_comp): Skip trivial components.
15838 2017-07-28 Richard Biener <rguenther@suse.de>
15840 * match.pd: Remove superfluous :c.
15841 * genmatch.c (simplify::id): Add member.
15842 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
15844 (current_id): New global.
15845 (dt_node::parent): Move from ...
15846 (dt_operand::parent): ... here. Add for_id member.
15847 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
15848 (decision_tree::find_node): Relax order requirement when
15849 merging DT_TRUE nodes to ones inbetween the current simplify
15850 and the one we try to merge with. Add diagnostic whenever
15851 we need to enforce pattern order by not merging.
15852 (decision_tree::insert): Set current_id.
15853 (decision_tree::print_node): Dump parent node and for_id.
15854 (parser::last_id): Add member.
15855 (parser::push_simplify): Assign unique id.
15856 (parser::parser): Initialize last_id.
15858 2017-07-28 Martin Liska <mliska@suse.cz>
15861 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
15862 gimple_build_debug_bind.
15864 2017-07-28 Richard Biener <rguenther@suse.de>
15866 PR tree-optimization/81502
15867 * match.pd: Add pattern combining BIT_INSERT_EXPR with
15869 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
15871 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
15872 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
15873 for BIT_FIELD_REF args.
15874 * fold-const.c (make_bit_field_ref): Likewise.
15875 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
15877 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15880 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
15881 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
15882 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
15883 Or it into SANITIZER_UNDEFINED.
15884 * ubsan.c: Include gimple-fold.h and varasm.h.
15885 (ubsan_expand_ptr_ifn): New function.
15886 (instrument_pointer_overflow): New function.
15887 (maybe_instrument_pointer_overflow): New function.
15888 (instrument_object_size): Formatting fix.
15889 (pass_ubsan::execute): Call instrument_pointer_overflow
15890 and maybe_instrument_pointer_overflow.
15891 * internal-fn.c (expand_UBSAN_PTR): New function.
15892 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
15893 * sanitizer.def (__ubsan_handle_pointer_overflow,
15894 __ubsan_handle_pointer_overflow_abort): New builtins.
15895 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
15896 * internal-fn.def (UBSAN_PTR): New internal function.
15897 * opts.c (sanitizer_opts): Add pointer-overflow.
15898 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
15899 * fold-const.c (build_range_check): Compute pointer range check in
15900 integral type if pointer arithmetics would be needed. Formatting
15903 2017-07-28 Martin Liska <mliska@suse.cz>
15906 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
15907 parameters that are of a variable-length.
15909 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
15911 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
15913 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
15914 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
15915 (CRT_CALL_STATIC_FUNCTION): Likewise.
15916 (ASM_DEFAULT_SPEC): New define.
15917 (ASM_SPEC32): Likewise.
15918 (ASM_SPEC64): Likewise.
15919 (ASM_SPEC_COMMON): Likewise.
15920 (ASM_SPEC): Likewise.
15921 (INVALID_64BIT): Likewise.
15922 (LINK_OS_DEFAULT_SPEC): Likewise.
15923 (LINK_OS_SPEC32): Likewise.
15924 (LINK_OS_SPEC64): Likewise.
15925 (POWERPC_LINUX): Likewise.
15926 (PTRDIFF_TYPE): Likewise.
15927 (RESTORE_FP_PREFIX): Likewise.
15928 (RESTORE_FP_SUFFIX): Likewise.
15929 (SAVE_FP_PREFIX): Likewise.
15930 (SAVE_FP_SUFFIX): Likewise.
15931 (SIZE_TYPE): Likewise.
15932 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
15933 (TARGET_64BIT): Likewise.
15934 (TARGET_64BIT): Likewise.
15935 (TARGET_AIX): Likewise.
15936 (WCHAR_TYPE_SIZE): Likewise.
15937 (WCHAR_TYPE): Undefine.
15938 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
15939 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
15940 (CPP_OS_RTEMS_SPEC): Delete.
15941 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
15942 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
15944 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
15946 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15948 PR tree-optimization/81578
15949 * tree-parloops.c (build_new_reduction): Bail out if
15950 reduction_code isn't one of the standard OpenMP reductions.
15951 Move the details printing after that decision.
15953 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15955 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
15956 related to reload_in_progress.
15957 (splat_input_operand): Likewise.
15958 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
15960 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
15962 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
15963 (TARGET_INSTANTIATE_DECLS): Likewise.
15964 (legitimate_indexed_address_p): Delete reload_in_progress code.
15965 (rs6000_debug_legitimate_address_p): Likewise.
15966 (rs6000_eliminate_indexed_memrefs): Likewise.
15967 (rs6000_emit_le_vsx_store): Likewise.
15968 (rs6000_emit_move_si_sf_subreg): Likewise.
15969 (rs6000_emit_move): Likewise.
15970 (register_to_reg_type): Likewise.
15971 (rs6000_pre_atomic_barrier): Likewise.
15972 (rs6000_machopic_legitimize_pic_address): Likewise.
15973 (rs6000_allocate_stack_temp): Likewise.
15974 (rs6000_address_for_fpconvert): Likewise.
15975 (rs6000_address_for_altivec): Likewise.
15976 (rs6000_secondary_memory_needed_rtx): Delete function.
15977 (rs6000_check_sdmode): Likewise.
15978 (rs6000_alloc_sdmode_stack_slot): Likewise.
15979 (rs6000_instantiate_decls): Likewise.
15980 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
15981 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
15982 Delete reload_in_progress.
15983 (*vec_reload_and_plus_<mptrsize>): Likewise.
15984 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
15985 (vsx_div_v2di): Likewise.
15986 (vsx_udiv_v2di): Likewise.
15988 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15990 * config/rs6000/rs6000.opt (mlra): Replace with stub.
15991 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
15992 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
15993 (rs6000_debug_reg_global): Delete print of LRA status.
15994 (rs6000_option_override_internal): Delete dead LRA related code.
15995 (rs6000_lra_p): Delete function.
15996 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
15998 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
16000 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
16001 * config/riscv/rtems.h: New file.
16003 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16004 Sudakshina Das <sudi.das@arm.com>
16006 * config/aarch64/aarch64.md
16007 (define_split for and<mode>3nr_compare): Move
16008 non aarch64_logical_operand to a register.
16009 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
16010 register immediate operand to a register.
16011 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
16013 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16015 PR middle-end/81564
16016 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
16018 2017-07-27 Richard Biener <rguenther@suse.de>
16020 PR tree-optimization/81573
16021 PR tree-optimization/81494
16022 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
16023 multi defuse cycle case.
16025 2017-07-27 Richard Biener <rguenther@suse.de>
16027 PR tree-optimization/81571
16028 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
16031 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
16033 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
16034 earlier and only if MASK_FPU is set. Adjust formatting.
16036 2017-07-27 Martin Liska <mliska@suse.cz>
16038 * opt-functions.awk: Add validation of value of Init.
16039 * optc-gen.awk: Pass new argument.
16041 2017-07-27 Martin Liska <mliska@suse.cz>
16043 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
16044 Fix wrong condition.
16046 2017-07-27 Martin Liska <mliska@suse.cz>
16048 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
16049 BBs and edges seen by autoFDO.
16051 2017-07-27 Richard Biener <rguenther@suse.de>
16053 PR tree-optimization/81502
16054 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
16055 with incompatible but same sized type.
16056 (execute_update_addresses_taken): Likewise.
16058 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
16060 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
16061 flag_tree_loop_vectorize rather than flag_tree_vectorize.
16063 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16066 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
16067 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
16068 Change s_operand to memory_operand.
16070 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
16072 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
16073 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
16074 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
16075 Emit instructions rather than returning an expression. Handle TFmode
16076 and KFmode by casting to TImode.
16077 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
16078 (rs6000_emit_le_vsx_store): Likewise.
16079 * config/rs6000/vsx.md (VSX_TI): New iterator.
16080 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
16081 (*vsx_le_undo_permute_<mode>): Likewise.
16082 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
16083 emit the split sequence.
16084 (*vsx_le_perm_store_<mode>): Likewise.
16086 2017-07-27 Jakub Jelinek <jakub@redhat.com>
16088 PR tree-optimization/81555
16089 PR tree-optimization/81556
16090 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
16091 if true, force CHANGED for the recursive invocation.
16092 (reassociate_bb): Remember original length of ops array, pass
16093 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
16095 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
16096 attributes for noipa attribute. For naked attribute use
16097 lookup_attribute first before lookup_attribute_spec.
16098 * final.c (rest_of_handle_final): Disable IPA RA for functions with
16100 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
16101 for functions with noipa attribute.
16102 (cgraph_externally_visible_p): Return true for functions with noipa
16104 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
16105 for functions with noipa attribute.
16106 * doc/extend.texi: Document noipa function attribute.
16107 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
16108 also for functions with noipa attribute.
16109 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
16111 2017-07-26 Andrew Pinski <apinski@cavium.com>
16113 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
16114 vec_unalign_load_cost and vec_unalign_store_cost.
16116 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
16118 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
16119 -mvsx-small-integer option.
16120 (ISA_3_0_MASKS_IEEE): Likewise.
16121 (OTHER_VSX_VECTOR_MASKS): Likewise.
16122 (POWERPC_MASKS): Likewise.
16123 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
16124 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
16125 code, only testing for DImode being allowed in non-VSX floating
16127 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
16128 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
16130 (rs6000_option_override_internal): Delete -mvsx-small-integer.
16131 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
16132 TARGET_P8_VECTOR test.
16133 (rs6000_secondary_reload_simple_move): Likewise.
16134 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
16135 since TARGET_P9_VECTOR was already tested.
16136 (rs6000_opt_masks): Remove -mvsx-small-integer.
16137 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
16138 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16140 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
16141 test for TARGET_VEXTRACTUB was used, and that uses
16143 (p9 extract splitter): Likewise.
16144 (vsx_extract_<mode>_di_p9): Likewise.
16145 (vsx_extract_<mode>_store_p9): Likewise.
16146 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
16147 for TARGET_P9_VECTOR was used. Delete code that is now dead with
16148 the elimination of TARGET_VSX_SMALL_INTEGER.
16149 (vsx_extract_<mode>_p8): Likewise.
16150 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
16151 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
16152 (vsx_set_<mode>_p9): Likewise.
16153 (vsx_set_v4sf_p9): Likewise.
16154 (vsx_set_v4sf_p9_zero): Likewise.
16155 (vsx_insert_extract_v4sf_p9): Likewise.
16156 (vsx_insert_extract_v4sf_p9_2): Likewise.
16157 * config/rs6000/rs6000.md (sign extend splitter): Change
16158 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
16159 (floatsi<mode>2_lfiwax_mem): Likewise.
16160 (floatunssi<mode>2_lfiwzx_mem): Likewise.
16161 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
16162 since a test for TARGET_P9_VECTOR was used.
16163 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16164 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
16165 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16166 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16167 TARGET_P8_VECTOR test.
16168 (fix_trunc<mode>si2_stfiwx): Likewise.
16169 (fix_trunc<mode>si2_internal): Likewise.
16170 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
16171 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16173 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16174 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16175 TARGET_P8_VECTOR test.
16176 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16177 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
16178 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16180 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16181 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16182 since a test for TARGET_P9_VECTOR was used.
16183 (splitter for loading small constants): Likewise.
16185 2017-07-26 Andrew Pinski <apinski@cavium.com>
16187 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16190 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16193 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16194 (fp_valid_at): Likewise.
16196 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16198 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16199 (qdf24xx_addrcost_table): Likewise.
16200 (cortexa57_tunings): Update to use generic_branch_cost.
16201 (cortexa72_tunings): Likewise.
16202 (cortexa73_tunings): Likewise.
16203 (qdf24xx_tunings): Likewise.
16205 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16207 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16208 (thunderx2t99_branch_cost): Likewise.
16209 (cortexa35_tunings): Update to use generic_branch_cost.
16210 (cortexa53_tunings): Likewise.
16211 (cortexa57_tunings): Likewise.
16212 (cortexa72_tunings): Likewise.
16213 (cortexa73_tunings): Likewise.
16214 (thunderx2t99_tunings): Likewise.
16216 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16218 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16219 (sparc_option_override): Honour MASK_FSMULD.
16220 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16221 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16222 * config/sparc/sparc.opt (mfsmuld): New option.
16223 * doc/invoke.texi (mfsmuld): Document option.
16225 2017-07-26 Marek Polacek <polacek@redhat.com>
16227 PR middle-end/70992
16228 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16230 2017-07-26 Richard Biener <rguenther@suse.de>
16232 * gimple-match-head.c (do_valueize): Return OP if valueize
16234 (get_def): New helper to get at the def stmt of a SSA name
16235 if valueize allows.
16236 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16237 do_valueize to get at the def stmt.
16238 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16240 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16242 PR middle-end/46932
16243 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16245 2017-07-26 Martin Liska <mliska@suse.cz>
16248 * function.c (expand_function_start): Make expansion of
16249 nonlocal_goto_save_area after parm_birth_insn.
16251 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16253 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16254 from all CPU target flags enable members.
16256 2017-07-26 Richard Biener <rguenther@suse.de>
16258 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16259 (decision_tree::gen): Make 'type' const.
16260 (write_predicate): Likewise.
16262 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16264 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16265 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16266 (rs6000_option_override_internal): Likewise.
16267 (rs6000_expand_vector_set): Likewise.
16268 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16269 (TARGET_UPPER_REGS_SF): Likewise.
16270 (TARGET_UPPER_REGS_DI): Likewise.
16271 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16272 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16273 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16274 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16275 (Splitters for DI constants in Altivec registers): Likewise.
16276 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16277 (vsx_set_v4sf_p9): Likewise.
16278 (vsx_set_v4sf_p9_zero): Likewise.
16279 (vsx_insert_extract_v4sf_p9): Likewise.
16280 (vsx_insert_extract_v4sf_p9_2): Likewise.
16282 2017-07-25 Carl Love <cel@us.ibm.com>
16284 * doc/extend.texi: Update the built-in documentation file for the
16285 existing built-in functions
16286 vector signed char vec_cnttz (vector signed char);
16287 vector unsigned char vec_cnttz (vector unsigned char);
16288 vector signed short vec_cnttz (vector signed short);
16289 vector unsigned short vec_cnttz (vector unsigned short);
16290 vector signed int vec_cnttz (vector signed int);
16291 vector unsigned int vec_cnttz (vector unsigned int);
16292 vector signed long long vec_cnttz (vector signed long long);
16293 vector unsigned long long vec_cnttz (vector unsigned long long);
16295 2017-07-25 Andrew Pinski <apinski@cavium.com>
16297 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16298 accesses where the use is for the first operand of a BIT_INSERT.
16300 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16303 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16304 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16306 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16308 * config/i386/gstabs.h: Delete.
16309 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16311 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16313 * config/i386/i386.c (ix86_decompose_address): Do not check for
16314 register RTX when looking at index_reg or base_reg.
16315 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16317 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16319 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16320 to update EH info here.
16322 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16324 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16326 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16328 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16330 2017-07-25 Torsten Duwe <duwe@suse.de>
16332 * common.opt: Introduce -fpatchable-function-entry
16333 command line option, and its variables function_entry_patch_area_size
16334 and function_entry_patch_area_start.
16335 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16336 including a two-value parser.
16337 * target.def (print_patchable_function_entry): New target hook.
16338 * targhooks.h (default_print_patchable_function_entry): New function.
16339 * targhooks.c (default_print_patchable_function_entry): Likewise.
16340 * toplev.c (process_options): Switch off IPA-RA if
16341 patchable function entries are being generated.
16342 * varasm.c (assemble_start_function): Look at the
16343 patchable-function-entry command line switch and current
16344 function attributes and maybe generate NOP instructions by
16345 calling the print_patchable_function_entry hook.
16346 * doc/extend.texi: Document patchable_function_entry attribute.
16347 * doc/invoke.texi: Document -fpatchable_function_entry
16348 command line option.
16349 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16351 * doc/tm.texi: Re-generate.
16353 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16356 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16357 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16359 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16361 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16363 2017-07-25 Richard Biener <rguenther@suse.de>
16365 PR tree-optimization/81455
16366 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16367 not walk in cycles when looking for guards.
16369 2017-07-25 Richard Biener <rguenther@suse.de>
16371 PR tree-optimization/81529
16372 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16373 when optimizing backedge uses.
16375 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16377 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16379 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16380 to dl_section_ref. On AIX, append an expression to subtract
16381 the size of the section length to dl_section_ref.
16383 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16385 * configure.ac: If any of the config.* scripts fail, exit 1.
16386 * configure: Regenerate.
16388 2017-07-25 Richard Biener <rguenther@suse.de>
16390 PR middle-end/81546
16391 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16392 of immediate uses, be more verbose on errors.
16394 2017-07-25 Richard Biener <rguenther@suse.de>
16396 PR tree-optimization/81510
16397 * tree-vect-loop.c (vect_is_simple_reduction): When the
16398 reduction stmt is not inside the loop bail out.
16400 2017-07-25 Richard Biener <rguenther@suse.de>
16402 PR tree-optimization/81303
16403 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16404 profitability check against LOOP_VINFO_NITERSM1.
16406 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16408 * domwalk.c (cmp_bb_postorder): Simplify.
16409 (sort_bbs_postorder): New function. Use it...
16410 (dom_walker::walk): ...here to optimize common cases.
16412 2017-07-25 Martin Liska <mliska@suse.cz>
16415 * ipa-visibility.c (function_and_variable_visibility): Make the
16416 redirection just on target that supports aliasing.
16417 Fix GNU coding style.
16419 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16422 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16424 * config/arm/rtems.h: Likewise.
16425 * config/bfin/rtems.h: Likewise.
16426 * config/i386/rtemself.h: Likewise.
16427 * config/lm32/rtems.h: Likewise.
16428 * config/m32c/rtems.h: Likewise.
16429 * config/m68k/rtemself.h: Likewise.
16430 * config/microblaze/rtems.h: Likewise.
16431 * config/mips/rtems.h: Likewise.
16432 * config/moxie/rtems.h: Likewise.
16433 * config/nios2/rtems.h: Likewise.
16434 * config/powerpcspe/rtems.h: Likewise.
16435 * config/rs6000/rtems.h: Likewise.
16436 * config/rtems.h: Likewise.
16437 * config/sh/rtems.h: Likewise.
16438 * config/sh/rtemself.h: Likewise.
16439 * config/sparc/rtemself.h: Likewise.
16441 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16444 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16445 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16446 * tree-ssa-structalias.c (alias_get_name): Same.
16448 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16451 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16452 instructions if no du chain is found.
16454 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16456 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16458 2017-07-25 Richard Biener <rguenther@suse.de>
16460 PR middle-end/81505
16461 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16464 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16466 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16467 upper-regs options.
16468 (ISA_2_7_MASKS_SERVER): Likewise.
16469 (ISA_3_0_MASKS_IEEE): Likewise.
16470 (OTHER_P8_VECTOR_MASKS): Likewise.
16471 (OTHER_VSX_VECTOR_MASKS): Likewise.
16472 (POWERPC_MASKS): Likewise.
16473 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
16474 duplicate list of options.
16475 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
16476 explicit -mupper-regs options.
16477 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
16478 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
16479 alias for -mupper-regs-df.
16480 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16481 (rs6000_init_hard_regno_mode_ok): Likewise.
16482 (rs6000_option_override_internal): Likewise.
16483 (rs6000_opt_masks): Likewise.
16484 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
16485 options in terms of whether -mvsx or -mpower8-vector was used.
16486 (TARGET_UPPER_REGS_DI): Likewise.
16487 (TARGET_UPPER_REGS_SF): Likewise.
16488 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
16489 -mupper-regs-* options.
16491 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16493 * passes.c (emergency_dump_function): Print some empty lines and a
16494 header before the RTL dump.
16496 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16498 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
16500 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
16503 * config/aarch64/aarch64.c (aarch64_classify_symbol):
16504 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
16506 2017-07-24 Carl Love <cel@us.ibm.com>
16508 * config/rs6000/rs6000-c.c: Add support for built-in functions
16509 vector float vec_extract_fp32_from_shorth (vector unsigned short);
16510 vector float vec_extract_fp32_from_shortl (vector unsigned short);
16511 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
16512 vec_extract_fp_from_shortl): Add defines for the two builtins.
16513 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
16514 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
16516 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
16517 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
16518 * doc/extend.texi: Update the built-in documentation file for the
16519 new built-in function.
16521 2017-07-24 Jakub Jelinek <jakub@redhat.com>
16524 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
16526 * doc/generic.texi: Likewise.
16527 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16528 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16530 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
16532 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
16533 (aarch64_mls_elt_merge<mode>): Likewise.
16535 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
16537 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16538 having __cxa_atexit.
16540 2017-07-23 Michael Collison <michael.collison@arm.com>
16542 * config/arm/arm.c (arm_option_override): Deprecate
16543 use of -mstructure-size-boundary.
16544 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
16545 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
16547 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16550 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16551 Reduce cost estimate for direct moves.
16553 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
16556 * config/i386/i386.c (ix86_option_override_internal): Disable
16557 BMI, BMI2 and TBM instructions for -m16.
16559 2017-07-21 Carl Love <cel@us.ibm.com>
16561 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16562 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16563 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16564 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16565 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16566 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16567 VMULOSW): New enum "unspec" values.
16568 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16569 altivec_vmulosw): New patterns.
16570 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16571 VMULOSW): Add definitions.
16573 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
16575 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
16576 (qdf24xx): Likewise.
16577 * config/aarch64/aarch64-options-extensions.def (rdma); New.
16578 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
16579 (AARCH64_FL_V8_1): Renumber.
16580 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
16581 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
16582 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
16583 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
16584 rdma to feature modifiers list.
16586 2017-07-21 Yury Gribov <tetra2005@gmail.com>
16588 PR middle-end/56727
16589 * ipa-visibility (function_and_variable_visibility): Convert
16590 recursive PLT call to direct call if appropriate.
16592 2017-07-21 Andrew Pinski <apinski@cavium.com>
16594 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
16595 operand 1 to see if the types precision matches.
16596 * fold-const.c (operand_equal_p): Likewise.
16598 2017-07-21 Richard Biener <rguenther@suse.de>
16600 PR tree-optimization/81303
16601 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
16602 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
16603 (vect_peeling_hash_get_lowest_cost): Adjust.
16604 (vect_enhance_data_refs_alignment): Likewise. Use
16605 vect_get_peeling_costs_all_drs to compute the penalty for no
16606 peeling to match up costs.
16608 2017-07-21 Richard Biener <rguenther@suse.de>
16610 PR tree-optimization/81500
16611 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
16612 we didn't identify a reduction path.
16614 2017-07-21 Tom de Vries <tom@codesourcery.com>
16615 Cesar Philippidis <cesar@codesourcery.com>
16617 PR gcov-profile/81442
16618 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
16621 2017-07-21 Tom de Vries <tom@codesourcery.com>
16624 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
16626 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
16627 nvptx_override_options_after_change.
16629 2017-07-21 Ulrich Drepper <drepper@redhat.com>
16631 * dwarf2out.c (output_file_names): Avoid double testing for
16632 dwarf_version >= 5.
16634 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
16636 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
16638 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16640 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
16642 (try_crossjump_to_edge): Do not punt on partitioned functions.
16644 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16646 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
16647 Put all BBs reachable only via paths crossing cold region to cold
16649 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
16651 2016-07-21 Richard Biener <rguenther@suse.de>
16653 PR tree-optimization/81303
16654 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
16655 into account prologue and epilogue iterations when raising
16656 min_profitable_iters to sth at least covering one vector iteration.
16658 2017-07-21 Tamar Christina <tamar.christina@arm.com>
16660 * config/arm/arm.c (arm_test_cpu_arch_dat):
16663 2017-07-20 Nathan Sidwell <nathan@acm.org>
16665 Remove TYPE_METHODS.
16666 * tree.h (TYPE_METHODS): Delete.
16667 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
16668 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
16669 (dbxout_type_methods): Scan TYPE_FIELDS.
16670 (dbxout_type): Don't check TYPE_METHODS here.
16671 * function.c (use_register_for_decl): Always ignore register for
16672 class types when not optimizing.
16673 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
16674 * tree.c (free_lang_data_in_type): Stitch out member functions and
16675 templates from TYPE_FIELDS.
16676 (build_distinct_type_copy, verify_type_variant,
16677 verify_type): Member fns are on TYPE_FIELDS.
16678 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
16679 * tree-pretty-print.c (dump_generic_node): Likewise.
16681 2017-07-20 Jakub Jelinek <jakub@redhat.com>
16684 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
16685 V2TImode and V4TImode.
16686 (ix86_expand_vector_extract): Likewise.
16687 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
16688 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
16689 (ssescalarmode): Handle V4TImode and V2TImode.
16690 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
16691 (*vec_extractv2ti, *vec_extractv4ti): New insns.
16692 (VEXTRACTI128_MODE): New mode iterator.
16693 (splitter for *vec_extractv?ti first element): New.
16694 (VEC_INIT_MODE): New mode iterator.
16695 (vec_init<mode>): Consolidate 3 expanders into one using
16696 VEC_INIT_MODE mode iterator.
16698 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16700 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
16701 non_spilled_static_chain_regno_p.
16703 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16705 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
16707 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
16709 * bb-reorder.c (connect_traces): Allow copying of blocks within
16712 2017-07-20 Richard Biener <rguenther@suse.de>
16714 * gimple.h (gimple_phi_result): Add gphi * overload.
16715 (gimple_phi_result_ptr): Likewise.
16716 (gimple_phi_arg): Likewise. Adjust index assert to only
16717 allow actual argument accesses rather than all slots available
16719 (gimple_phi_arg_def): Add gphi * overload.
16720 * tree-phinodes.c (make_phi_node): Initialize only actual
16722 (resize_phi_node): Clear memory not covered by old node,
16723 do not initialize excess argument slots.
16724 (reserve_phi_args_for_new_edge): Initialize new argument slot
16727 2017-07-20 Bin Cheng <bin.cheng@arm.com>
16729 PR tree-optimization/81388
16731 2016-07-21 Bin Cheng <bin.cheng@arm.com>
16733 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
16734 by removing computation of may_be_zero.
16736 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16737 Tom de Vries <tom@codesourcery.com>
16739 PR middle-end/81030
16740 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
16741 when gimple level profile disagrees with what RTL expander did.
16743 2017-07-20 Richard Biener <rguenther@suse.de>
16745 PR tree-optimization/61171
16746 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
16747 (vect_analyze_stmt): Add slp instance parameter.
16748 (vectorizable_reduction): Likewise.
16749 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
16750 (vect_is_simple_reduction): Deal with chains not detected
16751 as SLP reduction chain, specifically not properly associated
16752 chains containing a mix of plus/minus.
16753 (get_reduction_op): Remove.
16754 (get_initial_defs_for_reduction): Simplify, pass in whether
16755 this is a reduction chain, pass in the SLP node for the PHIs.
16756 (vect_create_epilog_for_reduction): Get the SLP instance as
16758 (vectorizable_reduction): Get the SLP instance as arg.
16759 During analysis remember the SLP node with the PHIs in the
16760 instance. Simplify getting at the vectorized reduction PHIs.
16761 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
16762 through SLP instance.
16763 (vect_slp_analyze_operations): Likewise.
16764 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
16765 (vect_transform_stmt): Likewise.
16767 2017-07-20 Tom de Vries <tom@codesourcery.com>
16769 PR tree-optimization/81489
16770 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
16771 read of phi arg location to before loop that modifies phi.
16773 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16775 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
16778 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16780 PR middle-end/81331
16781 * except.c (execute): Fix ordering issue.
16783 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16785 PR rtl-optimization/81423
16786 * combine.c (make_compound_operation_int): Don't try to optimize
16787 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
16789 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16791 PR rtl-optimization/81423
16792 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
16793 with a constant that is -1 in the truncated to mode.
16795 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16797 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
16798 (determine_unlikely_bbs): ... here.
16799 * predict.h (propagate_unlikely_bbs_forward): Declare.
16800 * cfgexpand.c (pass_expand::execute): Use it.
16801 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
16803 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
16804 propagate_unlikely_bbs_forward.
16806 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16808 PR middle-end/81331
16809 * except.c (maybe_add_nop_after_section_switch): New function.
16812 2017-07-19 Tom de Vries <tom@codesourcery.com>
16814 * gimple.h (gimple_phi_set_arg): Make assert more strict.
16816 2017-07-19 Tom de Vries <tom@codesourcery.com>
16818 * gimple.h (gimple_phi_arg): Make assert more strict.
16820 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
16822 * config.gcc (powerpc*-*-*): Add mmintrin.h.
16823 * config/rs6000/mmintrin.h: New file.
16824 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
16826 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16828 PR tree-optimization/81346
16829 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
16831 2017-07-19 Tom de Vries <tom@codesourcery.com>
16833 * config/nvptx/nvptx.md (VECIM): Add V2DI.
16835 2017-07-19 Tom de Vries <tom@codesourcery.com>
16837 * config/nvptx/nvptx-modes.def: Add V2DImode.
16838 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
16839 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
16840 (nvptx_output_mov_insn): Handle lack of mov.b128.
16841 (nvptx_print_operand): Handle 'H' and 'L' codes.
16842 (nvptx_vector_mode_supported): Allow V2DImode.
16843 (nvptx_preferred_simd_mode): New function.
16844 (nvptx_data_alignment): New function.
16845 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
16846 nvptx_preferred_simd_mode.
16847 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
16849 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
16851 2017-07-19 Tom de Vries <tom@codesourcery.com>
16853 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
16854 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
16855 (nvptx_vector_mode_supported): New function. Allow V2SImode.
16856 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
16857 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
16858 (mov<VECIM>_insn): New define_insn.
16859 (define_expand "mov<VECIM>): New define_expand.
16861 2017-07-19 Tom de Vries <tom@codesourcery.com>
16863 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
16865 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16867 PR tree-optimization/81346
16868 * fold-const.h (fold_div_compare, range_check_type): Declare.
16869 * fold-const.c (range_check_type): New function.
16870 (build_range_check): Use range_check_type.
16871 (fold_div_compare): No longer static, rewritten into
16872 a match.pd helper function.
16873 (fold_comparison): Don't call fold_div_compare here.
16874 * match.pd (X / C1 op C2): New optimization using fold_div_compare
16875 as helper function.
16877 2017-07-19 Nathan Sidwell <nathan@acm.org>
16879 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
16880 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
16881 * tree.c (find_decls_types_r, verify_type): Use
16882 TYPE_{MIN,MAX}_VALUE_RAW.
16883 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
16884 (hash_tree): Likewise.
16885 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16887 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16890 2017-07-18 Tom de Vries <tom@codesourcery.com>
16892 PR middle-end/81464
16893 * omp-expand.c (expand_omp_for_static_chunk): Handle
16894 equal-argument loop exit phi.
16896 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
16899 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
16900 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
16901 operand 2 predicate.
16902 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
16903 operand 2 predicate.
16904 (ror,rol -> rorx splitters): Use const_int_operand as
16905 operand 2 predicate.
16907 2017-06-18 Richard Biener <rguenther@suse.de>
16909 PR tree-optimization/81410
16910 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
16911 the gap in the ! slp_perm SLP case after each group.
16913 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16915 PR middle-end/81463
16916 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
16919 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16921 PR middle-end/81462
16922 * predict.c (set_even_probabilities): Cleanup; do not affect
16923 probabilities that are already known.
16924 (combine_predictions_for_bb): Call even when count is set.
16926 2017-07-18 Nathan Sidwell <nathan@acm.org>
16928 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
16931 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16934 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
16935 optimization for loop niter analysis.
16937 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
16940 * config/avr/avr.c (avr_optimize_casesi): Don't use
16941 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
16943 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
16945 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
16946 body_cost_vec from _vect_peel_extended_info.
16947 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
16948 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
16951 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16953 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
16955 2017-07-18 Richard Biener <rguenther@suse.de>
16957 PR tree-optimization/80620
16958 PR tree-optimization/81403
16959 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
16960 info when re-using a VN table entry.
16962 2017-07-18 Richard Biener <rguenther@suse.de>
16964 PR tree-optimization/81418
16965 * tree-vect-loop.c (vectorizable_reduction): Properly compute
16966 vectype_in. Verify that with lane-reducing reduction operations
16967 we have a single def-use cycle.
16969 2017-07-17 Carl Love <cel@us.ibm.com>
16971 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
16973 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16974 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16975 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16976 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16977 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16978 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16979 VMULOSW): New enum "unspec" values.
16980 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16981 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16982 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16983 altivec_vmulosw): New patterns.
16984 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16985 VMULOSW): Add definitions.
16986 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
16988 * config/alpha/alpha.c: Include predict.h.
16990 2017-07-17 Yury Gribov <tetra2005@gmail.com>
16992 * tree-vrp.c (compare_assert_loc): Fix comparison function
16993 to return predictable results.
16995 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16997 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
16999 (subdi3): Likewise.
17000 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
17001 * doc/invoke.texi (mexpand-adddi): Update text.
17003 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17005 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
17006 that also clobbers the CC register. The old expand code is moved
17008 (*arc_clzsi2): ... here.
17009 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
17010 the CC register. The old expand code is moved to ...
17011 (arc_ctzsi2): ... here.
17013 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17015 * config/arc/arc.opt (mindexed-loads): Use initial value
17016 TARGET_INDEXED_LOADS_DEFAULT.
17017 (mauto-modify-reg): Use initial value
17018 TARGET_AUTO_MODIFY_REG_DEFAULT.
17019 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17020 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17021 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17022 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17024 2017-07-17 Martin Liska <mliska@suse.cz>
17027 * opts.c (finish_options): Do not allow -fgnu-tm
17028 w/ -fsanitize={kernel-,}address. Say sorry.
17030 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17033 * tree-loop-distribution.c (classify_partition): Only assert on
17034 numer of iterations.
17035 (merge_dep_scc_partitions): Delete prameter. Update function call.
17036 (distribute_loop): Remove code handling loop with unknown niters.
17037 (pass_loop_distribution::execute): Skip loop with unknown niters.
17039 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17042 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
17043 function sort_partitions_by_post_order.
17045 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17047 PR tree-optimization/81374
17048 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
17049 the max index of basic blocks, rather than number of basic blocks.
17051 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17053 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
17055 (arc_legitimate_pic_operand_p): Likewise.
17056 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
17058 (arc_needs_pcl_p): Likewise.
17059 (arc_legitimate_pc_offset_p): Likewise.
17060 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
17061 function is also used in constrains.md.
17062 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
17063 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
17064 PLUS. Only return true/false in known cases, otherwise assert.
17065 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
17066 is already called in arc_legitimate_constant_p.
17067 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
17069 (LEGITIMATE_PIC_OPERAND_P): Use
17070 arc_raw_symbolic_reference_mentioned_p function.
17071 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
17076 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17077 Andrew Burgess <andrew.burgess@embecosm.com>
17079 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
17080 (arc_return_address_register): New function.
17081 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
17082 (arc_handle_fndecl_attribute): Add naked attribute.
17083 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17084 (TARGET_WARN_FUNC_RETURN): Likewise.
17085 (arc_allocate_stack_slots_for_args): New function.
17086 (arc_warn_func_return): Likewise.
17087 (machine_function): Change type fn_type.
17088 (arc_compute_function_type): Consider new naked function type,
17089 change function return type.
17090 (arc_must_save_register): Adapt to handle new
17091 arc_compute_function_type's return type.
17092 (arc_expand_prologue): Likewise.
17093 (arc_expand_epilogue): Likewise.
17094 (arc_return_address_regs): Delete.
17095 (arc_return_address_register): New function.
17096 (arc_epilogue_uses): Use above function.
17097 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
17098 (arc_function_type): Change encoding, add naked type.
17099 (ARC_INTERRUPT_P): Change to handle the new encoding.
17100 (ARC_FAST_INTERRUPT_P): Likewise.
17101 (ARC_NORMAL_P): Define.
17102 (ARC_NAKED_P): Likewise.
17103 (arc_compute_function_type): Delete prototype.
17104 * config/arc/arc.md (in_ret_delay_slot): Use
17105 arc_return_address_register function.
17106 (simple_return): Likewise.
17107 (p_return_i): Likewise.
17109 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17111 PR tree-optimization/81428
17112 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
17113 can't be built for those types.
17115 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17117 Remove stuff dead since r239246.
17119 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
17120 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
17121 (avr_inform_devices): Remove dead stuff.
17123 2017-07-17 Tamar Christina <tamar.christina@arm.com>
17125 * config/arm/arm_neon.h: Fix softp typo.
17127 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17129 PR tree-optimization/81365
17130 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
17131 aggregate moves onto bb predecessor edges, make sure there are no
17132 loads that could alias the lhs in between the start of bb and the
17135 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17138 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
17139 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
17140 [LSHIFTRT, outer_code = TRUNCATE]: Same.
17142 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17144 PR tree-optimization/81396
17145 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
17146 (init_symbolic_number): Initialize it to 1.
17147 (perform_symbolic_merge): Add n_ops from both operands into the new
17149 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
17150 without base_addr as useless if they need more than one operation.
17151 (bswap_replace): Handle !bswap case for NULL base_addr.
17153 2017-07-17 Tom de Vries <tom@codesourcery.com>
17156 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
17159 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
17161 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
17162 conditional builtin define __FIX_LEON3FT_B2BST.
17164 2017-07-17 Daniel Cederman <cederman@gaisler.com>
17166 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
17167 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
17170 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17172 PR rtl-optimization/81424
17173 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
17174 to remove potential trapping from operands if -fnon-call-exceptions.
17176 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17178 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
17179 profile_proability for scalling.
17180 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17182 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17184 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17186 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17188 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17189 fixpoint arithmetics.
17191 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17193 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17194 fixpoint arithmetics.
17196 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17198 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17199 fixpoint arithmetics.
17201 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17203 * profile-count.h (profile_probability::from_reg_br_prob_note,
17204 profile_probability::to_reg_br_prob_note): New functions.
17205 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17206 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17207 * predict.c (probability_reliable_p): Update.
17208 (edge_probability_reliable_p): Update.
17209 (br_prob_note_reliable_p): Update.
17210 (invert_br_probabilities): Update.
17211 (add_reg_br_prob_note): New function.
17212 (combine_predictions_for_insn): Update.
17213 * asan.c (asan_clear_shadow): Update.
17214 * cfgbuild.c (compute_outgoing_frequencies): Update.
17215 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17216 (update_br_prob_note): Update.
17217 (rtl_verify_edges): Update.
17218 (purge_dead_edges): Update.
17219 (fixup_reorder_chain): Update.
17220 * emit-rtl.c (try_split): Update.
17221 * ifcvt.c (cond_exec_process_insns): Update.
17222 (cond_exec_process_if_block): Update.
17223 (dead_or_predicable): Update.
17224 * internal-fn.c (expand_addsub_overflow): Update.
17225 (expand_neg_overflow): Update.
17226 (expand_mul_overflow): Update.
17227 * loop-doloop.c (doloop_modify): Update.
17228 * loop-unroll.c (compare_and_jump_seq): Update.
17229 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17230 * predict.h: Update.
17231 * reorg.c (mostly_true_jump): Update.
17233 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17234 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17235 * config/arc/arc.c: (emit_unlikely_jump): Update.
17236 * config/arm/arm.c: (emit_unlikely_jump): Update.
17237 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17238 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17239 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17240 (ix86_print_operand): Update.
17241 (ix86_split_fp_branch): Update.
17242 (predict_jump): Update.
17243 * config/ia64/ia64.c (ia64_print_operand): Update.
17244 * config/mmix/mmix.c (mmix_print_operand): Update.
17245 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17246 (rs6000_expand_split_stack_prologue): Update.
17247 * config/rs6000/rs6000.c: Update.
17248 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17249 (s390_expand_vec_movstr): Update.
17250 (s390_expand_cs_tdsi): Update.
17251 (s390_expand_split_stack_prologue): Update.
17252 * config/sh/sh.c (sh_print_operand): Update.
17253 (expand_cbranchsi4): Update.
17254 (expand_cbranchdi4): Update.
17255 * config/sparc/sparc.c (output_v9branch): Update.
17256 * config/spu/spu.c (get_branch_target): Update.
17257 (ea_load_store_inline): Update.
17258 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17259 * config/tilepro/tilepro.c: Update.
17261 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17263 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17264 (gimplify_save_expr): Likewise.
17266 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17268 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17270 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17272 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17274 * timevar.def (TV_IPA_FNSUMMARY): Define.
17276 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17278 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17279 to back store errata sensitive sequence from being generated.
17280 (sqrtdf2_fix): Likewise.
17282 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17284 * tree-ssa-threadupdate.c (compute_path_counts,
17285 update_joiner_offpath_counts): Use profile_probability.
17287 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17290 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17292 * config/arm/arm-c.c (arm_cpu_builtins): Define
17293 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17295 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17297 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17298 array entries to represent __ieee128 versions of the
17299 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17300 scalar_extract_sig, and scalar_insert_exp built-in functions.
17301 (altivec_resolve_overloaded_builtin): Add special case handling
17302 for the __builtin_scalar_insert_exp function, as represented by
17303 the P9V_BUILTIN_VEC_VSIEDP constant.
17304 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17305 exponent support for __ieee128 argument.
17306 (VSESQP): Add scalar extract signature support for __ieee128
17308 (VSTDCNQP): Add scalar test negative support for __ieee128
17310 (VSIEQP): Add scalar insert exponent support for __int128 argument
17311 with __ieee128 result.
17312 (VSIEQPF): Add scalar insert exponent support for __ieee128
17313 argument with __ieee128 result.
17314 (VSTDCQP): Add scalar test data class support for __ieee128
17316 (VSTDCNQP): Add overload support for scalar test negative with
17317 __ieee128 argument.
17318 (VSTDCQP): Add overload support for scalar test data class
17319 __ieee128 argument.
17320 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17321 UNSPEC_VSX_SXSIGDP.
17322 (UNSPEC_VSX_SIEXPQP): New constant.
17323 (xsxexpqp): New insn for VSX scalar extract exponent quad
17325 (xsxsigqp): New insn for VSX scalar extract significand quad
17327 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17328 precision with floating point argument.
17329 (xststdcqp): New expand for VSX scalar test data class quad
17331 (xststdcnegqp): New expand for VSX scalar test negative quad
17333 (xststdcqp): New insn to match expansions for VSX scalar test data
17334 class quad precision and VSX scalar test negative quad precision.
17335 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17336 special case operand checking to enforce that second operand of
17337 VSX scalar test data class with quad precision argument is a 7-bit
17339 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17340 prototypes and descriptions of __ieee128 versions of
17341 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17342 scalar_test_data_class, and scalar_test_neg built-in functions.
17344 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17346 PR tree-optimization/81162
17347 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17348 replace a negate with an add.
17350 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17352 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17354 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17356 * config/arm/arm-c.c (arm_cpu_builtins): Define
17357 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17359 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17361 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17362 (armv8-r): Set ARM Cortex-R52 as default CPU.
17363 * config/arm/arm-tables.opt: Regenerate.
17364 * config/arm/arm-tune.md: Regenerate.
17365 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17367 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17368 extension for -mcpu=cortex-r52.
17370 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17372 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17373 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17374 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17375 (fp-armv8): Define it as FP_ARMv8 only.
17376 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17377 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17379 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17380 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17381 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17382 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17383 than TARGET_FPU_ARMV8.
17384 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17385 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17386 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17388 * config/arm/neon.md (neon_vrint): Likewise.
17389 (neon_vcvt): Likewise.
17390 (neon_<fmaxmin_op><mode>): Likewise.
17391 (<fmaxmin><mode>3): Likewise.
17392 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17393 * config/arm/predicates.md (arm_cond_move_operator): Check against
17394 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17396 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17398 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17399 to top of function.
17401 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17403 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17404 loop in comment with memset.
17406 2017-07-14 Martin Liska <mliska@suse.cz>
17408 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17409 * dwarf2out.c (is_java): Remove the function.
17410 (output_pubname): Remove usage of the function.
17411 (lower_bound_default): Remove usage of DW_LANG_Java.
17412 (gen_compile_unit_die): Likewise.
17413 * gcc.c: Remove compiler defaults for .java and .zip files.
17414 * gimple-expr.c (remove_suffix): Change as there's no longer
17415 extension than 4-letter one.
17416 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17417 (gimplify_save_expr): Likewise.
17418 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17419 as it's possible even for other languages than Java.
17420 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17421 * lto-opts.c (lto_write_options): Remove reference to Java.
17422 * opts.c (strip_off_ending): Update file extension handling.
17423 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17424 * tree-eh.c (lower_resx): Likewise.
17425 * tree.c (free_lang_data_in_type): Remove dead code.
17426 (find_decls_types_r): Likewise.
17427 (build_common_builtin_nodes): Remove Java from a comment.
17428 (verify_type): Remove dead code.
17429 * varasm.c (assemble_external): Remove Java from a comment.
17431 2017-07-14 Martin Liska <mliska@suse.cz>
17433 * opts.c (finish_options): Add quotes.
17434 (common_handle_option): Likewise.
17436 2017-07-14 Martin Liska <mliska@suse.cz>
17438 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17439 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17440 Remove N_SO_PASCAL.
17441 * dwarf2out.c (lower_bound_default): Do not handle
17443 (gen_compile_unit_die): Likewise.
17444 * gcc.c: Remove default extension binding for GNU Pascal.
17445 * stmt.c: Remove Pascal language from a comment.
17446 * xcoffout.c: Likewise.
17448 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17451 * diagnostic-show-locus.c (fixit_cmp): New function.
17452 (layout::layout): Sort m_fixit_hints.
17453 (column_range::column_range): Assert that the values are valid.
17454 (struct char_span): New struct.
17455 (correction::overwrite): New method.
17456 (struct source_line): New struct.
17457 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17458 calls in terms of classes source_line and char_span, and
17459 correction::overwrite.
17460 (selftest::test_overlapped_fixit_printing_2): New function.
17461 (selftest::diagnostic_show_locus_c_tests): Call it.
17463 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17465 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17466 early if there is no lhs.
17468 2017-07-13 Martin Liska <mliska@suse.cz>
17470 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
17471 (gen_reference_type_die): Likewise.
17472 * stor-layout.c: Remove Pascal-related comment.
17474 2017-07-13 Martin Liska <mliska@suse.cz>
17476 * opts.c (finish_options): Add quotes to error messages.
17477 (parse_sanitizer_options): Likewise.
17479 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
17481 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
17483 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
17485 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
17487 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17489 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
17491 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
17493 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17496 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
17497 provides the hardware capability bits, define the macro
17498 __BUILTIN_CPU_SUPPORTS__.
17499 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
17500 if GLIBC does not provide the hardware capability bits. Add a
17501 gcc_unreachable call if the built-in cpu function is neither
17502 __builtin_cpu_is nor __builtin_cpu_supports.
17503 (rs6000_get_function_versions_dispatcher): Change the warning
17504 that an old GLIBC is used which does not export the capability
17505 bits to be an error.
17506 * doc/extend.texi (target_clones attribute): Document the
17507 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
17508 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
17509 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
17510 the macros defined by GCC if the newer GLIBC is available.
17512 2017-07-12 Jeff Law <law@redhat.com>
17514 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
17515 remaining includes slightly.
17516 * config/riscv/riscv-builtins.c: Include profile-count.h.
17518 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17521 * config/avr/avr.c (avr_set_current_function): In diagnostic
17522 messages: Quote keywords and (parts of) identifiers.
17523 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
17526 2017-07-12 Carl Love <cel@us.ibm.com>
17528 * config/rs6000/rs6000-c.c: Add support for built-in functions
17529 vector bool char vec_revb (vector bool char);
17530 vector bool short vec_revb (vector short char);
17531 vector bool int vec_revb (vector bool int);
17532 vector bool long long vec_revb (vector bool long long);
17533 * doc/extend.texi: Update the built-in documentation file for the
17534 new built-in functions.
17536 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17538 * config/s390/s390.md: Remove movcc splitter.
17540 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17542 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
17543 load/store on condition.
17545 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17548 * config/avr/avr.c (avr_encode_section_info)
17549 [progmem && !TREE_READONLY]: Error if progmem object needs
17552 2017-07-11 Michael Collison <michael.collison@arm.com>
17554 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
17557 2017-07-11 Carl Love <cel@us.ibm.com>
17559 * config/rs6000/rs6000-c.c: Add support for builtins
17560 vector unsigned int vec_parity_lsbb (vector signed int);
17561 vector unsigned int vec_parity_lsbb (vector unsigned int);
17562 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
17563 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
17564 vector unsigned long long vec_parity_lsbb (vector signed long long);
17565 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
17566 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
17567 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
17568 * doc/extend.texi: Update the built-in documentation file for the
17569 new built-in functions.
17571 2017-07-11 David Malcolm <dmalcolm@redhat.com>
17573 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
17574 (layout::m_primary_loc): New field.
17575 (layout::layout): Initialize new field. Move location filtering
17576 logic from here to...
17577 (layout::maybe_add_location_range): ...this new method. Add
17578 support for filtering to just the lines already specified by other
17580 (layout::will_show_line_p): New method.
17581 (gcc_rich_location::add_location_if_nearby): New method.
17582 (selftest::test_add_location_if_nearby): New test function.
17583 (selftest::diagnostic_show_locus_c_tests): Call it.
17584 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
17587 2017-07-11 Tom de Vries <tom@codesourcery.com>
17589 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
17590 (bb_first_real_insn): New function.
17591 (nvptx_single): Add extra initialization of broadcasted condition
17594 2017-07-11 Nathan Sidwell <nathan@acm.org>
17596 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
17598 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
17600 * doc/extend.texi (AVR Function Attributes): Remove weblink to
17601 Binutils doc as TEXI will mess them up.
17602 * doc/invoke.texi (AVR Options): Same here.
17604 2017-07-11 Daniel Cederman <cederman@gaisler.com>
17606 * config/sparc/sparc.opt (mfix-ut700): New option.
17607 (mfix-gr712rc): Likewise.
17608 (sparc_fix_b2bst): New variable.
17609 * doc/invoke.texi (SPARC options): Document them.
17610 (ARM options): Fix warnings.
17611 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
17612 instructions to prevent sequences that can trigger the store-store
17613 errata for certain LEON3FT processors.
17614 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
17615 (sparc_option_override): Set sparc_fix_b2bst appropriately.
17616 * config/sparc/sparc.md (fix_b2bst): New attribute.
17617 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
17619 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
17622 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
17624 (*rsqrtsf2_sse): Ditto.
17626 (div<mode>3): Macroize insn from divdf3 and divsf3
17627 using MODEF mode iterator.
17629 2017-07-10 Martin Sebor <msebor@redhat.com>
17631 PR tree-optimization/80397
17632 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
17633 instead of testing for equality to INTEGER_TYPE.
17635 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
17637 * config.gcc: Remove uclibc from arc target spec.
17639 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
17641 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
17643 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17646 * lto-wrapper.c (remove_option): New function.
17647 (merge_and_complain): Merge PIC/PIE options more realistically.
17649 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17651 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
17655 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
17656 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
17657 * config.in: Regenerate.
17658 * configure: Regenerate.
17659 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
17660 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
17661 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
17662 (TARGET_GASISR_PROLOGUES): ...target mask.
17663 * common/config/avr/avr-common.c
17664 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
17665 Set -mgas-isr-prologues.
17666 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
17667 INSERT_PASS_BEFORE for it.
17668 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
17669 * config/avr/avr.c (avr_option_override)
17670 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
17671 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
17672 (avr_attribute_table) <no_gccisr>: Add new function attribute.
17673 (avr_set_current_function) <is_no_gccisr>: Init machine field.
17674 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
17676 (make_avr_pass_pre_proep): New function.
17677 (emit_push_sfr) <treg>: Add argument to function and use it
17678 instead of TMP_REG.
17679 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
17680 and set machine->gasisr.yes.
17681 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
17682 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
17683 __gcc_isr.n_pushed to .L__stack_usage.
17684 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
17685 (avr_asm_final_postscan_insn): ...this new static function.
17686 * config/avr/avr.h (machine_function)
17687 <is_no_gccisr, use_L__stack_usage>: New fields.
17688 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
17689 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
17690 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
17691 (gasisr, *gasisr): New expander and insn.
17692 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17693 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
17694 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
17696 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
17698 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
17701 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17703 Move jump-tables out of .text again.
17706 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
17707 (ASM_OUTPUT_ADDR_VEC): New function.
17708 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
17709 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
17710 INSN_ADDRESSes as asm comment.
17711 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
17712 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
17713 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
17714 * config/avr/avr.md (*tablejump): Adjust comment.
17715 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
17716 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
17718 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
17719 (avr_output_addr_vec): New proto.
17720 (avr_log_t) <insn_addresses>: New field.
17722 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
17725 * config/i386/i386.c (ix86_function_arg_advance): Set
17726 outgoing_args_on_stack to true if there are outgoing arguments
17728 (ix86_function_arg): Likewise.
17729 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
17730 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
17731 * config/i386/i386.h (machine_function): Add
17732 outgoing_args_on_stack.
17734 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
17736 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17737 supporting pthreds.
17738 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
17740 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
17742 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
17743 (REAL_H): Remove $(MACHMODE_H).
17744 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
17746 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
17747 $(MACHMODE_H) and double-int.h.
17748 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
17750 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
17753 2017-07-07 Andrew Pinski <apinski@cavium.com>
17755 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
17756 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
17758 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17760 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
17761 Add warning if GCC was not configured to link against a GLIBC that
17762 exports the hardware capability bits.
17763 (make_resolver_func): Make resolver function private and not a
17764 COMDAT function. Create the name with clone_function_name instead
17765 of make_unique_name.
17768 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
17769 correct operand in doing the split.
17771 2017-07-07 Carl Love <cel@us.ibm.com>
17773 * config/rs6000/rs6000-c: Add support for built-in function
17774 vector unsigned short vec_pack_to_short_fp32 (vector float,
17776 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
17777 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
17778 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
17779 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
17780 (convert_4f32_8i16): Add define_expand.
17781 * doc/extend.texi: Update the built-in documentation file for the
17782 new built-in function.
17784 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17786 * config/sparc/m8.md: New file.
17787 * config/sparc/sparc.md: Include m8.md.
17789 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17791 * config/sparc/sparc.opt: New option -mvis4b.
17792 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
17793 (sparc_option_override): Handle VIS4B.
17794 (enum sparc_builtins): Define
17795 SPARC_BUILTIN_DICTUNPACK{8,16,32},
17796 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
17797 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
17798 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
17799 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
17800 (check_constant_argument): New function.
17801 (sparc_vis_init_builtins): Define builtins
17802 __builtin_vis_dictunpack{8,16,32},
17803 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
17804 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
17805 __builtin_vis_fpcmpde{8,16,32}shl and
17806 __builtin_vis_fpcmpur{8,16,32}shl.
17807 (sparc_expand_builtin): Check that the constant operands to
17808 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
17809 constant and in range.
17810 * config/sparc/sparc-c.c (sparc_target_macros): Handle
17812 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
17813 (SPARC_IMM5_P): Likewise.
17814 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
17815 (enabled): Handle vis4b.
17816 (UNSPEC_DICTUNPACK): New unspec.
17817 (UNSPEC_FPCMPSHL): Likewise.
17818 (UNSPEC_FPUCMPSHL): Likewise.
17819 (UNSPEC_FPCMPDESHL): Likewise.
17820 (UNSPEC_FPCMPURSHL): Likewise.
17821 (cpu_feature): New CPU feature `vis4b'.
17822 (dictunpack{8,16,32}): New insns.
17823 (FPCSMODE): New mode iterator.
17824 (fpcscond): New code iterator.
17825 (fpcsucond): Likewise.
17826 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
17827 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
17828 (fpcmpde{8,16,32}{si,di}shl): Likewise.
17829 (fpcmpur{8,16,32}{si,di}shl): Likewise.
17830 * config/sparc/constraints.md: Define constraints `q' for unsigned
17831 2-bit integer constants and `t' for unsigned 5-bit integer
17833 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
17835 (imm5_operand_dictunpack16): Likewise.
17836 (imm5_operand_dictunpack32): Likewise.
17837 (imm2_operand): Likewise.
17838 * doc/invoke.texi (SPARC Options): Document -mvis4b.
17839 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
17840 ditunpack* and fpcmp*shl builtins.
17842 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17844 * config.gcc: Handle m8 in --with-{cpu,tune} options.
17845 * config.in: Add HAVE_AS_SPARC6 define.
17846 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
17848 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
17850 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
17851 (CPP_CPU_SPEC): Handle m8.
17852 (ASM_CPU_SPEC): Likewise.
17853 * config/sparc/sparc-opts.h (enum processor_type): Add
17855 * config/sparc/sparc.c (m8_costs): New struct.
17856 (sparc_option_override): Handle TARGET_CPU_m8.
17857 (sparc32_initialize_trampoline): Likewise.
17858 (sparc64_initialize_trampoline): Likewise.
17859 (sparc_issue_rate): Likewise.
17860 (sparc_register_move_cost): Likewise.
17861 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
17862 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
17863 (ASM_CPU64_DEFAULT_SPEC): Likewise.
17864 (CPP_CPU_SPEC): Handle M8.
17865 (ASM_CPU_SPEC): Likewise.
17866 (AS_M8_FLAG): Define.
17867 * config/sparc/sparc.md: Add m8 to the cpu attribute.
17868 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
17869 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
17871 * configure: Regenerate.
17872 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
17875 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17877 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
17879 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
17880 ("*movdi_insn_sp32"): Do not set v3pipe.
17881 ("*movsi_insn"): Likewise.
17882 ("*movdi_insn_sp64"): Likewise.
17883 ("*movsf_insn"): Likewise.
17884 ("*movdf_insn_sp32"): Likewise.
17885 ("*movdf_insn_sp64"): Likewise.
17886 ("*zero_extendsidi2_insn_sp64"): Likewise.
17887 ("*sign_extendsidi2_insn"): Likewise.
17888 ("*mov<VM32:mode>_insn"): Likewise.
17889 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17890 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17891 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17892 ("<vlop:code><VL:mode>3"): Likewise.
17893 ("*not_<vlop:code><VL:mode>3"): Likewise.
17894 ("*nand<VL:mode>_vis"): Likewise.
17895 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
17896 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
17897 ("one_cmpl<VL:mode>2"): Likewise.
17898 ("faligndata<VM64:mode>_vis"): Likewise.
17899 ("alignaddrsi_vis"): Likewise.
17900 ("alignaddrdi_vis"): Likweise.
17901 ("alignaddrlsi_vis"): Likewise.
17902 ("alignaddrldi_vis"): Likewise.
17903 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17904 ("bmaskdi_vis"): Likewise.
17905 ("bmasksi_vis"): Likewise.
17906 ("bshuffle<VM64:mode>_vis"): Likewise.
17907 ("cmask8<P:mode>_vis"): Likewise.
17908 ("cmask16<P:mode>_vis"): Likewise.
17909 ("cmask32<P:mode>_vis"): Likewise.
17910 ("pdistn<P:mode>_vis"): Likewise.
17911 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17913 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17915 * config/sparc/sparc.md ("subtype"): New insn attribute.
17916 ("*wrgsr_sp64"): Set insn subtype.
17917 ("*rdgsr_sp64"): Likewise.
17918 ("alignaddrsi_vis"): Likewise.
17919 ("alignaddrdi_vis"): Likewise.
17920 ("alignaddrlsi_vis"): Likewise.
17921 ("alignaddrldi_vis"): Likewise.
17922 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17923 ("fexpand_vis"): Likewise.
17924 ("fpmerge_vis"): Likewise.
17925 ("faligndata<VM64:mode>_vis"): Likewise.
17926 ("bshuffle<VM64:mode>_vis"): Likewise.
17927 ("cmask8<P:mode>_vis"): Likewise.
17928 ("cmask16<P:mode>_vis"): Likewise.
17929 ("cmask32<P:mode>_vis"): Likewise.
17930 ("fchksm16_vis"): Likewise.
17931 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
17932 ("fmean16_vis"): Likewise.
17933 ("fp<plusminus_insn>64_vis"): Likewise.
17934 ("<plusminus_insn>v8qi3"): Likewise.
17935 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17936 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
17937 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
17938 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
17939 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
17940 ("*movqi_insn"): Likewise.
17941 ("*movhi_insn"): Likewise.
17942 ("*movsi_insn"): Likewise.
17943 ("movsi_pic_gotdata_op"): Likewise.
17944 ("*movdi_insn_sp32"): Likewise.
17945 ("*movdi_insn_sp64"): Likewise.
17946 ("movdi_pic_gotdata_op"): Likewise.
17947 ("*movsf_insn"): Likewise.
17948 ("*movdf_insn_sp32"): Likewise.
17949 ("*movdf_insn_sp64"): Likewise.
17950 ("*zero_extendhisi2_insn"): Likewise.
17951 ("*zero_extendqihi2_insn"): Likewise.
17952 ("*zero_extendqisi2_insn"): Likewise.
17953 ("*zero_extendqidi2_insn"): Likewise.
17954 ("*zero_extendhidi2_insn"): Likewise.
17955 ("*zero_extendsidi2_insn_sp64"): Likewise.
17956 ("ldfsr"): Likewise.
17957 ("prefetch_64"): Likewise.
17958 ("prefetch_32"): Likewise.
17959 ("tie_ld32"): Likewise.
17960 ("tie_ld64"): Likewise.
17961 ("*tldo_ldub_sp32"): Likewise.
17962 ("*tldo_ldub1_sp32"): Likewise.
17963 ("*tldo_ldub2_sp32"): Likewise.
17964 ("*tldo_ldub_sp64"): Likewise.
17965 ("*tldo_ldub1_sp64"): Likewise.
17966 ("*tldo_ldub2_sp64"): Likewise.
17967 ("*tldo_ldub3_sp64"): Likewise.
17968 ("*tldo_lduh_sp32"): Likewise.
17969 ("*tldo_lduh1_sp32"): Likewise.
17970 ("*tldo_lduh_sp64"): Likewise.
17971 ("*tldo_lduh1_sp64"): Likewise.
17972 ("*tldo_lduh2_sp64"): Likewise.
17973 ("*tldo_lduw_sp32"): Likewise.
17974 ("*tldo_lduw_sp64"): Likewise.
17975 ("*tldo_lduw1_sp64"): Likewise.
17976 ("*tldo_ldx_sp64"): Likewise.
17977 ("*mov<VM32:mode>_insn"): Likewise.
17978 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17979 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17981 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17983 * config/sparc/sparc.md ("type"): New insn type viscmp.
17984 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
17986 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
17987 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
17988 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17989 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
17991 ("n7_vis_logical_11cycle"): Likewise.
17992 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
17993 * config/sparc/niagara2.md ("niag3_vis": Likewise.
17994 * config/sparc/niagara.md ("niag_vis"): Likewise.
17995 * config/sparc/ultra3.md ("us3_fga"): Likewise.
17996 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
17998 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18000 * config/sparc/sparc.md: New instruction type `bmask'.
18001 (bmaskdi_vis): Use the `bmask' type.
18002 (bmasksi_vis): Likewise.
18003 * config/sparc/ultra3.md (us3_array): Likewise.
18004 * config/sparc/niagara7.md (n7_array): Likewise.
18005 * config/sparc/niagara4.md (n4_array): Likewise.
18006 * config/sparc/niagara2.md (niag2_vis): Likewise.
18007 (niag3_vis): Likewise.
18008 * config/sparc/niagara.md (niag_vis): Likewise.
18010 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18012 * ipa-comdats.c: Remove optimize check from gate.
18013 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
18014 for functions not optimized.
18015 (ipa_fn_summary_read): Skip optimize check.
18016 (ipa_fn_summary_write): Likewise.
18017 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
18019 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
18021 (can_inline_edge_p): Check flag_pcc_struct_return for match.
18022 (check_callers): Give up on caller which is not optimized.
18023 (inline_small_functions): Likewise.
18024 (ipa_inline): Do not give up when not optimizing.
18025 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
18026 away unoptimizes cdtors.
18027 (whole_program_function_and_variable_visibility): Do
18028 ipa_discover_readonly_nonaddressable_vars in LTO mode.
18029 * ipa.c (process_references): Do not check optimize.
18030 (symbol_table::remove_unreachable_nodes): Update optimize check.
18031 (set_writeonly_bit): Update optimize check.
18032 (pass_ipa_cdtor_merge::gate): Do not check optimize.
18033 (pass_ipa_single_use::gate): Remove.
18035 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18037 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
18038 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
18039 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
18040 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
18041 permute_load, permute_store, adjust_extract, adjust_splat,
18042 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
18043 replace_swap_with_copy, dump_swap_insn_table,
18044 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
18045 recombine_lvx_pattern, recombine_stvx_pattern,
18046 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
18047 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
18048 to file rs6000-p8swap.c.
18049 * config/rs6000/rs6000-p8swap.c: New file.
18050 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
18051 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
18052 and rs6000*-*-* targets.
18054 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18056 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
18058 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18060 * lto-wrapper.c (merge_and_complain): Do not merge
18061 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
18062 fsigned_zeros, ftrapping_math, fwrapv.
18063 (append_compiler_options): Do not track these options.
18064 (append_linker_options): Likewie
18066 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18068 * cgraphunit.c (cgraph_node::finalize_function): When
18069 !flag_toplevel_reorde set no_reorder flag.
18070 (varpool_node::finalize_decl): Likewise.
18071 (symbol_table::compile): Drop no toplevel reorder path.
18073 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18075 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
18076 edges; zero probability is not better than uninitialized.
18078 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18080 * asan.h (asan_sanitize_allocas_p): Declare.
18081 * asan.c (asan_sanitize_allocas_p): New function.
18082 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
18083 (handle_builtin_alloca): Likewise.
18084 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
18085 if !asan_sanitize_allocas_p.
18086 * params.def (asan-instrument-allocas): Add new option.
18087 * params.h (ASAN_PROTECT_ALLOCAS): Define.
18088 * opts.c (common_handle_option): Disable allocas sanitization for
18091 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18093 * asan.c: Include gimple-fold.h.
18094 (get_last_alloca_addr): New function.
18095 (handle_builtin_stackrestore): Likewise.
18096 (handle_builtin_alloca): Likewise.
18097 (asan_emit_allocas_unpoison): Likewise.
18098 (get_mem_refs_of_builtin_call): Add new parameter, remove const
18099 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
18100 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
18101 (instrument_builtin_call): Pass gimple iterator to
18102 get_mem_refs_of_builtin_call.
18103 (last_alloca_addr): New global.
18104 * asan.h (asan_emit_allocas_unpoison): Declare.
18105 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
18106 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
18107 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
18108 if function calls alloca.
18109 * gimple-fold.c (replace_call_with_value): Remove static keyword.
18110 * gimple-fold.h (replace_call_with_value): Declare.
18111 * internal-fn.c: Include asan.h.
18112 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
18113 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
18115 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18117 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
18118 (C_SELFTEST_FLAGS): New.
18119 (CPP_SELFTEST_FLAGS): New.
18120 (SELFTEST_DEPS): New, from deps of s-selftest.
18121 (C_SELFTEST_DEPS): New, from deps of s-selftest.
18122 (CPP_SELFTEST_DEPS): New.
18123 (selftest): Add dependency on s-selftest-c++.
18124 (s-selftest): Rename to...
18125 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
18126 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
18127 than SELFTEST_FLAGS.
18128 (selftest-gdb): Rename to...
18129 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
18131 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
18132 (selftest-valgrind): Rename to...
18133 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
18135 (selftest-valgrind): Reintroduce as an alias for
18136 selftest-c-valgrind.
18137 (s-selftest-c++): New.
18138 (selftest-c++-gdb): New.
18139 (selftest-c++-valgrind): New.
18141 2017-07-06 Olivier Hainque <hainque@adacore.com>
18143 * gcc.c (process_command): When deciding if undefined variables
18144 should be ignored when processing specs, accept "gcc -v" as well.
18146 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18148 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
18149 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
18151 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18153 * config/arm/arm-cpus.in (armv8-r): Add new entry.
18154 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
18155 * config/arm/arm-tables.opt: Regenerate.
18156 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
18158 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
18160 2017-07-06 Carl Love <cel@us.ibm.com>
18162 * ChangeLog: Clean up from mid air collision
18164 2017-07-06 Carl Love <cel@us.ibm.com>
18166 * config/rs6000/rs6000-c.c: Add support for built-in functions
18167 vector signed int vec_subc (vector signed int, vector signed int);
18168 vector signed __int128 vec_subc (vector signed __int128,
18169 vector signed __int128);
18170 vector unsigned __int128 vec_subc (vector unsigned __int128,
18171 vector unsigned __int128);
18172 vector signed int vec_sube (vector signed int, vector signed int,
18173 vector signed int);
18174 vector unsigned int vec_sube (vector unsigned int,
18175 vector unsigned int,
18176 vector unsigned int);
18177 vector signed __int128 vec_sube (vector signed __int128,
18178 vector signed __int128,
18179 vector signed__int128);
18180 vector unsigned __int128 vec_sube (vector unsigned __int128,
18181 vector unsigned __int128,
18182 vector unsigned __int128);
18183 vector signed int vec_subec (vector signed int, vector signed int,
18184 vector signed int);
18185 vector unsigned int vec_subec (vector unsigned int,
18186 vector unsigned int,
18187 vector unsigned int);
18188 vector signed __int128 vec_subec (vector signed __int128,
18189 vector signed __int128,
18190 vector signed__int128);
18191 vector unsigned __int128 vec_subec (vector unsigned __int128,
18192 vector unsigned __int128,
18193 vector unsigned __int128);
18194 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18195 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18196 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18197 BU_ALTIVEC_OVERLOAD_X definitions.
18198 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18199 * doc/extend.texi: Update the built-in documentation file for the new
18200 built-in functions.
18202 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18205 * diagnostic-show-locus.c (layout::layout): Use start and finish
18206 spelling location for the start and finish of each range.
18207 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18208 Add unused aspect param.
18209 * input.c (expand_location_1): Add "aspect" param, and use it
18210 to access the correct part of the location.
18211 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18213 (expand_location_to_spelling_point): Likewise.
18214 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18215 param, and pass it to expand_location_1.
18217 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18219 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18220 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18221 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18222 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18223 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18224 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18225 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18226 _mm_maskz_getmant_ss): New intrinsics.
18227 (__builtin_ia32_getexpss128_mask): Changed to ...
18228 __builtin_ia32_getexpss128_round ... this.
18229 (__builtin_ia32_getexpsd128_mask): Changed to ...
18230 __builtin_ia32_getexpsd128_round ... this.
18231 * config/i386/i386-builtin-types.def
18232 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18233 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18234 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18235 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18236 __builtin_ia32_getmantss_mask_round): New builtins.
18237 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18238 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18239 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18240 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18241 * config/i386/sse.md
18242 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18243 avx512f_sgetexp<mode><mask_scalar_name>
18244 <round_saeonly_scalar_name> ... this.
18245 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18246 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18247 vgetexp<ssescalarmodesuffix>
18248 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18249 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18250 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18251 avx512f_vgetmant<mode><mask_scalar_name>
18252 <round_saeonly_scalar_name> ... this.
18253 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18254 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18255 vgetmant<ssescalarmodesuffix>
18256 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18257 %0<mask_scalar_operand4>, %1, %2
18258 <round_saeonly_scalar_mask_op4>, %3} ... this.
18259 * config/i386/subst.md (mask_scalar_operand4,
18260 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18261 round_saeonly_scalar_nimm_predicate): New subst attributes.
18263 2017-07-06 Julia Koval <julia.koval@intel.com>
18265 * config/i386/i386.c (ix86_erase_embedded_rounding):
18266 Remove code for old rounding pattern.
18268 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18270 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18272 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18274 * doc/sourcebuild.texi (Test Directives, Variants of
18275 dg-require-support): Add documentation for dg-require-stack-check.
18277 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18279 * config/i386/subst.md (mask_scalar, round_scalar,
18280 round_saeonly_scalar): New meta-templates.
18281 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18282 round_scalar_mask_operand3, round_scalar_mask_op3,
18283 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18284 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18285 round_saeonly_scalar_constraint,
18286 round_saeonly_scalar_prefix): New subst attribute.
18287 * config/i386/sse.md
18288 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18289 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18290 <round_scalar_name> ... this.
18291 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18292 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18293 <round_scalar_name> ... this.
18294 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18295 <sse>_vm<code><mode>3<mask_scalar_name>
18296 <round_saeonly_scalar_name> ... this.
18297 (v<plusminus_mnemonic><ssescalarmodesuffix>
18298 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18299 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18300 v<plusminus_mnemonic><ssescalarmodesuffix>
18301 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18302 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18303 (v<multdiv_mnemonic><ssescalarmodesuffix>
18304 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18305 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18306 v<multdiv_mnemonic><ssescalarmodesuffix>
18307 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18308 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18309 (v<maxmin_float><ssescalarmodesuffix>
18310 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18311 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18312 v<maxmin_float><ssescalarmodesuffix>
18313 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18314 %0<mask_scalar_operand3>, %1, %<iptr>2
18315 <round_saeonly_scalar_mask_op3>} ... this.
18317 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18319 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18320 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18322 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18323 Alan Hayward <alan.hayward@arm.com>
18324 David Sherwood <david.sherwood@arm.com>
18326 * combine.c (simplify_if_then_else): Remove "enum" before
18328 * compare-elim.c (can_eliminate_compare): Likewise.
18329 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18331 (aarch64_lookup_simd_builtin_type): Likewise.
18332 (aarch64_simd_builtin_type): Likewise.
18333 (aarch64_init_simd_builtin_types): Likewise.
18334 (aarch64_simd_expand_args): Likewise.
18335 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18337 (aarch64_reverse_mask): Likewise.
18338 (aarch64_simd_emit_reg_reg_move): Likewise.
18339 (aarch64_gen_adjusted_ldpstp): Likewise.
18340 (aarch64_ccmp_mode_to_code): Likewise.
18341 (aarch64_operands_ok_for_ldpstp): Likewise.
18342 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18343 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18345 (aarch64_min_divisions_for_recip_mul): Likewise.
18346 (aarch64_reassociation_width): Likewise.
18347 (aarch64_get_condition_code_1): Likewise.
18348 (aarch64_simd_emit_reg_reg_move): Likewise.
18349 (aarch64_simd_attr_length_rglist): Likewise.
18350 (aarch64_reverse_mask): Likewise.
18351 (aarch64_operands_ok_for_ldpstp): Likewise.
18352 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18353 (aarch64_gen_adjusted_ldpstp): Likewise.
18354 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18356 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18357 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18358 (arm_lookup_simd_builtin_type): Likewise.
18359 (arm_simd_builtin_type): Likewise.
18360 (arm_init_simd_builtin_types): Likewise.
18361 (arm_expand_builtin_args): Likewise.
18362 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18363 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18364 (ft32_setup_incoming_varargs): Likewise.
18365 (ft32_function_arg): Likewise.
18366 (ft32_function_arg_advance): Likewise.
18367 (ft32_pass_by_reference): Likewise.
18368 (ft32_arg_partial_bytes): Likewise.
18369 (ft32_valid_pointer_mode): Likewise.
18370 (ft32_addr_space_pointer_mode): Likewise.
18371 (ft32_addr_space_legitimate_address_p): Likewise.
18372 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18374 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18375 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18376 (iamcu_alignment): Likewise.
18377 (canonicalize_vector_int_perm): Likewise.
18378 (ix86_noce_conversion_profitable_p): Likewise.
18379 (ix86_mpx_bound_mode): Likewise.
18380 (ix86_operands_ok_for_move_multiple): Likewise.
18381 * config/microblaze/microblaze-protos.h
18382 (microblaze_expand_conditional_branch_reg): Likewise.
18383 * config/microblaze/microblaze.c
18384 (microblaze_expand_conditional_branch_reg): Likewise.
18385 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18387 (rs6000_reassociation_width): Likewise.
18388 (rs6000_invalid_binary_op): Likewise.
18389 (fusion_p9_p): Likewise.
18390 (emit_fusion_p9_load): Likewise.
18391 (emit_fusion_p9_store): Likewise.
18392 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18394 (riscv_hard_regno_mode_ok_p): Likewise.
18395 (riscv_address_insns): Likewise.
18396 (riscv_split_symbol): Likewise.
18397 (riscv_legitimize_move): Likewise.
18398 (riscv_function_value): Likewise.
18399 (riscv_hard_regno_nregs): Likewise.
18400 (riscv_expand_builtin): Likewise.
18401 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18402 (riscv_build_integer): Likewise.
18403 (riscv_split_integer): Likewise.
18404 (riscv_legitimate_constant_p): Likewise.
18405 (riscv_cannot_force_const_mem): Likewise.
18406 (riscv_regno_mode_ok_for_base_p): Likewise.
18407 (riscv_valid_base_register_p): Likewise.
18408 (riscv_valid_offset_p): Likewise.
18409 (riscv_valid_lo_sum_p): Likewise.
18410 (riscv_classify_address): Likewise.
18411 (riscv_legitimate_address_p): Likewise.
18412 (riscv_address_insns): Likewise.
18413 (riscv_load_store_insns): Likewise.
18414 (riscv_force_binary): Likewise.
18415 (riscv_split_symbol): Likewise.
18416 (riscv_force_address): Likewise.
18417 (riscv_legitimize_address): Likewise.
18418 (riscv_move_integer): Likewise.
18419 (riscv_legitimize_const_move): Likewise.
18420 (riscv_legitimize_move): Likewise.
18421 (riscv_address_cost): Likewise.
18422 (riscv_subword): Likewise.
18423 (riscv_output_move): Likewise.
18424 (riscv_canonicalize_int_order_test): Likewise.
18425 (riscv_emit_int_order_test): Likewise.
18426 (riscv_function_arg_boundary): Likewise.
18427 (riscv_pass_mode_in_fpr_p): Likewise.
18428 (riscv_pass_fpr_single): Likewise.
18429 (riscv_pass_fpr_pair): Likewise.
18430 (riscv_get_arg_info): Likewise.
18431 (riscv_function_arg): Likewise.
18432 (riscv_function_arg_advance): Likewise.
18433 (riscv_arg_partial_bytes): Likewise.
18434 (riscv_function_value): Likewise.
18435 (riscv_pass_by_reference): Likewise.
18436 (riscv_setup_incoming_varargs): Likewise.
18437 (riscv_print_operand): Likewise.
18438 (riscv_elf_select_rtx_section): Likewise.
18439 (riscv_save_restore_reg): Likewise.
18440 (riscv_for_each_saved_reg): Likewise.
18441 (riscv_register_move_cost): Likewise.
18442 (riscv_hard_regno_mode_ok_p): Likewise.
18443 (riscv_hard_regno_nregs): Likewise.
18444 (riscv_class_max_nregs): Likewise.
18445 (riscv_memory_move_cost): Likewise.
18446 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18447 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18448 (rl78_addr_space_address_mode): Likewise.
18449 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18451 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18452 (rs6000_reassociation_width): Likewise.
18453 (rs6000_invalid_binary_op): Likewise.
18454 (fusion_p9_p): Likewise.
18455 (emit_fusion_p9_load): Likewise.
18456 (emit_fusion_p9_store): Likewise.
18457 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18458 (ok_for_simple_move_operands): Likewise.
18459 (ok_for_simple_move_strict_operands): Likewise.
18460 (ok_for_simple_arith_logic_operands): Likewise.
18461 (visium_legitimize_reload_address): Likewise.
18462 (visium_select_cc_mode): Likewise.
18463 (output_cbranch): Likewise.
18464 (visium_split_double_move): Likewise.
18465 (visium_expand_copysign): Likewise.
18466 (visium_expand_int_cstore): Likewise.
18467 (visium_expand_fp_cstore): Likewise.
18468 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18469 (visium_function_arg): Likewise.
18470 (visium_function_arg_advance): Likewise.
18471 (visium_libcall_value): Likewise.
18472 (visium_setup_incoming_varargs): Likewise.
18473 (visium_legitimate_constant_p): Likewise.
18474 (visium_legitimate_address_p): Likewise.
18475 (visium_legitimize_address): Likewise.
18476 (visium_secondary_reload): Likewise.
18477 (visium_register_move_cost): Likewise.
18478 (visium_memory_move_cost): Likewise.
18479 (prepare_move_operands): Likewise.
18480 (ok_for_simple_move_operands): Likewise.
18481 (ok_for_simple_move_strict_operands): Likewise.
18482 (ok_for_simple_arith_logic_operands): Likewise.
18483 (visium_function_value_1): Likewise.
18484 (rtx_ok_for_offset_p): Likewise.
18485 (visium_legitimize_reload_address): Likewise.
18486 (visium_split_double_move): Likewise.
18487 (visium_expand_copysign): Likewise.
18488 (visium_expand_int_cstore): Likewise.
18489 (visium_expand_fp_cstore): Likewise.
18490 (visium_split_cstore): Likewise.
18491 (visium_select_cc_mode): Likewise.
18492 (visium_split_cbranch): Likewise.
18493 (output_cbranch): Likewise.
18494 (visium_print_operand_address): Likewise.
18495 * expmed.c (flip_storage_order): Likewise.
18496 * expmed.h (emit_cstore): Likewise.
18497 (flip_storage_order): Likewise.
18498 * genrecog.c (validate_pattern): Likewise.
18499 * hsa-gen.c (gen_hsa_addr): Likewise.
18500 * internal-fn.c (expand_arith_overflow): Likewise.
18501 * ira-color.c (allocno_copy_cost_saving): Likewise.
18502 * lra-assigns.c (find_hard_regno_for_1): Likewise.
18503 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
18504 (process_invariant_for_inheritance): Likewise.
18505 * lra-eliminations.c (move_plus_up): Likewise.
18506 * omp-low.c (lower_oacc_reductions): Likewise.
18507 * simplify-rtx.c (simplify_subreg): Likewise.
18508 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
18509 (TARGET_CHKP_BOUND_MODE): Likewise..
18510 * targhooks.c (default_chkp_bound_mode): Likewise.
18511 (default_setup_incoming_vararg_bounds): Likewise.
18512 * targhooks.h (default_chkp_bound_mode): Likewise.
18513 (default_setup_incoming_vararg_bounds): Likewise.
18514 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
18515 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
18516 (have_whole_vector_shift): Likewise.
18517 * tree-vect-stmts.c (vectorizable_load): Likewise.
18518 * doc/tm.texi: Regenerate.
18520 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18522 Graceful degrade if Binutils PR21472 is not available.
18525 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
18526 .rodata in flash test fails.
18527 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
18528 * confgure: Regenerate.
18529 * config.in: Regenerate.
18530 * config/avr/avr.c (avr_asm_named_section)
18531 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
18532 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
18533 (avr_asm_init_sections): Same.
18535 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18537 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
18538 (fma<VH:mode>4_intrinsic): Likewise.
18539 (*fmsub<VCVTF:mode>4): Likewise.
18540 (*fmsub<VH:mode>4_intrinsic): Likewise.
18542 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18545 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
18546 Don't depend on "optimize > 0".
18547 (out_movhi_r_mr, out_movqi_mr_r): Same.
18548 (out_movhi_mr_r, out_movqi_r_mr): Same.
18549 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
18550 io_address_operand on "optimize > 0".
18552 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18554 * tree-loop-distribution.c: Add general explanantion on the pass.
18555 (generate_loops_for_partition): Mark distributed loop.
18556 (pg_add_dependence_edges): New parameter. Handle alias data
18557 dependence specially and record it in the parameter if asked.
18558 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
18559 (init_partition_graph_vertices, add_partition_graph_edge): New.
18560 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
18561 (free_partition_graph_vdata, build_partition_graph): New.
18562 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
18563 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
18564 (data_ref_segment_size, latch_dominated_by_data_ref): New.
18565 (compute_alias_check_pairs, version_loop_by_alias_check): New.
18566 (version_for_distribution_p, finalize_partitions): New.
18567 (distribute_loop): Handle alias data dependence specially. Factor
18568 out loop fusion code as functions and call these functions.
18570 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18572 * tree-loop-distribution.c (classify_partition): New parameter and
18573 better handle reduction statement.
18574 (rdg_build_partitions): Revise comment.
18575 (distribute_loop): Compute statements in all partitions and pass it
18576 to classify_partition.
18578 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18580 * tree-loop-distribution.c (enum partition_type): New.
18581 (struct partition): New field type.
18582 (partition_merge_into): Add parameter. Update partition type.
18583 (data_dep_in_cycle_p, update_type_for_merge): New functions.
18584 (build_rdg_partition_for_vertex): Compute partition type.
18585 (rdg_build_partitions): Dump partition type.
18586 (distribute_loop): Update calls to partition_merge_into.
18588 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18590 * tree-loop-distribution.c (struct ddr_hasher): New.
18591 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
18593 (classify_partition): Call get_data_dependence.
18594 (pg_add_dependence_edges): Ditto.
18595 (distribute_loop): Release data dependence hash table.
18597 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18599 * tree-loop-distribution.c (ref_base_address): Delete.
18600 (similar_memory_accesses): Rename ...
18601 (share_memory_accesses): ... to this. Check if partitions access
18602 the same memory reference.
18603 (distribute_loop): Call share_memory_accesses.
18605 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18607 * tree-loop-distribution.c (struct partition): New field recording
18608 its data reference.
18609 (partition_alloc, partition_free): Init and release data refs.
18610 (partition_merge_into): Merge data refs.
18611 (build_rdg_partition_for_vertex): Collect data refs for partition.
18612 (pg_add_dependence_edges): Change parameters from vector to bitmap.
18614 (distribute_loop): Remve data refs from vertice data of partition
18617 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18619 * tree-loop-distribution.c (params.h): Include header file.
18620 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
18621 (datarefs_vec): New global var.
18622 (create_rdg_vertices): Use datarefs_vec directly.
18623 (free_rdg): Don't free data references.
18624 (build_rdg): Update use. Don't free data references.
18625 (distribute_loop): Compute global variable for data references.
18626 Bail out if there are too many data references.
18628 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18630 * tree-loop-distribution.c (loop_nest): New global var.
18631 (build_rdg): Use loop directly, rather than loop nest.
18632 (pg_add_dependence_edges): Remove loop nest parameter. Use global
18634 (distribute_loop): Compute global variable loop nest. Update use.
18636 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18638 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
18639 (partition_merge_into): New parameter. Dump reason for fusion.
18640 (distribute_loop): Update use of partition_merge_into.
18642 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18644 * tree-loop-distribution.c (bb_top_order_index): New.
18645 (bb_top_order_index_size, bb_top_order_cmp): New.
18646 (stmts_from_loop): Use topological order.
18647 (pass_loop_distribution::execute): Compute and release topological
18648 order for basic blocks.
18650 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18652 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
18655 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18657 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
18658 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
18659 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
18660 * internal-fn.def (LOOP_DIST_ALIAS): New.
18661 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
18662 (fold_loop_internal_call): ... this.
18663 (vect_loop_dist_alias_call): New function.
18664 (set_uid_loop_bbs): Call fold_loop_internal_call.
18665 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
18668 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18671 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
18672 Require dead FLAGS_REG at the beginning of a peephole.
18674 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18677 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
18678 arguments in the call to __builtin_ia32_sbb_u32.
18679 (_subborrow_u64): Swap _X and _Y arguments in the call to
18680 __builtin_ia32_sbb_u64.
18682 2017-07-04 Jakub Jelinek <jakub@redhat.com>
18685 * tree-vrp.c (compare_assert_loc): Turn into a function template
18686 with stable template parameter. Only test if a->e is NULL,
18687 !a->e == !b->e has been verified already. Use e == NULL or
18688 e != NULL instead of e or ! e tests. If stable is true, don't use
18689 iterative_hash_expr, on the other side allow a or b or both NULL
18690 and sort the NULLs last.
18691 (process_assert_insertions): Sort using compare_assert_loc<false>
18692 instead of compare_assert_loc, later sort using
18693 compare_assert_loc<true> before calling process_assert_insertions_for
18694 in a loop. Use break instead of continue once seen NULL pointer.
18696 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18698 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18699 Cortex-R7 and Cortex-R8 processors.
18701 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18703 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
18704 uninitialized while src is not.
18706 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
18708 * common/config/arm/arm-common.c: Adjust include path for
18710 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
18711 (arm-cpu.h): Create in build directory. Adjust dependency rules.
18712 (arm-cpu-data.h): Likewise.
18713 (arm-cpu-cdata.h): Likewise.
18714 * config/arm/arm-cpu.h: Delete.
18715 * config/arm/arm-cpu-cdata.h: Delete.
18716 * config/arm/arm-cpu-data.h: Delete.
18718 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
18720 * config/arm/arm-cpus.in (cortex-a55): New.
18721 (cortex-a75): Likewise.
18722 (cortex-a75.cortex-a55): Likewise.
18723 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
18725 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
18726 * config/arm/arm-cpu-cdata.h: Regenerate.
18727 * config/arm/arm-cpu-data.h: Regenerate.
18728 * config/arm/arm-cpu.h: Regenerate.
18729 * config/arm/arm-tables.opt: Regenerate.
18730 * config/arm/arm-tune.md: Regenerate.
18732 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18734 * haifa-sched.c (sched_create_recovery_edges): Update profile.
18736 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18738 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
18741 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
18743 PR tree-optimization/81292
18744 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
18745 full_string_p, also call adjust_related_strinfos if the adjustment
18746 is simple, otherwise invalidate related strinfos.
18748 2017-07-04 Martin Liska <mliska@suse.cz>
18751 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
18752 newly created variable as DECL_IGNORED_P.
18754 2017-07-04 Martin Liska <mliska@suse.cz>
18757 * ipa-inline.c (inline_small_functions):
18758 Use xstrdup_for_dump.
18760 2017-07-04 Tom de Vries <tom@codesourcery.com>
18762 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
18764 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18767 * config/darwin.c (darwin_function_switched_text_sections):
18770 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
18772 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
18774 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18776 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
18778 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18780 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
18781 min_profitable_iters, and th as inclusive lower bounds.
18782 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
18783 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
18784 for min_profitable_iters and min_profitable_estimate.
18785 (vect_transform_loop): Treat th as an inclusive lower bound.
18786 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
18788 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18791 * config/darwin.c (darwin_function_switched_text_sections):
18792 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
18793 in two pieces, and suppress the use of buf.
18795 2017-07-03 Nathan Sidwell <nathan@acm.org>
18797 * hash-table.h (hash_table_mod1): Fix indentation.
18799 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18801 PR middle-end/81290
18802 * predict.c (force_edge_cold): Be more careful about propagation
18804 * profile-count.h (profile_probability::guessed,
18805 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
18807 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
18809 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
18811 * doc/invoke.texi (rcpc architecture extension): Document it.
18813 2017-07-03 Richard Biener <rguenther@suse.de>
18815 PR tree-optimization/60510
18816 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
18817 the scalar reduction PHI and use it.
18818 (vectorizable_reduction): Properly guard the single_defuse_cycle
18819 path for non-SLP reduction chains where we cannot use it.
18820 Rework reduc_def/index and vector type deduction. Rework
18821 vector operand gathering during reduction op code-gen.
18822 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
18823 chains dissolve the chain and leave it to non-SLP reduction
18826 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18828 * tree-data-ref.h (dr_alignment): Declare.
18829 * tree-data-ref.c (dr_alignment): New function.
18830 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
18831 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
18833 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
18835 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18837 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
18838 and base_misalignment fields.
18839 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
18840 * tree-data-ref.c: Include builtins.h.
18841 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
18842 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
18843 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
18844 * tree-vect-data-refs.c: Include tree-cfg.h.
18845 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
18846 fields instead of calculating an alignment here.
18847 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
18848 innermost_loop_behavior fields.
18850 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18852 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
18854 (DR_STEP_ALIGNMENT): New macro.
18855 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
18856 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
18857 (create_data_ref): Print it.
18858 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
18859 to tell whether the step preserves vector (mis)alignment.
18860 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18861 Move the check for an integer step and generalise to all INTEGER_CST.
18862 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
18863 Print the outer step alignment.
18865 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18867 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
18868 with offset_alignment.
18869 (DR_ALIGNED_TO): Delete.
18870 (DR_OFFSET_ALIGNMENT): New macro.
18871 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
18872 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
18873 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
18874 (create_data_ref): Likewise.
18875 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18876 (vect_analyze_data_refs): Likewise.
18877 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
18878 creating dummy innermost behavior.
18880 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18882 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
18883 with a "innermost_loop_behavior *" and refeence tree.
18884 * tree-data-ref.c (dr_analyze_innermost): Likewise.
18885 (create_data_ref): Update call accordingly.
18886 * tree-predcom.c (find_looparound_phi): Likewise.
18888 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18890 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
18891 fields with dr_wrt_vec_loop.
18892 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
18893 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
18894 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
18895 (vect_dr_behavior): New function.
18896 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18897 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
18898 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
18899 track whether the step preserves the misalignment.
18900 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18901 Use vect_dr_behavior.
18902 (vect_setup_realignment): Update call accordingly.
18903 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
18904 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
18905 call to vect_create_addr_base_for_vector_ref.
18906 (vect_create_cond_for_align_checks): Likewise.
18907 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
18908 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
18909 (vect_recog_mask_conversion_pattern): Likewise.
18910 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
18911 (new_stmt_vec_info): Remove redundant zeroing.
18913 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18915 * common/config/arm/arm-common.c (arm_be8_option): New function.
18916 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
18917 (ISA_ARMv6): Add isa_bit_be8.
18918 * config/arm/arm.h (arm_be8_option): Add prototype.
18919 (BE8_SPEC_FUNCTION): New define.
18920 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
18921 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
18922 (mlittle-endian): Similarly.
18923 (mbe8, mbe32): New options.
18924 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
18925 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
18927 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18929 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
18931 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18933 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
18934 (cleanup_tree_cfg_bb): Use it.
18935 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
18937 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
18939 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18942 * loop-doloop.c (add_test): Update profile.
18944 2017-07-03 Martin Liska <mliska@suse.cz>
18947 * sanopt.c (rewrite_usage_of_param): New function.
18948 (sanitize_rewrite_addressable_params): Likewise.
18949 (pass_sanopt::execute): Call rewrite_usage_of_param.
18951 2017-07-03 Richard Biener <rguenther@suse.de>
18953 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
18954 back to using VIEW_CONVERT_EXPR.
18956 2017-07-03 Martin Liska <mliska@suse.cz>
18959 * doc/extend.texi: Document when a resolver function is
18960 generated for target_clones.
18962 2017-07-03 Martin Liska <mliska@suse.cz>
18964 * asan.c (asan_emit_stack_protection): Unpoison just red zones
18965 and shadow memory of auto variables which are subject of
18966 use-after-scope sanitization.
18967 (asan_expand_mark_ifn): Add do set only when is_poison.
18969 2016-07-03 Richard Biener <rguenther@suse.de>
18971 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
18973 (vect_force_simple_reduction): Record reduction def -> phi mapping.
18974 (vectorizable_reduction): Perform reduction PHI creation when
18975 visiting a reduction PHI and adjust and simplify code generation
18976 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
18977 (vect_transform_loop): Visit reduction PHIs.
18978 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
18979 defs into the SLP tree.
18980 (vect_build_slp_tree): Reduction defs terminate the recursion.
18981 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
18983 (vect_get_vec_defs_for_stmt_copy): Export.
18984 (vect_get_vec_defs): Likewise.
18985 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
18987 (vect_get_vec_defs_for_stmt_copy): Declare.
18988 (vect_get_vec_defs): Likewise.
18990 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18992 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
18993 parameter with a "loop" parameter and use it instead of the
18994 loop containing DR_STMT. Don't check simple_iv when doing
18995 BB analysis. Describe the two analysis modes in the comment.
18997 2017-07-03 Tom de Vries <tom@codesourcery.com>
18999 PR tree-optimization/69468
19000 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
19001 (find_same_succ_bb): Handle ignore_edge_flags.
19003 2017-07-03 Tom de Vries <tom@codesourcery.com>
19005 PR tree-optimization/81192
19006 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
19008 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
19010 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
19012 2017-07-03 Tom de Vries <tom@codesourcery.com>
19014 PR tree-optimization/81192
19015 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
19016 BB_SAME_SUCC (bb) == NULL.
19018 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19020 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
19023 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19025 * dumpfile.c: Include profile-count.h
19026 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
19028 (insert_cond_bb): Update profile.
19029 * tree-cfg.h (insert_cond_bb): Update prototype.
19030 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
19031 * tree-dump.c: Do not include tree-cfg.
19033 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19035 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
19037 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19039 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
19042 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19044 * tree-complex.c (expand_complex_div_wide): update profile.
19046 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19047 Alan Hayward <alan.hayward@arm.com>
19048 David Sherwood <david.sherwood@arm.com>
19050 * Makefile.in (MACHMODE_H): Remove insn-modes.h
19051 (CORETYPES_H): New define.
19052 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
19053 (insn-modes-inline.h, s-modes-inline-h): New rules.
19054 (generated_files): Add insn-modes-inline.h.
19055 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
19056 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
19057 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
19058 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
19059 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
19060 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
19061 (build/gencodes.o, build/genconditions.o): Likewise.
19062 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
19063 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
19064 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
19065 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
19066 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
19067 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
19068 * coretypes.h: Include everything up to real.h for generators.
19069 Include insn-modes.h first. Include wide-int-print.h after
19070 wide-int.h. Include insn-modes-inline.h and then machmode.h.
19071 * machmode.h: Don't include insn-modes.h here.
19072 * function-tests.c: Remove includes of signop.h, machmode.h,
19073 double-int.h and wide-int.h.
19075 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
19077 * optc-save-gen.awk: Likewise.
19078 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
19079 * godump.c: Remove include of wide-int-print.h.
19080 * pretty-print.h: Likewise.
19081 * wide-int-print.cc: Likewise.
19082 * wide-int.cc: Likewise.
19083 * hash-map-tests.c: Remove include of signop.h.
19084 * hash-set-tests.c: Likewise.
19085 * rtl-tests.c: Likewise.
19086 * mkconfig.sh: Remove include of machmode.h.
19087 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
19089 (emit_insn_modes_inline_h): ...this new function. Emit the code
19090 into an insn-modes-inline.h header file, adding appropriate
19091 include guards and end comments.
19092 (emit_insn_modes_c_header): Remove include of machmode.h.
19093 (emit_min_insn_modes_c_header): Include coretypes.h rather than
19095 (main): Handle -i flag and call emit_insn_modes_inline_h when
19098 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19100 * tree-ssa-strlen.c (strinfo): Rename the length field to
19101 nonzero_chars. Add a full_string_p field.
19102 (compare_nonzero_chars, zero_length_string_p): New functions.
19103 (get_addr_stridx): Add an offset_out parameter.
19104 Use compare_nonzero_chars.
19105 (get_stridx): Update accordingly. Use compare_nonzero_chars.
19106 (new_strinfo): Update after above changes to strinfo.
19107 (set_endptr_and_length): Set full_string_p.
19108 (get_string_length): Update after above changes to strinfo.
19109 (unshare_strinfo): Update call to new_strinfo.
19110 (maybe_invalidate): Likewise.
19111 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
19112 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
19113 as a uhwi instead of an shwi. Update after above changes to
19114 strinfo and new_strinfo.
19115 (zero_length_string): Assert that chainsi contains full strings.
19116 Use zero_length_string_p. Update call to new_strinfo.
19117 (adjust_related_strinfos): Update after above changes to strinfo.
19118 Copy full_string_p from origsi.
19119 (adjust_last_stmt): Use zero_length_string_p.
19120 (handle_builtin_strlen): Update after above changes to strinfo and
19121 new_strinfo. Install the lhs as the string length if the previous
19122 entry didn't describe a full string.
19123 (handle_builtin_strchr): Update after above changes to strinfo
19125 (handle_builtin_strcpy): Likewise.
19126 (handle_builtin_strcat): Likewise.
19127 (handle_builtin_malloc): Likewise.
19128 (handle_pointer_plus): Likewise.
19129 (handle_builtin_memcpy): Likewise. Track nonzero characters
19130 that aren't necessarily followed by a nul terminator.
19131 (handle_char_store): Likewise.
19133 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19135 PR tree-optimization/80769
19136 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
19137 for malloc and calloc. Document the new invariant that all related
19138 strinfos have delayed lengths or none do.
19139 (verify_related_strinfos): Move earlier in file.
19140 (set_endptr_and_length): New function, split out from...
19141 (get_string_length): ...here. Also set the lengths of related
19143 (zero_length_string): Assert that chainsi has known (rather than
19145 (adjust_related_strinfos): Likewise.
19147 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19149 PR tree-optimization/81136
19150 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
19151 assert that two references with the same misalignment have the same
19152 compile-time misalignment if those compile-time misalignments
19155 2017-07-01 Andi Kleen <ak@linux.intel.com>
19157 * print-tree.c (print_node): Print all attributes.
19159 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19161 * cfg.c (scale_bbs_frequencies): New function.
19162 * cfg.h (scale_bbs_frequencies): Declare it.
19163 * cfgloopanal.c (single_likely_exit): Cleanup.
19164 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
19166 (scale_loop_profile): Likewise.
19167 (loop_version): Likewise.
19168 (create_empty_loop_on_edge): Update.
19169 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
19170 scale_loop_frequencies, scale_loop_profile, loopify,
19171 loop_version): Update prototypes.
19172 * modulo-sched.c (sms_schedule): Update.
19173 * predict.c (unlikely_executed_edge_p): Also check probability.
19174 (probably_never_executed_edge_p): Fix typo.
19175 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19176 * tree-parloops.c (gen_parallel_loop): Update.
19177 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
19178 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19179 * tree-ssa-loop-split.c (split_loop): Update.
19180 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19181 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19182 (vect_loop_versioning): Update.
19183 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19185 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19187 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19189 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19191 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19192 to keep profile consistent.
19194 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19196 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19197 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19198 * profile-count.h (max_safe_multiplier): Make unsigned.
19199 (profile_count::guessed_zero): New.
19201 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19203 * bb-reorder.c (fix_up_crossing_landing_pad,
19204 fix_crossing_conditional_branches): Use make_single_succ_edge
19205 to keep profile consistent.
19207 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19209 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19212 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19215 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19216 the right scopes, make sure cond_jump isn't preserved between multiple
19217 iterations. Search for fallthru edge whenever there are 3+ edges and
19218 use find_fallthru_edge for it.
19220 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19222 Patch by Alexander Monakov <amonakov@ispras.ru>
19223 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19224 probabilities consistently.
19226 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19228 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19229 emit_cmp_and_jump_insns.
19231 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19234 * tree-inline.c (expand_call_inline): Combine profile statuses.
19236 2017-06-30 Andrew Pinski <apinski@cavium.com>
19238 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19239 fold_stmt returned true.
19241 2017-06-30 Nathan Sidwell <nathan@acm.org>
19243 * ggc.h (empty_string): Delete.
19244 * cfgexpand.c (expand_asm_stmt): Use plain "".
19245 * optabs.c (expand_asm_memory_barrier): Likewise.
19246 * stringpool.c (empty_string): Delete.
19247 (digit_vector, digit_string): Delete.
19248 (ggc_alloc_string): Use plain "", don't optimize single digit
19249 strings. Use ggc_alloc_atomic.
19251 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19253 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19254 comparison set and one other set, use the cost of the non-comparison
19257 2017-06-30 Nathan Sidwell <nathan@acm.org>
19259 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19262 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19264 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19265 loops. Remove now unneeded calls to gimple_switch_set_label() that
19266 just set removed labels to NULL_TREE.
19268 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19270 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19271 (set_range_info): ...here. Only call set_range_info_raw if domain
19273 (set_nonzero_bits): Call set_range_info_raw.
19274 * tree-ssanames.h (set_range_info_raw): New.
19276 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19279 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19280 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19281 of nonimmediate_operand and <store_mask_constraint> instead of m
19282 for the input operand. For V8FI iterator, always split if input
19283 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19284 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19285 <store_mask_predicate> instead of register_operand and
19286 <store_mask_constraint> instead of v for the input operand. Make
19287 sure both operands aren't MEMs for if not <mask_applied>.
19289 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19291 * lto-wrapper.c (copy_file) Close both file descriptors before
19294 2017-06-30 Martin Liska <mliska@suse.cz>
19297 * multiple_target.c (create_dispatcher_calls): Make ifunc
19298 also for function that don't have calls or are not referenced.
19300 2017-06-30 Richard Biener <rguenther@suse.de>
19302 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19303 analyze the first scalar stmt. Move vector type computation
19304 for the BB case here from ...
19305 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19306 live operation processing in the SLP case properly.
19308 2017-06-30 Richard Biener <rguenther@suse.de>
19310 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19312 2017-06-30 Martin Liska <mliska@suse.cz>
19315 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19316 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19318 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19320 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19323 2017-06-30 Richard Biener <rguenther@suse.de>
19325 PR tree-optimization/81249
19326 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19327 condition reduction result to original scalar type.
19329 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19331 * profile-count.h (enum profile_quality): Fix typos and whitespace
19334 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19336 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19337 type for branch probabilities.
19339 2017-06-29 Julian Brown <julian@codesourcery.com>
19340 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19342 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19343 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19344 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19345 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19347 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19349 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19350 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19351 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19352 CC usage from generic code to here.
19353 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19354 CC usage into the target macros.
19356 2017-06-29 Maya Rashish <coypu@sdf.org>
19358 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19361 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19363 * arm/arm-builtins.c: Include profile-count.h
19364 * except.c (sjlj_emit_function_enter): Use
19365 profile_probability::unlikely.
19367 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19369 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19370 and tocrel_offset be pointer args rather than implicitly using
19372 (legitimate_constant_pool_address_p, rs6000_emit_move,
19373 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19374 tocrel_offset and use in toc_relative_expr_p call.
19375 (print_operand, print_operand_address): Use static tocrel_base_oac
19376 and tocrel_offset_oac.
19377 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19380 2017-06-29 Maya Rashish <coypu@sdf.org>
19382 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19384 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19386 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19387 objects, take into account only the alignment of 'op0' and 'mode1' if
19390 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19392 * ccmp.c (ccmp_tree_comparison_p): New function.
19393 (ccmp_candidate_p): Update to use above function.
19394 (get_compare_parts): New function.
19395 (expand_ccmp_next): Update to use new functions.
19396 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19398 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19399 take mode as argument.
19400 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19401 * expr.c (expand_expr_real_1): Pass mode as argument.
19403 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19405 * combine.c (combine_instructions): Print insns to dump_file, together
19408 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19410 * asan.c (asan_emit_stack_protection): Update.
19411 (create_cond_insert_point): Update.
19412 * auto-profile.c (afdo_propagate_circuit): Update.
19413 * basic-block.h (struct edge_def): Turn probability to
19414 profile_probability.
19415 (EDGE_FREQUENCY): Update.
19416 * bb-reorder.c (find_traces_1_round): Update.
19417 (better_edge_p): Update.
19418 (sanitize_hot_paths): Update.
19419 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19420 (make_single_succ_edge): Update.
19421 (check_bb_profile): Update.
19422 (dump_edge_info): Update.
19423 (update_bb_profile_for_threading): Update.
19424 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19425 probabilitycount to 0.
19426 * cfgbuild.c (compute_outgoing_frequencies): Update.
19427 * cfgcleanup.c (try_forward_edges): Update.
19428 (outgoing_edges_match): Update.
19429 (try_crossjump_to_edge): Update.
19430 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19431 (expand_gimple_tailcall): Update.
19432 (construct_init_block): Use make_single_succ_edge.
19433 (construct_exit_block): Use make_single_succ_edge.
19434 * cfghooks.c (verify_flow_info): Update.
19435 (redirect_edge_succ_nodup): Update.
19436 (split_edge): Update.
19437 (account_profile_record): Update.
19438 * cfgloopanal.c (single_likely_exit): Update.
19439 * cfgloopmanip.c (scale_loop_profile): Update.
19440 (set_zero_probability): Remove.
19441 (duplicate_loop_to_header_edge): Update.
19442 * cfgloopmanip.h (loop_version): Update prototype.
19443 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19444 (force_nonfallthru_and_redirect): Update.
19445 (update_br_prob_note): Update.
19446 (rtl_verify_edges): Update.
19447 (purge_dead_edges): Update.
19448 (rtl_lv_add_condition_to_bb): Update.
19449 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19450 * cgraphunit.c (init_lowered_empty_function): Update.
19451 (cgraph_node::expand_thunk): Update.
19452 * cilk-common.c: Include profile-count.h
19453 * dojump.c (inv): Remove.
19454 (jumpifnot): Update.
19455 (jumpifnot_1): Update.
19456 (do_jump_1): Update.
19458 (do_jump_by_parts_greater_rtx): Update.
19459 (do_compare_rtx_and_jump): Update.
19460 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19461 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19462 * dwarf2cfi.c: Include profile-count.h
19463 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19464 (sjlj_emit_dispatch_table): Likewise.
19465 * explow.c: Include profile-count.h
19466 * expmed.c (emit_store_flag_force): Update.
19467 (do_cmp_and_jump): Update.
19468 * expr.c (compare_by_pieces_d::generate): Update.
19469 (compare_by_pieces_d::finish_mode): Update.
19470 (emit_block_move_via_loop): Update.
19471 (store_expr_with_bounds): Update.
19472 (store_constructor): Update.
19473 (expand_expr_real_2): Update.
19474 (expand_expr_real_1): Update.
19475 * expr.h (try_casesi, try_tablejump): Update prototypes.
19476 * gimple-pretty-print.c (dump_probability): Update.
19477 (dump_profile): New.
19478 (dump_gimple_label): Update.
19479 (dump_gimple_bb_header): Update.
19480 * graph.c (draw_cfg_node_succ_edges): Update.
19481 * hsa-gen.c (convert_switch_statements): Update.
19482 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
19483 (find_if_case_1): Update.
19484 (find_if_case_2): Update.
19485 * internal-fn.c (expand_arith_overflow_result_store): Update.
19486 (expand_addsub_overflow): Update.
19487 (expand_neg_overflow): Update.
19488 (expand_mul_overflow): Update.
19489 (expand_vector_ubsan_overflow): Update.
19490 * ipa-cp.c (good_cloning_opportunity_p): Update.
19491 * ipa-split.c (split_function): Use make_single_succ_edge.
19492 * ipa-utils.c (ipa_merge_profiles): Update.
19493 * loop-doloop.c (add_test): Update.
19494 (doloop_modify): Update.
19495 * loop-unroll.c (compare_and_jump_seq): Update.
19496 (unroll_loop_runtime_iterations): Update.
19497 * lra-constraints.c (lra_inheritance): Update.
19498 * lto-streamer-in.c (input_cfg): Update.
19499 * lto-streamer-out.c (output_cfg): Update.
19500 * mcf.c (adjust_cfg_counts): Update.
19501 * modulo-sched.c (sms_schedule): Update.
19502 * omp-expand.c (expand_omp_for_init_counts): Update.
19503 (extract_omp_for_update_vars): Update.
19504 (expand_omp_ordered_sink): Update.
19505 (expand_omp_for_ordered_loops): Update.
19506 (expand_omp_for_generic): Update.
19507 (expand_omp_for_static_nochunk): Update.
19508 (expand_omp_for_static_chunk): Update.
19509 (expand_cilk_for): Update.
19510 (expand_omp_simd): Update.
19511 (expand_omp_taskloop_for_outer): Update.
19512 (expand_omp_taskloop_for_inner): Update.
19513 * omp-simd-clone.c (simd_clone_adjust): Update.
19514 * optabs.c (expand_doubleword_shift): Update.
19515 (expand_abs): Update.
19516 (emit_cmp_and_jump_insn_1): Update.
19517 (expand_compare_and_swap_loop): Update.
19518 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
19519 * predict.c (predictable_edge_p): Update.
19520 (edge_probability_reliable_p): Update.
19521 (set_even_probabilities): Update.
19522 (combine_predictions_for_insn): Update.
19523 (combine_predictions_for_bb): Update.
19524 (propagate_freq): Update.
19525 (estimate_bb_frequencies): Update.
19526 (force_edge_cold): Update.
19527 * profile-count.c (profile_count::dump): Add missing space into dump.
19528 (profile_count::debug): Add newline.
19529 (profile_count::differs_from_p): Explicitly convert to unsigned.
19530 (profile_count::stream_in): Update.
19531 (profile_probability::dump): New member function.
19532 (profile_probability::debug): New member function.
19533 (profile_probability::differs_from_p): New member function.
19534 (profile_probability::differs_lot_from_p): New member function.
19535 (profile_probability::stream_in): New member function.
19536 (profile_probability::stream_out): New member function.
19537 * profile-count.h (profile_count_quality): Rename to ...
19538 (profile_quality): ... this one.
19539 (profile_probability): New.
19540 (profile_count): Update.
19541 * profile.c (compute_branch_probabilities): Update.
19542 * recog.c (peep2_attempt): Update.
19543 * sched-ebb.c (schedule_ebbs): Update.
19544 * sched-rgn.c (find_single_block_region): Update.
19545 (compute_dom_prob_ps): Update.
19546 (schedule_region): Update.
19547 * sel-sched-ir.c (compute_succs_info): Update.
19548 * stmt.c (struct case_node): Update.
19549 (do_jump_if_equal): Update.
19550 (get_outgoing_edge_probs): Update.
19551 (conditional_probability): Update.
19552 (emit_case_dispatch_table): Update.
19553 (expand_case): Update.
19554 (expand_sjlj_dispatch_table): Update.
19555 (emit_case_nodes): Update.
19556 * targhooks.c: Update.
19557 * tracer.c (better_p): Update.
19558 (find_best_successor): Update.
19559 * trans-mem.c (expand_transaction): Update.
19560 * tree-call-cdce.c: Update.
19561 * tree-cfg.c (gimple_split_edge): Upate.
19562 (move_sese_region_to_fn): Upate.
19563 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
19564 * tree-eh.c (lower_resx): Upate.
19565 (cleanup_empty_eh_move_lp): Upate.
19566 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19567 * tree-inline.c (copy_edges_for_bb): Update.
19568 (copy_cfg_body): Update.
19569 * tree-parloops.c (gen_parallel_loop): Update.
19570 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
19571 (gimple_gen_time_profiler): Update.
19572 * tree-ssa-dce.c (remove_dead_stmt): Update.
19573 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19574 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
19575 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
19576 (unloop_loops): Update.
19577 (try_peel_loop): Update.
19578 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19579 * tree-ssa-loop-split.c (connect_loops): Update.
19580 (split_loop): Update.
19581 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19582 (hoist_guard): Update.
19583 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
19584 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
19585 (value_replacement): Update.
19586 * tree-ssa-reassoc.c (branch_fixup): Update.
19587 * tree-ssa-tail-merge.c (replace_block_by): Update.
19588 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
19589 (create_edge_and_update_destination_phis): Update.
19590 (compute_path_counts): Update.
19591 (recompute_probabilities): Update.
19592 (update_joiner_offpath_counts): Update.
19593 (freqs_to_counts_path): Update.
19594 (duplicate_thread_path): Update.
19595 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
19596 (struct switch_conv_info): Update.
19597 (gen_inbound_check): Update.
19598 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19599 (vect_do_peeling): Update.
19600 (vect_loop_versioning): Update.
19601 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19602 (optimize_mask_stores): Update.
19603 * ubsan.c (ubsan_expand_null_ifn): Update.
19604 * value-prof.c (gimple_divmod_fixed_value): Update.
19605 (gimple_divmod_fixed_value_transform): Update.
19606 (gimple_mod_pow2): Update.
19607 (gimple_mod_pow2_value_transform): Update.
19608 (gimple_mod_subtract): Update.
19609 (gimple_mod_subtract_transform): Update.
19610 (gimple_ic): Update.
19611 (gimple_stringop_fixed_value): Update.
19612 (gimple_stringops_transform): Update.
19613 * value-prof.h: Update.
19615 2017-06-29 Carl Love <cel@us.ibm.com>
19617 * config/rs6000/rs6000-c.c: Add support for built-in functions
19618 vector signed int vec_signed (vector float);
19619 vector signed long long vec_signed (vector double);
19620 vector signed int vec_signed2 (vector double, vector double);
19621 vector signed int vec_signede (vector double);
19622 vector signed int vec_signedo (vector double);
19623 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
19624 instruction generator.
19625 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19626 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
19627 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
19629 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
19630 vunsignede_v2df): Add define_expands.
19631 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
19632 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
19633 VEC_UNSIGNEDO): Add definitions.
19634 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19635 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
19636 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
19637 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
19638 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
19639 * config/rs6000/altivec.h (vec_signed, vec_signed2,
19640 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
19641 vec_unsignede, vec_unsignedo): Add builtin defines.
19642 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
19644 * doc/extend.texi: Update the built-in documentation file for the
19645 new built-in functions.
19647 2017-06-29 Richard Biener <rguenther@suse.de>
19649 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
19650 reduction chains to LOOP_VINFO_REDUCTIONS.
19651 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
19652 SLP reductions after processing reduction chains.
19654 2017-06-29 Nathan Sidwell <nathan@acm.org>
19656 * builtins.c (fold_builtin_FUNCTION): Use
19657 lang_hooks.decl_printable_name.
19659 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
19661 PR middle-end/81194
19662 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
19663 with only one label.
19664 * stmt.c (expand_case): Assert NCASES is greater than one.
19666 2017-06-29 Richard Biener <rguenther@suse.de>
19668 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
19670 (group_case_labels): Likewise.
19671 (find_taken_edge): Push sanity checking on val to workers...
19672 (find_taken_edge_cond_expr): ... here
19673 (find_taken_edge_switch_expr): ... and here, handle cases
19674 with just a default label.
19675 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
19676 (group_case_labels): Likewise.
19677 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
19678 group_case_labels does anything cleanup the CFG again.
19680 2017-06-29 Bin Cheng <bin.cheng@arm.com>
19682 PR tree-optimization/81196
19683 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
19684 exit condition comparing two IVs.
19686 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
19688 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
19689 profile to the dummy entry at the end of the list of architectures.
19690 * config/arm/arm-cpu-cdata.h: Regenerated.
19692 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19693 Michael Collison <michael.collison@arm.com>
19696 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
19698 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
19699 (*aarch64_reg_<mode>3_minus_mask): New pattern.
19700 (*aarch64_<optab>_reg_di3_mask2): New pattern.
19701 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
19702 of shift when the shift amount is masked with constant equal to
19703 the size of the mode.
19704 * config/aarch64/predicates.md (subreg_lowpart_operator): New
19707 2017-06-29 Martin Liska <mliska@suse.cz>
19709 * config/i386/i386.opt: Change range from [1,5] to [0,5].
19711 2017-06-29 Yury Gribov <tetra2005@gmail.com>
19714 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
19717 (edge_growth_cache_entry::edge_growth_cache_entry): New
19719 (reset_edge_growth_cache): Update to use constructor.
19721 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19723 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
19724 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19725 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19727 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
19729 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
19730 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
19732 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
19734 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
19735 (*-linux-uclibc*): Add t-uclibc tmake_file.
19736 * config/t-musl: New.
19737 * config/t-uclibc: New.
19739 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
19741 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
19743 (gen_comm_data): Emit architectural setting of arch_prof.
19744 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
19746 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
19747 (armv8-m.base, armv8-m.main): Likewise.
19748 * arm-protos.h (arm_build_target): Add profile field.
19749 (arch_option): Likewise.
19750 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
19752 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
19753 arm_active_target.profile.
19755 2017-06-28 Richard Biener <rguenther@suse.de>
19757 PR middle-end/81227
19758 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
19759 TYPE_OVERFLOW_WRAPS.
19760 * match.pd (negate_expr_p): Likewise.
19761 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
19762 fold_build2, not fold_binary.
19764 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19766 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
19767 Convert memory address to Pmode.
19768 (aarch64_print_operand): Assert MEM operands are always Pmode.
19770 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19773 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
19774 Remove redundant if.
19775 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
19776 * config/arm/aarch-common-protos.h
19777 (aarch_forward_to_shift_is_not_shifted_re): Remove.
19778 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
19780 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
19783 * multiple_target.c (create_dispatcher_calls): Set the default
19784 clone to be static, not public.
19786 2017-06-28 Richard Biener <rguenther@suse.de>
19788 * tree-vect-loop.c (vectorizable_reduction): Move special
19789 cond reduction IV var creation ...
19790 (vect_create_epilog_for_reduction): ... here. Remove induction_index
19791 parameter. Use STMT_VINFO_VECTYPE.
19792 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
19795 2017-06-28 Martin Liska <mliska@suse.cz>
19798 * ipa-visibility.c (non_local_p): Handle visibility.
19800 2017-06-28 Martin Liska <mliska@suse.cz>
19803 * common.opt: Add IntegerRange to various options.
19804 * opt-functions.awk (integer_range_info): New function.
19805 * optc-gen.awk: Add integer_range_info to cl_options struct.
19806 * opts-common.c (decode_cmdline_option): Handle
19807 CL_ERR_INT_RANGE_ARG.
19808 (cmdline_handle_error): Likewise.
19809 * opts.c (print_filtered_help): Show valid interval in
19810 when --help is provided.
19811 * opts.h (struct cl_option): Add range_min and range_max fields.
19812 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
19814 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
19816 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
19817 (x * C EQ/NE y * C): New transformation.
19819 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
19821 * genmultilib (combination_space): Accept '+' in option names.
19823 2017-06-28 Martin Liska <mliska@suse.cz>
19826 * asan.c (instrument_derefs): Bail out inner references
19827 that are hard register variables.
19829 2017-06-28 Jakub Jelinek <jakub@redhat.com>
19832 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
19833 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
19835 2017-06-28 Richard Biener <rguenther@suse.de>
19837 * tree-vectorizer.h (vect_get_vec_defs): Remove.
19838 (vect_get_slp_defs): Adjust.
19839 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
19841 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
19843 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
19844 get_initial_defs_for_reduction instead of vect_get_vec_defs.
19845 (vectorizable_reduction): Adjust.
19846 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
19848 (vect_get_slp_defs): Likewise.
19849 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
19850 (vectorizable_bswap): Adjust.
19851 (vectorizable_call): Likewise.
19852 (vectorizable_conversion): Likewise.
19853 (vectorizable_assignment): Likewise.
19854 (vectorizable_shift): Likewise.
19855 (vectorizable_operation): Likewise.
19856 (vectorizable_store): Likewise.
19857 (vectorizable_condition): Likewise.
19858 (vectorizable_comparison): Likewise.
19860 2017-06-28 Michael Collison <michael.collison@arm.com>
19863 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
19865 (arm_gen_movmemqi): Removed unused variable 'i'.
19866 Convert 'for' loop into 'while' loop.
19867 (arm_expand_prologue): Remove last unnecessary set of insn.
19868 (thumb_pop): Remove unused variable 'pushed_words'.
19869 (thumb_exit): Remove last unnecessary set of regs_to_pop.
19871 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19873 * config/s390/predicates.md: Use s390_rel_address_ok_p.
19874 * config/s390/s390-protos.h: Add prototype of
19875 s390_rel_address_ok_p.
19876 * config/s390/s390.c (s390_got_symbol): New function.
19877 (s390_rel_address_ok_p): New function.
19878 (legitimize_pic_address): Use s390_rel_address_ok_p.
19879 (s390_load_got): Use s390_got_symbol.
19880 (s390_option_override): Issue error if
19881 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
19882 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
19884 * config/s390/s390.opt: New option mpic-data-is-text-relative.
19886 2017-06-27 Andrew Pinski <apinski@cavium.com>
19888 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
19889 (X * copysign (1.0, X)): New pattern.
19890 (X * copysign (1.0, -X)): New pattern.
19891 (copysign (-1.0, CST)): New pattern.
19893 2017-06-27 Joseph Myers <joseph@codesourcery.com>
19895 * genmultilib (combination_space): Remove variable.
19896 Validate reuse rules against regular expression for any sequence
19897 of multilib options in any order.
19899 2017-06-27 Michael Collison <michael.collison@arm.com>
19901 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
19902 call aarch64_split_simd_combine.
19903 * (aarch64_combine_internal<mode>): Delete pattern.
19904 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
19905 Allow register and subreg operands.
19907 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19909 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
19910 specific need, just fallback on defaults.
19911 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
19913 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19914 Olivier Hainque <hainque@adacore.com>
19916 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
19918 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
19919 targets. Pick a default if no particular attempt applied.
19920 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
19923 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19925 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
19926 (x86_64-wrs-vxworks7): Likewise.
19928 2017-06-27 Marek Polacek <polacek@redhat.com>
19931 * ubsan.c (instrument_null): Check get_base_address's result for null.
19933 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19935 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
19937 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19939 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
19940 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
19941 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
19942 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
19943 New function types.
19944 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
19945 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
19946 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
19947 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
19948 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
19949 BUILT_IN_FEUPDATEENV): New builtins.
19950 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
19951 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
19952 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
19953 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
19955 (builtin_structptr_types): Adjust size.
19956 * tree.c (builtin_structptr_types): Add four entries.
19958 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19959 Olivier Hainque <hainque@adacore.com>
19961 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
19962 (TLS_SYM): New local macro, forcing reference to __tls__ on
19963 link command lines for VxWorks 7 RTPs, triggering initialization
19965 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
19966 OS features TLS support, true for RTPs on VxWorks 7.
19967 * config/vxworks.c (vxworks_override_options): Setup emutls
19970 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19972 * predict.c (test_prediction_value_range): Use -1U instead of -1
19973 to avoid narrowing conversion warning.
19974 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
19975 to avoid narrowing conversion warning.
19976 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
19978 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
19980 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19982 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
19983 64bit configurations.
19984 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
19985 (SIZE_TYPE): Likewise.
19986 * config/vxworks.c (vxworks_emutls_var_fields): Use
19987 long_unsigned_type_node instead of unsigned_type_node as the offset
19988 field type, which is "pointer" mode in emutls.c.
19990 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19993 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
19995 PR middle-end/81207
19996 * gimple-fold.c (replace_call_with_call_and_fold): Handle
19997 gimple_vuse copying separately from gimple_vdef copying.
19999 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20001 * value-prof.c (free_hist): Remove call to memset and the enclosing if
20004 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20005 Olivier Hainque <hainque@adacore.com>
20007 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
20008 for all vxworks7 targets.
20009 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
20010 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
20011 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
20012 variations for VX6/VX7 and 32/64bits later on in ...
20013 (VXWORKS_LIB_SPEC): Leverage new macros.
20014 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
20015 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
20017 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20019 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
20020 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
20022 2017-06-26 Carl Love <cel@us.ibm.com>
20024 * config/rs6000/rs6000-c.c: Add support for built-in functions
20025 vector bool char vec_reve (vector bool char);
20026 vector signed char vec_reve (vector signed char);
20027 vector unsigned char vec_reve (vector unsigned char);
20028 vector bool int vec_reve (vector bool int);
20029 vector signed int vec_reve (vector signed int);
20030 vector unsigned int vec_reve (vector unsigned int);
20031 vector bool long long vec_reve (vector bool long long);
20032 vector signed long long vec_reve (vector signed long long);
20033 vector unsigned long long vec_reve (vector unsigned long long);
20034 vector bool short vec_reve (vector bool short);
20035 vector signed short vec_reve (vector signed short);
20036 vector double vec_reve (vector double);
20037 vector float vec_reve (vector float);
20038 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
20039 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
20040 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
20041 (altivec_vreve): New pattern.
20042 * config/rs6000/altivec.h (vec_reve): New define.
20043 * doc/extend.texi (vec_rev): Update the built-in documentation file
20044 for the new built-in functions.
20046 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20048 PR tree-optimization/71815
20049 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
20051 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
20053 (slsr_process_phi): Likewise.
20054 (replace_uncond_cands_and_profitable_phis): Don't replace a
20055 multiply candidate with a stride of 1 (copy or cast).
20056 (phi_incr_cost): Call uses_consumed_by_stmt rather than
20058 (lowest_cost_path): Likewise.
20059 (total_savings): Likewise.
20061 2017-06-26 Richard Biener <rguenther@suse.de>
20064 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
20065 Use def_builtin_pure for all gather builtins.
20067 2017-06-26 Richard Biener <rguenther@suse.de>
20069 PR tree-optimization/81203
20070 * tree-tailcall.c (find_tail_calls): Do not move stmts into
20071 non-dominating BBs.
20073 2017-06-26 Marek Polacek <polacek@redhat.com>
20076 * doc/invoke.texi: Document -Wmultistatement-macros.
20078 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
20080 * doc/sourcebuild.texi (ARM-specific attributes): Document new
20081 arm_neon_ok_no_float_abi effective target.
20083 2017-06-26 Richard Biener <rguenther@suse.de>
20085 PR tree-optimization/80928
20086 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
20087 (copy_bbs): Set BB_DUPLICATED flag early.
20088 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
20090 (execute_on_shrinking_pred): Likewise.
20091 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
20092 BB_DUPLICATED blocks.
20093 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
20094 iterate over all PHIs considering removal of *gsi.
20096 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
20098 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
20101 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20103 * config/rs6000/rs6000-string.c: (expand_block_clear,
20104 do_load_for_compare, select_block_compare_mode,
20105 compute_current_alignment, expand_block_compare,
20106 expand_strncmp_align_check, expand_strn_compare,
20107 expand_block_move, rs6000_output_load_multiple)
20108 Move functions related to string/block move/compare
20109 to a separate file.
20110 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
20111 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
20112 for this function which is now used in two files.
20113 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
20114 * config.gcc: Add rs6000-string.o to extra_objs for
20115 targets powerpc*-*-* and rs6000*-*-*.
20117 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
20120 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
20121 32-bit, since indexed is not valid for DImode.
20122 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
20123 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
20124 (define_peephole2 for Altivec d-form load): Add 32-bit support.
20125 (define_peephole2 for Altivec d-form store): Likewise.
20128 * multiple_target.c (create_dispatcher_calls): Only create the
20129 dispatcher call if the function is the default clone of a
20130 versioned function.
20132 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
20134 PR middle-end/80902
20135 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
20136 a call, force the call to not be a tail call.
20138 2017-06-23 Jeff Law <law@redhat.com>
20140 * doc/contrib.texi: Add entry for Steven Pemberton's work on
20143 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
20145 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
20146 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
20147 handling for early expansion of vector shifts (sl,sr,sra,rl).
20148 (builtin_function_type): Add vector shift right instructions
20149 to the unsigned argument list.
20151 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
20153 rtl-optimizatoin/79286
20154 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
20155 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
20156 trap. PIC register plus a const unspec without offset can never trap.
20158 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
20160 * tree.h (builtin_structptr_type): New type.
20161 (builtin_structptr_types): Declare new array.
20162 * tree.c (builtin_structptr_types): New array.
20163 (free_lang_data, build_common_tree_nodes): Use it.
20165 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
20168 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
20171 2017-06-22 Matt Turner <mattst88@gmail.com>
20173 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
20174 Lake models to skylake case. Assume skylake for unknown
20175 models with clflushopt.
20177 2017-06-22 Jeff Law <law@redhat.com>
20179 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20180 frame sizes that do not satisfy aarch64_uimm12_shift.
20182 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20184 * profile-count.h (apply_probability,
20185 apply_scale, probability_in): Fix checks for zero.
20187 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20189 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20190 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20192 2016-06-22 Richard Biener <rguenther@suse.de>
20194 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20195 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20196 REDUC_MAX_EXPR support.
20197 (vectorizable_reduction): Likewise.
20198 (vect_create_epilog_for_reduction): Likewise.
20200 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20202 * match.pd (A / (1 << B) -> A >> B): New.
20203 * generic-match-head.c: Include optabs-tree.h.
20204 * gimple-match-head.c: Likewise.
20205 * optabs-tree.h (target_supports_op_p): New.
20206 * optabs-tree.c (target_supports_op_p): New.
20208 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20210 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20211 $gcc_cv_ld --help output.
20212 (gcc_cv_ld_demangle): Likewise.
20213 (gcc_cv_ld_eh_frame_hdr): Likewise.
20214 (gcc_cv_ld_pie): Likewise.
20215 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20216 (gcc_cv_ld_buildid): Likewise.
20217 (gcc_cv_ld_sysroot): Likewise.
20218 (ld_bndplt_support): Likewise.
20219 (ld_pushpopstate_support): Likewise.
20220 * configure: Regenerate.
20221 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20223 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20226 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20227 operands indexes to avoid gap between operands and match_dups.
20229 2017-06-21 Andrew Pinski <apinski@cavium.com>
20231 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20232 Increment Arith_shift and Arith_shift_reg by 1.
20233 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20235 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20236 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20237 (aarch64_strip_extend): Add new argument and test for it.
20238 (aarch64_cheap_mult_shift_p): New function.
20239 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20240 add a cost if it is true.
20241 Update calls to aarch64_strip_extend.
20242 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20244 2017-06-21 Andrew Pinski <apinski@cavium.com>
20246 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20248 (thunderxt88): Likewise.
20249 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20250 (thunderx_prefetch_tune): New variable.
20251 (thunderx2t99_prefetch_tune): Update for the correct values.
20252 (thunderxt88_tunings): New variable.
20253 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20254 generic_prefetch_tune.
20255 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20257 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20259 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20260 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20261 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20262 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20263 (aarch64_atomic_cas<mode>, GPI): Likewise.
20265 2017-06-21 Martin Liska <mliska@suse.cz>
20267 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20268 statements on cold and hot labels.
20269 * predict.c (tree_estimate_probability_bb): Remove the
20270 prediction from this place.
20272 2017-06-21 Martin Liska <mliska@suse.cz>
20274 PR tree-optimization/79489
20275 * gimplify.c (maybe_add_early_return_predict_stmt): New
20277 (gimplify_return_expr): Call the function.
20278 * predict.c (tree_estimate_probability_bb): Remove handling
20280 * predict.def: Update comment about early return predictor.
20281 * gimple-predict.h (is_gimple_predict): New function.
20282 * predict.def: Change default value of early return to 66.
20283 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20285 * passes.def: Put pass_strip_predict_hints to the beginning of
20288 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20290 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20291 FUNCTION_DECL declarations.
20292 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20294 (dwaf2out_decl): Likewise.
20295 * godump.c (go_early_global_decl): Skip call to the real debug hook
20296 for FUNCTION_DECL declarations.
20297 * passes.c (rest_of_decl_compilation): Skip call to the
20298 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20299 -fdump-go-spec is passed.
20301 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20303 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20304 Reorder for compactness.
20305 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20306 (def_builtin_pure, def_builtin_pure2): New functions.
20307 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20309 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20311 * match.pd (nop_convert): New predicate.
20312 ((A +- CST1) +- CST2): Allow some NOP conversions.
20314 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20317 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20318 with ctors/dtors if GOVD_SHARED is set.
20320 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20322 * config/aarch64/aarch64.md (movti_aarch64):
20323 Emit mov rather than orr.
20324 (movtf_aarch64): Likewise.
20325 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20326 Emit mov rather than orr.
20328 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20330 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20331 Swap alternatives, make integer dup more expensive.
20333 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20335 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20336 Return true for non-tls symbols.
20338 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20340 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20341 (cortex-a75): Likewise.
20342 (cortex-a75.cortex-a55): Likewise.
20343 * config/aarch64/aarch64-tune.md: Regenerate.
20344 * doc/invoke.texi (-mtune): Document new values for -mtune.
20346 2017-06-21 Tom de Vries <tom@codesourcery.com>
20348 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20349 stack_size feature.
20350 (Effective-Target Keywords, Other attributes): Suggest using
20351 dg-add-options stack_size feature to get stack limit in stack_size
20352 effective target documentation.
20354 2017-06-21 Julian Brown <julian@codesourcery.com>
20355 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20357 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20358 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20359 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20361 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20362 attribute type list for neon_multiply.
20363 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20364 attribute type list for neon_multiply.
20365 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20366 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20367 attribute type list for neon_multiply.
20368 * config/arm/types.md (crypto_pmull): Add.
20369 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20370 attribute type list.
20372 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20374 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20377 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20379 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20380 to make sure not to dereference a NULL cost_classes_ptr pointer.
20382 2017-06-20 Carl Love <cel@us.ibm.com>
20384 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20385 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20386 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20387 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20388 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20389 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20390 VMULOSW): New enum "unspec" values.
20391 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20392 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20393 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20394 altivec_vmulosw): New patterns.
20395 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20396 VMULOSW): Add definitions.
20398 2017-06-20 Julia Koval <julia.koval@intel.com>
20400 * config/i386/i386.c: Fix rounding expand for new pattern.
20401 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20403 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20405 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20406 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20408 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20410 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20413 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20415 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20416 architecture, then by implementer ID.
20417 * config/aarch64/aarch64-tune.md: Regenerate.
20419 2017-06-20 Richard Biener <rguenther@suse.de>
20421 PR middle-end/81097
20422 * fold-const.c (split_tree): Fold to type before negating.
20424 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20426 * diagnostic-show-locus.c
20427 (selftest::test_fixit_deletion_affecting_newline): New function.
20428 (selftest::diagnostic_show_locus_c_tests): Call it.
20430 2017-06-20 Andreas Schwab <schwab@suse.de>
20433 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20436 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20438 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20439 __ARM_FEATURE_COPROC according to support.
20441 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20443 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20444 Rewritten to avoid overflow for > 32-bit pointers.
20447 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20448 by removing enum keyword.
20449 (ubsan_type_descriptor): Likewise. Formatting fix.
20452 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
20453 splitter): Require TARGET_SSE2 in the condition.
20455 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20458 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20459 for doing vector set of SFmode on ISA 3.0.
20460 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20461 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20463 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20464 SFmode value into a V4SF variable that was extracted from another
20465 V4SF variable without converting the element to double precision
20466 and back to single precision vector format.
20467 (vsx_insert_extract_v4sf_p9_2): Likewise.
20469 2017-06-19 Jakub Jelinek <jakub@redhat.com>
20471 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
20472 in UWHI to avoid undefined overflow.
20475 * ubsan.h (enum ubsan_encode_value_phase): New.
20476 (ubsan_encode_value): Change second argument to
20477 enum ubsan_encode_value_phase with default value of
20478 UBSAN_ENCODE_VALUE_GENERIC.
20479 * ubsan.c (ubsan_encode_value): Change second argument to
20480 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
20481 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
20482 create_tmp_var_raw instead of create_tmp_var and use a
20484 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
20485 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
20486 ubsan_encode_value callers.
20489 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
20490 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
20491 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
20493 2017-06-19 Richard Biener <rguenther@suse.de>
20495 PR middle-end/81118
20496 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
20497 estimates if we changed anything.
20499 2017-06-19 Richard Biener <rguenther@suse.de>
20501 PR tree-optimization/80887
20502 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
20503 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
20504 simplified lookups, then reset mprts_hook.
20505 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
20507 (try_to_simplify): Likewise.
20509 2017-06-19 Martin Liska <mliska@suse.cz>
20512 * gimplify.c (gimplify_switch_expr):
20513 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
20515 2017-06-19 Martin Liska <mliska@suse.cz>
20517 * doc/install.texi: Document that PGO runs in 4 stages.
20519 2017-06-19 Martin Liska <mliska@suse.cz>
20522 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
20523 to dispatcher function name.
20524 * multiple_target.c (replace_function_decl): New function.
20525 (create_dispatcher_calls): Redirect both edges and references.
20527 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
20529 * profile-count.c (profile_count::dump): Dump quality.
20530 (profile_count::differs_from_p): Update for unsigned val.
20531 * profile-count.h (profile_count_quality): New enum.
20532 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
20534 2017-06-19 Richard Biener <rguenther@suse.de>
20536 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
20537 struct function as arg.
20538 (estimate_numbers_of_iterations): Export overload with loop arg.
20539 (free_numbers_of_iterations_estimates_loop): Use an overload of
20540 free_numbers_of_iterations_estimates instead.
20541 * tree-cfg.c (remove_bb): Adjust.
20542 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
20543 * tree-parloops.c (gen_parallel_loop): Likewise.
20544 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
20546 (tree_unroll_loops_completely): Likewise.
20547 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
20548 Use an overload instead and export.
20549 (estimated_loop_iterations): Adjust.
20550 (max_loop_iterations): Likewise.
20551 (likely_max_loop_iterations): Likewise.
20552 (estimate_numbers_of_iterations): Take struct function as arg
20554 (loop_exits_before_overflow): Adjust.
20555 (free_numbers_of_iterations_estimates_loop): Use an overload.
20556 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
20557 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
20559 2017-06-19 Richard Biener <rguenther@suse.de>
20562 * ipa-prop.c (find_constructor_constant_at_offset): Handle
20563 RANGE_EXPR conservatively.
20565 2017-06-16 Carl Love <cel@us.ibm.com>
20567 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20568 definitions for vec_float, vec_float2, vec_floato,
20569 vec_floate built-ins.
20570 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
20571 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
20573 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
20574 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
20575 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
20576 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
20577 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
20578 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
20579 vec_floato): Add builtin defines.
20580 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
20581 Update the built-in documentation file for the new built-in
20584 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20586 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
20587 (mthumb): Mark as the negative of -marm.
20589 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20591 * doc/invoke.texi (ARM Options, -mcpu): Document supported
20593 (ARM Options, -mtune): Document that this accepts the same
20594 extension options as -mcpu.
20595 (ARM Options, -mfpu): Document addition of -mfpu=auto.
20597 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20599 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
20600 permitted extensions.
20602 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20604 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
20605 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
20606 (armv8-m.main): Add option +nodsp.
20607 * config/arm/arm-cpu-cdata.h: Regenerated.
20609 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20611 * config/arm/t-fuchsia: New file.
20612 * config.gcc (arm*-*-fuchsia*): Use it.
20614 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20616 * config/arm/t-symbian: Rewrite for new option infrastructure.
20618 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20620 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
20621 (MULTILIB_REQUIRED): Likewise.
20623 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20625 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
20626 (MULTILIB_RESUE): Likewise.
20627 (MULTILIB_MATCHES): Likewise.
20628 (MULTLIB_REQUIRED): Likewise.
20630 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20632 * config/arm/t-rtems: Rewrite for new option framework.
20634 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20636 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
20637 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
20638 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
20639 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
20640 * config/arm/t-multilib: ... here.
20641 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
20642 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
20643 armv7-a and armv8*-a when A-profile libraries have not been built.
20644 * config/arm/t-rmprofile: Rewrite.
20646 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20648 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
20649 with a backslash. Remove the backslash after substituting unescaped
20651 * doc/fragments.texi (MULTILIB_REUSE): Document it.
20653 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20655 * config.gcc: (arm*-*-*): When building a-profile libraries, force
20656 the driver to pass through the default setting of -mfloat-abi.
20657 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
20659 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
20660 (all_feat_combs): New rule.
20661 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
20663 * config/arm/t-aprofile: Rewrite.
20665 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20667 * config/arm/arm.h (FPUTYPE_AUTO): Define.
20668 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
20669 fpu is not specified by the user/command-line.
20670 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
20671 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
20672 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
20673 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
20674 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
20675 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
20677 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20679 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
20680 * config/arm/t-arm-elf: Rewritten.
20682 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20684 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
20685 have some floating-point instructions.
20686 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
20687 (TARGET_MAYBE_HARD_FLOAT): New macro.
20688 * config/arm/arm-builtins.c (arm_init_builtins): Use
20689 TARGET_MAYBE_HARD_FLOAT.
20690 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
20692 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20694 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
20695 (configargs.h): Include it.
20696 (arm_print_hint_for_fpu_option): New function.
20697 (arm_parse_fpu_option): New function.
20698 (candidate_extension): New class.
20699 (arm_canon_for_multilib): New function.
20700 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
20701 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
20702 (ARCH_CANONICAL_SPECS): New macro.
20703 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
20705 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20707 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
20708 are set after handling multilib fragments. Set target_cpu_default2
20711 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20713 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
20715 (arm*-*-*): Set target_cpu_default2 to a quoted string.
20716 * config/arm/parsecpu.awk (check_cpu): Validate any extension
20718 (check_arch): Likewise.
20719 * config/arm/arm.c (arm_configure_build_target): Handle
20720 TARGET_CPU_DEFAULT being a string constant. Scan any feature
20721 options in the default.
20723 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20725 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
20726 when an option is an alias of another.
20727 * config/arm/parsecpu.awk (optalias): New parser token.
20728 (gen_comm_data): Mark non-alias options as such. Emit entries
20729 for extension aliases.
20730 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
20731 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
20732 (armv6kz, armv6zk, armv6t2): Likewise.
20733 (armv7): Make vfpv3-d16 an alias.
20734 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
20736 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
20737 Sort in canonical order.
20738 (armv8-a): Sort in canonical order.
20739 (armv8.1-a, armv8.2-a): Likewise.
20740 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
20742 (cortex-a9): Sort in canonical order.
20743 * config/arm/arm.c (selftests.h): Include it.
20744 (arm_test_cpu_arch_data): New function.
20745 (arm_run_self_tests): New function.
20746 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
20747 (targetm): Move declaration to the end of the file.
20748 * arm-cpu-cdata.h: Regenerated.
20750 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20752 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
20753 call to target_mode_check describing the type of option passed.
20754 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
20755 (arm_target_thumb_only): Use arm_parse_arch_option_name or
20756 arm_parse_cpu_option_name to match parameters against list of
20758 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
20759 arm_arch_core_flags data structure.
20760 * config/arm/arm-cpu_cdata.h: Regenerated.
20762 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20764 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
20766 (arm_print_hint_for_cpu_option): Likewise.
20767 (arm_print_hint_for_arch_option): Likewise.
20768 (arm_parse_cpu_option_name): Likewise.
20769 (arm_parse_arch_option_name): Likewise.
20770 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
20771 of entries in the all_fpus list.
20772 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
20773 (arm_parse_cpu_option_name): Declare.
20774 (arm_parse_arch_option_name): Declare.
20775 (arm_parse_option_features): Declare.
20776 (arm_intialize_isa): Declare.
20777 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
20779 (gen_comm_data): ... here. Make definitions non-static.
20780 * config/arm/arm-cpu-data.h: Regenerated.
20781 * config/arm/arm-cpu-cdata.h: Regenerated.
20783 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20785 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
20786 (cpu_arch_extension): New structure.
20787 (cpu_arch_option, arch_option, cpu_option): New structures.
20788 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
20789 architecture types.
20790 (gen_data): Generate new format data tables.
20791 * config/arm/arm.c (cpu_tune): New structure.
20792 (cpu_option, processors): Delete.
20793 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
20794 (arm_print_hint_for_cpu_option): ... this and ...
20795 (arm_print_hint_for_arch_option): ... this.
20796 (arm_parse_arch_cpu_name): Delete. Replace with ...
20797 (arm_parse_cpu_option_name): ... this and ...
20798 (arm_parse_arch_option_name): ... this.
20799 (arm_unrecognized_feature): Change type of target parameter to
20801 (arm_parse_arch_cpu_features): Delete. Replace with ...
20802 (arm_parse_option_features): ... this.
20803 (arm_configure_build_target): Rework to use new configuration data
20805 (arm_print_tune_info): Rework for new configuration data tables.
20806 * config/arm/arm-cpu-data.h: Regenerated.
20807 * config/arm/arm-cpu.h: Regenerated.
20809 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20811 * Makefile.in (OBJS): Move sbitmap.o from here ...
20812 (OBJS-libcommon): ... to here.
20814 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20816 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
20817 (ISA_ALL_CRYPTO): New macro.
20818 (ISA_ALL_SIMD): New macro
20819 (ISA_ALL_FP): New macro.
20820 * config/arm/arm.c (fpu_bitlist): Update initializer.
20821 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
20823 (arm9e): Add fpu. Add option for nofp
20824 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
20825 (arm926ej-s, arm1026ej-s): Likewise.
20826 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
20827 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
20828 neon-fp16, neon-vfpv4, nofp and nosimd.
20829 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
20830 (cortex-a8): Add fpu. Add option for nofp.
20831 (cortex-a9): Add fpu. Add options for nosimd and nofp.
20832 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
20833 (cortex-r4f): Add fpu.
20834 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
20835 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
20837 (cortex-r8): Likewise.
20838 (cortex-m4): Add fpu. Add option for nofp.
20839 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
20840 (cortex-a17.cortex-a7): Likewise.
20841 (cortex-a32): Add fpu. Add options for crypto and nofp.
20842 (cortex-a35, cortex-a53): Likewise.
20843 (cortex-a57): Add fpu. Add option for crypto.
20844 (cortex-a72, cortex-a73): Likewise.
20845 (exynos-m1): Likewise.
20846 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
20847 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
20848 (cortex-m33): Add fpu. Add option for nofp.
20849 * config/arm/arm-cpu-cdata.h: Regenerated
20850 * config/arm/arm-cpu-data.h: Regenerated.
20852 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20854 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
20855 (armv5te, armv5tej): Likewise.
20856 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
20857 (armv7): Add options fp and vfpv3-d16.
20858 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
20859 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
20861 (armv7ve): Likewise.
20862 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
20863 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
20864 (armv8-a): Add nocrypto option.
20865 (armv8.1-a, armv8.2-a): Likewise.
20866 (armv8-m.main): add options fp, fp.dp and nofp.
20868 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20870 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
20872 (armv8-a+crc): Delete.
20873 (armv8.1-a): Add options simd, crypto and nofp.
20874 (armv8.2-a): Add options fp16, simd, crypto and nofp.
20875 (armv8.2-a+fp16): Delete.
20876 (armv8-m.main): Add option dsp.
20877 (armv8-m.main+dsp): Delete.
20878 (cortex-a8): Add fpu. Add nofp option.
20879 (cortex-a9): Add fpu. Add nofp and nosimd options.
20880 * config/arm/parsecpu.awk (gen_data): Generate option tables and
20881 link to main cpu and architecture data structures.
20882 (gen_comm_data): Only put isa attributes from the main architecture
20884 (option): New statement for architecture and CPU entries.
20885 * arm.c (struct cpu_option): New structure.
20886 (struct processors): Add entry for options.
20887 (arm_unrecognized_feature): New function.
20888 (arm_parse_arch_cpu_name): Ignore any characters after the first
20890 (arm_parse_arch_cpu_feature): New function.
20891 (arm_configure_build_target): Separate out any CPU and architecture
20892 features and parse separately. Don't error out if -mfpu=auto is
20893 used with only an architecture string.
20894 (arm_print_asm_arch_directives): New function.
20895 (arm_file_start): Call it.
20896 * config/arm/arm-cpu-cdata.h: Regenerated.
20897 * config/arm/arm-cpu-data.h: Likewise.
20898 * config/arm/arm-tables.opt: Likewise.
20900 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20902 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
20903 assembler when it is not -mfpu=auto.
20905 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20907 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
20908 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
20909 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
20910 (ASM_CPU_SPEC): Rewrite.
20911 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
20912 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
20913 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
20914 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
20915 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
20916 copied string is NUL-terminated. Also strip any characters prefixed
20918 (arm_rewrite_selected_arch): New function.
20919 (arm_rewrite_march): New function.
20921 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20923 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
20924 (x_arm_cpu_string, x_arm_tune_string): Likewise.
20925 (march, mcpu, mtune): Convert to string-based options.
20926 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
20927 (arm_parse_arch_cpu_name): New function.
20928 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
20929 identify selected architecture or CPU.
20930 (arm_option_save): New function.
20931 (TARGET_OPTION_SAVE): Redefine.
20932 (arm_option_restore): Restore string options.
20933 (arm_option_print): Print string options.
20935 2017-06-16 Martin Sebor <msebor@redhat.com>
20937 PR tree-optimization/80933
20938 PR tree-optimization/80934
20939 * builtins.c (fold_builtin_3): Do not handle bcmp here.
20940 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
20941 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
20942 (gimple_fold_builtin): Call them.
20944 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20946 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
20947 as unlikely; update profile.
20949 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20951 * predict.c (force_edge_cold): Handle declaring edges impossible
20954 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20956 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
20958 (try_unroll_loop_completely): Fix reporting.
20960 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20962 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
20964 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
20967 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
20968 if given a non-constant argument for an intrinsic which requires a
20971 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20973 * profile.c (compare_freqs): New function.
20974 (branch_prob): Sort edge list.
20975 (find_spanning_tree): Assume that the list is priority sorted.
20977 2017-06-16 Richard Biener <rguenther@suse.de>
20979 PR tree-optimization/81090
20980 * passes.def (pass_record_bounds): Remove.
20981 * tree-pass.h (make_pass_record_bounds): Likewise.
20982 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
20983 make_pass_record_bounds): Likewise.
20984 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
20985 not free niter estimates at the beginning but at the end.
20986 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
20988 2017-06-16 Richard Biener <rguenther@suse.de>
20990 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
20991 initializer to workaround ICE in host GCC 4.8.
20993 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20995 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
20997 (clone_inlined_nodes): Update.
20999 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21001 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
21002 prefetch settings, and enable prefetching by default at -O3.
21004 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21006 * config/aarch64/aarch64.c (aarch64_override_options_internal):
21007 Set flag_prefetch_loop_arrays according to tuning data.
21009 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21011 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
21012 New tune structure.
21013 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
21014 [Unrelated to main purpose of the patch] Place the pointer field last
21015 to enable type checking errors when tune structure are wrongly merged.
21016 * config/aarch64/aarch64.c (generic_prefetch_tune,)
21017 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
21018 (thunderx2t99_prefetch_tune): New tune constants.
21019 (tune_params *_tunings): Update all tunings (no functional change).
21020 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
21021 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
21022 from tunings structures.
21024 2017-06-16 Jakub Jelinek <jakub@redhat.com>
21027 * ubsan.c (instrument_null): Add T argument, use it instead
21028 of computing it based on IS_LHS.
21029 (instrument_object_size): Likewise.
21030 (pass_ubsan::execute): Adjust instrument_null and
21031 instrument_object_size callers to pass gimple_get_lhs or
21032 gimple_assign_rhs1 result to it. Use instrument_null instead of
21033 calling get_base_address and instrument_mem_ref. Handle
21034 aggregate call arguments for object-size sanitization.
21036 2017-06-16 Yury Gribov <tetra2005@gmail.com>
21038 PR tree-optimization/81089
21039 * tree-vrp.c (is_masked_range_test): Validate operands of
21042 2017-06-15 Martin Sebor <msebor@redhat.com>
21045 * dumpfile.c (dump_register): Avoid calling memset to initialize
21046 a class with a default ctor.
21047 * gcc.c (struct compiler): Remove const qualification.
21048 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
21049 * hash-table.h: Ditto.
21050 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
21052 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
21053 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
21055 * params.h (struct param_info): Make struct members non-const.
21056 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
21057 with default initialization.
21058 * vec.h (vec_copy_construct, vec_default_construct): New helper
21060 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
21061 with vec_copy_construct.
21062 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
21063 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
21064 * doc/invoke.texi (-Wclass-memaccess): Document.
21066 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21068 * emit-rtl.h (is_leaf): Update comment about local
21069 register allocator.
21071 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21074 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
21075 for a variable to have a section before checking if the section has a
21077 Set section to.persistent if persistent attribute is set.
21078 Warn if .persistent attribute is used on an automatic variable.
21080 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
21082 PR rtl-optimization/80474
21083 * reorg.c (update_block): Do not ignore instructions in a delay slot.
21085 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
21087 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
21090 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
21092 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
21093 (casesi): Emit bounds checking as RTL.
21094 (casesi_internal_mips16_<mode>): Remove bounds checking.
21096 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
21098 * config/xtensa/xtensa.c (xtensa_option_override): Append
21099 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
21100 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
21101 xtensa_doloop_hooks): Define unconditionally.
21102 (xtensa_reorg_loops): Only call reorg_loops in the presence of
21104 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
21105 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
21106 for it in xtensa_option_override.
21107 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
21108 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
21110 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
21112 * doc/cppopts.texi: Document '-' special value to -MF.
21114 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
21116 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
21117 (cortex_a53_fconst): Likewise.
21118 (cortex_a53_fpmul): Likewise.
21119 (cortex_a53_f_load_64): Likewise.
21120 (cortex_a53_f_load_many): Likewise.
21121 (cortex_a53_advsimd_alu): Likewise.
21122 (cortex_a53_advsimd_alu_q): Likewise.
21123 (cortex_a53_advsimd_mul): Likewise.
21124 (cortex_a53_advsimd_mul_q): Likewise.
21125 (fpmac bypass): Add new bypass for fpmac-fpmac case.
21126 Add missing fmul, r2f_cvt and fconst cases.
21128 2017-06-14 Richard Biener <rguenther@suse.de>
21130 PR middle-end/81088
21131 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
21133 (fold_binary_loc): When associating do not treat pre-existing
21134 TREE_OVERFLOW on literal constants as a reason to allow
21135 TREE_OVERFLOW on associated literal constants.
21137 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21139 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
21140 (MASK_FEATURES): New macro.
21141 * config/sparc/sparc.c (sparc_option_override): Remove the special
21142 handling of -mfpu and generalize it to all MASK_FEATURES switches.
21144 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21146 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
21147 a division of 0 if non-call exceptions are enabled.
21149 2017-06-14 Andrew Pinski <apinski@cavium.com>
21150 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21153 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
21154 Improve vector initialization code gen for only variable case.
21156 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21158 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
21160 2017-06-14 Richard Biener <rguenther@suse.de>
21162 PR tree-optimization/81083
21163 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
21166 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21168 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
21169 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
21170 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
21171 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
21172 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
21173 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
21175 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21177 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
21178 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21180 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21182 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21184 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21186 * config/rs6000/t-rtems: Don't handle SPE.
21188 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21190 * config/rs6000/t-linux: Don't handle SPE.
21192 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21194 * config/rs6000/eabispe.h: Delete file.
21196 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21198 * config/rs6000/t-spe: Delete file.
21200 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21202 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21203 (rs6000_legitimate_offset_address_p): Return false for anything in
21204 V2SImode or V2SFmode.
21206 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21208 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21209 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21210 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21212 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21213 (rs6000_legitimate_offset_address_p): Ditto.
21214 (rs6000_emit_move): Ditto.
21215 (rs6000_init_builtins): Remove V4HI_type_node.
21217 2017-06-13 Martin Liska <mliska@suse.cz>
21220 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21221 (gate_asan): Likewise.
21222 * asan.h (asan_no_sanitize_address_p): Remove the function.
21223 (sanitize_flags_p): New function.
21224 * builtins.def: Fix coding style.
21225 * common.opt: Use renamed enum value.
21226 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21227 * doc/extend.texi: Document no_sanitize attribute.
21228 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21229 to SANITIZE_UNDEFINED_NONDEFAULT.
21230 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21231 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21232 Use sanitize_flags_p.
21233 * gimplify.c (gimplify_function_tree): Likewise.
21234 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21235 * opts.c (parse_no_sanitize_attribute): New function.
21236 (common_handle_option): Use renamed enum value.
21237 * opts.h (parse_no_sanitize_attribute): Declare.
21238 * tree.c (sanitize_flags_p): New function.
21239 * tree.h: Declared here.
21240 * tsan.c: Use sanitize_flags_p.
21241 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21242 (instrument_mem_ref): Likewise.
21243 (instrument_bool_enum_load): Likewise.
21244 (do_ubsan_in_current_function): Remove the function.
21245 (pass_ubsan::execute): Use sanitize_flags_p.
21246 * ubsan.h: Remove do_ubsan_in_current_function
21247 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21248 (dump_function_to_file): Use it here.
21250 2017-06-13 Martin Jambor <mjambor@suse.cz>
21252 PR tree-optimization/80803
21253 PR tree-optimization/81063
21254 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21255 (propagate_subaccesses_across_link): Enqueue subtree whenever
21256 necessary instead of relying on the caller.
21258 2017-06-13 Martin Jambor <mjambor@suse.cz>
21260 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21261 that have a first_link.
21262 (sort_and_splice_var_accesses): Do not check first_link before
21264 (subtree_mark_written_and_enqueue): Likewise.
21265 (propagate_all_subaccesses): Likewise and do not stop at first
21266 parent with a first_link.
21268 2017-06-13 Martin Jambor <mjambor@suse.cz>
21270 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21273 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21275 * match.pd: New pattern.
21277 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21279 * tree-vrp.c (is_masked_range_test): New function.
21280 (register_edge_assert_for): Determine ranges for
21283 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21285 PR tree-optimization/67328
21286 * fold-const.c (maskable_range_p): New function.
21287 (build_range_check): Generate bittests if possible.
21289 2017-06-13 Martin Liska <mliska@suse.cz>
21291 * gimple-pretty-print.c (dump_probability): Add new argument.
21292 (dump_edge_probability): Dump both probability and count.
21293 (dump_gimple_label): Likewise.
21294 (dump_gimple_bb_header): Likewise.
21296 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21299 * config/avr/avr-devices.c: Fix indentation.
21300 * config/avr/gen-avr-mmcu-specs.c: Dito.
21302 2017-06-13 Richard Biener <rguenther@suse.de>
21304 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21305 instead get vector type from stmt_info.
21306 (vectorizable_reduction): Adjust. Remove dead code.
21308 2017-06-13 Richard Biener <rguenther@suse.de>
21310 PR middle-end/81065
21311 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21312 case of C * (x * C2 + C3).
21313 (fold_addr_of_array_ref_difference): Properly fold index difference.
21315 2017-06-12 David S. Miller <davem@davemloft.net>
21318 * config/sparc/sparc.md (return expander): Emit frame blockage if
21319 function uses alloca.
21321 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21323 * combine.c (make_field_assignment): Check len rather than the mode
21324 precision when calling force_to_mode.
21326 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21328 Support multilibs and devices that see flash in RAM address range.
21331 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21332 (avr_mcu_t) <flash_pm_offset>: New field.
21333 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21334 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21335 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21336 (AVR_TINY_PM_OFFSET): Remove macro.
21337 * config/avr/avr.opt (-mshort-calls): New option.
21338 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21339 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21340 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21341 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21342 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21343 instead of avr_arch->have_jmp_call.
21344 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21345 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21346 avr_arch->flash_pm_offset to define.
21347 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21348 new field flash_pm_offset. Add entry for avrxmega3.
21349 (avr_texinfo): Add entry for avrxmega3.
21350 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21351 attiny212, attiny214,
21352 attiny412, attiny414, attiny416, attiny417,
21353 attiny814, attiny816, attiny817,
21354 attiny1614, attiny1616, attiny1617,
21355 attiny3214, attiny3216, attiny3217.
21356 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21357 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21358 (avr_print_operand_address) [AVR_TINY]: Same.
21359 (avr_asm_init_sections) <readonly_data_section>: Only patch
21360 callback if avr_arch->flash_pm_offset = 0.
21361 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21362 for rodata if avr_arch->flash_pm_offset != 0.
21363 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21364 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21365 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21366 (m_options): Append opt_rcall.
21367 (m_dirnames): Append dir_rcall.
21368 * config/avr/t-multilib: Regenerate.
21370 * configure.ac [target=avr]: Check whether avrxmega3 default
21371 linker description file works as needed.
21372 * configure: Regenerate.
21373 * doc/avr-mmcu.texi: Regenerate.
21374 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21375 <__AVR_ARCH__>: Document avrxmega3 and 103.
21376 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21377 <__AVR_SHORT_CALLS__>: Document it.
21378 <__AVR_PM_BASE_ADDRESS__>: Document it.
21379 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21380 (AVR Variable Attributes) <progmem>: Document this is
21381 not needed for avrxmega3.
21382 (AVR Named Address Spaces) <__flash>: Dito.
21384 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21386 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21388 2017-06-12 Doug Rupp <rupp@adacore.com>
21390 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21391 Append vxworks-stdint.h to the tm_file list.
21392 * config/vxworks-stdint.h: New file.
21394 2017-06-12 Martin Liska <mliska@suse.cz>
21396 PR tree-optimization/81041
21397 * tree-profile.c (gimple_gen_ic_func_profiler):
21398 Create an extra BB in profile-generate
21399 (gimple_gen_time_profiler): Likewise.
21401 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21403 PR tree-optimization/81003
21404 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21405 (update_range_test): Use it instead of force_gimple_operand_gsi.
21407 2017-06-12 Richard Biener <rguenther@suse.de>
21409 PR tree-optimization/81053
21410 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21411 with backedge value not defined in loop. Simplify def stmt
21414 2017-06-11 Tom de Vries <tom@codesourcery.com>
21417 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21419 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21420 nvptx_cannot_force_const_mem.
21422 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21424 * opts.c (finish_options): Move test for flag_split_stack after
21425 it has been initialized.
21427 2017-06-11 Jason Merrill <jason@redhat.com>
21429 * tree.h (id_equal): New.
21430 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21431 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21432 instead of strcmp of IDENTIFIER_POINTER.
21434 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21436 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21437 (mark_all_inlined_calls_cdtor): Fix formating.
21438 (inline_transform): Rescale profile before inlining.
21440 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21442 * cgraph.h (cgraph_edge::clone): Update prototype.
21443 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21444 (cgraph_node::create_clone): Update.
21445 (cgraph_node::create_version_clone): Update.
21446 * tree-inline.c (copy_bb): Update.
21447 (expand_call_inline): Update.
21449 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21451 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21452 factored out from ...
21453 (rs6000_emit_prologue): ... here.
21455 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21457 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21458 factored out from ...
21459 (rs6000_emit_prologue): ... here.
21461 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21463 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21465 (handle_missing_profiles): Fix computation of tp_first_run.
21466 (counts_to_freqs): Do not touch freqs when count is 0.
21468 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21470 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
21473 2017-06-10 Tom de Vries <tom@codesourcery.com>
21475 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21476 attributes): Document signal effective target.
21478 2017-06-10 Tom de Vries <tom@codesourcery.com>
21480 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21481 Document effective target stack_size.
21483 2017-06-09 David Malcolm <dmalcolm@redhat.com>
21485 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
21486 to the edit_context if they can be auto-applied.
21488 2017-06-9 Ian Lance Taylor <iant@golang.org>
21490 * opts.c (finish_options): If -fsplit-stack, disable implicit
21491 -forder-blocks-and-partition.
21492 * doc/invoke.texi (Optimize Options): Document that when using
21493 -fsplit-stack -forder-blocks-and-partition is not implicitly
21496 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21498 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
21499 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
21500 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
21501 * builtins.def (abort, trap, unreachable): Declare cold.
21502 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
21503 * tree-core.h (ECF_COLD): New.
21504 * tree.c (set_call_expr_flags): Handle ECF_COLD.
21505 (build_common_builtin_nodes): Mark unreachable and abort as cold.
21507 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21509 * predict.c (unlikely_executed_stmt_p): Cleanup.
21511 2017-06-09 Richard Biener <rguenther@suse.de>
21513 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
21514 model if the ref is always written to.
21516 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21518 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
21520 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21522 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
21525 2017-06-09 Tom de Vries <tom@codesourcery.com>
21528 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
21529 "target cannot support label values" when encountering LABEL_REF.
21531 2017-06-09 Martin Liska <mliska@suse.cz>
21533 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
21534 (gimple_gen_ic_func_profiler): Emit direct comparison
21535 of __gcov_indirect_call_callee with NULL.
21536 (gimple_gen_time_profiler): Change probability from
21537 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
21539 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21541 * profile.c (edge_gcov_counts): Turn to pointer.
21542 (compute_branch_probabilities, compute_branch_probabilities): Update.
21543 (branch_prob): Do not clear edge_gcov_count.
21544 * profile.h (edge_gcov_counts): Turn to pointer.
21545 (edge_gcov_count): Update.
21547 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21549 * gimple.h (gimple_check_failed): Mark cold.
21551 2017-06-09 Richard Biener <rguenther@suse.de>
21553 PR tree-optimization/66623
21554 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
21555 refactor check_reduction into two parts, properly computing
21556 whether we have to check reduction validity for outer loop
21559 2017-06-09 Richard Biener <rguenther@suse.de>
21561 PR tree-optimization/79483
21562 * graphite-scop-detection.c (order): New global.
21563 (get_order): Compute bb to order mapping that satisfies code
21564 generation constraints.
21565 (cmp_pbbs): New helper.
21566 (build_scops): Start domwalk at entry block, sort generated
21569 2017-06-09 Richard Biener <rguenther@suse.de>
21571 PR middle-end/81007
21572 * ipa-polymorphic-call.c
21573 (ipa_polymorphic_call_context::restrict_to_inner_class):
21574 Skip FIELD_DECLs with error_mark_node type.
21575 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
21578 2017-06-09 Martin Liska <mliska@suse.cz>
21580 * predict.c (struct branch_predictor): New struct.
21581 (test_prediction_value_range): New test.
21582 (predict_c_tests): New function.
21583 * selftest-run-tests.c (selftest::run_tests): Run the function.
21584 * selftest.h: Declare new tests.
21586 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
21589 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
21590 gen_add3_insn did not fail.
21591 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
21592 r0, construct that number in a temporary reg and add that reg to r0.
21593 If asked to put the result in r0 as well, fail.
21595 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
21597 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21598 for early expansion of vec_eqv.
21600 2017-06-08 Jakub Jelinek <jakub@redhat.com>
21602 PR middle-end/81005
21603 * ubsan.c (instrument_null): Avoid pointless code temporary.
21604 (pass_ubsan::execute): Instrument aggregate arguments of calls.
21606 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
21610 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
21613 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
21616 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21618 * predict.c (unlikely_executed_edge_p): Move ahead.
21619 (probably_never_executed_edge_p): Use it.
21621 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21623 PR middle-end/79988
21624 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
21625 gimple_call_builtin_p call.
21627 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21629 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
21630 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
21631 rtl_check_failed_type2, rtl_check_failed_code1,
21632 rtl_check_failed_code2, rtl_check_failed_code_mode,
21633 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
21634 rtvec_check_failed_bounds, rtl_check_failed_flag,
21635 _fatal_insn_not_found, _fatal_insn): Likewise.
21636 * tree.h (tree_contains_struct_check_failed,
21637 tree_check_failed, tree_not_check_failed,
21638 tree_class_check_failed, tree_range_check_failed,
21639 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
21640 tree_vec_elt_check_failed, phi_node_elt_check_failed,
21641 tree_operand_check_failed, omp_clause_check_failed,
21642 omp_clause_operand_check_failed, omp_clause_range_check_failed):
21645 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21647 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
21648 flag_branch_probabilities.
21649 * ipa-inline.c (edge_badness): Likewise.
21650 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
21651 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
21652 * predict.c (maybe_hot_frequency_p): Likewise.
21653 (probably_never_executed): Likewise.
21654 * sched-ebb.c (schedule_ebbs): Likewise.
21655 * sched-rgn.c (find_single_block_region): Likewise.
21656 * tracer.c (tail_duplicate): Likewise.
21658 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21660 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
21661 longer requires x_flag_profile_use.
21663 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21665 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
21666 instead of flag_reorder_blocks_and_partition.
21667 * dbxout.c (dbxout_function_end): Likewise.
21668 * dwarf2out.c (gen_subprogram_die): Likewise.
21669 * haifa-sched.c (sched_create_recovery_edges): Likewise.
21670 * hw-doloop.c (reorg_loops): Likewise.
21671 * varasm.c (assemble_start_function,
21672 assemble_end_function): Likewise.
21673 (decide_function_section): Do not check for
21674 flag_reorder_blocks_and_partition.
21676 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21678 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21680 (chkp_get_hard_register_fake_addr_expr): Ditto.
21681 (chkp_build_addr_expr): Add check for hard reg case.
21682 (chkp_parse_array_and_component_ref): Ditto.
21683 (chkp_find_bounds_1): Ditto.
21684 (chkp_process_stmt): Don't generate bounds store for
21687 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21689 * predict.c (maybe_hot_bb_p): Do not check profile status.
21690 (maybe_hot_edge_p): Likewise.
21691 (probably_never_executed): Check for zero counts even if profile
21693 (unlikely_executed_edge_p): New function.
21694 (unlikely_executed_stmt_p): New function.
21695 (unlikely_executed_bb_p): New function.
21696 (set_even_probabilities): Use unlikely predicates.
21697 (combine_predictions_for_bb): Likewise.
21698 (predict_paths_for_bb): Likewise.
21699 (predict_paths_leading_to_edge): Likewise.
21700 (determine_unlikely_bbs): New function.
21701 (estimate_bb_frequencies): Use it.
21702 (compute_function_frequency): Use zero counts even if profile is
21704 * profile-count.h: Fix typo.
21706 2017-08-08 Julia Koval <julia.koval@intel.com>
21708 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
21709 _mm512_mask_cvtsepi16_storeu_epi8,
21710 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
21711 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
21712 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
21713 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
21714 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
21715 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
21716 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
21717 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
21718 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
21719 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
21720 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
21721 __builtin_ia32_pmovuswb256mem_mask,
21722 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
21723 __builtin_ia32_pmovwb512mem_mask): New builtins.
21725 2017-08-08 Julia Koval <julia.koval@intel.com>
21727 PR target/73350,80862
21728 * config/i386/subst.md (round): Fix round pattern.
21729 * config/i386/i386.c (ix86_erase_embedded_rounding):
21730 Fix erasing rounding for the fixed pattern.
21732 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21734 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
21736 2017-06-08 Martin Liska <mliska@suse.cz>
21738 PR gcov-profile/80911
21739 * gcov.c (block_info::block_info): New constructor.
21741 2017-06-07 Carl Love <cel@us.ibm.com>
21743 * config/rs6000/rs6000-c: The return type of the following
21744 built-in functions was implemented as int not long long. Fix sign
21745 of return value for the unsigned version of vec_mulo and vec_mule.
21746 vector unsigned long long vec_bperm (vector unsigned long long,
21747 vector unsigned char)
21748 vector signed long long vec_mule (vector signed int,
21750 vector unsigned long long vec_mule (vector unsigned int,
21751 vector unsigned int)
21752 vector signed long long vec_mulo (vector signed int,
21754 vector unsigned long long vec_mulo (vector unsigned int,
21755 vector unsigned int)
21756 * doc/extend.texi: Fix the documentation for the built-in
21759 2017-06-07 Carl Love <cel@us.ibm.com>
21762 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
21765 2017-06-07 Carl Love <cel@us.ibm.com>
21767 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
21768 support, Generate doublehv for signed int/float for BE case only.
21770 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
21772 * doc/invoke.texi (mcx16): Rewrite.
21774 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21776 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
21777 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
21778 *mov<mode>_softfloat, and an anonymous splitter): Use
21779 nonimmediate_operand instead of rs6000_nonimmediate_operand.
21781 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21783 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
21785 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21786 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
21787 (rs6000_debug_reg_global): Adjust.
21788 (rs6000_init_hard_regno_mode_ok): Adjust.
21789 (rs6000_dbx_register_number): Adjust.
21790 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
21791 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21792 Remove SPE_ACC and SPEFSCR.
21793 (REG_ALLOC_ORDER): Ditto.
21794 (FRAME_POINTER_REGNUM): Change to 111.
21795 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
21796 (REG_CLASS_NAMES): Ditto.
21797 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
21798 (REGISTER_NAMES): Ditto.
21799 (ADDITIONAL_REG_NAMES): Ditto.
21800 (rs6000_reg_names): Ditto.
21801 * config/rs6000/rs6000.md: Renumber some register number
21804 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21806 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
21808 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21809 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
21811 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
21812 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21813 Delete the SPE high registers.
21814 (REG_ALLOC_ORDER): Ditto.
21815 (enum reg_class): Remove SPE_HIGH_REGS.
21816 (REG_CLASS_NAMES): Ditto.
21817 (REG_CLASS_CONTENTS): Delete the SPE high registers.
21818 (REGISTER_NAMES): Ditto.
21819 (rs6000_reg_names): Ditto.
21820 * doc/tm.texi.in: Remove SPE as example.
21821 * doc/tm.texi: Regenerate.
21823 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21825 * config/rs6000/8540.md (ppc8540_brinc): Delete.
21826 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
21827 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
21828 * config/rs6000/rs6000.md (type): Remove "brinc".
21830 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21832 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
21833 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
21834 * config/rs6000/linuxspe.h: Delete file.
21835 * config/rs6000/rs6000.md: Don't include spe.md.
21836 * config/rs6000/spe.h: Delete file.
21837 * config/rs6000/spe.md: Delete file.
21838 * config/rs6000/t-rs6000: Remove spe.md.
21840 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21842 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
21843 (reg_or_none500mem_operand): Delete.
21844 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
21845 instead of reg_or_none500mem_operand.
21847 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21849 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21850 handling of SPE flags.
21851 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
21853 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21855 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
21857 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
21858 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
21859 paired_divv2sf3): Similar.
21860 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
21861 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
21862 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
21864 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
21865 Rename the paired_* instruction patterns.
21866 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
21868 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
21869 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
21870 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
21871 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
21872 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
21873 PAIRED_VECTOR_MODE.
21874 (struct machine_function): Delete field spe_insn_chain_scanned_p.
21875 (spe_func_has_64bit_regs_p): Delete.
21876 (spe_expand_predicate_builtin): Delete.
21877 (spe_expand_evsel_builtin): Delete.
21878 (TARGET_DWARF_REGISTER_SPAN): Do not define.
21879 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
21880 (invalid_e500_subreg): Delete.
21881 (rs6000_legitimize_address): Always force_reg op2 as well, for
21882 paired single memory accesses.
21883 (rs6000_member_type_forces_blk): Delete.
21884 (rs6000_spe_function_arg): Delete.
21885 (rs6000_expand_unop_builtin): Delete SPE handling.
21886 (rs6000_expand_binop_builtin): Ditto.
21887 (spe_expand_stv_builtin): Delete.
21888 (bdesc_2arg_spe): Delete.
21889 (spe_expand_builtin): Delete.
21890 (spe_expand_predicate_builtin): Delete.
21891 (spe_expand_evsel_builtin): Delete.
21892 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
21893 (spe_init_builtins): Delete.
21894 (spe_func_has_64bit_regs_p): Delete.
21895 (savres_routine_name): Delete "info" parameter. Adjust callers.
21896 (rs6000_emit_stack_reset): Ditto.
21897 (rs6000_dwarf_register_span): Delete.
21898 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
21899 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
21900 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
21902 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
21904 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
21905 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
21906 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
21907 mulv2sf3, divv2sf3): Delete expanders.
21909 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21911 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
21913 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21915 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
21916 * config/rs6000/rs6000.c: Ditto.
21918 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21920 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
21921 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
21922 comparison_operator.
21924 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21926 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
21927 * config/rs6000/rs6000.opt: Ditto.
21928 * config/rs6000/t-rtems: Ditto.
21930 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21932 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
21933 TARGET_E500_SINGLE by 0, simplify.
21934 * config/rs6000/rs6000.c: Ditto.
21935 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
21936 (spe_build_register_parallel): Delete.
21937 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
21938 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
21939 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
21940 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
21941 (E500_CONVERT): Delete.
21942 * config/rs6000/spe.md: Remove many patterns and all define_constants.
21944 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21946 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
21947 * config/rs6000/dfp.md: Ditto.
21948 (negdd2, *negdd2_fpr): Merge.
21949 (absdd2, *absdd2_fpr): Merge.
21950 (negtd2, *negtd2_fpr): Merge.
21951 (abstd2, *abstd2_fpr): Merge.
21952 * config/rs6000/e500.h: Delete file.
21953 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
21954 TARGET_FPRS by 1 and simplify.
21955 * config/rs6000/rs6000-c.c: Ditto.
21956 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
21957 TARGET_DF_SPE by 0.
21958 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
21960 * config/rs6000/rs6000.md: Ditto.
21961 (floatdidf2, *floatdidf2_fpr): Merge.
21962 (move_from_CR_gt_bit): Delete.
21963 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
21964 (E500_CR_IOR_COMPARE): Delete.
21965 (All patterns that require !TARGET_FPRS): Delete.
21966 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
21968 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21970 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
21972 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21974 * graphds.c (add_edge): Intitialize edge's attached data.
21975 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
21976 pointer parameter. Call pointed function on each edge during
21977 graph traversing. Skip traversing the edge when the function
21979 (graphds_dfs, graphds_scc): Ditto.
21980 (for_each_edge): New parameter. Pass the new parameter to callback
21982 * graphds.h (skip_edge_callback): New function pointer type.
21983 (graphds_dfs, graphds_scc): New function pointer parameter.
21984 (graphds_edge_callback, for_each_edge): New parameter.
21986 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21988 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
21989 out code checking if runtime alias check is possible to below ...
21990 Call the new function.
21991 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
21992 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
21994 2017-06-07 Marek Polacek <polacek@redhat.com>
21997 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
21998 TYPE_OVERFLOW_WRAPS check.
22000 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22002 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
22003 if versioning is required.
22004 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
22005 peeling with the check for versioning.
22007 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22009 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
22010 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
22011 Set true to new parameter if new ssa variable is defined.
22012 (vect_gen_vector_loop_niters): Refactor. Set range information
22013 for the new vector loop bound variable.
22014 (vect_do_peeling): Ditto.
22016 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22018 * tree-affine.c (ssa.h): Include header file.
22019 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
22020 has wrapping overflow behavior.
22022 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22024 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
22026 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22028 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
22029 (tree_to_aff_combination): ... here.
22031 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22033 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
22034 reg_pressure model function.
22035 (ivopts_global_cost_for_size): Delete.
22036 (determine_set_costs, iv_ca_recount_cost): Call new model function
22037 ivopts_estimate_reg_pressure.
22039 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22041 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
22042 expensive than udiv. Remove floating point cases from mod.
22044 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22046 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
22047 Increase idiv cost.
22049 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22051 * config/aarch64/aarch64.md
22052 (copysignsf3): Fix mask generation.
22054 2017-06-07 Jakub Jelinek <jakub@redhat.com>
22056 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
22058 (class dump_manager): Add register_dumps method.
22059 * dumpfile.c: Include langhooks.h.
22060 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
22061 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
22062 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
22063 (dump_manager::dump_register): Start with 512 entries instead of 32.
22064 (dump_manager::register_dumps): New method.
22065 * toplev.c (general_init): Instead of invoking register_dumps
22066 langhook, invoke register_dumps method on the dump manager.
22067 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
22070 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
22072 * doc/md.texi: Clarify the restrictions on a define_insn condition.
22073 Say that # requires an associated define_split to exist, and that
22074 the define_split must be suitable for use after register allocation.
22076 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22078 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
22079 (compute_outgoing_frequencies): Also initialize zero counts.
22080 (find_many_sub_basic_blocks): Do not produce uninitialized profile
22081 around loops; preserve more of profile when nothing changes.
22083 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
22085 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
22087 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
22088 * config/arm/arm-cpu-cdata.h: Regenerate.
22089 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
22090 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
22091 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
22092 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
22093 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
22095 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
22097 * config/arm/t-rmprofile: Likewise.
22098 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
22100 2017-06-06 David S. Miller <davem@davemloft.net>
22103 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
22104 blockage if function uses alloca.
22106 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22108 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
22109 New "uid" fields to hold pretty-print IDs of group and ref.
22110 Memory references are now identified as <group_id>:<ref_id>
22111 instead of using [random] addresses.
22112 (dump_mem_details): Simplify, no functional change.
22113 (dump_mem_ref): Simplify and make output more concise.
22114 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
22115 (find_or_create_group): Initialize group uid.
22116 (record_ref): Initialize ref uid. Improve debug output.
22117 (prune_group_by_reuse, should_issue_prefetch_p,)
22118 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
22119 (mark_nontemporal_store, determine_loop_nest_reuse):
22120 Improve debug output.
22122 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22124 * dbgcnt.def (prefetch): New debug counter.
22125 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
22126 (schedule_prefetches): Stop issueing prefetches if debug counter
22129 2017-06-06 Tom de Vries <tom@codesourcery.com>
22131 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
22132 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
22134 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22136 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
22137 Use aarch64_reg_or_zero predicate for operand 4.
22138 (aarch64_compare_and_swap<mode> define_insn_and_split):
22139 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
22140 (aarch64_store_exclusive<mode>): Likewise for operand 2.
22142 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22144 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
22145 (arm_compute_save_core_reg_mask): This.
22146 (thumb1_compute_save_reg_mask): Rename into ...
22147 (thumb1_compute_save_core_reg_mask): This.
22148 (arm_compute_save_reg0_reg12_mask): Adapt comment.
22149 (arm_compute_frame_layout): Likewise.
22151 2017-06-06 Richard Biener <rguenther@suse.de>
22153 PR tree-optimization/80974
22154 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
22155 keep or clear leaders SSA info.
22157 2017-06-06 Tom de Vries <tom@codesourcery.com>
22159 * config/nvptx/nvptx.c (split_mode_p): New function.
22160 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
22162 2017-06-06 Tom de Vries <tom@codesourcery.com>
22164 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
22166 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22169 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
22172 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22174 * shrink-wrap.c (handle_simple_exit): Update profile.
22175 (try_shrink_wrapping): Upate profile.
22177 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22179 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22180 (tree_guess_outgoing_edge_probabilities): New.
22181 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22182 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22184 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22186 * ipa-split.c (split_function): Initialize return bb profile.
22188 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22190 * profile.c (compute_branch_probabilities): Also initialize
22191 EXIT_BLOCK profile.
22193 2017-06-06 Richard Biener <rguenther@suse.de>
22195 PR tree-optimization/80928
22196 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22197 (vect_analyze_loop_operations): Properly guard analysis for
22199 (vect_transform_loop): Likewise.
22200 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22201 (vect_model_induction_cost): Do not cost for pure SLP.
22202 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22203 of induction in inner loop vectorization.
22204 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22205 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22206 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22208 (vect_analyze_slp_cost_1): Cost induction.
22209 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22210 (vect_get_slp_vect_defs): Likewise.
22211 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22212 (vect_transform_stmt): Handle SLP reductions.
22213 * tree-vectorizer.h (vectorizable_induction): Adjust.
22215 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22217 * config/rs6000/rs6000.c (make_resolver_func): Update
22218 init_lowered_empty_function call.
22220 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22222 * doc/invoke.texi: Document the -fprofile-abs-path option.
22223 * common.opt (fprofile-abs-path): New option.
22224 * gcov-io.h (gcov_write_filename): Declare.
22225 * gcov-io.c (gcov_write_filename): New function.
22226 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22227 * profile.c (output_location): Likewise.
22229 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22231 * shring-wrap.c: Revert accidental commit.
22233 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22235 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22237 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22239 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22241 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22242 profile in callgraph edge.
22243 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22244 (apply_scale): Likewise.
22245 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22246 Also scale profile when inlining function with zero profile.
22247 (initialize_cfun): Update exit block profile even when it is zero.
22248 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22249 when profile is read.
22251 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22253 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22254 (CLONE_*): New constants to define the processors we can generate
22255 code for with the target_clone attribute.
22256 (rs6000_clone_map): New array to identify which clone processors
22257 the current program is running on.
22258 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22259 target_clone attribute.
22260 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22261 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22262 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22263 (cpu_expand_builtin): Add support for target_clone attribute.
22264 (rs6000_valid_attribute_p): Allow "default" attribute.
22265 (get_decl_name): New debug function to simplify printing the
22266 current function name in debugging statements.
22267 (rs6000_clone_priority): New functions to support the target_clone
22268 attribute, and be able to generate code to switch between ISA 2.05
22269 through ISA 3.0 (power6 through power9).
22270 (rs6000_compare_version_priority): Likewise.
22271 (rs6000_get_function_versions_dispatcher): Likewise.
22272 (make_resolver_func): Likewise.
22273 (add_condition_to_bb): Likewise.
22274 (dispatch_function_versions): Likewise.
22275 (rs6000_generate_version_dispatcher_body): Likewise.
22276 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22277 (fusion_gpr_load_p): Fix a spacing issue.
22278 * doc/extend.texi (Common Function Attributes): Document that the
22279 PowerPC supports the target_clone attribute.
22281 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22283 * config/arm/arm.h: explain F symbol found in description of ARM
22284 register allocation in its legend.
22286 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22288 * config/mips/frame-header-opt.c: Include profile-count.h.
22289 * config/riscv/riscv.c: Include profile-count.h
22291 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22293 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22295 (sm_set_flag_if_changed): Add bbs field.
22296 (execute_sm_if_changed_flag_set): Pass BBS.
22297 (execute_sm): Update.
22299 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22301 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22304 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22306 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22307 (peephole2): New peephole2 to emit the above.
22308 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22310 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22312 * config/aarch64/aarch64.c (define_peephole2 above
22313 *sub_<shift>_<mode>): New peephole.
22315 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22317 * config/i386/i386.c (make_resolver_func): Update.
22318 * Makefile.in: Add profile-count.h and profile-count.o
22319 * auto-profile.c (afdo_indirect_call): Update to new API.
22320 (afdo_set_bb_count): Update.
22321 (afdo_propagate_edge): Update.
22322 (afdo_propagate_circuit): Update.
22323 (afdo_calculate_branch_prob): Update.
22324 (afdo_annotate_cfg): Update.
22325 * basic-block.h: Include profile-count.h
22326 (struct edge_def): Turn count to profile_count.
22327 (struct basic_block_def): Likewie.
22328 (REG_BR_PROB_BASE): Move to profile-count.h
22329 (RDIV): Move to profile-count.h
22330 * bb-reorder.c (max_entry_count): Turn to profile_count.
22331 (find_traces): Update.
22332 (rotate_loop):Update.
22333 (connect_traces):Update.
22334 (sanitize_hot_paths):Update.
22335 * bt-load.c (migrate_btr_defs): Update.
22336 * cfg.c (RDIV): Remove.
22337 (init_flow): Use alloc_block.
22338 (alloc_block): Uninitialize count.
22339 (unchecked_make_edge): Uninitialize count.
22340 (check_bb_profile): Update.
22341 (dump_edge_info): Update.
22342 (dump_bb_info): Update.
22343 (update_bb_profile_for_threading): Update.
22344 (scale_bbs_frequencies_int): Update.
22345 (scale_bbs_frequencies_gcov_type): Update.
22346 (scale_bbs_frequencies_profile_count): New.
22347 * cfg.h (update_bb_profile_for_threading): Update.
22348 (scale_bbs_frequencies_profile_count): Declare.
22349 * cfgbuild.c (compute_outgoing_frequencies): Update.
22350 (find_many_sub_basic_blocks): Update.
22351 * cfgcleanup.c (try_forward_edges): Update.
22352 (try_crossjump_to_edge): Update.
22353 * cfgexpand.c (expand_gimple_tailcall): Update.
22354 (construct_exit_block): Update.
22355 * cfghooks.c (verify_flow_info): Update.
22356 (dump_bb_for_graph): Update.
22357 (split_edge): Update.
22358 (make_forwarder_block): Update.
22359 (duplicate_block): Update.
22360 (account_profile_record): Update.
22361 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22362 (get_estimated_loop_iterations): Update.
22363 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22364 (single_likely_exit): Update.
22365 * cfgloopmanip.c (scale_loop_profile): Update.
22367 (set_zero_probability): Update.
22368 (lv_adjust_loop_entry_edge): Update.
22369 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22370 (purge_dead_edges): Update.
22371 (rtl_account_profile_record): Update.
22372 * cgraph.c (cgraph_node::create): Uninitialize count.
22373 (symbol_table::create_edge): Uninitialize count.
22374 (cgraph_update_edges_for_call_stmt_node): Update.
22375 (cgraph_edge::dump_edge_flags): Update.
22376 (cgraph_node::dump): Update.
22377 (cgraph_edge::maybe_hot_p): Update.
22378 * cgraph.h: Include profile-count.h
22379 (create_clone), create_edge, create_indirect_edge): Update.
22380 (cgraph_node): Turn count to profile_count.
22381 (cgraph_edge0: Likewise.
22382 (make_speculative, clone): Update.
22383 (create_edge): Update.
22384 (init_lowered_empty_function): Update.
22385 * cgraphclones.c (cgraph_edge::clone): Update.
22386 (duplicate_thunk_for_node): Update.
22387 (cgraph_node::create_clone): Update.
22388 * cgraphunit.c (cgraph_node::analyze): Update.
22389 (cgraph_node::expand_thunk): Update.
22390 * final.c (dump_basic_block_info): Update.
22391 * gimple-streamer-in.c (input_bb): Update.
22392 * gimple-streamer-out.c (output_bb): Update.
22393 * graphite.c (print_global_statistics): Update.
22394 (print_graphite_scop_statistics): Update.
22395 * hsa-brig.c: Include basic-block.h.
22396 * hsa-dump.c: Include basic-block.h.
22397 * hsa-gen.c (T sum_slice): Update.
22398 (convert_switch_statements):Update.
22399 * hsa-regalloc.c: Include basic-block.h.
22400 * ipa-chkp.c (chkp_produce_thunks): Update.
22401 * ipa-cp.c (struct caller_statistics): Update.
22402 (init_caller_stats): Update.
22403 (gather_caller_stats): Update.
22404 (ipcp_cloning_candidate_p): Update.
22405 (good_cloning_opportunity_p): Update.
22406 (get_info_about_necessary_edges): Update.
22407 (dump_profile_updates): Update.
22408 (update_profiling_info): Update.
22409 (update_specialized_profile): Update.
22410 (perhaps_add_new_callers): Update.
22411 (decide_about_value): Update.
22412 (ipa_cp_c_finalize): Update.
22413 * ipa-devirt.c (struct odr_type_warn_count): Update.
22414 (struct decl_warn_count): Update.
22415 (struct final_warning_record): Update.
22416 (possible_polymorphic_call_targets): Update.
22417 (ipa_devirt): Update.
22418 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22419 * ipa-icf.c (sem_function::merge): Update.
22420 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22421 * ipa-inline.c (compute_uninlined_call_time): Update.
22422 (compute_inlined_call_time): Update.
22423 (want_inline_small_function_p): Update.
22424 (want_inline_self_recursive_call_p): Update.
22425 (edge_badness): Update.
22426 (lookup_recursive_calls): Update.
22427 (recursive_inlining): Update.
22428 (inline_small_functions): Update.
22429 (dump_overall_stats): Update.
22430 (dump_inline_stats): Update.
22431 * ipa-profile.c (ipa_profile_generate_summary): Update.
22432 (ipa_propagate_frequency): Update.
22433 (ipa_profile): Update.
22434 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22435 * ipa-utils.c (ipa_merge_profiles): Update.
22436 * loop-doloop.c (doloop_modify): Update.
22437 * loop-unroll.c (report_unroll): Update.
22438 (unroll_loop_runtime_iterations): Update.
22439 * lto-cgraph.c (lto_output_edge): Update.
22440 (lto_output_node): Update.
22441 (input_node): Update.
22442 (input_edge): Update.
22443 (merge_profile_summaries): Update.
22444 * lto-streamer-in.c (input_cfg): Update.
22445 * lto-streamer-out.c (output_cfg): Update.
22446 * mcf.c (create_fixup_graph): Update.
22447 (adjust_cfg_counts): Update.
22448 (sum_edge_counts): Update.
22449 * modulo-sched.c (sms_schedule): Update.
22450 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22451 * predict.c (maybe_hot_count_p): Update.
22452 (probably_never_executed): Update.
22453 (dump_prediction): Update.
22454 (combine_predictions_for_bb): Update.
22455 (propagate_freq): Update.
22456 (handle_missing_profiles): Update.
22457 (counts_to_freqs): Update.
22458 (rebuild_frequencies): Update.
22459 (force_edge_cold): Update.
22460 * predict.h: Include profile-count.h
22461 (maybe_hot_count_p, counts_to_freqs): UPdate.
22462 * print-rtl-function.c: Do not include cfg.h
22463 * print-rtl.c: Include basic-block.h
22464 * profile-count.c: New file.
22465 * profile-count.h: New file.
22466 * profile.c (is_edge_inconsistent): Update.
22467 (correct_negative_edge_counts): Update.
22468 (is_inconsistent): Update.
22469 (set_bb_counts): Update.
22470 (read_profile_edge_counts): Update.
22471 (compute_frequency_overlap): Update.
22472 (compute_branch_probabilities): Update; Initialize and deinitialize
22474 (branch_prob): Update.
22475 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
22476 (edge_gcov_count): New.
22477 (bb_gcov_count): New.
22478 * shrink-wrap.c (try_shrink_wrapping): Update.
22479 * tracer.c (better_p): Update.
22480 * trans-mem.c (expand_transaction): Update.
22481 (ipa_tm_insert_irr_call): Update.
22482 (ipa_tm_insert_gettmclone_call): Update.
22483 * tree-call-cdce.c: Update.
22484 * tree-cfg.c (gimple_duplicate_sese_region): Update.
22485 (gimple_duplicate_sese_tail): Update.
22486 (gimple_account_profile_record): Update.
22487 (execute_fixup_cfg): Update.
22488 * tree-inline.c (copy_bb): Update.
22489 (copy_edges_for_bb): Update.
22490 (initialize_cfun): Update.
22491 (freqs_to_counts): Update.
22492 (copy_cfg_body): Update.
22493 (expand_call_inline): Update.
22494 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22495 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
22496 (try_unroll_loop_completely): Update.
22497 (try_peel_loop): Update.
22498 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22499 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
22500 * tree-ssa-loop-split.c (connect_loops): Update.
22501 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
22502 * tree-ssa-reassoc.c (branch_fixup): Update.
22503 * tree-ssa-tail-merge.c (replace_block_by): Update.
22504 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
22505 (compute_path_counts): Update.
22506 (update_profile): Update.
22507 (recompute_probabilities): Update.
22508 (update_joiner_offpath_counts): Update.
22509 (estimated_freqs_path): Update.
22510 (freqs_to_counts_path): Update.
22511 (clear_counts_path): Update.
22512 (ssa_fix_duplicate_block_edges): Update.
22513 (duplicate_thread_path): Update.
22514 * tree-switch-conversion.c (case_bit_test_cmp): Update.
22515 (struct switch_conv_info): Update.
22516 * tree-tailcall.c (decrease_profile): Update.
22517 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22518 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22519 * value-prof.c (check_counter): Update.
22520 (gimple_divmod_fixed_value): Update.
22521 (gimple_mod_pow2): Update.
22522 (gimple_mod_subtract): Update.
22523 (gimple_ic_transform): Update.
22524 (gimple_stringop_fixed_value): Update.
22525 * value-prof.h (gimple_ic): Update.
22527 2017-06-02 Carl Love <cel@us.ibm.com>
22529 * config/rs6000/rs6000-c: Add support for built-in functions
22530 vector double vec_doublee (vector signed int);
22531 vector double vec_doublee (vector unsigned int);
22532 vector double vec_doublee (vector float);
22533 vector double vec_doubleh (vector signed int);
22534 vector double vec_doubleh (vector unsigned int);
22535 vector double vec_doubleh (vector float);
22536 vector double vec_doublel (vector signed int);
22537 vector double vec_doublel (vector unsigned int);
22538 vector double vec_doublel (vector float);
22539 vector double vec_doubleo (vector signed int);
22540 vector double vec_doubleo (vector unsigned int);
22541 vector double vec_doubleo (vector float);.
22542 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
22543 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
22545 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
22546 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
22547 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
22549 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
22550 vec_doublel, vec_doubleh.
22551 * doc/extend.texi: Update the built-in documentation file for the
22552 new built-in functions.
22554 2017-06-02 David Malcolm <dmalcolm@redhat.com>
22557 * ipa-inline-analysis.c (free_growth_caches): Set
22558 edge_removal_hook_holder to NULL after removing it.
22560 2017-06-02 Sudakshina Das <sudi.das@arm.com>
22562 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
22563 comparision with zero.
22565 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22566 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22567 for early expansion of vec_min and vec_max builtins.
22568 (builtin_function_type): Add min/max unsigned variants to those
22569 identified as having unsigned arguments.
22571 2017-06-02 Olivier Hainque <hainque@adacore.com>
22573 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
22575 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22577 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
22578 Use VALL_F16 iterator rather than VALL.
22580 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22582 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
22583 Emit CBNZ inside loop when doing a strong exchange and comparing
22584 against zero. Generate the CC flags after the loop.
22586 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
22588 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
22589 (dl_section_ref): New.
22590 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
22591 On AIX, append an expression to subtract the size of the
22592 section length to dl_section_ref.
22594 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22596 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22597 for early expansion of vector absolute builtins.
22599 2017-06-02 Richard Biener <rguenther@suse.de>
22601 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
22602 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
22604 2017-06-02 Richard Biener <rguenther@suse.de>
22606 PR tree-optimization/80948
22607 * tree-tailcall.c (find_tail_calls): Track stmts to move in
22608 stmt order as well.
22610 2017-06-02 Richard Biener <rguenther@suse.de>
22612 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
22614 * tree-vect-stmts.c (process_use): Do not mark backedge defs
22615 for inductions as relevant.
22617 2017-06-02 Richard Biener <rguenther@suse.de>
22619 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
22620 (vectorizable_induction): ... this. Remove dead code.
22622 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
22624 * builtins. (expand_builtin_alloca): Remove second parameter and
22625 infer its value from the first parameter instead.
22626 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
22628 2017-06-02 Jakub Jelinek <jakub@redhat.com>
22630 PR rtl-optimization/80903
22631 * loop-doloop.c (add_test): Unshare sequence.
22633 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
22635 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
22637 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
22639 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
22641 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
22642 xlogue_layout::get_instance, logue_layout::xlogue_layout,
22643 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
22644 (xlogue_layout::get_stub_rtx): Make static.
22645 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
22646 (xlogue_layout::compute_stub_managed_regs): Rename to...
22647 (xlogue_layout::count_stub_managed_regs): ...this.
22648 (xlogue_layout::is_stub_managed_reg): New function.
22649 (xlogue_layout::m_stub_names): Rename to...
22650 (xlogue_layout::s_stub_names): ...this, make static.
22651 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
22652 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
22653 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
22654 xlogue_layout::s_stub_names): Instantiate statics.
22655 (stub_managed_regs): Remove.
22656 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
22657 (disable_call_ms2sysv_xlogues): Rename to...
22658 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
22659 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
22661 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
22662 change after reload_completed.
22663 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
22665 (ix86_expand_prologue): Likewise.
22666 (ix86_expand_epilogue): Likewise.
22667 (ix86_expand_split_stack_prologue): Likewise.
22668 (ix86_compute_frame_layout): Remove frame parameter ...
22669 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
22670 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
22672 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
22673 (ix86_frame): Move from here ...
22674 * config/i386/i386.h (ix86_frame): ... to here.
22675 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
22676 complete ix86_frame data structure instead. Remove some_ld_name.
22678 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22680 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
22681 symbols that hold a DECL_VALUE_EXPR.
22683 2017-06-01 Martin Jambor <mjambor@suse.cz>
22685 PR tree-optimization/80898
22686 * tree-sra.c (process_subtree_disqualification): Removed.
22687 (disqualify_candidate): Do not acll
22688 process_subtree_disqualification.
22689 (subtree_mark_written_and_enqueue): New function.
22690 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
22691 RHS has been disqualified and re-queue LHS if necessary. Apart
22692 from that, ignore disqualified RHS.
22694 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22696 * config/s390/s390.c (s390_emit_epilogue): Disable early return
22697 address fetch for z10 or later.
22699 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22701 * config/arc/arc.md (tst_movb): Add guard when splitting.
22703 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22705 * config/arc/arc.c (arc_can_eliminate): Test against
22706 arc_frame_pointer_needed.
22708 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22710 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
22711 to prevent store reordering.
22712 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
22713 (type): Add block type.
22714 (stack_tie): Define special instruction to be used in
22717 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22719 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
22720 constraint. It is not valid for the pattern.
22721 (noncommutative_binary_comparison): Likewise.
22723 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22725 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
22728 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22730 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
22731 be used by the reg-alloc.
22733 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22735 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
22736 reg-alloc when having mul64 or mul32x16 instructions.
22737 (mulsidi3): Likewise.
22738 (umulsidi3): Likewise.
22739 (mulsi32x16): New pattern.
22740 (mulsi64): Likewise.
22741 (mulsidi64): Likewise.
22742 (umulsidi64): Likewise.
22743 (MUL32x16_REG): Define.
22744 (mul64_600): Use MUL32x16_REG.
22745 (mac64_600): Likewise.
22746 (umul64_600): Likewise.
22747 (umac64_600): Likewise.
22749 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22751 * config/arc/arc.md (mulsi3_700): Make it commutative.
22753 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
22755 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
22757 (*sign_extendsidi2_insn): Likewise for movstosw.
22759 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22761 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
22762 the type of the input discriminant value. Convert the
22763 discriminant value of signedness vary.
22765 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
22767 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
22768 Add to -Wall section.
22770 2017-06-01 Richard Biener <rguenther@suse.de>
22772 PR middle-end/66313
22773 * fold-const.c (fold_plusminus_mult_expr): If the factored
22774 factor may be zero use a wrapping type for the inner operation.
22775 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
22776 and handle moved defs.
22777 (process_assignment): Properly guard the unary op case. Return a
22778 tri-state indicating that moving the stmt before the call may allow
22779 to continue. Pass through to_move.
22780 (find_tail_calls): Handle moving unrelated defs before
22783 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
22786 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
22787 splitter result in the canonical way.
22789 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
22791 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
22792 also for 32bit target. Update insn attributes.
22793 (zero-extendsidi2 splitter): Allow all registers for operand 1.
22795 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
22797 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
22798 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
22799 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
22800 (_mm_maskz_min_ss): New intrinsics.
22802 2017-05-31 Martin Liska <mliska@suse.cz>
22804 * tree-vect-loop.c (vect_create_epilog_for_reduction):
22805 Change comment style to one we normally use.
22806 (vectorizable_reduction): Likewise.
22807 (vectorizable_induction): Likewise.
22808 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
22809 (vectorizable_call): Likewise.
22810 (vectorizable_simd_clone_call): Likewise.
22811 (vectorizable_conversion): Likewise.
22812 (vectorizable_assignment): Likewise.
22813 (vectorizable_shift): Likewise.
22814 (vectorizable_operation): Likewise.
22815 (vectorizable_store): Likewise.
22816 (vectorizable_load): Likewise.
22817 * tree-vectorizer.h: Likewise.
22819 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
22821 * passes.c (emergency_dump_function): New.
22822 * tree-pass.h (emergency_dump_function): Declare.
22823 * plugin.c (plugins_internal_error_function): Remove.
22824 * plugin.h (plugins_internal_error_function): Remove declaration.
22825 * toplev.c (internal_error_function): New static function. Use it...
22826 (general_init): ...here.
22828 2017-05-31 Graham Markall <graham.markall@embecosm.com>
22830 * config/arc/arc.c (arc_print_operand): Handle constant operands.
22831 (arc_rtx_costs): Add costs for new patterns.
22832 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
22833 * config/arc/predicates.md: Add _1_2_3_operand predicate.
22835 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22837 * tree-ssa-strlen.c (get_next_strinfo): New function.
22838 (get_stridx_plus_constant): Use it.
22839 (zero_length_string): Likewise.
22840 (adjust_related_strinfos): Likewise.
22841 (adjust_last_stmt): Likewise.
22843 2017-05-31 Richard Biener <rguenther@suse.de>
22846 * config/i386/i386.c (ix86_expand_builtin): Remove assert
22847 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
22849 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22851 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
22852 loop_vinfo argument and use of dependence distance vectors.
22853 Check instead whether the two references differ only in their
22854 initial value and assume that they have the same alignment if the
22855 difference is a multiple of the vector alignment.
22856 (vect_analyze_data_refs_alignment): Update call accordingly.
22858 2017-05-31 Martin Liska <mliska@suse.cz>
22861 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
22863 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22865 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
22866 (create_intersect_range_checks): Move from ...
22867 * tree-data-ref.c (create_intersect_range_checks_index)
22868 (create_intersect_range_checks): ... to here.
22869 (create_runtime_alias_checks): New function factored from ...
22870 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
22871 here. Call above function.
22872 * tree-data-ref.h (create_runtime_alias_checks): New function.
22874 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22876 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
22877 segment length for dr_b and compute it in wide_int.
22879 2017-05-31 Richard Biener <rguenther@suse.de>
22881 PR tree-optimization/80906
22882 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
22883 and pass through iv_map.
22884 (copy_bb_and_scalar_dependences): Adjust.
22885 (translate_pending_phi_nodes): Likewise.
22886 (copy_loop_close_phi_args): Handle code-generating IVs instead
22889 2017-05-30 David Malcolm <dmalcolm@redhat.com>
22891 * diagnostic-color.c (color_dict): Add "type-diff".
22892 (parse_gcc_colors): Update comment.
22893 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
22894 -fdiagnostics-show-template-tree and -fno-elide-type.
22895 (GCC_COLORS): Add type-diff to example.
22897 (-fdiagnostics-show-template-tree): New.
22898 (-fno-elide-type): New.
22899 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
22900 the pp_format_decoder callback. Call any m_format_postprocessor's
22902 (pretty_printer::pretty_printer): Initialize
22903 m_format_postprocessor.
22904 (pretty_printer::~pretty_printer): Delete any
22905 m_format_postprocessor.
22906 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
22907 (class format_postprocessor): New class.
22908 (struct pretty_printer::format_decoder): Document the new parameters.
22909 (struct pretty_printer::m_format_postprocessor): New field.
22910 * tree-diagnostic.c (default_tree_printer): Update for new
22911 bool and const char ** params.
22912 * tree-diagnostic.h (default_tree_printer): Likewise.
22914 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
22916 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
22917 (lwa_operand): Delete rs6000_gen_cell_microcode test.
22918 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22919 rs6000_gen_cell_microcode code.
22920 (rs6000_final_prescan_insn): Delete.
22921 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
22922 "warn-cell-microcode" entries.
22923 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
22924 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
22925 throughout. Change cc_reg_not_micro_cr0_operand to
22926 cc_reg_not_cr0_operand throughout.
22927 (*extendhi<mode>2_noload): Delete.
22928 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
22929 (mwarn-cell-microcode): Delete.
22930 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
22931 -mgen-cell-microcode and -mwarn-cell-microcode.
22933 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
22936 * config/i386/constraints.md (Yd): New constraint.
22938 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
22939 and (?Yd, r) alternatives. Update insn attributes.
22940 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
22941 and (?*Yd, r) alternatives. Update insn attributes.
22942 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
22944 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
22946 * gimplify.c (gimplify_modify_expr): Don't create a
22947 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
22950 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
22952 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
22954 2017-05-30 Richard Biener <rguenther@suse.de>
22956 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
22957 and reduc_def fields.
22958 (STMT_VINFO_REDUC_TYPE): New define.
22959 (STMT_VINFO_REDUC_DEF): Likewise.
22960 (vect_force_simple_reduction): Adjust prototype.
22961 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
22962 (vect_is_simple_reduction): Remove check_reduction argument.
22963 (vect_force_simple_reduction): Adjust and set
22964 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22965 (vectorizable_reduction): Do not re-do reduction analysis
22966 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22967 * tree-parloops.c (gather_scalar_reductions): Adjust.
22969 2017-05-30 Richard Biener <rguenther@suse.de>
22971 PR middle-end/80901
22972 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
22975 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
22977 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
22978 Introduce unknown_misalignment parameter and remove vf.
22979 (vect_peeling_hash_get_lowest_cost):
22980 Pass unknown_misalignment parameter.
22981 (vect_enhance_data_refs_alignment):
22982 Fix unsupportable data ref treatment.
22984 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22986 * tree-vect-data-refs.c (vect_get_data_access_cost):
22987 Workaround for SLP handling.
22988 (vect_enhance_data_refs_alignment):
22989 Compute costs for doing no peeling at all, compare to the best
22990 peeling costs so far and avoid peeling if cheaper.
22992 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22994 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
22995 Return peeling info and set costs to zero for unlimited cost
22997 (vect_enhance_data_refs_alignment): Also inspect all datarefs
22998 with unknown misalignment. Compute and costs for unknown
22999 misalignment, compare them to the costs for known misalignment
23000 and choose the cheapest for peeling.
23002 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23004 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
23005 (vect_get_peeling_costs_all_drs): Create function.
23006 (vect_peeling_hash_get_lowest_cost):
23007 Use vect_get_peeling_costs_all_drs.
23008 (vect_peeling_supportable): Create function.
23009 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
23011 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23013 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
23014 DR_HAS_NEGATIVE_STEP.
23015 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
23016 (vect_enhance_data_refs_alignment): Use.
23017 (vect_duplicate_ssa_name_ptr_info): Use.
23018 * tree-vectorizer.h (dr_misalignment): Use.
23019 (known_alignment_for_access_p): Use.
23021 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23024 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
23026 (has_section_name): New function.
23028 2017-05-30 Martin Liska <mliska@suse.cz>
23031 * auto-profile.c (get_function_decl_from_block): Fix
23034 2017-05-30 Richard Biener <rguenther@suse.de>
23036 PR middle-end/80876
23037 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
23039 2017-05-30 Martin Liska <mliska@suse.cz>
23041 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
23042 * dumpfile.h (struct dump_file_info): Remove ctors.
23044 2017-05-30 Martin Liska <mliska@suse.cz>
23046 * predict.def: Fix GNU coding style.
23048 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23050 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23051 Mark 'to' argument with ATTRIBUTE_UNUSED.
23053 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23055 * config/xtensa/xtensa.c (xtensa_emit_call): Use
23056 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
23057 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
23060 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
23062 * doc/install.texi (Options specification): Restore entry of
23063 --enable-sjlj-exceptions.
23065 2017-05-27 Michael Eager <eager@eagercon.com>
23068 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
23070 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
23072 * config/microblaze/microblaze.h
23073 (FIXED_REGISTERS): Update in macro.
23074 (CALL_USED_REGISTERS): Update in macro.
23076 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
23078 * doc/install.texi: Add links to macOS binary distributions.
23080 2017-05-27 Jakub Jelinek <jakub@redhat.com>
23084 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23086 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23088 2017-05-26 Martin Liska <mliska@suse.cz>
23090 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
23092 2017-05-26 Martin Liska <mliska@suse.cz>
23094 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
23095 always leading ';; '.
23096 (dump_bb_info): Likewise.
23097 (brief_dump_cfg): Likewise.
23098 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
23099 * dumpfile.c: Remove usage of TDF_VERBOSE.
23100 * dumpfile.h (enum dump_kind): Likewise.
23101 (dump_gimple_bb_header): Do not use TDF_COMMENT.
23102 * print-tree.c (debug_verbose): Remove.
23103 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
23104 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
23105 * tree-diagnostic.c (default_tree_printer): Replace
23106 TDF_DIAGNOSTIC with TDF_SLIM.
23108 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23110 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
23111 in parameter loop, rather than loop_vinfo.
23112 (create_intersect_range_checks): Ditto.
23113 (vect_create_cond_for_alias_checks): Update call to above functions.
23115 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23117 PR tree-optimization/80815
23118 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
23119 for merging runtime alias checks. Handle negative DR_STEPs.
23121 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23123 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
23125 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
23126 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
23127 out code pruning runtime alias checks.
23128 * tree-data-ref.c (prune_runtime_alias_test_list): New function
23129 factored out from above.
23130 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23132 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23134 (prune_runtime_alias_test_list): New decalaration.
23136 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23138 * tree-vect-data-refs.c (compare_tree): Rename and move ...
23139 * tree-data-ref.c (data_ref_compare_tree): ... to here.
23140 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
23141 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
23142 (operator==, comp_dr_with_seg_len_pair): Ditto.
23143 (vect_prune_runtime_alias_test_list): Ditto.
23145 2017-05-26 Martin Liska <mliska@suse.cz>
23148 * params.def: Bound partial-inlining-entry-probability param.
23150 2017-05-26 Marek Polacek <polacek@redhat.com>
23153 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
23156 2017-05-26 Richard Biener <rguenther@suse.de>
23158 PR tree-optimization/80842
23159 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
23162 2017-05-26 Richard Biener <rguenther@suse.de>
23164 PR tree-optimization/80844
23165 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
23167 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
23169 * doc/md.texi (Machine Constraints): Update x86 family
23170 machine constraints section to match 'config/i386/constraints.md'.
23172 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
23174 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
23176 2017-05-25 Nathan Sidwell <nathan@acm.org>
23178 * doc/invoke.texi (--enable-languages): Update documentation.
23180 2017-05-25 Martin Liska <mliska@suse.cz>
23182 * dumpfile.c: Add TDF_FOLDING.
23183 * dumpfile.h (enum dump_kind): Likewise.
23184 * genmatch.c (dt_simplify::gen_1): Use it.
23186 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23188 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23190 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23192 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23193 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23195 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23197 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23198 * match.pd (X == C): Rewrite it here.
23199 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23200 with_certain_nonzero_bits2): New predicates.
23201 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23203 2017-05-24 Nathan Sidwell <nathan@acm.org>
23205 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23208 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23211 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23213 * config/powerpcspe: New port. Files are copied from the rs6000
23214 port, with "rs6000" in filenames replaced by "powerpcspe".
23216 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23218 PR rtl-optimization/80754
23219 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23221 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23223 * config/sparc/sparc.md (length): Return the correct value for -mflat
23224 sibcalls to match output_sibcall.
23226 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23230 * config/rs6000/rs6000.c (struct machine_function): Add new field
23232 (rs6000_get_separate_components): Init that field, use it.
23233 (rs6000_components_for_bb): Use the field.
23235 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23237 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23239 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23241 PR middle-end/80823
23242 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23244 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23247 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23248 address against address_operand predicate.
23249 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23251 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23253 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23254 parameters passed indirectly.
23256 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23258 * config/i386/i386.md (*movdi_internal): Remove SSE4
23259 alternative 18 (?r, *v). Update insn attributes.
23260 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23261 Update insn attributes.
23262 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23263 Update insn attributes.
23264 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23265 alternative 1 (r, v). Remove isa attribute.
23266 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23267 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23268 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23270 2017-05-23 Tom de Vries <tom@codesourcery.com>
23272 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23275 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23277 * cgraphunit.c (symbol_table::process_new_functions): Update.
23278 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23279 (inline_generate_summary): Rename to ...
23280 (ipa_fn_summary_generate): ... this one.
23281 (inline_read_summary): Rename to ...
23282 (ipa_fn_summary_read): ... this one.
23283 (inline_write_summary): Rename to ...
23284 (ipa_fn_summary_write): ... this one.
23285 (inline_free_summary): Rename to ...
23286 (ipa_free_fn_summary): ... this one.
23287 (pass_data_local_fn_summary, pass_local_fn_summary,
23288 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23289 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23290 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23291 make_pass_ipa_fn_summary): New.
23292 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23293 inline_write_summary, inline_free_summary): Remove.
23294 (ipa_free_fn_summary) : New.
23295 * ipa-inline.c (ipa_inline): Update.
23296 (pass_ipa_inline): Do not generate summaries.
23297 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23299 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23300 and add pass_ipa_fn_summary.
23301 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23303 (make_pass_inline_parameters): Remove.
23305 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23307 * omp-low.c (struct omp_context): Remove "default_kind" member.
23310 * omp-offload.c (execute_oacc_device_lower): Remove the
23311 parallelism dimensions function attributes for unparallelized
23312 OpenACC kernels constructs.
23314 2017-05-23 Martin Liska <mliska@suse.cz>
23316 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23318 (cgraph_edge::make_speculative): Likewise.
23319 (cgraph_edge::resolve_speculation): Likewise.
23320 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23321 (cgraph_node::dump): Likewise.
23322 * cgraph.h: Likewise.
23323 * cgraphunit.c (analyze_functions): Likewise.
23324 (symbol_table::compile): Likewise.
23325 * ipa-cp.c (print_all_lattices): Likewise.
23326 (determine_versionability): Likewise.
23327 (initialize_node_lattices): Likewise.
23328 (ipcp_verify_propagated_values): Likewise.
23329 (estimate_local_effects): Likewise.
23330 (update_profiling_info): Likewise.
23331 (create_specialized_node): Likewise.
23332 (perhaps_add_new_callers): Likewise.
23333 (decide_about_value): Likewise.
23334 (decide_whether_version_node): Likewise.
23335 (identify_dead_nodes): Likewise.
23336 (ipcp_store_bits_results): Likewise.
23337 * ipa-devirt.c (dump_targets): Likewise.
23338 (ipa_devirt): Likewise.
23339 * ipa-icf.c (sem_item::dump): Likewise.
23340 (sem_function::equals): Likewise.
23341 (sem_variable::equals): Likewise.
23342 (sem_item_optimizer::read_section): Likewise.
23343 (sem_item_optimizer::execute): Likewise.
23344 (congruence_class::dump): Likewise.
23345 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23346 (dump_inline_summary): Likewise.
23347 (estimate_node_size_and_time): Likewise.
23348 (inline_analyze_function): Likewise.
23349 * ipa-inline-transform.c (inline_call): Likewise.
23350 * ipa-inline.c (report_inline_failed_reason): Likewise.
23351 (want_early_inline_function_p): Likewise.
23352 (edge_badness): Likewise.
23353 (update_edge_key): Likewise.
23354 (inline_small_functions): Likewise.
23355 * ipa-profile.c (ipa_profile): Likewise.
23356 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23357 (ipa_make_edge_direct_to_target): Likewise.
23358 (remove_described_reference): Likewise.
23359 (ipa_impossible_devirt_target): Likewise.
23360 (propagate_controlled_uses): Likewise.
23361 (ipa_print_node_params): Likewise.
23362 (ipcp_transform_function): Likewise.
23363 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23364 (propagate_pure_const): Likewise.
23365 * ipa-reference.c (generate_summary): Likewise.
23366 (read_write_all_from_decl): Likewise.
23367 (propagate): Likewise.
23368 (ipa_reference_read_optimization_summary): Likewise.
23369 * ipa-utils.c (ipa_merge_profiles): Likewise.
23370 * ipa.c (walk_polymorphic_call_targets): Likewise.
23371 (symbol_table::remove_unreachable_nodes): Likewise.
23372 (ipa_single_use): Likewise.
23373 * passes.c (execute_todo): Likewise.
23374 * predict.c (drop_profile): Likewise.
23375 * symtab.c (symtab_node::get_dump_name): New function.
23376 (symtab_node::dump_name): Likewise.
23377 (symtab_node::dump_asm_name): Likewise.
23378 (symtab_node::dump_references): Likewise.
23379 (symtab_node::dump_referring): Likewise.
23380 (symtab_node::dump_base): Likewise.
23381 (symtab_node::debug_symtab): Likewise.
23382 * tree-sra.c (convert_callers_for_node): Likewise.
23383 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23384 * value-prof.c (init_node_map): Likewise.
23386 2017-05-23 Martin Liska <mliska@suse.cz>
23388 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23389 and symtab_node::debug_symtab to symbol_table::debug.
23390 * cgraphunit.c (analyze_functions): Use the renamed function.
23391 (symbol_table::compile): Likewise.
23392 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23393 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23394 * passes.c (execute_todo): Likewise.
23395 * symtab.c (symbol_table::dump): New function.
23396 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23398 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23400 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23401 that nonconst implies exec.
23403 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23405 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23406 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23407 (inline_edge_summary_vec): Turn into ...
23408 (ipa_call_summaries): ... this one.
23409 (redirect_to_unreachable, edge_set_predicate,
23410 evaluate_properties_for_edge, inline_summary_alloc,
23411 reset_ipa_call_summary, reset_inline_summary,
23412 inline_summary_t::duplicate): Update.
23413 (inline_edge_duplication_hook): Turn to ...
23414 (ipa_call_summary_t::duplicate): ... this one.
23415 (inline_edge_removal_hook): Turn to ...
23416 (ipa_call_summary_t::remove): ... this one.
23417 (dump_inline_edge_summary): Turn to ...
23418 (dump_ipa_call_summary): ... this one.
23419 (estimate_function_body_sizes): Update.
23420 (inline_update_callee_summaries): Update.
23421 (remap_edge_change_prob): Update.
23422 (remap_edge_summaries): Update.
23423 (inline_merge_summary): Update.
23424 (do_estimate_edge_time): Update.
23425 (inline_generate_summary): Update.
23426 (inline_read_section): Update.
23427 (inline_read_summary): Update.
23428 (inline_free_summary): Update.
23429 * ipa-inline.c (can_inline_edge_p): Update.
23430 (compute_inlined_call_time): Update.
23431 (want_inline_small_function_p): Update.
23432 (edge_badness): Update.
23433 (early_inliner): Update.
23434 * ipa-inline.h (inline_edge_summary): Turn to ...
23435 (ipa_call_summary): ... this one.
23436 (ipa_call_summary_t): New class.
23437 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23438 (ipa_call_summaries): New.
23439 (inline_edge_summary): Remove.
23440 (estimate_edge_growth): Update.
23441 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23442 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23443 * ipa-split.c (execute_split_functions): Update.
23444 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23446 2017-05-23 Tom de Vries <tom@codesourcery.com>
23448 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23449 attributes): Document rdrand effective target.
23451 2017-05-23 Tom de Vries <tom@codesourcery.com>
23453 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23454 attributes): Sort alphabetically.
23456 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23458 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23460 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23463 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23464 V2DF/V2DI splat into two separate patterns, one that handles
23465 registers, and the other that only handles memory. Drop support
23466 for splatting from a GPR on ISA 2.07 and then splitting the
23467 splat into direct move and splat.
23468 (vsx_splat_<mode>_reg): Likewise.
23469 (vsx_splat_<mode>_mem): Likewise.
23471 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
23473 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
23475 2017-05-22 Jakub Jelinek <jakub@redhat.com>
23477 PR middle-end/80809
23478 * omp-low.c (finish_taskreg_remap): New function.
23479 (finish_taskreg_scan): If unit size of ctx->record_type
23480 is non-constant, unshare the size expression and replace
23481 decls in it with possible outer var refs.
23483 PR middle-end/80809
23484 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
23485 GOVD_SHARED rather than GOVD_PRIVATE with it.
23486 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
23487 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
23489 PR middle-end/80853
23490 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
23491 as last argument to build_outer_var_ref for pointer bases of array
23492 section reductions.
23494 2017-05-19 Martin Sebor <msebor@redhat.com>
23496 * print-tree.c (print_node): Print DECL_READ_P flag.
23498 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23500 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
23501 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
23502 * cgraph.c: Likewise.
23503 * cgraphunit.c: Likewise.
23504 * gengtype.c: Likewise.
23505 * ipa-cp.c: Likewise.
23506 * ipa-devirt.c: Likewise.
23507 * ipa-icf.c: Likewise.
23508 * ipa-predicate.c: Likewise.
23509 * ipa-profile.c: Likewise.
23510 * ipa-prop.c: Likewise.
23511 * ipa-split.c: Likewise.
23513 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
23514 edge_predicate_pool, dump_inline_hints,
23515 inline_summary::account_size_time, redirect_to_unreachable,
23516 edge_set_predicate, set_hint_predicate,
23517 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
23518 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
23519 inline_summary_t::remove, remap_hint_predicate_after_duplication,
23520 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
23521 ipa_call_summary_t::remove, initialize_growth_caches,
23522 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
23523 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
23524 mark_modified, unmodified_parm_1, unmodified_parm,
23525 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
23526 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
23527 compute_bb_predicates, will_be_nonconstant_expr_predicate,
23528 will_be_nonconstant_predicate, record_modified_bb_info,
23529 get_minimal_bb, record_modified, param_change_prob,
23530 phi_result_unknown_predicate, predicate_for_phi_result,
23531 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
23532 estimate_function_body_sizes, compute_inline_parameters,
23533 compute_inline_parameters_for_curren, pass_data_inline_parameters,
23534 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
23535 inline_update_callee_summaries, remap_edge_change_prob,
23536 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
23537 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
23538 inline_analyze_function, inline_summary_t::insert,
23539 inline_generate_summary, read_ipa_call_summary, inline_read_section,
23540 inline_read_summary, write_ipa_call_summary, inline_write_summary,
23541 inline_free_summary): Move to ipa-fnsummary.h
23542 (predicate_t): Remove.
23543 * ipa-fnsummary.c: New file.
23544 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
23545 (enum inline_hints_vals, inline_hints, agg_position_info,
23546 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
23547 inline_summaries, ipa_call_summary, ipa_call_summary_t,
23548 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
23549 dump_inline_summary, dump_inline_hints, inline_generate_summary,
23550 inline_read_summary, inline_write_summary, inline_free_summary,
23551 inline_analyze_function, initialize_inline_failed,
23552 inline_merge_summary, inline_update_overall_summary,
23553 compute_inline_parameters): Move to ipa-fnsummary.h
23554 * ipa-fnsummary.h: New file.
23555 * ipa-inline-transform.h: Include ipa-inline.h.
23556 * ipa-inline.c: LIkewise.
23558 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23560 * ipa-inline.c (edge_badness): Use inlined_time instead of
23561 inline_summaries->get.
23563 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23565 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
23567 2017-05-22 Nathan Sidwell <nathan@acm.org>
23569 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
23570 (fdump-lang): Document 'raw' option.
23571 * dumpfile.h (TDI_tu): Delete.
23572 * dumpfile.c (dump_files): Remove translation-unit.
23573 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
23575 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
23577 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
23578 command option from $(AWK) call.
23579 * config/avr/genmultilib.awk: Simplify and rewrite so that it
23580 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
23581 [FORMAT]: Remove handling of variable.
23582 * config/avr/t-multilib: Regenerate.
23584 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23586 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
23588 (dump_inline_summary): Do not print self_time.
23589 (estimate_function_body_sizes): Do not set self_time.
23590 (compute_inline_parameters): Likewise.
23591 (inline_read_section, inline_write_summary): Do not stream self_time.
23592 * ipa-inline.h (inline_summary): Drop self_time.
23594 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23596 * ipa-inline-analysis.c (account_size_time): Rename to ...
23597 (inline_summary::account_size_time): ... this one.
23598 (reset_ipa_call_summary): Turn to ...
23599 (ipa_call_summary::reset): ... this one.
23600 (reset_inline_summary): Turn to ...
23601 (inline_summary::reset): ... this one.
23602 (inline_summary_t::remove): Update.
23603 (inline_summary_t::duplicate): Update.
23604 (ipa_call_summary_t::remove): Update.
23605 (dump_inline_summary): Update.
23606 (estimate_function_body_sizes): Update.
23607 (compute_inline_parameters): Update.
23608 (estimate_node_size_and_time): Update.
23609 (inline_merge_summary): Update.
23610 (inline_update_overall_summary): Update.
23611 (inline_read_section): Update.
23612 (inline_write_summary): Update.
23613 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
23614 add account_size_time and reset member functions.
23615 (ipa_call_summary): Add reset function.
23616 * ipa-predicate.h (predicate::operator &): Constify.
23618 2017-05-22 Richard Biener <rguenther@suse.de>
23620 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
23622 2017-05-19 Jason Merrill <jason@redhat.com>
23624 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
23626 2017-05-19 Marek Polacek <polacek@redhat.com>
23629 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
23630 TYPE_OVERFLOW_WRAPS checks.
23632 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
23634 * tree-core.h (enum omp_clause_default_kind): Add
23635 "OMP_CLAUSE_DEFAULT_PRESENT".
23636 * tree-pretty-print.c (dump_omp_clause): Handle it.
23637 * gimplify.c (enum gimplify_omp_var_data): Add
23638 "GOVD_MAP_FORCE_PRESENT".
23639 (gimplify_adjust_omp_clauses_1): Map it to
23640 "GOMP_MAP_FORCE_PRESENT".
23641 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
23643 * gimplify.c (oacc_default_clause): Clarify.
23645 2017-05-19 Nathan Sidwell <nathan@acm.org>
23647 LANG_HOOK_REGISTER_DUMPS
23648 * toplev.c (general_init): Call register dump lang hook.
23649 * doc/invoke.texi: Document -fdump-lang option family.
23650 * dumpfile.c (dump_files): Remove class dump here.
23651 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
23652 * dumpfile.h (tree_dump_index): Remove TDI_class.
23653 * langhooks-def.h (lhd_register_dumps): Declare.
23654 (LANG_HOOKS_REGISTER_DUMPS): Define.
23655 (LANG_HOOKS_INITIALIZER): Add it.
23656 * langhooks.c (lhd_register_dumps): Define.
23657 * langhooks.h (struct lang_hooks): Add register_dumps.
23659 2017-05-19 Nathan Sidwell <nathan@acm.org>
23661 * context.h (context::set_passes): New.
23662 * context.c (context::context): Do not create pass manager.
23663 * toplev.c (general_init): Create pass manager here.
23665 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
23667 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
23668 use this splitter if two add or or instructions would also work for
23669 the constant we want to generate.
23671 2017-05-19 Richard Biener <rguenther@suse.de>
23674 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
23675 predicate evaluation.
23677 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23679 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
23681 * ipa-inline.c (want_inline_small_function_p): Do not cast to
23684 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23686 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23687 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23688 (inline_edge_summary_vec): Turn into ...
23689 (ipa_call_summaries): ... this one.
23690 (redirect_to_unreachable, edge_set_predicate,
23691 evaluate_properties_for_edge, inline_summary_alloc,
23692 reset_ipa_call_summary, reset_inline_summary,
23693 inline_summary_t::duplicate): Update.
23694 (inline_edge_duplication_hook): Turn to ...
23695 (ipa_call_summary_t::duplicate): ... this one.
23696 (inline_edge_removal_hook): Turn to ...
23697 (ipa_call_summary_t::remove): ... this one.
23698 (dump_inline_edge_summary): Turn to ...
23699 (dump_ipa_call_summary): ... this one.
23700 (estimate_function_body_sizes): Update.
23701 (inline_update_callee_summaries): Update.
23702 (remap_edge_change_prob): Update.
23703 (remap_edge_summaries): Update.
23704 (inline_merge_summary): Update.
23705 (do_estimate_edge_time): Update.
23706 (inline_generate_summary): Update.
23707 (inline_read_section): Update.
23708 (inline_read_summary): Update.
23709 (inline_free_summary): Update.
23710 * ipa-inline.c (can_inline_edge_p): Update.
23711 (compute_inlined_call_time): Update.
23712 (want_inline_small_function_p): Update.
23713 (edge_badness): Update.
23714 (early_inliner): Update.
23715 * ipa-inline.h (inline_edge_summary): Turn to ...
23716 (ipa_call_summary): ... this one.
23717 (ipa_call_summary_t): New class.
23718 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23719 (ipa_call_summaries): New.
23720 (inline_edge_summary): Remove.
23721 (estimate_edge_growth): Update.
23722 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23723 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23724 * ipa-split.c (execute_split_functions): Update.
23725 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23727 2017-05-19 Richard Biener <rguenther@suse.de>
23729 PR middle-end/80764
23730 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
23732 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
23734 * config/rs6000/rs6000.c (struct machine_function): Add field
23735 fpr_is_wrapped_separately.
23736 (rs6000_get_separate_components): Use 64 components. Handle the
23737 new FPR components.
23738 (rs6000_components_for_bb): Handle the FPR components.
23739 (rs6000_emit_prologue_components): Handle the FPR components.
23740 (rs6000_emit_epilogue_components): Handle the FPR components.
23741 (rs6000_set_handled_components): Handle the FPR components.
23742 (rs6000_emit_prologue): Don't output prologue code for those FPRs
23743 that are already separately shrink-wrapped.
23744 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
23745 that are already separately shrink-wrapped.
23747 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23750 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
23753 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
23754 (define_peephole2 for Altivec d-form load): Add peepholes to catch
23755 cases where the register allocator uses a move and an offsettable
23756 memory operation to/from a FPR register on ISA 2.06/2.07.
23757 (define_peephole2 for Altivec d-form store): Likewise.
23759 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
23762 * config/i386/mmx.md (*mov<mode>_internal): Enable
23763 alternatives 11, 12, 13 and 14 also for 32bit targets.
23764 Remove alternatives 15, 16, 17 and 18.
23765 * config/i386/sse.md (vec_concatv2di): Change
23766 alternative (!x, *y) to (x, ?!*Yn).
23768 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
23770 * dumpfile.h (enum dump_kind): Remove stray comma.
23772 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23774 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
23775 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
23776 predicate::num_conditions
23777 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
23778 (CHANGED): turn into predicate::changed.
23779 (agg_position_info): Move to ipa-predicate.h
23780 (add_condition, predicate::add_clause, predicate::operator &=,
23781 predicate::or_with, predicate::evaluate, predicate::probability,
23782 dump_condition, dump_clause, predicate::dump,
23783 predicate::remap_after_duplication, predicate::remap_after_inlining,
23784 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
23785 (evaluate_conditions_for_known_args): Update.
23786 (set_cond_stmt_execution_predicate): Update.
23787 * ipa-inline.h: Include ipa-predicate.h
23788 (condition, inline_param_summary, conditions, agg_position_info,
23789 predicate): Move to ipa-predicate.h
23790 * ipa-predicate.c: New file.
23791 * ipa-predicate.h: New file.
23793 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
23795 * final.c (leaf_function_p): Check we are not in a sequence.
23797 2017-05-18 Martin Liska <mliska@suse.cz>
23799 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
23800 * dumpfile.c (dump_register): Use new enum dump_kind.
23801 (get_dump_file_name): Likewise.
23802 (dump_enable_all): Likewise.
23803 (dump_switch_p_1): Likewise.
23804 (enable_rtl_dump_file): Remove usage of TDF_RTL.
23805 * dumpfile.h (enum dump_kind): New enum type.
23806 (struct dump_file_info): Create constructor and
23807 format fields and comments.
23808 * passes.c (pass_manager::register_one_dump_file):
23810 * statistics.c (statistics_early_init): Likewise.
23811 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
23812 TDF_TREE with TDF_SLIM.
23813 (gather_memory_references_ref): Likewise.
23815 2017-05-18 Martin Liska <mliska@suse.cz>
23817 * vec.h (struct vnull): Use it.
23819 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23821 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
23822 (true_predicate, false_predicate, true_predicate_p,
23823 false_predicate_p): Remove.
23824 (single_cond_predicate, not_inlined_predicate): Turn to member function
23826 (add_condition): Update.
23827 (add_clause): Turn to...
23828 (predicate::add_clause): ... this one; update; allow passing NULL
23830 (and_predicates): Turn to ...
23831 (predicate::operator &=): ... this one.
23832 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
23833 (or_predicates): Turn to ...
23834 (predicate::or_with): ... this one.
23835 (evaluate_predicate): Turn to ...
23836 (predicate::evaluate): ... this one.
23837 (predicate_probability): Turn to ...
23838 (predicate::probability): ... this one.
23839 (dump_condition): Update.
23840 (dump_predicate): Turn to ...
23841 (predicate::dump): ... this one.
23842 (account_size_time): Update.
23843 (edge_set_predicate): Update.
23844 (set_hint_predicate): UPdate.
23845 (evaluate_conditions_for_known_args): Update.
23846 (evaluate_properties_for_edge): Update.
23847 (remap_predicate_after_duplication): Turn to...
23848 (predicate::remap_after_duplication): ... this one.
23849 (remap_hint_predicate_after_duplication): Update.
23850 (inline_summary_t::duplicate): UPdate.
23851 (dump_inline_edge_summary): Update.
23852 (dump_inline_summary): Update.
23853 (set_cond_stmt_execution_predicate): Update.
23854 (set_switch_stmt_execution_predicate): Update.
23855 (compute_bb_predicates): Update.
23856 (will_be_nonconstant_expr_predicate): Update.
23857 (will_be_nonconstant_predicate): Update.
23858 (phi_result_unknown_predicate): Update.
23859 (predicate_for_phi_result): Update.
23860 (array_index_predicate): Update.
23861 (estimate_function_body_sizes): Update.
23862 (estimate_node_size_and_time): Update.
23863 (estimate_ipcp_clone_size_and_time): Update.
23864 (remap_predicate): Rename to ...
23865 (predicate::remap_after_inlining): ... this one.
23866 (remap_hint_predicate): Update.
23867 (inline_merge_summary): Update.
23868 (inline_update_overall_summary): Update.
23869 (estimate_size_after_inlining): Update.
23870 (read_predicate): Rename to ...
23871 (predicate::stream_in): ... this one.
23872 (read_inline_edge_summary): Update.
23873 (write_predicate): Rename to ...
23874 (predicate::stream_out): ... this one.
23875 (write_inline_edge_summary): Update.
23876 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
23877 (clause_t): Turn to uint32_t
23878 (predicate): Turn to class; implement constructor and operators
23880 (size_time_entry): Update.
23881 (inline_summary): Update.
23882 (inline_edge_summary): Update.
23884 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
23886 * fold-const.c (fold_binary_loc): Move transformation...
23887 * match.pd (C - X CMP X): ... here.
23889 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
23891 * config/sparc/sparc.c (sparc_option_override): Set function
23892 alignment for -mcpu=niagara7 to 64 to match the I$ line.
23893 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
23895 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
23897 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
23899 2017-05-18 Marek Polacek <polacek@redhat.com>
23902 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
23903 (pass_ubsan::execute): Call gimple_assign_single_p instead of
23904 gimple_assign_load_p.
23906 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
23908 PR middle-end/80692
23909 * real.c (do_compare): Give decimal_do_compare preference over
23910 comparing just the signs.
23912 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
23914 * doc/md.texi (Canonicalization of Instructions): Describe the
23915 canonical form of instructions that inherently set a condition
23918 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
23920 PR middle-end/80775
23921 * tree-cfg.c: Move deletion of unreachable case statements to after
23922 the merging of consecutive case labels.
23924 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
23926 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
23927 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
23928 restoring of callee-saved registers.
23930 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
23932 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
23933 * config/visium/visium.c (single_set_and_flags): Likewise.
23934 * config/visium/visium.md (Substitutions): Likewise.
23936 2017-05-17 Martin Liska <mliska@suse.cz>
23938 * cfg.c: Introduce dump_flags_t type and
23939 use it instead of int type.
23941 * cfghooks.c: Likewise.
23942 * cfghooks.h (struct cfg_hooks): Likewise.
23943 * cfgrtl.c: Likewise.
23944 * cfgrtl.h: Likewise.
23945 * cgraph.c (cgraph_node::get_body): Likewise.
23946 * coretypes.h: Likewise.
23947 * domwalk.c: Likewise.
23948 * domwalk.h: Likewise.
23949 * dumpfile.c (struct dump_option_value_info): Likewise.
23950 (dump_enable_all): Likewise.
23951 (dump_switch_p_1): Likewise.
23952 (opt_info_switch_p): Likewise.
23953 * dumpfile.h (enum tree_dump_index): Likewise.
23954 (struct dump_file_info): Likewise.
23955 * genemit.c: Likewise.
23956 * generic-match-head.c: Likewise.
23957 * gengtype.c (open_base_files): Likewise.
23958 * gimple-pretty-print.c: Likewise.
23959 * gimple-pretty-print.h: Likewise.
23960 * graph.c (print_graph_cfg): Likewise.
23961 * graphite-scop-detection.c (dot_all_sese): Likewise.
23962 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
23963 * loop-unroll.c (report_unroll): Likewise.
23964 * passes.c (pass_manager::register_one_dump_file): Likewise.
23965 * print-tree.c: Likewise.
23966 * statistics.c: Likewise.
23967 * tree-cfg.c: Likewise.
23968 * tree-cfg.h: Likewise.
23969 * tree-dfa.c: Likewise.
23970 * tree-dfa.h: Likewise.
23971 * tree-dump.c (dump_function): Likewise.
23972 * tree-dump.h (struct dump_info): Likewise.
23973 * tree-pretty-print.c: Likewise.
23974 * tree-pretty-print.h: Likewise.
23975 * tree-ssa-live.c: Likewise.
23976 * tree-ssa-live.h: Likewise.
23977 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
23978 * tree-vect-loop.c: Likewise.
23979 * tree-vect-slp.c: Likewise.
23981 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
23982 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23984 PR tree-optimization/80457
23985 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
23986 of all arguments to a statement as scalar_to_vec operations.
23987 (vectorizable_call): Adjust call to vect_model_simple_cost for
23989 (vectorizable_conversion): Likewise.
23990 (vectorizable_assignment): Likewise.
23991 (vectorizable_shift): Likewise.
23992 (vectorizable_operation): Likewise.
23993 (vectorizable_comparison): Likewise.
23994 (vect_is_simple_cond): Record the def types for operands.
23995 (vectorizable_condition): Likewise, call vect_model_simple_cost.
23996 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
23997 for statement argument count.
23999 2017-05-16 Carl Love <cel@us.ibm.com>
24001 * config/rs6000/rs6000-c: Add support for built-in functions
24002 vector unsigned long long vec_bperm (vector unsigned long long,
24003 vector unsigned char)
24004 vector signed long long vec_mule (vector signed int,
24006 vector unsigned long long vec_mule (vector unsigned int,
24007 vector unsigned int)
24008 vector signed long long vec_mulo (vector signed int,
24010 vector unsigned long long vec_mulo (vector unsigned int,
24011 vector unsigned int)
24012 vector signed char vec_sldw (vector signed char,
24013 vector signed char,
24015 vector unsigned char vec_sldw (vector unsigned char,
24016 vector unsigned char,
24018 vector signed short vec_sldw (vector signed short,
24019 vector signed short,
24021 vector unsigned short vec_sldw (vector unsigned short,
24022 vector unsigned short,
24024 vector signed int vec_sldw (vector signed int,
24027 vector unsigned int vec_sldw (vector unsigned int,
24028 vector unsigned int,
24030 vector signed long long vec_sldw (vector signed long long,
24031 vector signed long long,
24033 vector unsigned long long vec_sldw (vector unsigned long long,
24034 vector unsigned long long,
24036 * config/rs6000/rs6000-c: Add support for built-in functions
24037 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
24038 * config/rs6000/altivec.h: Add defintion for vec_sldw.
24039 * doc/extend.texi: Update the built-in documentation for the
24040 new built-in functions.
24042 2017-05-16 Marek Polacek <polacek@redhat.com>
24046 * tree.c (save_expr): Don't fold the expression.
24048 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
24050 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
24051 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
24052 and (?*y,m). Update insn attributes.
24054 2017-05-16 Martin Liska <mliska@suse.cz>
24056 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
24057 flags argument of print_gimple_stmt, print_gimple_expr,
24058 print_generic_stmt and print_generic_expr.
24059 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
24060 * coretypes.h: Likewise.
24061 * except.c (dump_eh_tree): Likewise.
24062 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
24063 * gimple-pretty-print.h: Likewise.
24064 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
24065 (backprop::push_to_worklist): Likewise.
24066 (backprop::pop_from_worklist): Likewise.
24067 (backprop::process_use): Likewise.
24068 (backprop::intersect_uses): Likewise.
24069 (note_replacement): Likewise.
24070 * gimple-ssa-store-merging.c
24071 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
24072 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
24073 (pass_store_merging::execute): Likewise.
24074 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
24075 (ssa_base_cand_dump_callback): Likewise.
24076 (dump_incr_vec): Likewise.
24077 (replace_refs): Likewise.
24078 (replace_mult_candidate): Likewise.
24079 (create_add_on_incoming_edge): Likewise.
24080 (create_phi_basis): Likewise.
24081 (insert_initializers): Likewise.
24082 (all_phi_incrs_profitable): Likewise.
24083 (introduce_cast_before_cand): Likewise.
24084 (replace_one_candidate): Likewise.
24085 * gimplify.c (gimplify_expr): Likewise.
24086 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
24087 (set_rename): Likewise.
24088 (rename_uses): Likewise.
24089 (copy_loop_phi_nodes): Likewise.
24090 (add_close_phis_to_merge_points): Likewise.
24091 (copy_loop_close_phi_args): Likewise.
24092 (copy_cond_phi_args): Likewise.
24093 (graphite_copy_stmts_from_block): Likewise.
24094 (translate_pending_phi_nodes): Likewise.
24095 * graphite-poly.c (print_pdr): Likewise.
24096 (dump_gbb_cases): Likewise.
24097 (dump_gbb_conditions): Likewise.
24098 (print_scop_params): Likewise.
24099 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
24100 (build_cross_bb_scalars_use): Likewise.
24101 (gather_bbs::before_dom_children): Likewise.
24102 * hsa-dump.c (dump_hsa_immed): Likewise.
24103 * ipa-cp.c (print_ipcp_constant_value): Likewise.
24104 (get_replacement_map): Likewise.
24105 * ipa-inline-analysis.c (dump_condition): Likewise.
24106 (estimate_function_body_sizes): Likewise.
24107 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
24108 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
24109 * ipa-prop.c (ipa_dump_param): Likewise.
24110 (ipa_print_node_jump_functions_for_edge): Likewise.
24111 (ipa_modify_call_arguments): Likewise.
24112 (ipa_modify_expr): Likewise.
24113 (ipa_dump_param_adjustments): Likewise.
24114 (ipa_dump_agg_replacement_values): Likewise.
24115 (ipcp_modif_dom_walker::before_dom_children): Likewise.
24116 * ipa-pure-const.c (check_stmt): Likewise.
24117 (pass_nothrow::execute): Likewise.
24118 * ipa-split.c (execute_split_functions): Likewise.
24119 * omp-offload.c (dump_oacc_loop_part): Likewise.
24120 (dump_oacc_loop): Likewise.
24121 * trans-mem.c (tm_log_emit): Likewise.
24122 (tm_memopt_accumulate_memops): Likewise.
24123 (dump_tm_memopt_set): Likewise.
24124 (dump_tm_memopt_transform): Likewise.
24125 * tree-cfg.c (gimple_verify_flow_info): Likewise.
24126 (print_loop): Likewise.
24127 * tree-chkp-opt.c (chkp_print_addr): Likewise.
24128 (chkp_gather_checks_info): Likewise.
24129 (chkp_get_check_result): Likewise.
24130 (chkp_remove_check_if_pass): Likewise.
24131 (chkp_use_outer_bounds_if_possible): Likewise.
24132 (chkp_reduce_bounds_lifetime): Likewise.
24133 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
24134 (chkp_mark_completed_bounds): Likewise.
24135 (chkp_register_incomplete_bounds): Likewise.
24136 (chkp_mark_invalid_bounds): Likewise.
24137 (chkp_maybe_copy_and_register_bounds): Likewise.
24138 (chkp_build_returned_bound): Likewise.
24139 (chkp_get_bound_for_parm): Likewise.
24140 (chkp_build_bndldx): Likewise.
24141 (chkp_get_bounds_by_definition): Likewise.
24142 (chkp_generate_extern_var_bounds): Likewise.
24143 (chkp_get_bounds_for_decl_addr): Likewise.
24144 * tree-chrec.c (chrec_apply): Likewise.
24145 * tree-data-ref.c (dump_data_reference): Likewise.
24146 (dump_subscript): Likewise.
24147 (dump_data_dependence_relation): Likewise.
24148 (analyze_overlapping_iterations): Likewise.
24149 * tree-inline.c (expand_call_inline): Likewise.
24150 (tree_function_versioning): Likewise.
24151 * tree-into-ssa.c (dump_defs_stack): Likewise.
24152 (dump_currdefs): Likewise.
24153 (dump_names_replaced_by): Likewise.
24154 (dump_update_ssa): Likewise.
24155 (update_ssa): Likewise.
24156 * tree-object-size.c (pass_object_sizes::execute): Likewise.
24157 * tree-parloops.c (build_new_reduction): Likewise.
24158 (try_create_reduction_list): Likewise.
24159 (ref_conflicts_with_region): Likewise.
24160 (oacc_entry_exit_ok_1): Likewise.
24161 (oacc_entry_exit_single_gang): Likewise.
24162 * tree-pretty-print.h: Likewise.
24163 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
24164 (get_scalar_evolution): Likewise.
24165 (add_to_evolution): Likewise.
24166 (get_loop_exit_condition): Likewise.
24167 (analyze_evolution_in_loop): Likewise.
24168 (analyze_initial_condition): Likewise.
24169 (analyze_scalar_evolution): Likewise.
24170 (instantiate_scev): Likewise.
24171 (number_of_latch_executions): Likewise.
24172 (gather_chrec_stats): Likewise.
24173 (final_value_replacement_loop): Likewise.
24174 (scev_const_prop): Likewise.
24175 * tree-sra.c (dump_access): Likewise.
24176 (disqualify_candidate): Likewise.
24177 (create_access): Likewise.
24178 (reject): Likewise.
24179 (maybe_add_sra_candidate): Likewise.
24180 (create_access_replacement): Likewise.
24181 (analyze_access_subtree): Likewise.
24182 (analyze_all_variable_accesses): Likewise.
24183 (sra_modify_assign): Likewise.
24184 (initialize_constant_pool_replacements): Likewise.
24185 (find_param_candidates): Likewise.
24186 (decide_one_param_reduction): Likewise.
24187 (replace_removed_params_ssa_names): Likewise.
24188 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24189 * tree-ssa-copy.c (dump_copy_of): Likewise.
24190 (copy_prop_visit_cond_stmt): Likewise.
24191 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24192 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24193 (record_equivalences_from_stmt): Likewise.
24194 * tree-ssa-dse.c (compute_trims): Likewise.
24195 (delete_dead_call): Likewise.
24196 (delete_dead_assignment): Likewise.
24197 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24198 (forward_propagate_into_cond): Likewise.
24199 (pass_forwprop::execute): Likewise.
24200 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24201 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24203 (move_computations_worker): Likewise.
24204 (execute_sm): Likewise.
24205 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24206 (remove_exits_and_undefined_stmts): Likewise.
24207 (remove_redundant_iv_tests): Likewise.
24208 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24209 (adjust_iv_update_pos): Likewise.
24210 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24211 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24212 (value_replacement): Likewise.
24213 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24214 * tree-ssa-pre.c (print_pre_expr): Likewise.
24215 (get_representative_for): Likewise.
24216 (create_expression_by_pieces): Likewise.
24217 (insert_into_preds_of_block): Likewise.
24218 (eliminate_insert): Likewise.
24219 (eliminate_dom_walker::before_dom_children): Likewise.
24220 (eliminate): Likewise.
24221 (remove_dead_inserted_code): Likewise.
24222 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24223 * tree-ssa-reassoc.c (get_rank): Likewise.
24224 (eliminate_duplicate_pair): Likewise.
24225 (eliminate_plus_minus_pair): Likewise.
24226 (eliminate_not_pairs): Likewise.
24227 (undistribute_ops_list): Likewise.
24228 (eliminate_redundant_comparison): Likewise.
24229 (update_range_test): Likewise.
24230 (optimize_range_tests_var_bound): Likewise.
24231 (optimize_vec_cond_expr): Likewise.
24232 (rewrite_expr_tree): Likewise.
24233 (rewrite_expr_tree_parallel): Likewise.
24234 (linearize_expr): Likewise.
24235 (break_up_subtract): Likewise.
24236 (linearize_expr_tree): Likewise.
24237 (attempt_builtin_powi): Likewise.
24238 (attempt_builtin_copysign): Likewise.
24239 (transform_stmt_to_copy): Likewise.
24240 (transform_stmt_to_multiply): Likewise.
24241 (dump_ops_vector): Likewise.
24242 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24243 (print_scc): Likewise.
24244 (set_ssa_val_to): Likewise.
24245 (visit_reference_op_store): Likewise.
24246 (visit_use): Likewise.
24247 (sccvn_dom_walker::before_dom_children): Likewise.
24248 (run_scc_vn): Likewise.
24249 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24251 (expr_hash_elt::print): Likewise.
24252 (const_and_copies::pop_to_marker): Likewise.
24253 (const_and_copies::record_const_or_copy_raw): Likewise.
24254 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24255 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24256 (dump_predicates): Likewise.
24257 (find_uninit_use): Likewise.
24258 (warn_uninitialized_phi): Likewise.
24259 (pass_late_warn_uninitialized::execute): Likewise.
24260 * tree-ssa.c (verify_vssa): Likewise.
24261 (verify_ssa): Likewise.
24262 (maybe_optimize_var): Likewise.
24263 * tree-vrp.c (dump_value_range): Likewise.
24264 (dump_all_value_ranges): Likewise.
24265 (dump_asserts_for): Likewise.
24266 (register_edge_assert_for_2): Likewise.
24267 (vrp_visit_cond_stmt): Likewise.
24268 (vrp_visit_switch_stmt): Likewise.
24269 (vrp_visit_stmt): Likewise.
24270 (vrp_visit_phi_node): Likewise.
24271 (simplify_cond_using_ranges_1): Likewise.
24272 (fold_predicate_in): Likewise.
24273 (evrp_dom_walker::before_dom_children): Likewise.
24274 (evrp_dom_walker::push_value_range): Likewise.
24275 (evrp_dom_walker::pop_value_range): Likewise.
24276 (execute_early_vrp): Likewise.
24278 2017-05-16 Richard Biener <rguenther@suse.de>
24280 * dwarf2out.c (loc_list_from_tree_1): Do not create
24281 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24283 2017-05-16 Richard Biener <rguenther@suse.de>
24285 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24287 (note_variable_value_in_expr): If we resolved the decl ref
24288 do not push to the stack.
24290 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24292 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24293 operations in fast-math mode.
24294 (vaddq_f16): Likewise.
24295 (vmul_f16): Likewise.
24296 (vmulq_f16): Likewise.
24297 (vsub_f16): Likewise.
24298 (vsubq_f16): Likewise.
24299 * config/arm/neon.md (add<mode>3): New.
24301 (fma:<VH:mode>3): New. Also remove outdated comment.
24304 2017-05-16 Martin Liska <mliska@suse.cz>
24308 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24309 (odr_types_equivalent_p): Likewise.
24311 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24313 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24315 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24318 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24319 non-interunit SSE move alternatives with '?'.
24320 (zero-extendsidi peephole2): New peephole to skip intermediate
24321 general register in SSE zero-extend sequence.
24323 2017-05-15 Jeff Law <law@redhat.com>
24325 * reorg.c (relax_delay_slots): Create a new variable to hold
24326 the temporary target rather than clobbering TARGET_LABEL.
24328 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24329 missing argument to extract_bit_field call.
24330 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24332 2017-05-15 Martin Liska <mliska@suse.cz>
24335 * gcc.c (process_command): Do not allow empty argument of -o option.
24337 2017-05-15 Renlin Li <renlin.li@arm.com>
24339 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24340 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24341 * config/aarch64/constraints.md (Usf): Add long call check.
24342 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24343 (call_value): Likewise.
24344 (sibcall): Likewise.
24345 (sibcall_value): Likewise.
24347 (call_value_insn): New.
24348 (sibcall_insn): Update rtx pattern.
24349 (sibcall_value_insn): Likewise.
24350 (call_internal): Remove.
24351 (call_value_internal): Likewise.
24352 (sibcall_internal): Likewise.
24353 (sibcall_value_internal): Likewise.
24354 (call_reg): Likewise.
24355 (call_symbol): Likewise.
24356 (call_value_reg): Likewise.
24357 (call_value_symbol): Likewise.
24359 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24362 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24364 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24366 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24367 compatible with CCGOCmode and with CCZmode.
24369 2017-05-14 Martin Sebor <msebor@redhat.com>
24371 PR middle-end/77671
24372 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24373 (gimple_fold_builtin_snprintf): Same.
24374 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24375 (gimple_fold_builtin_snprintf): Same.
24376 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24377 of character types.
24378 (is_call_safe): New function.
24379 (try_substitute_return_value): Call it.
24380 (try_simplify_call): New function.
24381 (pass_sprintf_length::handle_gimple_call): Call it.
24383 2017-05-14 Martin Sebor <msebor@redhat.com>
24385 PR middle-end/80669
24386 * builtins.c (expand_builtin_stpncpy): Simplify.
24388 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24390 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24391 * config/i386/i386.h
24392 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24393 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24394 (struct machine_function): Add new members call_ms2sysv,
24395 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24396 (struct machine_frame_state): New fields sp_realigned and
24397 sp_realigned_offset.
24398 * config/i386/i386.c
24399 (enum xlogue_stub): New enum.
24400 (enum xlogue_stub_sets): New enum.
24401 (class xlogue_layout): New class.
24402 (struct ix86_frame): New fields stack_realign_allocate_offset,
24403 stack_realign_offset and outlined_save_offset. Modify comments to
24404 detail stack layout when using out-of-line stubs.
24405 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24406 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24407 -mcall-ms2sysv-xlogues.
24408 (stub_managed_regs): New static variable.
24409 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24410 registers managed by out-of-line stub.
24411 (disable_call_ms2sysv_xlogues): New function.
24412 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24413 m->call_ms2sysv when appropriate and compute frame layout for
24415 (sp_valid_at, fp_valid_at): New inline functions.
24416 (choose_basereg): New function.
24417 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24419 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24420 Use align parameter of choose_baseaddr to generated aligned SSE movs
24422 (pro_epilogue_adjust_stack): Modify to track
24423 machine_frame_state::sp_realigned.
24424 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24425 (ix86_nsaved_sseregs): Likewise.
24426 (ix86_emit_save_regs): Likewise.
24427 (ix86_emit_save_regs_using_mov): Likewise.
24428 (ix86_emit_save_sse_regs_using_mov): Likewise.
24429 (get_scratch_register_on_entry): Likewise.
24430 (gen_frame_set): New function.
24431 (gen_frame_load): Likewise.
24432 (gen_frame_store): Likewise.
24433 (emit_outlined_ms2sysv_save): Likewise.
24434 (emit_outlined_ms2sysv_restore): Likewise.
24435 (ix86_expand_prologue): Modify stack re-alignment code and call
24436 emit_outlined_ms2sysv_save when appropriate.
24437 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24438 parameter rtx_insn *insn, which allows the function to be used to only
24439 generate the notes.
24440 (ix86_expand_epilogue): Modify validity checks of frame and stack
24441 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24442 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24443 * config/i386/predicates.md
24444 (save_multiple): New predicate.
24445 (restore_multiple): Likewise.
24446 * config/i386/sse.md
24447 (save_multiple<mode>): New pattern.
24448 (save_multiple_realign<mode>): Likewise.
24449 (restore_multiple<mode>): Likewise.
24450 (restore_multiple_and_return<mode>): Likewise.
24451 (restore_multiple_leave_return<mode>): Likewise.
24452 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
24454 2017-05-14 Julia Koval <julia.koval@intel.com>
24456 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24457 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24458 (__builtin_ia32_xsetbv): New builtins.
24459 * config/i386/i386.c (ix86_expand_special_args_builtin):
24461 (ix86_expand_builtin): Special expand for new intrinsics.
24462 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24463 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24464 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24466 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24468 * cfganal.c (inverted_post_order_compute): Change argument type
24470 * cfganal.h (inverted_post_order_compute): Adjust prototype.
24471 * df-core.c (rest_of_handle_df_initialize): Adjust.
24472 (rest_of_handle_df_finish): Likewise.
24473 (df_analyze_1): Likewise.
24474 (df_analyze): Likewise.
24475 (loop_inverted_post_order_compute): Change argument to be a vec *.
24476 (df_analyze_loop): Adjust.
24477 (df_get_n_blocks): Likewise.
24478 (df_get_postorder): Likewise.
24479 * df.h (struct df_d): Change field to be a vec.
24480 * lcm.c (compute_laterin): Adjust.
24481 (compute_available): Likewise.
24482 * lra-lives.c (lra_create_live_ranges_1): Likewise.
24483 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
24484 * tree-ssa-pre.c (compute_antic): Likewise.
24486 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24488 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
24489 (depth_first_search::depth_first_search): Change structure init
24490 function to this constructor.
24491 (depth_first_search::add_bb): Rename function to this member.
24492 (depth_first_search::execute): Likewise.
24493 (flow_dfs_compute_reverse_finish): Adjust.
24495 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24497 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
24498 (longest_simple_path): Likewise.
24499 * shrink-wrap.c (spread_components): Likewise.
24500 (disqualify_problematic_components): Likewise.
24501 (emit_common_heads_for_components): Likewise.
24502 (emit_common_tails_for_components): Likewise.
24503 (insert_prologue_epilogue_for_components): Likewise.
24505 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24507 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
24510 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24512 * df-core.c (df_set_blocks): Start using auto_bitmap.
24513 (df_compact_blocks): Likewise.
24514 * df-problems.c (df_rd_confluence_n): Likewise.
24515 * df-scan.c (df_insn_rescan_all): Likewise.
24516 (df_process_deferred_rescans): Likewise.
24517 (df_update_entry_block_defs): Likewise.
24518 (df_update_exit_block_uses): Likewise.
24519 (df_entry_block_bitmap_verify): Likewise.
24520 (df_exit_block_bitmap_verify): Likewise.
24521 (df_scan_verify): Likewise.
24522 * lra-constraints.c (lra_constraints): Likewise.
24523 (undo_optional_reloads): Likewise.
24524 (lra_undo_inheritance): Likewise.
24525 * lra-remat.c (calculate_gen_cands): Likewise.
24526 (do_remat): Likewise.
24527 * lra-spills.c (assign_spill_hard_regs): Likewise.
24528 (spill_pseudos): Likewise.
24529 * tree-ssa-pre.c (bitmap_set_and): Likewise.
24530 (bitmap_set_subtract_values): Likewise.
24532 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24534 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
24535 management with auto_bitmap.
24536 (fix_inter_tick): Likewise.
24537 (fix_recovery_deps): Likewise.
24538 * ira.c (add_store_equivs): Likewise.
24539 (find_moveable_pseudos): Likewise.
24540 (split_live_ranges_for_shrink_wrap): Likewise.
24541 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
24542 (rtx_reuse_manager::seen_def_p): Likewise.
24543 (rtx_reuse_manager::set_seen_def): Likewise.
24544 * print-rtl.h (class rtx_reuse_manager): Likewise.
24546 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24548 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
24550 (migrate_btr_def): Likewise.
24551 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
24552 * df-core.c (loop_post_order_compute): Likewise.
24553 (loop_inverted_post_order_compute): Likewise.
24554 * hsa-common.h: Likewise.
24555 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
24556 * init-regs.c (initialize_uninitialized_regs): Likewise.
24557 * ipa-inline.c (resolve_noninline_speculation): Likewise.
24558 (inline_small_functions): Likewise.
24559 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
24560 * ira.c (combine_and_move_insns): Likewise.
24561 (build_insn_chain): Likewise.
24562 * loop-invariant.c (find_invariants): Likewise.
24563 * lower-subreg.c (propagate_pseudo_copies): Likewise.
24564 * predict.c (tree_predict_by_opcode): Likewise.
24565 (predict_paths_leading_to): Likewise.
24566 (predict_paths_leading_to_edge): Likewise.
24567 (estimate_loops_at_level): Likewise.
24568 (estimate_loops): Likewise.
24569 * shrink-wrap.c (try_shrink_wrapping): Likewise.
24570 (spread_components): Likewise.
24571 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
24572 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
24573 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
24574 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
24575 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
24576 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
24577 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
24578 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
24579 (mark_threaded_blocks): Likewise.
24580 (thread_through_all_blocks): Likewise.
24581 * tree-ssa.c (verify_ssa): Likewise.
24582 (execute_update_addresses_taken): Likewise.
24583 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
24585 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24587 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
24589 (post_order_compute): Likewise.
24590 (inverted_post_order_compute): Likewise.
24591 (pre_and_rev_post_order_compute_fn): Likewise.
24593 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24595 * genrecog.c (int_set::int_set): Explicitly construct our
24596 auto_vec base class.
24597 * vec.h (auto_vec::auto_vec): New constructor.
24599 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24601 * bitmap.h (class auto_bitmap): New constructor taking
24602 bitmap_obstack * argument.
24604 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24606 * bitmap.h (class auto_bitmap): Change type of m_bits to
24607 bitmap_head, and adjust ctor / dtor and member operators.
24609 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
24611 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
24612 when returned register mode doesn't match original mode.
24614 2017-05-12 Jeff Law <law@redhat.com>
24615 Jakub Jelinek <jakub@redhat.com>
24617 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
24618 we look for cc setter after the compare-elim changes.
24619 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
24620 within the vector to match what compare-elim now expects.
24621 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
24622 (xorsi3_flags, one_cmplsi2_flags): Likewise.
24624 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
24625 after the compare-elim changes.
24626 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
24627 the vector to match what compare-elim now expects.
24628 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
24629 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
24630 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
24631 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
24632 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
24634 * config/visium/visium.c (single_set_and_flags): Fix where
24635 we look for cc setter after the compare-elim changes.
24636 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
24637 with the vector to match what compare-elim now expects.
24638 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
24639 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
24640 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
24641 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
24642 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
24643 (neg<mode>2_insn_set_overflow): Likewise.
24645 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
24647 PR middle-end/79794
24648 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
24649 maybe_expand_insn call, set ops[0].target. If still set after call,
24650 set alt_rtl. Add extra arg to recursive calls.
24651 (extract_bit_field): Add alt_rtl argument. Pass to
24653 * expmed.h (extract_bit_field): Fix prototype.
24654 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
24655 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
24656 to extract_bit_field_calls.
24657 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
24658 Pass alt_rtl to extract_bit_field calls.
24659 * calls.c (store_unaligned_arguments_into_psuedos)
24660 load_register_parameters): Pass extra NULL to extract_bit_field calls.
24661 * optabs.c (maybe_legitimize_operand): Clear op->target when call
24663 * optabs.h (struct expand_operand): Add target bitfield.
24665 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24667 * compare-elim.c (try_eliminate_compare): Canonicalize
24668 operation with embedded compare to
24669 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
24670 (set (reg) (operation)].
24672 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
24674 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24677 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
24678 cost of adding a carry flag for ADC instruction.
24679 [case MINUS]: Ignore the cost of subtracting a carry flag
24680 for SBB instruction.
24682 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
24684 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
24686 * config/rs6000/bmiintrin.h: New file.
24687 * config/rs6000/bmi2intrin.h: New file.
24688 * config/rs6000/x86intrin.h: New file.
24690 2017-05-12 Jeff Law <law@redhat.com>
24692 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
24695 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
24697 PR middle-end/80707
24698 * tree-cfg.c: Remove cfg edges of unreachable case statements.
24700 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24702 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24703 early expansion of vector divide builtins.
24704 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
24705 builtins identified as having unsigned arguments.
24707 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24709 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
24710 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
24711 expansion of vector logical operations (and, andc, or, xor,
24714 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24716 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
24717 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
24719 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24721 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24722 early GIMPLE expansion of vector multiplies.
24724 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24726 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
24727 TARGET_HAVE_MOVT conditional.
24728 (movt splitter): Likewise.
24730 2017-05-12 Richard Biener <rguenther@suse.de>
24732 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
24733 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24734 Fold all stmts not inplace.
24736 2017-05-12 Richard Biener <rguenther@suse.de>
24738 PR tree-optimization/80713
24739 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
24740 inserted_exprs bit for not removed stmts.
24742 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
24744 PR middle-end/69921
24745 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
24746 parallelized" attribute for parallelized OpenACC kernels.
24747 * omp-offload.c (execute_oacc_device_lower): Use it.
24749 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
24750 Set "oacc kernels" attribute.
24751 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
24752 parameter. Adjust all users.
24753 (oacc_fn_attrib_kernels_p): Remove function.
24754 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
24755 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
24756 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
24757 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
24758 assert "oacc kernels" attribute is set.
24760 2017-05-11 Carl Love <cel@us.ibm.com>
24762 * config/rs6000/rs6000-c: Add support for built-in functions
24763 vector unsigned char vec_popcnt (vector signed char)
24764 vector unsigned char vec_popcnt (vector unsigned char)
24765 vector unsigned short vec_popcnt (vector signed short)
24766 vector unsigned short vec_popcnt (vector unsigned short)
24767 vector unsigned int vec_popcnt (vector signed int)
24768 vector unsigned int vec_popcnt (vector unsigned int)
24769 vector unsigned long long vec_popcnt (vector signed long long)
24770 vector unsigned long long vec_popcnt (vector unsigned long long)
24771 vector signed long long vec_slo (vector signed long long,
24772 vector signed char)
24773 vector signed long long vec_slo (vector signed long long,
24774 vector unsigned char)
24775 vector unsigned long long vec_slo (vector unsigned long long,
24776 vector signed char)
24777 vector unsigned long long vec_slo (vector unsigned long long,
24778 vector unsigned char)
24779 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
24780 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
24781 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
24782 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
24783 * doc/extend.texi: Update the built-in documentation file for the
24784 new built-in functions.
24786 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
24788 * attribs.h (sorted_attr_string): Move machine independent
24789 functions for target clone support from the i386 port to common
24790 code. Rename ix86_function_versions to common_function_versions.
24791 Rename make_name to make_unique_name.
24792 (common_function_versions): Likewise.
24793 (make_unique_name): Likewise.
24794 (make_dispatcher_decl): Likewise.
24795 (is_function_default_version): Likewise.
24796 * attribs.c (attr_strcmp): Likewise.
24797 (sorted_attr_string): Likewise.
24798 (common_function_versions): Likewise.
24799 (make_unique_name): Likewise.
24800 (make_dispatcher_decl): Likewise.
24801 (is_function_default_version): Likewise.
24802 * config/i386/i386.c (attr_strcmp): Likewise.
24803 (sorted_attr_string): Likewise.
24804 (ix86_function_versions): Likewise.
24805 (make_name): Likewise.
24806 (make_dispatcher_decl): Likewise.
24807 (is_function_default_version): Likewise.
24808 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
24810 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24813 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
24814 Account for direct move costs for vec_construct of integer
24817 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
24820 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
24821 (UNSPEC_STX_ATOMIC): Ditto.
24822 (loaddi_via_sse): New insn.
24823 (storedi_via_sse): Ditto.
24824 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
24825 Update corresponding peephole2 patterns.
24826 (atomic_storedi_fpu): Ditto.
24828 2017-05-11 Julia Koval <julia.koval@intel.com>
24830 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
24831 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
24833 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
24834 (__builtin_ia32_rsqrt14ss_mask): New builtins.
24835 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
24837 2017-05-11 Nathan Sidwell <nathan@acm.org>
24839 * graphite-poly.c: Include dumpfile.h.
24841 * dumpfle.h (dump_function): Declare here ...
24842 * tree-dump.h (dump_function): ... not here.
24843 * dumpfile.c: #include tree-cfg.h.
24844 (dump_function): Move here from ...
24845 * tree-dump.c (dump_function): ... here.
24846 * gimplify.c: #include splay-tree.h, not tree-dump.h.
24847 * graphite-poly.c: Don't include tree-dump.h.
24848 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
24849 * print-tree.c: Likewise.
24850 * stor-layout.c: Likewise.
24851 * tree-nested.c: Likewise.
24853 * dumpfile.c (dump_start): Use TDF_FLAGS.
24854 (dump_enable_all): Fix TDF_KIND check thinko.
24856 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
24858 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24859 array entries to represent two legal parameterizations of the
24860 overloaded __builtin_cmpb function, as represented by the
24861 P6_OV_BUILTIN_CMPB constant.
24862 (altivec_resolve_overloaded_builtin): Add special case handling
24863 for the __builtin_cmpb function, as represented by the
24864 P6_OV_BUILTIN_CMPB constant.
24865 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
24866 (BU_P6_64BIT_2): New macro.
24867 (BU_P6_OVERLOAD_2): New macro
24868 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
24869 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
24870 (CMPB): Add overload support to represent both 32-bit and 64-bit
24871 compare-bytes function.
24872 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
24873 support for TARGET_CMPB.
24874 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
24875 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24876 documentation of the __builtin_cmpb overloaded built-in function.
24878 2017-05-11 Richard Biener <rguenther@suse.de>
24880 PR tree-optimization/80705
24881 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
24882 bases are not vectorizable.
24884 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24886 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
24887 when counting register pressure.
24889 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24891 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
24892 (iv_ca_more_deps): Renamed to ...
24893 (iv_ca_compare_deps): ... this.
24894 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
24896 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24898 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
24900 (determine_group_iv_costs): ... here.
24901 (find_inv_vars_cb): Record inv var if it's not recorded before.
24903 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24905 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
24906 (get_shiftadd_cost): Ditto.
24908 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24910 * tree-ssa-address.c: Include header file.
24911 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
24913 (add_to_parts): Refactor.
24914 (addr_to_parts): New parameter. Update use of move_hint_to_base.
24915 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
24918 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24920 PR tree-optimization/53090
24921 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
24923 (extract_cond_operands): Detect condition with IV on both sides
24924 and return COMP_IV_EXPR_2.
24925 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
24926 (rewrite_use_compare): Simplify by removing call to function
24927 extract_cond_operands.
24929 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24931 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
24932 (extract_cond_operands): Detect condition comparing against non-
24933 invariant bound and return appropriate enum value.
24934 (find_interesting_uses_cond): Update use of extract_cond_operands.
24935 Handle its return value accordingly.
24936 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
24938 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24940 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
24941 nonlinear iv_use computation in loop invariant sensitive way.
24943 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24945 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
24946 (find_iv_candidates): Call relate_compare_use_with_all_cands.
24948 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24950 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
24951 (dump_cand): Support iv_cand.inv_exprs.
24952 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
24954 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
24957 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24959 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
24961 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
24962 as local function. Include necessary header files.
24963 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
24965 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24967 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
24969 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24971 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
24972 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
24973 RSHIFT_EXPR and BIT_NOT_EXPR.
24975 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24977 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
24978 (adjust_setup_cost): New parameter supporting round up adjustment.
24979 (struct address_cost_data): Delete.
24980 (force_expr_to_var_cost): Don't bound cost with spill_cost.
24981 (split_address_cost, ptr_difference_cost): Delete.
24982 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
24983 (struct ainc_cost_data): New struct.
24984 (get_address_cost_ainc): New function.
24985 (get_address_cost, get_computation_cost): Reimplement.
24986 (determine_group_iv_cost_address): Record inv_expr for all uses of
24988 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
24989 (iv_ca_has_deps): Reimplemented to ...
24990 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
24992 (iv_ca_extend): Call iv_ca_more_deps.
24994 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24996 * tree-ssa-address.c (struct mem_address): Move to header file.
24997 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
24998 * tree-ssa-address.h (struct mem_address): Move from C file.
24999 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
25001 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25003 * tree-affine.h (aff_combination_type): New interface.
25004 (aff_combination_zero_p): Remove static.
25005 (aff_combination_const_p): New interface.
25006 (aff_combination_singleton_var_p): New interfaces.
25008 2017-05-11 Richard Biener <rguenther@suse.de>
25010 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25011 Skip unreachable blocks and destinations.
25012 (eliminate): Move stmt removal and fixup ...
25013 (fini_eliminate): ... here. Skip inserted exprs.
25014 (pass_pre::execute): Move fini_pre after fini_eliminate.
25015 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
25016 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
25017 PRE to get rid of dead code that has invalid SSA form and
25018 split critical edges again.
25020 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25022 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25024 2017-05-11 Richard Biener <rguenther@suse.de>
25026 * passes.c (execute_function_todo): Verify loops if they are
25027 said to be up-to-date.
25028 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
25029 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
25031 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
25034 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
25035 handle calling assemble_external ourself.
25038 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
25039 modes with zero size. Enhance comment.
25041 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25043 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
25044 built-ins for vec_xl and vec_xst with short and char pointer
25047 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25049 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
25050 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
25051 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
25052 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
25053 (_mm_maskz_min_round_ss): New intrinsics.
25054 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
25055 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
25056 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
25057 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
25058 (__builtin_ia32_minss_mask_round): New builtins.
25059 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25060 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25061 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
25063 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
25064 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
25066 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
25069 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25071 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
25072 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
25073 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
25074 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
25075 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
25076 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
25077 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
25078 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25079 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25080 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
25081 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
25082 (__builtin_ia32_mulss_mask_round): New builtins.
25083 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25084 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25085 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
25087 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
25088 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25090 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25093 2017-05-10 Julia Koval <julia.koval@intel.com>
25095 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
25096 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
25097 (_mm256_setr_m128i): New intrinsics.
25099 2017-05-10 Julia Koval <julia.koval@intel.com>
25101 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
25102 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
25103 (_mm_maskz_rcp14_ss): New intrinsics.
25104 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
25105 (__builtin_ia32_rcp14ss_mask): New builtins.
25106 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
25108 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
25110 PR tree-optimization/51513
25111 * tree-cfg.c (gimple_seq_unreachable_p): New function.
25112 (assert_unreachable_fallthru_edge_p): Use it.
25113 (group_case_labels_stmt): Likewise.
25114 * tree-cfg.h: Prototype it.
25115 * stmt.c: Include cfghooks.h and tree-cfg.h.
25116 (emit_case_dispatch_table) <gap_label>: New local variable.
25117 Use it to fill dispatch table gaps.
25118 Test for default_label before updating probabilities.
25119 (expand_case) <default_label>: Remove unneeded initialization.
25120 Test for unreachable default case statement and remove its edge.
25121 Set default_label accordingly.
25122 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
25124 2017-05-10 Carl Love <cel@us.ibm.com>
25126 * config/rs6000/rs6000-c: Add support for built-in functions
25127 vector signed char vec_neg (vector signed char)
25128 vector signed short int vec_neg (vector short int)
25129 vector signed int vec_neg (vector signed int)
25130 vector signed long long vec_neg (vector signed long long)
25131 vector float vec_neg (vector float)
25132 vector double vec_neg (vector double)
25133 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
25135 * config/rs6000/altivec.h: Add define for vec_neg
25136 * doc/extend.texi: Update the built-in documentation for the
25137 new built-in functions.
25139 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25141 PR tree-optimization/77644
25142 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
25144 2017-05-10 Nathan Sidwell <nathan@acm.org>
25146 * dumpfile.h (TDI_lang_all): New.
25147 (TDF_KIND): New. Renumber others
25148 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
25150 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
25152 (get_dump_file_name): Adjust suffix generation.
25153 (dump_enable_all): Use TDF_KIND.
25154 * doc/invoke.texi (-fdump-lang-all): Document.
25156 * dumpfile.h: Tabify.
25158 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
25161 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
25162 Move member access before delete.
25164 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
25166 * tree-inline.c (expand_call_inline): Split block at stmt
25169 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
25172 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
25173 are now unused after splitting mov{sf,sd}_hardfloat.
25174 (f32_lr2): Likewise.
25175 (f32_lm): Likewise.
25176 (f32_lm2): Likewise.
25177 (f32_li): Likewise.
25178 (f32_li2): Likewise.
25179 (f32_lv): Likewise.
25180 (f32_sr): Likewise.
25181 (f32_sr2): Likewise.
25182 (f32_sm): Likewise.
25183 (f32_sm2): Likewise.
25184 (f32_si): Likewise.
25185 (f32_si2): Likewise.
25186 (f32_sv): Likewise.
25187 (f32_dm): Likewise.
25188 (f32_vsx): Likewise.
25189 (f32_av): Likewise.
25190 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25191 For movsf, order stores so the VSX stores occur before the GPR
25192 store which encourages the register allocator to use a traditional
25193 FPR instead of a GPR. For movsd, order the stores so that the GPR
25194 store comes before the VSX stores to allow the power6 to work.
25195 This is due to the power6 not having a 32-bit integer store
25196 instruction from a FPR.
25197 (movsf_hardfloat): Likewise.
25198 (movsd_hardfloat): Likewise.
25200 2017-05-09 Martin Sebor <msebor@redhat.com>
25202 PR translation/80280
25203 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25206 PR translation/80280
25207 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25208 data member added in r247778.
25209 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25211 2017-05-09 Nathan Sidwell <nathan@acm.org>
25213 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25215 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25218 2017-05-09 Marek Polacek <polacek@redhat.com>
25220 * doc/invoke.texi: Fix typo.
25222 2017-05-09 Richard Biener <rguenther@suse.de>
25224 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25225 (vrp_val_is_min): Likewise.
25226 (set_value_range_to_value): Likewise.
25227 (set_value_range_to_nonnegative): Likewise.
25228 (gimple_assign_nonzero_p): Likewise.
25229 (gimple_stmt_nonzero_p): Likewise.
25230 (vrp_int_const_binop): Likewise. Remove unreachable case.
25231 (adjust_range_with_scev): Adjust comments.
25232 (compare_range_with_value): Likewise.
25233 (extract_range_from_phi_node): Likewise.
25234 (test_for_singularity): Likewise.
25236 2017-05-09 Richard Biener <rguenther@suse.de>
25238 * tree-vrp.c (get_single_symbol): Add assert that we don't
25239 get overflowed constants as invariant part.
25240 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25241 checks. Use wi::cmp instead of recursing for integer constants.
25242 (compare_values): Just ignore whether we assumed undefined
25243 overflow instead of failing the compare.
25244 (extract_range_for_var_from_comparison_expr): Add comment before the
25245 TREE_NO_WARNING sets.
25246 (test_for_singularity): Likewise.
25247 (extract_range_from_comparison): Do not disable optimization
25248 when we assumed undefined overflow.
25249 (extract_range_basic): Remove init of unused var.
25251 2017-05-09 Richard Biener <rguenther@suse.de>
25253 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25254 (extract_range_from_multiplicative_op_1): Adjust.
25255 (extract_range_from_binary_expr_1): Use int_const_binop.
25257 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25260 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25261 rs6000_store_data_bypass_p in seven define_bypass directives and
25262 in several comments.
25263 * config/rs6000/rs6000-protos.h: Add prototype for
25264 rs6000_store_data_bypass_p function.
25265 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25266 function implements slightly different (rs6000-specific) semantics
25267 than store_data_bypass_p, returning false rather than aborting
25268 with assertion error when arguments do not satisfy the
25269 requirements of store data bypass.
25270 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25271 rs6000_store_data_bypass_p.
25273 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25275 * config/xtensa/xtensa-protos.h
25276 (xtensa_initial_elimination_offset): New declaration.
25277 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25278 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25279 macro definition, add case for FRAME_POINTER_REGNUM when
25280 FRAME_GROWS_DOWNWARD.
25281 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25282 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25283 xtensa_initial_elimination_offset.
25285 2017-05-08 Nathan Sidwell <nathan@acm.org>
25287 * doc/invoke.texi: Alphabetize -fdump options.
25289 2017-05-08 Martin Sebor <msebor@redhat.com>
25291 PR translation/80280
25292 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25294 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25296 * target.def (compute_frame_layout): New optional target hook.
25297 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25298 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25299 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25301 * reload1.c (verify_initial_elim_offsets): Likewise.
25302 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25303 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25304 (arm_get_frame_offsets): Split up into this ...
25305 (arm_compute_frame_layout): ... and this function.
25307 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25309 * config/aarch64/constraints.md (Usa): New constraint.
25310 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25312 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25314 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25315 with_multilib_list after it has been checked.
25317 2017-05-08 Richard Biener <rguenther@suse.de>
25319 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25320 (bitmap_set_subtract_values): Likewise.
25322 2017-05-08 Richard Biener <rguenther@suse.de>
25324 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25325 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25327 (gimple_stmt_nonzero_warnv_p): Rename to ...
25328 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25330 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25331 (extract_range_basic): Adjust, do not disable propagation on
25332 strict overflow sensitive simplification.
25333 (vrp_visit_cond_stmt): Likewise.
25335 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25337 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25338 body size unconditionally.
25340 2017-05-07 Jeff Law <law@redhat.com>
25343 2017-05-06 Jeff Law <law@redhat.com>
25344 PR tree-optimization/78496
25345 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25348 PR tree-optimization/78496
25349 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25350 (simplify_stmt_using_ranges): Call it.
25351 (vrp_dom_walker::before_dom_children): Extract equivalences
25352 from an ASSERT_EXPR with an equality comparison against a
25355 2017-05-06 Jeff Law <law@redhat.com>
25357 PR tree-optimization/78496
25358 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25361 PR tree-optimization/78496
25362 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25363 (simplify_stmt_using_ranges): Call it.
25364 (vrp_dom_walker::before_dom_children): Extract equivalences
25365 from an ASSERT_EXPR with an equality comparison against a
25368 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25370 * lra-constraints.c (lra_copy_reg_equiv): New function.
25371 (split_reg): Use it to copy equivalence information from the
25372 original register to the spill register.
25374 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25376 PR rtl-optimization/75964
25377 * simplify-rtx.c (simplify_const_relational_operation): Remove
25378 invalid handling of comparisons of integer ABS.
25380 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25382 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25383 initialize to zero.
25384 (init_regs): Remove declaration.
25385 (function_arg_advance_32): Initialize error_p as boolean variable.
25387 2017-05-05 Nathan Sidwell <nathan@acm.org>
25389 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25390 lines. Use for (;;).
25392 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25394 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25395 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25396 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25397 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25398 VF=2 that require versioning.
25400 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25402 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25405 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25407 * diagnostic.h (diagnostic_override_option_index): Convert from
25408 macro to inline function.
25410 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25412 * diagnostic.c (last_module_changed_p): New function.
25413 (set_last_module): New function.
25414 (diagnostic_report_current_module): Convert macro usage to
25415 the above functions.
25416 * diagnostic.h (diagnostic_context::last_module): Strengthen
25417 from const line_map * to const line_map_ordinary *.
25418 (diagnostic_last_module_changed): Delete macro.
25419 (diagnostic_set_last_module): Delete macro.
25421 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25423 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25424 with diagnostic_report_diagnostic.
25425 (diagnostic_n_impl_richloc): Likewise.
25426 * diagnostic.h (report_diagnostic): Delete macro.
25427 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25428 with diagnostic_report_diagnostic.
25429 * substring-locations.c (format_warning_va): Likewise.
25431 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25433 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25434 save/restor of format_spec. Move option-printing code to...
25435 (print_option_information): ...this new function, and
25436 reimplement by simply printing to the pretty_printer,
25437 rather than appending to the format string.
25439 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25441 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25442 handling logic into...
25443 (update_effective_level_from_pragmas): ...this new function.
25445 2017-05-04 Andrew Waterman <andrew@sifive.com>
25447 * config/riscv/riscv.opt (mstrict-align): New option.
25448 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25449 (SLOW_UNALIGNED_ACCESS): Define.
25450 (riscv_slow_unaligned_access): Declare.
25451 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
25453 (riscv_slow_unaligned_access): New variable.
25454 (rocket_tune_info): Set slow_unaligned_access to true.
25455 (optimize_size_tune_info): Set slow_unaligned_access to false.
25456 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25457 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25458 (riscv_option_override): Set riscv_slow_unaligned_access.
25459 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25461 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25463 * config/riscv/riscv.md: Unify indentation.
25465 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25470 * config/rs6000/rs6000.md (u code attribute): Add FIX and
25472 (extendsi<mode>2): Add support for doing sign extension via
25473 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
25474 don't have ISA 3.0 instructions.
25475 (extendsi<mode>2 splitter): Likewise.
25476 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
25477 generate the normal insns since SImode can now go in vector
25478 registers. Disallow the special UNSPECs needed for previous
25479 machines to hide SImode being used. Add new insns
25480 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
25481 (fix_trunc<mode>si2_stfiwx): Likewise.
25482 (fix_trunc<mode>si2_internal): Likewise.
25483 (fixuns_trunc<mode>si2): Likewise.
25484 (fixuns_trunc<mode>si2_stfiwx): Likewise.
25485 (fctiw<u>z_<mode>_smallint): Likewise.
25486 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
25487 of floating point to 32-bit integer from doing a direct move to
25488 the GPR registers to do a store.
25489 (fctiwz_<mode>): Break long line.
25491 2017-05-05 Bin Cheng <bin.cheng@arm.com>
25493 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
25494 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
25495 (addr_list, addr_offset_valid_p): New.
25496 (split_address_groups): Check offset validity with above function.
25497 (gt-tree-ssa-loop-ivopts.h): Include header file.
25499 2017-05-05 Nathan Sidwell <nathan@acm.org>
25501 * config.gcc (arm*-*-*): Add missing 'fi'.
25503 2017-05-05 Steve Ellcey <sellcey@cavium.com>
25505 * doc/invoke.texi (-fopt-info): Explicitly say order of options
25506 included in -fopt-info does not matter.
25507 * doc/optinfo.texi (-fopt-info): Fix description of default
25508 behavour. Explicitly say order of options included in -fopt-info
25511 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25513 * config.gcc: Allow combinations of aprofile and rmprofile values for
25514 --with-multilib-list.
25515 * config/arm/t-multilib: New file.
25516 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
25517 variables. Remove setting of ISA and floating-point ABI in
25518 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
25519 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
25520 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
25521 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
25523 * config/arm/t-rmprofile: Likewise except for the matches changes.
25524 * doc/install.texi (--with-multilib-list): Document the combination of
25525 aprofile and rmprofile values and warn about pitfalls in doing that.
25527 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25529 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
25530 (movdi_aarch64): Likewise.
25532 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25534 PR tree-optimization/80632
25535 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
25537 (build_arrays): Initialize it for virtual phis.
25538 (fix_phi_nodes): Use it for virtual phis.
25540 PR tree-optimization/80558
25541 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
25542 [x, y] op z into [x op, y op z] for op & or | if conditions
25545 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
25546 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25549 * config/arm/arm.md (use_literal_pool): Remove.
25550 (64-bit immediate split): No longer takes cost into consideration
25551 if arm_disable_literal_pool is enabled.
25552 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
25553 used when arm_disable_literal_pool is enabled.
25554 (arm_max_const_double_inline_cost): Remove use of
25555 arm_disable_literal_pool.
25556 (push_minipool_fix): Add assert.
25557 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
25558 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
25559 (no_literal_pool_sf_immediate): New.
25561 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25563 PR tree-optimization/80613
25564 * tree-ssa-dce.c (propagate_necessity): Remove cases for
25565 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
25567 2017-05-05 Richard Biener <rguenther@suse.de>
25569 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
25571 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
25573 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
25574 of this flag from insn conditions due to removal from r247495.
25576 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25578 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
25580 (arm_early_store_addr_dep_ptr): Likewise.
25581 * config/arm/aarch-common-protos.h
25582 (arm_early_load_addr_dep_ptr): Add prototype.
25583 (arm_early_store_addr_dep_ptr): Likewise.
25584 * config/arm/cortex-a53.md: Add new bypasses.
25586 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25588 * tree.c (next_type_uid): Change type to unsigned.
25589 (type_hash_canon): Decrement back next_type_uid if
25590 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
25591 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
25594 2017-05-04 Martin Sebor <msebor@redhat.com>
25596 * builtins.c: Fix a trivial typo in a comment.
25598 PR middle-end/79234
25599 * builtins.c (check_sizes): Adjust to handle reading past the end.
25600 Avoid printing excessive upper bound of ranges. Use %E to print
25601 tree nodes instead of converting them to %wu.
25602 (expand_builtin_memchr): New function.
25603 (compute_dest_size): Rename...
25604 (compute_objsize): ...to this.
25605 (expand_builtin_memcpy): Adjust.
25606 (expand_builtin_mempcpy): Adjust.
25607 (expand_builtin_strcat): Adjust.
25608 (expand_builtin_strcpy): Adjust.
25609 (check_strncat_sizes): Adjust.
25610 (expand_builtin_strncat): Adjust.
25611 (expand_builtin_strncpy): Adjust and simplify.
25612 (expand_builtin_memset): Adjust.
25613 (expand_builtin_bzero): Adjust.
25614 (expand_builtin_memcmp): Adjust.
25615 (expand_builtin): Handle memcmp.
25616 (maybe_emit_chk_warning): Check strncat just once.
25618 2017-05-04 Martin Sebor <msebor@redhat.com>
25620 PR preprocessor/79214
25621 PR middle-end/79222
25622 PR middle-end/79223
25623 * builtins.c (check_sizes): Add inlining context and issue
25624 warnings even when -Wno-system-headers is set.
25625 (check_strncat_sizes): Same.
25626 (expand_builtin_strncat): Same.
25627 (expand_builtin_memmove): New function.
25628 (expand_builtin_stpncpy): Same.
25629 (expand_builtin): Handle memmove and stpncpy.
25631 2017-05-04 Bin Cheng <bin.cheng@arm.com>
25633 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
25634 which is not used any more.
25636 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25638 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
25640 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25642 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
25643 (cortexa53_tunings): Likewise.
25644 (cortexa57_tunings): Likewise.
25645 (cortexa72_tunings): Likewise.
25646 (cortexa73_tunings): Likewise.
25648 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25650 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
25651 Set loop alignment to 8.
25653 2017-05-04 Martin Sebor <msebor@redhat.com>
25655 PR translation/80280
25656 * builtins.c (expand_builtin_object_size): Add missing quoting to
25657 %D and like directives.
25658 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
25659 (hsa_type_for_tree_type): Same.
25660 (verify_function_arguments): Same.
25661 * symtab.c (symbol_table::change_decl_assembler_name): Same.
25662 * varasm.c (get_section): Same.
25665 2017-05-04 Martin Sebor <msebor@redhat.com>
25667 PR translation/80280
25668 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
25670 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25672 * config/aarch64/aarch64.c (generic_addrcost_table):
25673 Change HI/TI mode setting.
25675 2017-05-04 Martin Jambor <mjambor@suse.cz>
25677 PR tree-optimization/80622
25678 * tree-sra.c (comes_initialized_p): New function.
25679 (build_accesses_from_assign): Only set write lazily when
25680 comes_initialized_p is false.
25681 (analyze_access_subtree): Use comes_initialized_p.
25682 (propagate_subaccesses_across_link): Assert !comes_initialized_p
25683 instead of testing for PARM_DECL.
25685 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25687 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
25688 constraint on operand 0 to allow more general addressing modes.
25689 Adjust output template.
25690 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
25692 * config/aarch64/aarch64-protos.h
25693 (aarch64_address_valid_for_prefetch_p): Declare prototype.
25694 * config/aarch64/constraints.md (Dp): New address constraint.
25695 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
25698 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
25700 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
25701 update use of estimate_ipcp_clone_size_and_time.
25702 (estimate_local_effects): Update use of
25703 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
25704 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
25705 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
25706 Return nonspecialized time.
25708 2017-05-04 Richard Biener <rguenther@suse.de>
25710 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
25711 for the last VUSE which def dominates the PHI. Directly call
25713 (get_continuation_for_phi_1): Remove.
25715 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
25717 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
25718 to explain the use of truncating division. Cap the number of
25719 iterations to the maximum given by nb_iterations_upper_bound,
25722 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25724 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
25725 * configure: Regenerate.
25726 * config.in: Regenerate.
25727 * config/i386/driver-mingw32.c: new file.
25728 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
25729 * config.host: Link driver-mingw32.o on MinGW host.
25730 * doc/install.texi: Document new --enable-mingw-wildcard configure
25733 2017-05-04 Marek Polacek <polacek@redhat.com>
25735 PR tree-optimization/80612
25736 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
25738 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25739 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
25741 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
25742 (movt splitter): Likewise.
25743 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
25744 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
25745 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
25746 block for Thumb-1 with MOVT.
25747 (thumb2_legitimate_address_p): Move code block ...
25748 (can_avoid_literal_pool_for_label_p): ... into this new function.
25749 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
25751 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
25752 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
25753 "M-profile targets with the MOVT instruction".
25755 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25757 * config/arm/arm-builtins.c (arm_init_builtins): Rename
25758 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
25759 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
25761 2017-05-04 Martin Liska <mliska@suse.cz>
25763 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
25764 variable cond_code.
25766 2017-05-04 Richard Biener <rguenther@suse.de>
25768 * tree.c (array_at_struct_end_p): Handle arrays at struct
25769 end with flexarrays more conservatively. Refactor and treat
25770 arrays of arrays or aggregates more strict. Fix
25771 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
25772 * tree.c (array_at_struct_end_p): Adjust prototype.
25773 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
25774 * gimple-fold.c (get_range_strlen): Likewise.
25775 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
25777 2017-05-04 Richard Biener <rguenther@suse.de>
25779 PR tree-optimization/31130
25780 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
25782 (supports_overflow_infinity): Likewise.
25783 (is_negative_overflow_infinity): Likewise.
25784 (is_positive_overflow_infinity): Likewise.
25785 (is_overflow_infinity): Likewise.
25786 (stmt_overflow_infinity): Likewise.
25787 (overflow_infinity_range_p): Likewise.
25788 (usable_range_p): Remove as always returning true.
25789 (make_overflow_infinity): Remove.
25790 (negative_overflow_infinity): Likewise.
25791 (positive_overflow_infinity): Likewise.
25792 (avoid_overflow_infinity): Likewise.
25793 (set_value_range): Adjust accordingly.
25794 (set_value_range_to_nonnegative): Likewise, remove now unused
25795 overflow_infinity arg.
25796 (vrp_operand_equal_p): Adjust.
25797 (update_value_range): Likewise.
25798 (range_int_cst_singleton_p): Likewise.
25799 (operand_less_p): Likewise.
25800 (compare_values_warnv): Likewise.
25801 (extract_range_for_var_from_comparison_expr): Likewise.
25802 (vrp_int_const_binop): Likewise.
25803 (zero_nonzero_bits_from_vr): Likewise.
25804 (extract_range_from_multiplicative_op_1): Likewise.
25805 (extract_range_from_binary_expr_1): Likewise.
25806 (extract_range_from_unary_expr): Likewise.
25807 (extract_range_from_comparison): Likewise.
25808 (extract_range_basic): Likewise.
25809 (adjust_range_with_scev): Likewise.
25810 (compare_ranges): Likewise.
25811 (compare_range_with_value): Likewise.
25812 (dump_value_range): Likewise.
25813 (test_for_singularity): Likewise, remove strict_overflow_p parameter
25815 (simplify_cond_using_ranges): Adjust.
25817 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25819 * brig-builtins.def: Added a builtin for class_f64.
25820 * builtin-types.def: Added a builtin type needed by class_f64.
25822 2017-05-03 Jason Merrill <jason@redhat.com>
25824 * timevar.def: Add TV_CONSTEXPR.
25826 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25828 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
25830 2017-05-03 Martin Jambor <mjambor@suse.cz>
25832 * ipa-prop.c (ipa_update_after_lto_read): Removed.
25833 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
25834 * ipa-cp.c (ipcp_propagate_stage): Do not call
25835 ipa_update_after_lto_read.
25836 * ipa-inline.c (ipa_inline): Likewise.
25838 2017-05-03 Martin Jambor <mjambor@suse.cz>
25840 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
25841 tag. Added a default constructor and a destructor.
25842 (ipa_edge_args_sum_t): New class;
25843 (ipa_edge_args_sum): Declare.
25844 (ipa_edge_args_vector): Remove declaration.
25845 (IPA_EDGE_REF): Use ipa_edge_args_sum.
25846 (ipa_free_edge_args_substructures): Remove declaration.
25847 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
25848 (ipa_edge_args_info_available_for_edge_p): Likewise.
25849 * ipa-prop.c (ipa_edge_args_vector): Removed.
25850 (edge_removal_hook_holder): Likewise.
25851 (edge_duplication_hook_holder): Likewise.
25852 (ipa_edge_args_sum): New variable.
25853 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
25854 ipa_edge_args_vector.
25855 (ipa_free_edge_args_substructures): Likewise.
25856 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
25857 ipa_edge_args_vector.
25858 (ipa_edge_removal_hook): Turned into method
25859 ipa_edge_args_sum_t::remove.
25860 (ipa_edge_duplication_hook): Turned into method
25861 ipa_edge_args_sum_t::duplicate.
25862 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
25863 registering edge hooks.
25864 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
25865 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
25866 ipa_edge_args_sum instead of ipa_edge_args_vector.
25867 * ipa-profile.c (ipa_profile): Likewise.
25869 2017-05-03 Martin Jambor <mjambor@suse.cz>
25871 * symbol-summary.h (function_summary): New method exists.
25872 (function_summary::symtab_removal): Deallocate through release.
25873 (call_summary): New class.
25874 (gt_ggc_mx): New overload.
25875 (gt_pch_nx): Likewise.
25876 (gt_pch_nx): Likewise.
25878 2017-05-03 Jeff Law <law@redhat.com>
25880 PR tree-optimization/78496
25881 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
25882 from simplify_cond_using_ranges. Split off code to walk
25883 backwards through casts into ...
25884 (simplify_cond_using_ranges_2): New function.
25885 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
25886 (execute_vrp): After identifying jump threads, call
25887 simplify_cond_using_ranges_2.
25889 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
25892 * ipa-inline.h (inline_summary): Add ctor.
25893 (create_ggc): Do not use ggc_cleared_alloc.
25895 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
25896 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25898 * gcc.c (handle_braces): Support escaping in switch matching
25900 * doc/invoke.texi (Spec Files): Document it.
25901 Remove superfluous @code markup in items.
25903 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25905 * diagnostic-show-locus.c (struct column_range): New struct.
25906 (get_affected_columns): New function.
25907 (get_printed_columns): New function.
25908 (struct correction): New struct.
25909 (correction::ensure_capacity): New function.
25910 (correction::ensure_terminated): New function.
25911 (struct line_corrections): New struct.
25912 (line_corrections::~line_corrections): New dtor.
25913 (line_corrections::add_hint): New function.
25914 (layout::print_trailing_fixits): Reimplement in terms of the new
25916 (selftest::test_overlapped_fixit_printing): New function.
25917 (selftest::diagnostic_show_locus_c_tests): Call it.
25919 2017-05-03 Nathan Sidwell <nathan@acm.org>
25921 Canonicalize canonical type hashing
25922 * tree.h (type_hash_canon_hash): Declare.
25923 * tree.c (type_hash_list, attribute_hash_list): Move into
25924 type_hash_canon_hash.
25925 (build_type_attribute_qual_variant): Break out hash code calc into
25926 type_hash_canon_hash.
25927 (type_hash_canon_hash): New. Generic type hash computation.
25928 (build_range_type_1, build_array_type_1, build_function_type,
25929 build_method_type_directly, build_offset_type, build_complex_type,
25930 make_vector_type): Call it.
25932 2017-05-03 Richard Biener <rguenther@suse.de>
25934 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
25935 When all DRs have unknown misaligned do not always peel
25936 when there is a store but apply the same costing model as if
25937 there were only loads.
25939 2017-05-03 Richard Biener <rguenther@suse.de>
25942 PR tree-optimization/80492
25943 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25944 compare_base_decls returning dont-know properly.
25946 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
25948 * config/arm/iterators.md (CCSI): New mode iterator.
25949 (arch): New mode attribute.
25950 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
25951 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
25952 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
25953 code iterator for success result mode.
25954 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
25955 the corresponding new insn generators.
25957 2017-05-03 Bin Cheng <bin.cheng@arm.com>
25960 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25961 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25963 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
25965 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
25966 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
25967 (DDR_A): Wrap DDR argument in brackets.
25968 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
25969 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
25970 (DDR_REVERSED_P): Likewise.
25972 2017-05-03 Jakub Jelinek <jakub@redhat.com>
25974 PR tree-optimization/79472
25975 * tree-switch-conversion.c (struct switch_conv_info): Add
25976 contiguous_range and default_case_nonstandard fields.
25977 (collect_switch_conv_info): Compute contiguous_range and
25978 default_case_nonstandard fields, don't clear final_bb if
25979 contiguous_range and only the default case doesn't have the required
25981 (check_all_empty_except_final): Set default_case_nonstandard instead
25982 of failing if contiguous_range and the default case doesn't have empty
25984 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
25985 and only the default case doesn't have the required constants. Skip
25987 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
25988 if default_case_nonstandard.
25989 (build_constructors): Build constant 1 just once. Assert that default
25990 values aren't inserted in between cases if contiguous_range. Skip
25992 (build_arrays): Skip virtual phis.
25993 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
25994 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
25995 Handle virtual phis.
25996 (gen_inbound_check): Handle default_case_nonstandard case.
25997 (process_switch): Adjust check_final_bb caller. Call
25998 gather_default_values with the first non-default case instead of
25999 default case if default_case_nonstandard.
26001 2017-05-02 Nathan Sidwell <nathan@acm.org>
26003 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
26004 check. Fix formatting.
26006 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
26008 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
26009 errors when comparing specialized and unspecialized times.
26011 2017-05-02 David Malcolm <dmalcolm@redhat.com>
26013 * diagnostic-show-locus.c
26014 (layout::should_print_annotation_line_p): Make private.
26015 (layout::print_annotation_line): Make private.
26016 (layout::annotation_line_showed_range_p): Make private.
26017 (layout::show_ruler): Make private.
26018 (layout::print_source_line): Make private. Pass in line and
26019 line_width, rather than calling location_get_source_line. Drop
26021 (layout::print_leading_fixits): New method.
26022 (layout::print_any_fixits): Rename to...
26023 (layout::print_trailing_fixits): ...this, and make private.
26024 Don't print newline fixits.
26025 (diagnostic_show_locus): Move logic for printing one row into...
26026 (layout::print_line): ...this new function. Move the
26027 location_get_source_line call and error-handling from
26028 print_source_line to here. Call print_leading_fixits, and rename
26029 print_any_fixits to print_trailing_fixits.
26030 (selftest::test_fixit_insert_containing_newline): Update now that
26031 newlines are partially supported.
26032 (selftest::test_fixit_insert_containing_newline_2): New test.
26033 (selftest::test_fixit_replace_containing_newline): Update comments.
26034 (selftest::diagnostic_show_locus_c_tests): Call the new test.
26035 * edit-context.c (class added_line): New class.
26036 (class edited_line): Describe newline handling in comment.
26037 (edited_line::actually_edited_p): New method.
26038 (edited_line::print_content): Delete redundant decl.
26039 (edited_line::m_predecessors): New field.
26040 (edited_file::print_content): Call edited_line::print_content.
26041 (edited_file::print_diff): Update to support newlines.
26042 (edited_file::print_diff_hunk): Likewise.
26043 (edited_file::print_run_of_changed_lines): New function.
26044 (edited_file::print_diff_line): Convert to...
26045 (print_diff_line): ...this.
26046 (edited_file::get_effective_line_count): New function.
26047 (edited_line::edited_line): Initialize new field m_predecessors.
26048 (edited_line::~edited_line): Clean up m_predecessors.
26049 (edited_line::apply_fixit): Handle newlines.
26050 (edited_line::get_effective_line_count): New function.
26051 (edited_line::print_content): New function.
26052 (edited_line::print_diff_lines): New function.
26053 (selftest::test_applying_fixits_insert_containing_newline): New
26055 (selftest::test_applying_fixits_replace_containing_newline): New
26057 (selftest::insert_line): New function.
26058 (selftest::test_applying_fixits_multiple_lines): Add example of
26060 (selftest::edit_context_c_tests): Call the new tests.
26062 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26064 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
26065 parameter cand. Update dump information.
26066 (get_computation_cost): Update uses.
26068 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26070 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
26071 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
26072 (get_computation_at, rewrite_use_address): Update use of
26073 get_computation_aff.
26075 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26077 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
26078 (get_computation): Delete.
26079 (get_computation_cost): Implement like get_computation_cost_at.
26080 Use get_computation_at.
26081 (get_computation_cost_at): Delete.
26082 (rewrite_use_nonlinear_expr): Use get_computation_at.
26083 (rewrite_use_compare, remove_unused_ivs): Ditto.
26085 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26087 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
26089 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26091 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
26092 (ivopts_global_cost_for_size): Rename parameter and update uses.
26093 (iv_ca_recount_cost): Update uses.
26094 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
26095 candidates seperately in n_invs and n_cands.
26096 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
26098 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26100 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
26101 (find_inv_vars_cb): New.
26102 (find_depends): Renamed to ...
26103 (find_inv_vars): ... this.
26104 (add_candidate_1, force_var_cost): Call find_inv_vars.
26105 (split_address_cost, determine_group_iv_cost_cond): Ditto.
26107 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26109 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
26110 inv_vars. Add inv_exprs.
26111 (struct iv_cand): Rename depends_on to inv_vars.
26112 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
26113 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
26114 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
26115 (dump_cand): Dump inv_vars.
26116 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
26117 (record_invariant, find_depends, add_candidate_1): Ditto.
26118 (set_group_iv_cost, force_var_cost): Ditto.
26119 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
26120 (get_computation_cost_at, get_computation_cost): Ditto.
26121 (determine_group_iv_cost_generic): Ditto.
26122 (determine_group_iv_cost_address): Ditto.
26123 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
26124 (determine_group_iv_costs): Ditto.
26125 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
26126 (iv_ca_set_remove_invariants): Renamed to ...
26127 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
26128 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
26129 (iv_ca_set_add_invariants): Renamed to ...
26130 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
26131 (iv_ca_set_cp): Use iv_ca_set_add_invs.
26132 (iv_ca_has_deps): Support inv_vars and inv_exprs.
26133 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
26134 (create_new_ivs): Remove useless dump.
26136 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26138 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
26140 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
26141 (iv_ca_set_no_cp, create_new_iv): Ditto.
26143 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26145 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26147 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26149 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
26150 function tree_check2.
26152 2017-05-02 Martin Liska <mliska@suse.cz>
26154 * doc/gcov.texi: Add missing preposition.
26155 * gcov.c (function_info::function_info): Properly fill up
26156 all member variables.
26158 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26160 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
26162 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26164 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
26166 2017-05-02 Martin Liska <mliska@suse.cz>
26169 * lto-streamer-in.c (lto_read_tree_1): Remove
26170 LTO_STREAMER_DEBUG.
26171 * lto-streamer.c (struct tree_hash_entry): Likewise.
26172 (struct tree_entry_hasher): Likewise.
26173 (tree_entry_hasher::hash): Likewise.
26174 (tree_entry_hasher::equal): Likewise.
26175 (lto_streamer_init): Likewise.
26176 (lto_orig_address_map): Likewise.
26177 (lto_orig_address_get): Likewise.
26178 (lto_orig_address_remove): Likewise.
26179 * lto-streamer.h: Likewise.
26180 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26181 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26183 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26185 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26186 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26187 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26188 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26189 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26190 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26191 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26192 (mm_maskz_sub_ss): New intrinsics.
26193 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26194 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26195 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26196 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26197 (__builtin_ia32_subss_mask_round): New builtins.
26198 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26199 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26200 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26202 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26203 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26205 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26208 2017-05-02 Martin Jambor <mjambor@suse.cz>
26210 PR tree-optimization/78687
26211 * tree-sra.c (access): New field parent.
26212 (process_subtree_disqualification): New function.
26213 (disqualify_candidate): Call it.
26214 (build_accesses_from_assign): Reset write flag if creating an
26216 (build_access_subtree): Fill in parent field and also prpagate
26217 down grp_write flag.
26218 (create_artificial_child_access): New parameter set_grp_write, set
26219 grp_write to its value.
26220 (propagate_subaccesses_across_link): Also propagate grp_write flag
26222 (propagate_all_subaccesses): Push the closest parent back to work
26223 queue if add_access_to_work_queue returned true.
26225 2017-05-02 Richard Biener <rguenther@suse.de>
26227 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26228 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26229 -fstrict-overflow documentation.
26230 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26231 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26232 flag_strict_overflow.
26233 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26234 * lto-opts.c (lto_write_options): Do not stream it.
26235 * lto-wrapper.c (merge_and_complain): Do not handle it.
26236 * opts.c (default_options_table): Do not set -fstrict-overflow.
26237 (finish_options): Likewise do not clear it when sanitizing.
26238 * simplify-rtx.c (simplify_const_relational_operation): Do not
26239 test flag_strict_overflow.
26241 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26243 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26244 using enabled attribute.
26245 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26246 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26247 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26248 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26249 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26250 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26251 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26252 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26253 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26254 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26256 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26258 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26260 2017-05-02 Richard Biener <rguenther@suse.de>
26262 PR tree-optimization/80591
26264 2017-04-10 Richard Biener <rguenther@suse.de>
26266 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26269 2017-05-02 Richard Biener <rguenther@suse.de>
26271 PR tree-optimization/80549
26272 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26273 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26274 headers if they do not have a preheader.
26276 2017-05-02 Martin Liska <mliska@suse.cz>
26279 * common.opt: Fix typo.
26280 * doc/invoke.texi: Likewise.
26282 2017-05-01 Jan Beulich <jbeulich@suse.com>
26284 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26285 swapping, add (x,x,m,x,n) alternative.
26287 2017-05-01 Nathan Sidwell <nathan@acm.org>
26289 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26290 unnecessary unadjusted_alignment check.
26292 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26295 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26297 * gimplify.c (gimplify_cilk_detach): New function.
26298 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26299 * tree-core.h: Document EXPR_CILK_SPAWN.
26300 * tree.h (EXPR_CILK_SPAWN): Define.
26302 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26304 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26305 to use new fixit_hint representation, using the "replace" logic.
26306 (get_line_span_for_fixit_hint): Likewise.
26307 (layout::print_any_fixits): Likewise.
26308 (selftest::test_one_liner_many_fixits): Rename to...
26309 (selftest::test_one_liner_many_fixits_1): ...this, and update
26310 comment and expected output to reflect that the multiple fix-it
26311 hints are now consolidated into one insertion.
26312 (selftest::test_one_liner_many_fixits_2): New test.
26313 (selftest::test_diagnostic_show_locus_one_liner): Update for
26315 (selftest::test_fixit_consolidation): Update for fix-it API
26317 * diagnostic.c (print_parseable_fixits): Likewise.
26318 * edit-context.c (edited_line::m_line_events): Convert from
26319 auto_vec <line_event *> to auto_vec <line_event>.
26320 (class line_event): Convert from abstract base class to a concrete
26321 class, taking over the role of replace_event.
26322 (class insert_event): Delete.
26323 (class replace_event): Rename to class line_event. Convert to
26325 (edit_context::add_fixits): Reimplement.
26326 (edit_context::apply_insert): Delete.
26327 (edit_context::apply_replace): Rename to...
26328 (edit_context::apply_fixit): ...this. Convert to half-open range.
26329 (edited_file::apply_insert): Delete.
26330 (edited_file::apply_replace): Rename to...
26331 (edited_file::apply_fixit): ...this.
26332 (edited_line::~edited_line): Drop deletion of events.
26333 (edited_line::apply_insert): Delete.
26334 (edited_line::apply_replace): Rename to...
26335 (edited_line::apply_fixit): ...this. Convert to half-open range.
26336 Update for change to type of m_line_events.
26337 * edit-context.h (edit_context::apply_insert): Delete.
26338 (edit_context::apply_replace): Rename to...
26339 (edit_context::apply_fixit): ...this.
26341 2017-05-01 Martin Sebor <msebor@redhat.com>
26343 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26346 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26349 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26350 __get_cpuid_max returns 0.
26351 (__get_cpuid_count): Ditto.
26353 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26355 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26356 replacement expression is another instance of one of its arguments.
26358 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26361 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26362 check for stack push/pop autoinc.
26363 * config/i386/i386.c (ix86_agi_dependent): Return false
26364 if the only reason why modified_in_p returned true is that
26365 addr is SP based and set_insn is a push or pop.
26367 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26369 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26372 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26375 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26376 (account_size_time): Use two predicates - exec_pred and
26378 (evaluate_conditions_for_known_args): Compute both clause and
26380 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26381 (inline_summary_t::duplicate): Update.
26382 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26384 (compute_inline_parameters): Likewise.
26385 (estimate_edge_size_and_time): Update caluclation of time.
26386 (estimate_node_size_and_time): Compute both time and nonspecialized
26388 (estimate_ipcp_clone_size_and_time): Update.
26389 (inline_merge_summary): Update.
26390 (do_estimate_edge_time): Update.
26391 (do_estimate_edge_size): Update.
26392 (do_estimate_edge_hints): Update.
26393 (inline_read_section, inline_write_summary): Stream both new predicates.
26394 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26396 (compute_inlined_call_time): Cleanup.
26397 (big_speedup_p): Update.
26398 (edge_badness): Update.
26399 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26400 (size_time_entry): Replace predicate by exec_predicate and
26401 nonconst_predicate.
26402 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26403 (estimate_edge_time): Return also nonspec_time.
26404 (reset_edge_growth_cache): Update.
26406 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26408 PR rtl-optimization/80491
26409 * ifcvt.c (noce_process_if_block): When looking for x setter
26410 with missing else_bb, don't check only the insn right before
26411 cond_earliest, but look for the last insn that x is modified in
26412 within the same bb.
26414 PR rtl-optimization/80491
26415 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26417 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26419 PR tree-optimization/80487
26420 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26422 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26424 PR tree-optimization/79697
26425 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26426 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26427 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26429 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26430 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26432 2017-04-28 Martin Sebor <msebor@redhat.com>
26434 PR tree-optimization/80523
26435 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26436 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26438 (maybe_warn, format_directive, parse_directive): Use new functions.
26439 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26441 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26443 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26445 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26447 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26448 target_header_dir): Set correctly.
26449 * configure: Regenerated.
26450 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26451 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26452 instead of SYSTEM_HEADER_DIR.
26454 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
26456 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26457 (estimate_local_effects): Likewise.
26458 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26459 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26460 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26461 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26462 do_estimate_edge_time, estimate_edge_time): Likewise.
26463 * ipa-inline-analysis.c (estimate_node_size_and_time,
26464 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26465 (estimate_time_after_inlining): Remove.
26467 2017-04-28 Martin Liska <mliska@suse.cz>
26469 * doc/gcov.texi: Enhance documentation of gcov.
26471 2017-04-28 Martin Liska <mliska@suse.cz>
26473 * doc/gcov.texi: Sort options in alphabetic order.
26474 * doc/gcov-dump.texi: Likewise.
26475 * doc/gcov-tool.texi: Likewise.
26476 * gcov.c (print_usage): Likewise.
26477 * gcov-dump.c (print_usage): Likewise.
26478 * gcov-tool.c (print_merge_usage_message): Likewise.
26479 (print_rewrite_usage_message): Likewise.
26480 (print_overlap_usage_message): Likewise.
26482 2017-04-28 Martin Liska <mliska@suse.cz>
26484 PR gcov-profile/53915
26485 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
26487 2017-04-28 Martin Liska <mliska@suse.cz>
26489 PR gcov-profile/79891
26490 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
26491 is marked by compiler as living on a line.
26492 (get_cycles_count): Remove usage of the union.
26493 (output_intermediate_file): Likewise.
26494 (find_source): Fix GNU coding style.
26495 (accumulate_line_counts): Remove old non-all block mode.
26496 (output_lines): Remove usage of the union.
26497 * profile.c (output_location): Include all BBs, even if
26498 belonging to a same line (and file) as a previous BB.
26500 2017-04-28 Martin Liska <mliska@suse.cz>
26502 * gcov.c (process_args): Handle new argument 'w'.
26503 (read_graph_file): Assign ID to BBs.
26504 (output_branch_count): Display BB # if verbose flag is set.
26505 (output_lines): Likewise for arcs.
26506 (print_usage): Add '--verbose' option help.
26507 * doc/gcov.texi: Document --verbose (-w) option.
26509 2017-04-28 Martin Liska <mliska@suse.cz>
26511 * gcov.c (struct block_location_info): New struct.
26512 (process_file): Fill up the new structure.
26513 (read_graph_file): Replace usage of encoding by the newly added
26515 (add_line_counts): Likewise.
26516 (accumulate_line_counts): Remove usage of the union.
26517 (function_info::function_info): New function.
26518 (function_info::~function_info): Likewise.
26519 (process_file): Call delete instead of release_function.
26520 (release_function): Release the function.
26521 (release_structures): Call delete instead of release_function.
26522 (solve_flow_graph): Replace usage of num_blocks.
26523 (find_exception_blocks): Likewise.
26524 (output_lines): Fix GNU coding style.
26526 2017-04-28 Martin Liska <mliska@suse.cz>
26529 * coverage.c (coverage_remove_note_file): New function.
26530 * coverage.h: Declare the function.
26531 * toplev.c (finalize): Clean if an error has been seen.
26533 2017-04-28 Martin Liska <mliska@suse.cz>
26535 PR gcov-profile/80031
26536 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
26537 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
26538 * gcov.c (read_graph_file): Read just number of blocks.
26539 * profile.c (branch_prob): Do not stream 0 flags per a basic
26542 2017-04-28 Martin Liska <mliska@suse.cz>
26544 * gcov-dump.c (tag_*): Add new argument to declarations.
26545 (dump_gcov_file): Likewise.
26546 (tag_blocks): Add and use new argument depth.
26547 (tag_arcs): Likewise.
26548 (tag_lines): Likewise.
26549 (tag_counters): Likewise.
26550 (tag_summary): Likewise.
26551 (dump_working_sets): Use depth to do a proper indentation.
26553 2017-04-28 Jakub Jelinek <jakub@redhat.com>
26556 * cgraph.h (symtab_node::debug_symtab): No longer inline.
26557 * symtab.c (symtab_node::debug_symtab): Move definition here.
26559 2017-04-28 Richard Biener <rguenther@suse.de>
26561 * lto-streamer.h (LTO_major_version): Bump to 7.
26563 2017-04-28 Richard Biener <rguenther@suse.de>
26565 * tree-vrp.c (assert_info): New struct.
26566 (add_assert_info): New helper.
26567 (register_edge_assert_for_2): Refactor to add asserts to a vector
26569 (register_edge_assert_for_1): Likewise.
26570 (register_edge_assert_for): Likewise.
26571 (finish_register_edge_assert_for): New helper actually registering
26572 asserts where live on edge.
26573 (find_conditional_asserts): Adjust.
26574 (find_switch_asserts): Likewise.
26575 (evrp_dom_walker::try_find_new_range): Generalize.
26576 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
26578 2017-04-27 Marek Polacek <polacek@redhat.com>
26581 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
26582 arg10 and arg11 to itype.
26584 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
26586 * doc/extend.texi (Object Size Checking): Improve grammar.
26588 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
26591 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
26592 that the logic for permitting reciprocal estimates matches that
26595 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26598 * tree.c (type_cache_hasher::equal): Only compare
26599 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
26600 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
26601 non-aggregate element types.
26602 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
26603 about the flag on ARRAY_TYPEs in the comment, formatting fix.
26605 2017-04-27 Richard Biener <rguenther@suse.de>
26607 PR middle-end/80533
26608 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
26609 stripping ARRAY_REFs from MEM_EXPR make sure we're not
26610 keeping a reference to a trailing array.
26612 2017-04-27 Richard Biener <rguenther@suse.de>
26614 PR middle-end/80539
26615 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
26616 being in loop-closed SSA form conservatively.
26617 (chrec_fold_multiply_poly_poly): Likewise.
26619 2017-04-27 Tamar Christina <tamar.christina@arm.com>
26621 PR middle-end/79665
26622 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
26623 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
26625 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26628 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
26629 (aarch64_function_arg_alignment): Return unsigned int again, but still
26630 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
26631 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26632 Don't emit -Wpsabi note.
26633 (aarch64_function_arg_boundary): Likewise.
26634 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26637 2017-04-26 Nathan Sidwell <nathan@acm.org>
26639 * tree.h (crc32_unsigned_n): Declare.
26640 (crc32_unsigned, crc32_unsigned): Make inline.
26641 * tree.c (crc32_unsigned_bits): Replace with ...
26642 (crc32_unsigned_n): ... this.
26643 (crc32_unsigned, crc32_byte): Remove.
26644 (crc32_string): Remove unnecessary braces.
26646 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26648 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
26649 * ipa-inline-analysis.c (MAX_TIME): Remove.
26650 (account_size_time): Use sreal for time.
26651 (dump_inline_summary): Update.
26652 (estimate_function_body_sizes): Update.
26653 (estimate_edge_size_and_time): Update.
26654 (estimate_calls_size_and_time): Update.
26655 (estimate_node_size_and_time): Update.
26656 (inline_merge_summary): Update.
26657 (inline_update_overall_summary): Update.
26658 (estimate_time_after_inlining): Update.
26659 (inline_read_section): Update.
26660 (inline_write_summary): Update.
26661 * ipa-inline.c (compute_uninlined_call_time): Update.
26662 (compute_inlined_call_time): Update.
26663 (recursive_inlining): Update.
26664 (inline_small_functions): Update.
26665 (dump_overall_stats): Update.
26666 * ipa-inline.h: Include sreal.h.
26667 (size_time_entry): Turn time to sreal.
26668 (inline_summary): Turn self_time nad time to sreal.
26670 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26672 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
26674 (sreal::stream_out, sreal::stream_in): New.
26675 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
26677 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26679 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
26682 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
26685 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
26687 (dimode_scalar_chain::compute_convert_gain): Ditto.
26688 (dimode_scalar_chain::make_vector_copies): Ditto.
26689 (dimode_scalar_chain::convert_reg): Ditto.
26690 (dimode_scalar_chain::convert_insn): Ditto.
26691 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
26692 (VI248_AVX512BW_1): New mode iterator.
26693 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
26694 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
26697 2017-04-25 Martin Sebor <msebor@redhat.com>
26699 PR tree-optimization/80497
26700 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
26701 constants are representable in HOST_WIDE_INT.
26702 (parse_directive): Ditto.
26704 2017-04-25 Martin Sebor <msebor@redhat.com>
26707 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
26708 (new_zero_array): Adjust signature.
26709 (dom_info::dom_init): Used unsigned rather that size_t.
26710 (dom_info::dom_info): Same.
26712 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26713 Jakub Jelinek <jakub@redhat.com>
26716 * config/arm/arm.c: Include gimple.h.
26717 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26718 returns negative, increment ncrn only if it returned positive.
26719 (arm_needs_doubleword_align): Return int instead of bool,
26720 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
26721 members, but if there is any such non-FIELD_DECL
26722 > PARM_BOUNDARY aligned decl, return -1 instead of false.
26723 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26724 returns negative, increment nregs only if it returned positive.
26725 (arm_setup_incoming_varargs): Likewise.
26726 (arm_function_arg_boundary): Emit -Wpsabi note if
26727 arm_needs_doubleword_align returns negative, return
26728 DOUBLEWORD_ALIGNMENT only if it returned positive.
26730 2017-04-25 Marek Polacek <polacek@redhat.com>
26733 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26734 first argument to type.
26736 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
26739 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
26740 type checks to test for compatibility instead of equality.
26742 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26743 Jakub Jelinek <jakub@redhat.com>
26746 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
26748 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
26749 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
26750 the alignment computation, but return their maximum in warn_alignment.
26751 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26752 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
26754 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
26755 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26758 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26760 * config/arc/simdext.md (dmpyh): Fix typo.
26762 2017-04-25 Richard Biener <rguenther@suse.de>
26764 PR tree-optimization/80492
26765 * alias.c (compare_base_decls): Handle registers with asm
26766 specification conservatively.
26767 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26768 compare_base_decls returning dont-know properly.
26770 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26772 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
26773 (legitimate_offset_address_p): New function.
26774 (arc_legitimate_address_p): Use above function.
26776 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26778 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
26780 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26782 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
26783 ACCH registers whenever they are available.
26785 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26787 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
26788 double regs fix when not used.
26790 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26792 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
26794 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
26795 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
26797 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26799 * config/arc/arc.c (arc_output_addsi): Check for h-register class
26800 when emitting short ADD instructions.
26802 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26804 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
26806 (cmpsi_cc_c_insn): Likewise.
26807 (cbranchsi4_scratch): Compute proper instruction length using
26808 compact_hreg_operand.
26809 * config/arc/predicates.md (compact_hreg_operand): New predicate.
26811 2017-04-25 Richard Biener <rguenther@suse.de>
26813 PR middle-end/80509
26814 * passes.c (pass_manager::pass_manager): Initialize
26815 m_name_to_pass_map.
26817 2017-04-25 Richard Biener <rguenther@suse.de>
26819 PR tree-optimization/79201
26820 * tree-ssa-sink.c (statement_sink_location): Handle calls.
26822 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26825 * config/s390/vector.md: Split MEM->GPR vector moves for
26826 non-s_operand addresses.
26828 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26831 * config/s390/predicates.md (reload_const_wide_int_operand): New
26833 * config/s390/s390.md ("movti"): Remove d/P alternative.
26834 ("movti_bigconst"): New pattern definition.
26836 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26839 * s390-protos.h (s390_expand_cs_hqi): Removed.
26840 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
26841 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
26842 modes as well as CCZ1mode and CCZmode.
26843 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
26844 signature of s390_emit_compare_and_swap.
26845 (s390_expand_cs_hqi): Likewise, make static.
26846 (s390_expand_cs_tdsi): Generate an explicit compare before trying
26847 compare-and-swap, in some cases.
26848 (s390_expand_cs): Wrapper function.
26849 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
26851 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
26852 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
26853 patterns for small and large integers. Forbid symref memory operands.
26854 Move expander to s390.c. Require cc register.
26855 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
26856 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
26857 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
26858 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
26859 symref memory operands. Remove CC mode and call s390_match_ccmode
26861 ("atomic_exchange<mode>"): Allow and implement all integer modes.
26863 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26865 * config/s390/s390.md (define_peephole2): New peephole to help
26866 combining the load-and-test pattern with volatile memory.
26868 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26870 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
26871 with CCZmode for TARGET_Z196.
26873 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26875 PR rtl-optimization/80501
26876 * combine.c (make_compound_operation_int): Set subreg_code to SET
26877 even for AND with mask of the sign bit of mode.
26879 PR rtl-optimization/80500
26880 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
26881 sum's initial value.
26883 2017-04-25 Julian Brown <julian@codesourcery.com>
26884 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26886 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
26888 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
26890 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
26892 2017-04-25 Julian Brown <julian@codesourcery.com>
26893 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26895 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
26896 (thunderx2t99_sha): New Reservation.
26898 2017-04-25 Julian Brown <julian@codesourcery.com>
26899 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26901 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
26902 type for 1-element load.
26904 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
26906 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
26908 2017-04-24 Martin Jambor <mjambor@suse.cz>
26910 PR tree-optimization/80293
26911 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
26912 char arrays not totally scalarizable if it is false.
26913 (analyze_all_variable_accesses): Pass correct value in the new
26914 parameter. Add a statistics counter.
26916 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
26918 PR middle-end/79931
26919 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
26921 2017-04-24 Richard Biener <rguenther@suse.de>
26923 PR tree-optimization/80494
26924 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
26925 out for complex types.
26927 2017-04-24 Richard Biener <rguenther@suse.de>
26929 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
26930 * tree-ssa-sccvn.c (print_scc): Print SCC size.
26931 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
26932 (DFS): Adjust and never fail.
26933 (sccvn_dom_walker::fail): Remove.
26934 (sccvn_dom_walker::before_dom_children): Adjust.
26935 (run_scc_vn): Likewise and never fail.
26936 * tree-ssa-pre.c (pass_pre::execute): Adjust.
26937 (pass_fre::execute): Likewise.
26939 2017-04-24 Richard Biener <rguenther@suse.de>
26941 PR tree-optimization/79725
26942 * tree-ssa-sink.c (statement_sink_location): Return whether
26943 failure reason was zero uses. Move that check later.
26944 (sink_code_in_bb): Deal with zero uses by removing the stmt
26947 2017-04-24 Richard Biener <rguenther@suse.de>
26950 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
26951 pointer-based references.
26953 2017-04-24 Richard Biener <rguenther@suse.de>
26956 * pass_manager.h (pass_manager::operator new): Remove.
26957 (pass_manager::operator delete): Likewise.
26958 * passes.c (pass_manager::operator new): Remove.
26959 (pass_manager::operator delete): Likewise.
26960 (pass_manager::pass_manager): Zero individual pass members.
26962 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
26965 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
26966 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
26967 Check "XEXP (src, 1)" operand here.
26968 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
26969 Check "XEXP (src, 1)" operand here.
26970 (dimode_scalar_chain::make_vector_copies): Detect count register
26971 of a shift instruction. Zero extend count register from QImode
26972 to DImode to satisfy vector shift pattern count operand predicate.
26973 Substitute vector shift count operand with a DImode copy.
26974 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
26977 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
26979 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
26980 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26981 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
26982 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26983 (UNSPEC_NOREX_MEM): Remove definition.
26985 2017-04-21 Richard Biener <rguenther@suse.de>
26987 PR tree-optimization/79547
26988 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26989 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
26990 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
26991 without any constraints.
26993 2017-04-21 Richard Biener <rguenther@suse.de>
26995 PR tree-optimization/78847
26996 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
26998 2017-04-21 Richard Biener <rguenther@suse.de>
27000 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
27001 (build_distinct_type_copy): Likewise.
27002 (build_variant_type_copy): Likewise.
27003 * tree.c (build_qualified_type): Pass down mem-stat info.
27004 (build_distinct_type_copy): Likewise.
27005 (build_variant_type_copy): Likewise.
27007 2017-04-21 Richard Biener <rguenther@suse.de>
27009 PR tree-optimization/80237
27010 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
27012 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
27013 for a simplified result.
27015 2016-04-21 Richard Biener <rguenther@suse.de>
27017 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
27018 sth as strict as a simple_iv but a chrec without symbols and an
27019 operand defined in the loop we are peeling (and not some subloop).
27020 (propagate_constants_for_unrolling): Propagate all constants.
27022 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27025 * config/i386/i386.c (print_reg): Remove assert for disalowed
27026 regno values, call output_operand_lossage instead.
27028 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27031 * config/i386/constraints.md (Yc): New register constraint.
27032 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
27033 Use Yc constraint for alternative 2 of operand 0. Remove
27034 preferred_for_speed attribute.
27036 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27038 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
27039 lastprivate clauses in SIMT case.
27041 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
27043 * doc/invoke.texi (-Wextra-semi): Document new warning option.
27045 2017-04-20 Richard Biener <rguenther@suse.de>
27047 PR tree-optimization/57796
27048 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
27049 as N scalar stores.
27050 (vect_model_load_cost): Cost gathers as N scalar loads.
27052 2017-04-20 Richard Biener <rguenther@suse.de>
27054 * ggc-page.c (ggc_allocated_p): Rename to ...
27055 (safe_lookup_page_table_entry): ... this and return the lookup
27057 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
27059 2017-04-20 Richard Biener <rguenther@suse.de>
27061 PR tree-optimization/80453
27062 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
27063 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
27064 from the conditions.
27065 (vn_phi_eq): Pass them down.
27066 (vn_phi_lookup): Record them.
27067 (vn_phi_insert): Likewise.
27069 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
27071 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
27072 uninitialized variable warning to avoid buffer overrun.
27074 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27077 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
27078 is suppressed for '{ 0 }' in C.
27080 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27082 * BASE-VER: Set to 8.0.0.
27084 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
27086 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
27087 priority .init_array and .fini_array section with SECTION_NOTYPE
27090 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27092 PR middle-end/80423
27093 * tree.h (build_array_type): Add typeless_storage default argument.
27094 * tree.c (type_cache_hasher::equal): Also compare
27095 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
27096 (build_array_type): Add typeless_storage argument, set
27097 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
27099 (build_nonshared_array_type): Adjust build_array_type_1 caller.
27100 (build_array_type): Likewise. Add typeless_storage argument.
27102 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27103 Jakub Jelinek <jakub@redhat.com>
27105 PR tree-optimization/80426
27106 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
27107 operation on symbolic operands, also compute the overflow for the
27108 invariant part when the operation degenerates into a negation.
27110 2017-04-19 Jakub Jelinek <jakub@redhat.com>
27113 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
27114 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
27117 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
27119 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
27122 * config/avr/avr.c (tree.h): Include it.
27123 (cgraph.h): Include it.
27124 (avr_encode_section_info): Don't warn for uninitialized progmem
27125 variable if it's just an alias.
27127 2017-04-19 Richard Biener <rguenther@suse.de>
27130 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
27131 when needed by AutoPGO.
27133 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
27136 * doc/lto.texi: Remove an extra 'that'.
27138 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
27140 PR rtl-optimization/80429
27141 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
27142 are only used in debug insns.
27144 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27145 Vladimir Makarov <vmakarov@redhat.com>
27147 * config/sparc/predicates.md (input_operand): Add comment. Return
27148 true for any memory operand when LRA is in progress.
27149 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
27151 2017-04-18 Jeff Law <law@redhat.com>
27154 * mips.md ({return,simple_return}_internal): Do not overwrite
27157 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27159 PR tree-optimization/80443
27160 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
27161 instead of adding 1, subtract -1 and similarly instead of subtracting
27164 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
27166 PR rtl-optimization/80357
27167 * haifa-sched.c (tmp_bitmap): New variable.
27168 (model_recompute): Handle duplicate use records.
27169 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
27170 (free_global_sched_pressure_data): Free it.
27172 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
27175 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27176 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27177 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27178 instead of SYSTEM_HEADER_DIR.
27180 2017-04-18 Jeff Law <law@redhat.com>
27182 PR middle-end/80422
27183 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27184 predecessors after walking up the insn chain.
27186 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27189 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27190 sizetype type into debug info.
27192 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27195 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27196 unneeded test for TARGET_UPPER_REGS_SF.
27197 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27199 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27202 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27203 instead of gsi_after_labels.
27205 2017-04-18 Jeff Law <law@redhat.com>
27207 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27211 2017-04-13 Jeff Law <law@redhat.com>
27212 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27213 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27215 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27218 * config/avr/avr.c (intl.h): Include it.
27219 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27221 2017-04-18 Martin Liska <mliska@suse.cz>
27223 PR gcov-profile/78783
27224 * gcov-tool.c (gcov_output_files): Validate that destination
27225 file is either removed by the tool or by a user.
27227 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27228 Guy Benyei <guybe@mellanox.com>
27230 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27231 block, and do not negate it, the stored id is already negative.
27233 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27235 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27237 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27240 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27241 masks of options that should be turned off if the VSX vector
27242 options are turned off.
27243 (OTHER_P8_VECTOR_MASKS): Likewise.
27244 (OTHER_VSX_VECTOR_MASKS): Likewise.
27245 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27246 rs6000_disable_incompatible_switches to validate no type switches
27248 (rs6000_incompatible_switch): New function to disallow turning on
27249 other vector options if -mno-vsx, -mno-power8-vector, or
27250 -mno-power9-vector are specified.
27252 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27254 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27256 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27258 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27259 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27260 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27261 (ARG_POINTER_CFA_OFFSET): Likewise.
27263 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27265 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27266 conditions to take advantage of various optimizations.
27268 2017-04-13 Jeff Law <law@redhat.com>
27270 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27271 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27272 (zero_extendsidi2_dext): Likewise.
27274 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27277 * fold-const.c (fold_ternary_loc): Revert
27278 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27281 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27283 PR rtl-optimization/80343
27284 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27285 new scratch pseudo.
27287 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27290 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27291 to ubsan_encode_value.
27293 2017-04-13 Jeff Law <law@redhat.com>
27295 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27296 appearing in DEBUG_INSNs.
27298 2017-04-13 Martin Liska <mliska@suse.cz>
27300 PR gcov-profile/80413
27301 * gcov-io.c (gcov_write_string): Copy to buffer just when
27302 allocated size is greater than zero.
27304 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27307 * dwarf2out.c (decls_for_scope): Ignore declarations of
27308 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27310 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27313 * ipa-visibility.c (non_local_p): Fix typos.
27314 (localize_node): When localizing symbol in same comdat group,
27315 dissolve the group only when we know external symbols are going
27317 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27319 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27321 PR tree-optimization/79390
27322 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27323 order does not result in usable sequence, retry with reversed operand
27329 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27330 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27331 op0 instead of fold_convert_loc (loc, type, arg0).
27333 2017-04-12 Jeff Law <law@redhat.com>
27335 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27336 has a delay slot in the generated code.
27338 * config/cris/cris.md (cris_preferred_reload_class): Return
27339 GENNONACR_REGS rather than GENERAL_REGS.
27341 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27344 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27345 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27346 signedness of the result type.
27348 2017-04-12 Richard Biener <rguenther@suse.de>
27349 Jeff Law <law@redhat.com>
27351 PR tree-optimization/80359
27352 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27353 trim stores to TARGET_MEM_REFs.
27355 2017-04-12 Richard Biener <rguenther@suse.de>
27357 PR tree-optimization/79390
27358 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27359 threading case even more.
27361 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27364 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27365 for quad_address_p for TImode, instead of just not indexed_address.
27367 2017-04-12 Richard Biener <rguenther@suse.de>
27368 Bernd Edlinger <bernd.edlinger@hotmail.de>
27370 PR middle-end/79671
27371 * alias.c (component_uses_parent_alias_set_from): Handle
27372 TYPE_TYPELESS_STORAGE.
27373 (get_alias_set): Likewise.
27374 * tree-core.h (tree_type_common): Add typeless_storage flag.
27375 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27376 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27377 for types containing members with TYPE_TYPELESS_STORAGE.
27378 (place_field): Likewise.
27379 (layout_type): Likewise for ARRAY_TYPE.
27380 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27381 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27382 TYPE_TYPELESS_STORAGE.
27383 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27385 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27388 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27389 first argument to type.
27391 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27395 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27396 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27397 (rs6000_expand_binop_builtin): Likewise.
27398 (rs6000_expand_ternop_builtin): Likewise; also add missing
27399 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27400 vshasigma built-ins.
27401 * doc/extend.texi: Document that vec_xxpermdi's third argument
27402 must be a constant.
27404 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27406 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27407 Use shift_const cost parameter when calculating gain of STV shifts.
27409 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27411 PR rtl-optimization/70478
27412 * lra-constraints.c (process_alt_operands): Check memory for
27413 disfavoring memory insn operand.
27415 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27417 PR middle-end/80100
27418 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27419 left shift in unsigned HOST_WIDE_INT type.
27421 PR rtl-optimization/80385
27422 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27423 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27426 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27427 if they have any depend clauses.
27429 2017-04-11 Martin Liska <mliska@suse.cz>
27432 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27433 * ipa-split.c (split_function): Create a local comdat symbol
27434 if caller is in a comdat group.
27436 2017-04-11 Martin Liska <mliska@suse.cz>
27439 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27442 2017-04-11 Martin Sebor <msebor@redhat.com>
27444 PR middle-end/80364
27445 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27446 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27448 (directive::set_width, directive::set_precision, format_character):
27450 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
27453 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
27456 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27457 conflict, set target->arch_name instead of target->cpu_name.
27459 2017-04-11 Richard Biener <rguenther@suse.de>
27461 PR tree-optimization/80374
27462 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27463 build_zero_cst, remove fold_convertible_p check again.
27465 2017-04-11 Martin Liska <mliska@suse.cz>
27468 * ubsan.c (instrument_object_size): Do not instrument register
27471 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27474 * config/i386/i386-builtin-types.def
27475 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
27476 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
27477 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
27478 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
27479 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
27480 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
27481 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
27482 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
27483 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
27484 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
27485 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
27486 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
27487 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
27488 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
27489 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
27490 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
27491 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
27492 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
27493 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
27494 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
27495 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
27496 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
27497 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
27498 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
27499 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
27500 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
27501 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
27502 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
27503 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
27504 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
27505 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
27506 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
27507 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
27508 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
27509 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
27510 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
27511 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
27512 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
27513 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
27514 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
27515 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
27516 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
27517 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
27518 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
27519 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
27520 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
27522 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
27523 flag to second_arg_count, handle 4 argument function type _COUNT
27524 aliases, handle second_arg_count on second argument rather than last.
27526 2017-04-10 Jeff Law <law@redhat.com>
27528 PR tree-optimization/80374
27529 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
27530 record anything if we can not convert integer_zero_node to the
27533 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27536 * config/rs6000/rs6000.c (rs6000_option_override_internal):
27537 Enhance special handling given to the TARGET_P9_MINMAX option in
27538 relation to certain other options.
27540 2017-04-10 Bin Cheng <bin.cheng@arm.com>
27542 PR tree-optimization/80153
27543 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
27544 remove POINTER_PLUS_EXPR's base part directly, rather than through
27547 2017-04-10 Richard Biener <rguenther@suse.de>
27548 Bin Cheng <bin.cheng@arm.com>
27550 PR tree-optimization/80153
27551 * tree-affine.c (aff_combination_to_tree): Get base pointer from
27552 the first element of pointer type aff_tree. Build result expr in
27554 (add_elt_to_tree): Convert to type unconditionally. Remove other
27555 fold_convert calls.
27556 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
27557 (rewrite_use_nonlinear_expr): Check invariant using iv information.
27559 2017-04-10 Richard Biener <rguenther@suse.de>
27561 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27564 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
27566 PR rtl-optimization/70478
27567 * lra-constraints.c (curr_small_class_check): New.
27568 (update_and_check_small_class_inputs): New.
27569 (process_alt_operands): Update curr_small_class_check. Disfavor
27570 alternative insn memory operands. Check available regs for small
27573 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27576 * config/mips/mips.opt (-mvirt): Update description.
27577 * doc/invoke.texi (-mvirt): Likewise.
27579 2017-04-10 Richard Biener <rguenther@suse.de>
27581 PR middle-end/80362
27582 * fold-const.c (fold_binary_loc): Look at unstripped ops when
27583 looking for NEGATE_EXPR in -A / -B to A / B folding.
27585 2017-04-10 Martin Liska <mliska@suse.cz>
27587 PR gcov-profile/80224
27588 * gcov.c (print_usage): Fix usage string.
27589 (get_gcov_intermediate_filename): Remove.
27590 (output_gcov_file): Use both for normal and intermediate format.
27591 (generate_results): Do not initialize special file for
27592 intermediate format.
27594 2017-04-10 Richard Biener <rguenther@suse.de>
27596 PR tree-optimization/80304
27597 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
27600 2017-04-10 Nathan Sidwell <nathan@acm.org>
27603 * config/rs6000/rs6000.c (rs6000_vector_type): New.
27604 (rs6000_init_builtins): Use it.
27606 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27608 * config/arm/arm.md (<mrc>): Add mode to SET source.
27609 (<mrrc>): Likewise.
27611 2017-04-10 Richard Biener <rguenther@suse.de>
27613 PR middle-end/80344
27614 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
27616 2017-04-10 Jakub Jelinek <jakub@redhat.com>
27619 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
27620 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
27621 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
27622 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
27623 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
27624 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
27625 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
27626 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
27627 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
27628 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
27629 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
27630 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
27631 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
27632 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
27633 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
27634 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
27635 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
27636 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
27637 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
27638 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
27639 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
27640 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
27641 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
27643 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
27645 PR rtl-optimization/70478
27646 * lra-constraints.c: Reverse the last patch.
27648 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
27650 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
27651 Add comment for WCHAR_T.
27653 2017-04-08 Martin Liska <mliska@suse.cz>
27656 2017-04-07 Martin Liska <mliska@suse.cz>
27659 * ipa-split.c (split_function): Add function part to a same comdat
27662 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27665 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
27667 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
27669 * rs6000/rs6000.c (vec_load_pendulum): Rename...
27670 (vec_pairing): ...to this.
27671 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
27672 (rs6000_sched_init): Adjust for name change.
27673 (struct rs6000_sched_context): Likewise.
27674 (rs6000_init_sched_context): Likewise.
27675 (rs6000_set_sched_context): Likewise.
27677 2017-04-07 Jakub Jelinek <jakub@redhat.com>
27683 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
27685 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
27686 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
27687 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
27689 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
27691 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
27693 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27695 PR rtl-optimization/70703
27696 * ira-color.c (update_conflict_hard_regno_costs): Use
27697 int64_t instead of HOST_WIDE_INT.
27699 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27701 PR rtl-optimization/70478
27702 * lra-constraints.c (process_alt_operands): Disfavor alternative
27703 insn memory operands.
27705 2017-04-07 Jeff Law <law@redhat.com>
27707 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
27708 CALL and NOTE_INSN_CALL_ARG_LOCATION.
27710 2017-04-07 Martin Liska <mliska@suse.cz>
27713 * config/aarch64/aarch64.c (aarch64_process_target_attr):
27714 Show error message instead of an ICE.
27716 2017-04-07 Martin Liska <mliska@suse.cz>
27719 * ipa-split.c (split_function): Add function part to a same comdat
27722 2017-04-07 Richard Biener <rguenther@suse.de>
27724 PR middle-end/80341
27725 * tree.c (get_unwidened): Also handle ! for_type case for
27727 * convert.c (do_narrow): Split out from ...
27728 (convert_to_integer_1): ... here. Do not pass final truncation
27729 type to get_unwidened for TRUNC_DIV_EXPR.
27731 2017-04-07 Richard Biener <rguenther@suse.de>
27733 * tree-affine.c (wide_int_ext_for_comb): Take type rather
27735 (aff_combination_const): Adjust.
27736 (aff_combination_scale): Likewise.
27737 (aff_combination_add_elt): Likewise.
27738 (aff_combination_add_cst): Likewise.
27739 (aff_combination_convert): Likewise.
27740 (add_elt_to_tree): Likewise. Remove unused argument.
27741 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
27743 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
27745 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
27747 * config/arm/arm.c (arm_default_short_enums): Use
27748 ARM_DEFAULT_SHORT_ENUMS.
27749 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
27751 2017-04-06 Jakub Jelinek <jakub@redhat.com>
27754 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
27755 members with redundant out-of-class redeclaration.
27757 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27760 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
27761 * config/i386/i386.md (*zero_extendsidi2):
27762 Add (?*x,*x) and (?*v,*v) alternatives.
27764 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27767 * config/i386/i386.c (ix86_expand_builtin)
27768 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
27769 mode from insn data. Convert operands to insn operand mode.
27770 Copy operands that don't satisfy insn predicate to a register.
27772 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
27774 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
27777 2017-04-06 Richard Biener <rguenther@suse.de>
27779 PR tree-optimization/80334
27780 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
27781 preserve alignment of accesses.
27783 2017-04-06 Richard Biener <rguenther@suse.de>
27785 PR tree-optimization/80262
27786 * tree-sra.c (build_ref_for_offset): Preserve address-space
27788 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
27789 Drop useless address-space information on MEM_REF offsets.
27791 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
27793 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
27795 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27797 PR rtl-optimization/70703
27798 * ira-color.c (update_conflict_hard_regno_costs): Use
27799 HOST_WIDE_INT instead of long.
27801 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
27804 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
27805 not defined for x86_64 target. Add -mmmx target option when __SSE2__
27807 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
27808 for x86_64 target. Handle -m3dnowa option.
27810 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27812 PR rtl-optimization/70703
27813 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
27814 (update_conflict_hard_regno_costs): Use long instead of unsigned
27815 arithmetic for cost calculation.
27817 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27818 Bernd Edlinger <bernd.edlinger@hotmail.de>
27821 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
27824 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
27827 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
27828 ptr_mode with Pmode throughout.
27829 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
27830 into probe_stack_range and use DImode.
27832 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
27835 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
27836 call_eh_return is true.
27838 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27840 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
27841 Initialize last_match_fntype_index.
27843 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27846 * tree-nvr.c: Include internal-fn.h.
27847 (pass_return_slot::execute): Ignore internal calls without
27850 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27851 Richard Biener <rguenther@suse.de>
27854 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
27855 captures used multiple times, except for the last use.
27856 * generic-match-head.c: Include gimplify.h.
27858 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27860 PR tree-optimization/79390
27861 * target.h (struct noce_if_info): Declare.
27862 * targhooks.h (default_noce_conversion_profitable_p): Declare.
27863 * target.def (noce_conversion_profitable_p): New target hook.
27864 * ifcvt.h (struct noce_if_info): New type, moved from ...
27865 * ifcvt.c (struct noce_if_info): ... here.
27866 (noce_conversion_profitable_p): Renamed to ...
27867 (default_noce_conversion_profitable_p): ... this. No longer
27869 (noce_try_store_flag_constants, noce_try_addcc,
27870 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
27871 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
27872 instead of noce_conversion_profitable_p.
27873 * config/i386/i386.c: Include ifcvt.h.
27874 (ix86_option_override_internal): Don't override
27875 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
27876 (ix86_noce_conversion_profitable_p): New function.
27877 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
27878 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
27879 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
27880 * doc/tm.texi: Regenerated.
27882 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27884 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
27887 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
27890 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
27891 instructions for small multiply cores.
27893 2017-04-04 Jeff Law <law@redhat.com>
27895 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
27897 (mips_expand_vec_perm_const): Initialize elements in orig_perm
27898 that are not set by the loop over the elements.
27900 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27903 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
27904 int mode, convert_modes it to mode as unsigned, otherwise use
27905 lowpart_subreg to mode rather than SImode.
27906 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
27907 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
27908 Use DImode instead of SImode for the shift count operand.
27909 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
27912 2017-04-04 Richard Biener <rguenther@suse.de>
27914 PR middle-end/80281
27915 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
27916 arithmetic done for the negate or the plus. Simplify.
27917 (A - (-B) -> A + B): Likewise.
27918 * fold-const.c (split_tree): Make sure to not negate pointers.
27920 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
27922 PR rtl-optimization/60818
27923 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
27924 a compare of comparisons with the thing compared if this results
27925 in a different machine mode.
27927 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
27929 * alias.c (base_alias_check): Fix typo in comment.
27930 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
27931 * cgraphunit.c (symbol_table::compile): Likewise.
27932 * collect2.c (maybe_run_lto_and_relink): Likewise.
27933 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
27934 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
27935 * config/avr/avr.c (avr_map_op_t): Likewise.
27936 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
27937 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
27938 * config/epiphany/epiphany.md (movcc): Likewise.
27939 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
27940 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
27942 * config/mips/mips.c (mips_save_restore_reg): Likewise.
27943 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
27944 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
27945 * config/sh/sh.c (sh_rtx_costs): Likewise.
27946 * fold-const.c (fold_truth_andor): Likewise.
27947 * genautomata.c (collapse_flag): Likewise.
27948 * gengtype.h (struct type::u::s): Likewise.
27949 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
27950 * input.c (FORMAT_AMOUNT): Likewise.
27951 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
27952 (known_aggs_to_agg_replacement_list): Likewise.
27953 * ipa-inline-analysis.c: Likewise.
27954 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
27955 * ipa-polymorphic-call.c
27956 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
27957 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
27958 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
27960 * modulo-sched.c (apply_reg_moves): Likewise.
27961 * omp-expand.c (build_omp_regions_1): Likewise.
27962 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
27963 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
27964 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
27965 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
27966 * value-prof.c: Likewise.
27967 * var-tracking.c (val_reset): Likewise.
27969 2017-04-03 Richard Biener <rguenther@suse.de>
27971 PR tree-optimization/80275
27972 * fold-const.c (split_address_to_core_and_offset): Handle
27975 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
27977 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
27978 descriptors is at least equal to that of functions.
27980 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27982 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
27984 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27987 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
27988 (mov<IMOD4:mode>): New expander.
27989 (*mov<IMOD4:mode>_internal): New insn and split pattern.
27991 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
27993 PR rtl-optimization/79405
27994 * fwprop.c (propagations_left): New variable.
27995 (forward_propagate_into): Decrement it.
27996 (fwprop_init): Initialize it.
27997 (fw_prop): If the variable has reached zero, stop propagating.
27998 (fwprop_addr): Ditto.
28000 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28003 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
28004 a FUNCTION_DECL, pass it as decl instead of origin to
28007 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
28009 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
28012 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
28015 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
28016 TARGET_VSX_SMALL_INTEGER.
28018 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28020 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28021 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
28023 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28025 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
28026 extraction from odd-numbered MSA register.
28028 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28030 PR middle-end/80173
28031 * expmed.c (store_bit_field_1): Don't attempt to create
28032 a word subreg out of hard registers wider than word if they
28033 have HARD_REGNO_NREGS of 1 for their mode.
28035 PR middle-end/80163
28036 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
28037 conversions to integer types wider than word and pointer.
28040 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
28041 (rtx_equal_for_cselib_p): Pass 0 to it.
28042 * cselib.c (cselib_hasher::equal): Likewise.
28043 (rtx_equal_for_cselib_1): Add depth argument. If depth
28044 is 128, don't look up VALUE locs and punt. Increment
28045 depth in recursive calls when walking VALUE locs.
28047 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
28049 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
28050 (make_gcov_file_name): Use the canonical path name for generating
28052 (read_line): Fix handling of files with ascii null bytes.
28054 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
28056 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
28057 to initialise a vector register instead
28058 of using a const_int.
28060 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28062 PR translation/80189
28063 * gimplify.c (omp_default_clause): Use %qs instead of %s in
28064 diagnostic messages.
28066 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
28069 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
28070 (dfp_diex_<mode>): Update mode of operand 1.
28071 * doc/extend.texi (dxex, dxexq): Document change to return type.
28072 (diex, diexq): Document change to argument type.
28074 2017-03-30 Martin Jambor <mjambor@suse.cz>
28077 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
28078 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
28079 it reflects the signature changes performed at the callee side.
28080 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
28081 to cgraph_build_function_type_skip_args.
28082 (build_function_decl_skip_args): Adjust call to the above function.
28084 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28087 * config/i386/sse.md
28088 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
28089 register as dest whenever it is a MEM not rtx_equal_p to the
28090 corresponding dup operand, and when forcing into reg move the
28091 reg into the memory afterwards.
28092 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
28093 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
28094 for the force_reg mode.
28095 (avx512vl_vextractf128<mode>): Use register as dest either
28096 always when a MEM, or when it is a MEM not rtx_equal_p to the
28097 corresponding dup operand, or even not when it is a CONST_VECTOR
28098 depending on the mode and lo vs. hi.
28099 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
28101 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
28102 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
28103 Likewise. Require that operands[2] is even.
28104 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
28105 Remove extraneous parens. Require that operands[2] is a multiple
28107 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
28108 operands[0] is a MEM if <mask_applied>, the predicates/constraints
28109 disallow memory then.
28111 2017-03-30 Richard Biener <rguenther@suse.de>
28113 PR tree-optimization/77498
28114 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
28115 to non-constants over backedges.
28117 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
28119 PR rtl-optimization/80233
28120 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
28121 as last_combined_insn. Do not test for BARRIER_P separately.
28123 2017-03-29 Andreas Schwab <schwab@suse.de>
28126 * calls.c (prepare_call_address): Convert funexp to Pmode before
28127 copying to temp reg.
28129 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28131 PR tree-optimization/80158
28132 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
28133 Handle possible future case of more than one alternate
28135 (replace_rhs_if_not_dup): Likewise.
28136 (replace_one_candidate): Likewise.
28138 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
28140 PR rtl-optimization/80193
28141 * ira.c (ira): Do not check allocation for LRA.
28143 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
28145 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
28146 (nvptx_output_simt_exit): Declare.
28147 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
28148 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
28149 (init_softstack_frame): Move initialization of crtl->is_leaf to...
28150 (nvptx_declare_function_name): ...here. Emit declaration of local
28151 memory space buffer for omp_simt_enter insn.
28152 (nvptx_output_unisimt_switch): New.
28153 (nvptx_output_softstack_switch): New.
28154 (nvptx_output_simt_enter): New.
28155 (nvptx_output_simt_exit): New.
28156 * config/nvptx/nvptx.h (struct machine_function): New fields
28157 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
28158 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
28159 (UNSPECV_SIMT_EXIT): Ditto.
28160 (omp_simt_enter_insn): New insn.
28161 (omp_simt_enter): New expansion.
28162 (omp_simt_exit): New insn.
28163 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
28165 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
28166 (expand_GOMP_SIMT_ENTER_ALLOC): New.
28167 (expand_GOMP_SIMT_EXIT): New.
28168 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
28169 (GOMP_SIMT_ENTER_ALLOC): Ditto.
28170 (GOMP_SIMT_EXIT): Ditto.
28171 * target-insns.def (omp_simt_enter): New insn.
28172 (omp_simt_exit): Ditto.
28173 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
28175 (lower_rec_simd_input_clauses): Implement SIMT privatization.
28176 (lower_rec_input_clauses): Likewise.
28177 (lower_lastprivate_clauses): Handle SIMT privatization.
28179 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28180 (ompdevlow_adjust_simt_enter): New.
28181 (find_simtpriv_var_op): New.
28182 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28183 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28185 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28186 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28187 (copy_decl_for_dup_finish): Ditto.
28189 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28191 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28194 * config/i386/i386.c (ix86_option_override_internal): Always
28195 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28197 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28199 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28201 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28203 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28204 mark new edge's irreducible flag accordign to it.
28205 (vect_do_peeling): Check loop preheader edge's irreducible flag
28206 and pass it to function slpeel_add_loop_guard.
28208 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28210 PR tree-optimization/80218
28211 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28212 Update block frequencies and counts.
28214 2017-03-28 Richard Biener <rguenther@suse.de>
28216 PR tree-optimization/78644
28217 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28218 of a simplification result we may not use it at all.
28220 2017-03-28 Richard Biener <rguenther@suse.de>
28223 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28224 without arguments, generate default definition of a SSA name.
28226 2017-03-28 Richard Biener <rguenther@suse.de>
28228 PR middle-end/80222
28229 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28230 TYPE_REF_CAN_ALIAS_ALL references.
28231 * fold-const.c (fold_indirect_ref_1): Likewise.
28233 2017-03-28 Martin Liska <mliska@suse.cz>
28236 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28237 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28239 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28240 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28242 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28243 (EXTRA_SPECS): Define.
28244 (SUBTARGET_EXTRA_SPECS): Likewise.
28245 (SUBTARGET_CPP_SPEC): Likewise.
28246 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28247 SUBTARGET_EXTRA_SPECS.
28248 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28250 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28252 * config/arc/simdext.md (vst64_insn): Update pattern.
28253 (vld32wh_insn): Likewise.
28254 (vld32wl_insn): Likewise.
28255 (vld64_insn): Likewise.
28256 (vld32_insn): Likewise.
28258 2017-03-28 Marek Polacek <polacek@redhat.com>
28261 * fold-const.c (fold_comparison): Use protected_set_expr_location
28262 instead of SET_EXPR_LOCATION.
28264 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28266 * tree.c (add_expr): Avoid name lookup warning.
28268 2017-03-27 Jeff Law <law@redhat.com>
28270 PR tree-optimization/80216
28271 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28272 function name. Limit recursion depth.
28273 (record_temporary_equivalences): Corresponding changes.
28275 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28277 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28280 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28283 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28285 * cfgcleanup.c (reg_note_cfa_p): New array.
28286 (insns_have_identical_cfa_notes): New function.
28287 (old_insns_match_p): Don't cross-jump in between /f
28288 and non-/f instructions. If both i1 and i2 are frame related,
28289 verify all CFA notes, their order and content.
28291 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28294 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28295 HImode and SImode with zero extend to DImode to one insn.
28296 (bswap<mode>2_extenddi): Likewise.
28297 (bswapsi2_extenddi): Likewise.
28298 (bswaphi2_extendsi): Likewise.
28299 (bswaphi2): Combine bswap HImode and SImode into one insn.
28300 Separate memory insns from swapping register.
28301 (bswapsi2): Likewise.
28302 (bswap<mode>2): Likewise.
28303 (bswaphi2_internal): Delete, no longer used.
28304 (bswapsi2_internal): Likewise.
28305 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28306 store, and gpr<-gpr swap insns.
28307 (bswap<mode>2_store): Likewise.
28308 (bswaphi2_reg): Register only splitter, combine with the splitter.
28309 (bswaphi2 splitter): Likewise.
28310 (bswapsi2_reg): Likewise.
28311 (bswapsi2 splitter): Likewise.
28312 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28313 the insns into load, store, and register/register insns.
28314 (bswapdi2_ldbrx): Likewise.
28315 (bswapdi2_load): Likewise.
28316 (bswapdi2_store): Likewise.
28317 (bswapdi2_reg): Likewise.
28319 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28321 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28322 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28324 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28327 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28329 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28330 special handling for target option conflicts between dform
28331 options (-mpower9-dform, -mpower9-dform-vector,
28332 -mpower9-dform-scalar) and -mno-direct-move.
28334 2017-03-27 Richard Biener <rguenther@suse.de>
28336 PR tree-optimization/80181
28337 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28339 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28341 * config/arc/predicates.md (move_double_src_operand): Replace the
28342 call to move_double_src_operand with a call to address_operand.
28344 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28346 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28347 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28348 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28350 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28352 * config/arc/predicates.md (long_immediate_loadstore_operand):
28353 Consider scaled addresses cases.
28355 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28357 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28358 restored when in interrupt.
28359 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28360 doesn't have delay slot.
28362 2017-03-27 Richard Biener <rguenther@suse.de>
28365 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28366 inlined thunk clones.
28368 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28371 * asan.c (instrument_derefs): Copy over last operand from
28372 original COMPONENT_REF to the new COMPONENT_REF with
28373 DECL_BIT_FIELD_REPRESENTATIVE.
28374 * ubsan.c (instrument_object_size): Likewise.
28376 2017-03-27 Richard Biener <rguenther@suse.de>
28378 PR tree-optimization/80170
28379 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28380 sure DR/SCEV didnt fold in constants we do not see when looking
28381 at the reference base alignment.
28383 2017-03-27 Richard Biener <rguenther@suse.de>
28385 PR middle-end/80171
28386 * gimple-fold.c (fold_ctor_reference): Properly guard against
28387 NULL return value from canonicalize_constructor_val.
28389 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28392 * config/i386/i386.c (ix86_expand_builtin)
28393 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28394 flags reg setting and flags reg using instructions.
28395 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28396 clobbering instructions to zero extend op2.
28398 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28400 * doc/install.texi (Configuration) <--with-aix-soname>:
28401 Update link to AIX ld.
28403 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28405 PR rtl-optimization/80160
28406 PR rtl-optimization/80159
28407 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28408 reg_alternate_class into account.
28410 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28413 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28414 to consider in curr_insn_transform.
28416 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28418 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28419 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28420 and emit_mode_inner.
28422 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28424 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28425 argument to the overloaded builtin variants. Use the new flag to
28426 deprecate certain builtin variants.
28427 * config/s390/s390-builtin-types.def: Add new builtin types.
28428 * config/s390/s390-builtins.h: Support new flags field for
28429 overloaded builtins.
28430 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28431 (s390_macro_to_expand): Enable vector float data type.
28432 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28433 builtins by incrementing the __VEC__ version number.
28434 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28436 (s390_resolve_overloaded_builtin): Emit error messages depending
28437 on the builtin flags.
28438 * config/s390/s390.c (s390_expand_builtin): Support additional
28439 flags argument. Change error message to match the messages
28440 emitted in s390-c.c.
28441 * config/s390/s390.md: New UNSPEC_* constants.
28442 (op_type): Add new instruction types.
28443 * config/s390/vecintrin.h: Add new builtins and test data class
28445 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28446 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28447 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28448 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28449 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28450 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28452 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
28453 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
28454 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
28455 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28457 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28458 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28459 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28460 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28462 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28463 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28464 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28465 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28466 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28467 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28468 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28470 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
28471 ("vec_scatter_element<V_HW_4:mode>_DI")
28472 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
28473 ("vec_fpint<mode>", "vflls")
28474 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
28475 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
28476 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
28477 ("*vec_cmphe<mode>_cc"): ... these.
28479 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
28480 mode constant instead of magic value.
28482 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28484 * config/s390/s390.c (s390_expand_vec_compare): Support other
28485 vector floating point modes than just V2DF.
28486 (s390_expand_vcond): Likewise.
28487 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
28488 (s390_cannot_change_mode_class): Prevent mode changes between TF
28489 and V1TF in vector registers.
28490 * config/s390/s390.md (DF, SF): New mode attributes.
28491 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
28492 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
28493 SFmode support for VRs.
28494 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
28496 (VFT, VF_HW): New mode iterators.
28497 (vw, sdx): New mode attributes.
28498 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
28499 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
28500 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
28501 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
28502 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
28503 also the new vector floating point modes. Renaming to ...
28505 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
28506 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
28507 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
28508 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
28509 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
28510 ("vec_unordered<mode>"): ... these.
28512 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
28513 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
28514 ("*vec_extendv2df"): New insn definitions.
28516 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28518 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
28519 ("mulditi3_2", "*muldi3_sign"): New patterns.
28520 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
28521 rename the pattern definition.
28523 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28525 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
28527 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
28529 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28531 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
28532 instruction if possible.
28533 * config/s390/vector.md (vec_halfnumelts): New mode
28535 ("*vec_vllezlf<mode>"): New pattern.
28537 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28539 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
28540 ("popcountv4si2", "popcountv2di2"): Rename to ...
28541 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
28542 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
28544 ("popcount<mode>2_vxe"): New pattern.
28546 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28548 * common/config/s390/s390-common.c (processor_flags_table): Add
28550 * config.gcc: Add arch12.
28551 * config/s390/driver-native.c (s390_host_detect_local_cpu):
28552 Default to arch12 for unknown CPU model numbers.
28553 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
28554 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
28555 PROCESSOR_max sanity check.
28556 * config/s390/s390-opts.h (enum processor_type): Add
28558 * config/s390/s390.c (processor_table): Add arch12.
28559 (s390_expand_builtin): Add check for B_VXE flag.
28560 (s390_issue_rate): Add PROCESSOR_ARCH12.
28561 (s390_get_sched_attrmask): Likewise.
28562 (s390_get_unit_mask): Likewise.
28563 (s390_sched_score): Enable z13 scheduling for arch12.
28564 (s390_sched_reorder): Likewise.
28565 (s390_sched_variable_issue): Likewise.
28566 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
28568 (s390_tune_attr): Use z13 scheduling also for arch12.
28569 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
28570 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
28571 (TARGET_VXE_P): New macros.
28572 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
28573 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
28574 * config/s390/s390.opt: Add arch12 as processor_type.
28576 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28578 * config/s390/s390.md
28579 ("fixuns_truncdddi2", "fixuns_trunctddi2")
28580 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
28581 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
28583 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
28584 Rename expanders to ...
28586 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
28587 ("fixuns_truncdddi2_emu"): ... these.
28589 ("fixuns_trunc<mode>si2_emu"): New expander.
28591 ("*fixuns_truncdfdi2_z13"): Rename to ...
28592 ("*fixuns_truncdfdi2_vx"): ... this.
28594 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28596 * config/s390/2964.md: Remove the single element vector compare
28597 instructions which are no longer used.
28598 * config/s390/s390.c (s390_select_ccmode): Remove handling of
28600 (s390_canonicalize_comparison): Remove handling of DFmode
28602 (s390_expand_vec_compare_scalar): Remove function.
28603 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
28604 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
28606 ("*cmp<mode>_ccs"): Add wfcdb instruction.
28608 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28610 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
28612 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
28613 will anyway by matched by mov<mode>_64dfp.
28615 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28617 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
28618 vlef/vstef. Add missing operand to vleif.
28620 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28622 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
28623 pair for all vector types with 64 bit elements.
28624 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
28625 * config/s390/vector.md (V_HW_64): ... here.
28626 (V_128_NOSINGLE): New mode iterator.
28627 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
28628 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
28629 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
28630 ("*vec_load_pairv2di"): Change to ...
28631 ("*vec_load_pair<mode>"): ... this one.
28633 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28635 * config/s390/constraints.md: Add comments.
28636 (jKK): Reject element sizes > 8 bytes.
28637 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
28639 * config/s390/s390.md: Add the s_operand checks formerly in
28640 s390_split_ok_p to various splitters where they are still
28642 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
28643 for 128 bit vectors. Plus two splitters.
28645 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28647 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
28650 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28653 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
28654 error if the boundary argument is not constant.
28656 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28658 PR rtl-optimization/80112
28659 * loop-doloop.c (doloop_condition_get): Don't check condition
28660 if cmp isn't SET with IF_THEN_ELSE src.
28662 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28664 PR tree-optimization/80158
28665 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
28666 replacing a candidate statement, also replace it for the
28667 candidate's alternate interpretation.
28668 (replace_rhs_if_not_dup): Likewise.
28669 (replace_one_candidate): Likewise.
28671 2017-03-24 Richard Biener <rguenther@suse.de>
28673 PR tree-optimization/80167
28674 * graphite-isl-ast-to-gimple.c
28675 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
28677 (translate_isl_ast_to_gimple::get_rename): Likewise.
28679 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28681 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
28682 handling of certain combinations of target options, including the
28683 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
28684 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
28686 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28689 * config/arm/arm.md (*load_multiple): Add reload_completed to
28690 matching condition.
28692 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28693 Richard Biener <rguenth@suse.de>
28695 PR tree-optimization/79908
28696 PR tree-optimization/80136
28697 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28698 been cast away, gimplify_and_add suffices.
28700 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
28702 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
28704 2017-03-23 Richard Biener <rguenther@suse.de>
28706 PR tree-optimization/80032
28707 * gimplify.c (gimple_push_cleanup): Forced unconditional
28708 cleanups still have to go to the conditional_cleanups
28711 2017-03-22 Jakub Jelinek <jakub@redhat.com>
28713 PR tree-optimization/80072
28714 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
28716 (next_operand_entry_id): Change type to unsigned int.
28717 (sort_by_operand_rank): Make sure to return the right return value
28718 even if unsigned fields are bigger than INT_MAX.
28719 (struct oecount): Change cnt and id type to unsigned int.
28720 (oecount_hasher::equal): Formatting fix.
28721 (oecount_cmp): Make sure to return the right return value
28722 even if unsigned fields are bigger than INT_MAX.
28723 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
28726 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
28727 TREE_READONLY on result if writing it more than once.
28730 * doc/invoke.texi (-fsanitize=thread): Document that with
28731 -fnon-call-exceptions atomics are not able to throw
28735 * tsan.c: Include tree-eh.h.
28736 (instrument_builtin_call): Call maybe_clean_eh_stmt or
28737 maybe_clean_or_replace_eh_stmt where needed.
28738 (instrument_memory_accesses): Add cfg_changed argument.
28739 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
28740 if it returned true.
28741 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
28743 PR rtl-optimization/63191
28744 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
28745 wrapper function, moved the whole old content into ...
28746 (ix86_delegitimize_address_1): ... this. New inline function.
28747 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
28748 true as last argument instead of ix86_delegitimize_address.
28750 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28752 * config/aarch64/aarch64.c (generic_branch_cost): Copy
28753 cortexa57_branch_cost.
28755 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28757 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
28759 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28762 * doc/md.texi (Constraints): Document wA constraint.
28763 * config/rs6000/constraints.md (wA): New.
28764 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
28765 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
28766 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
28767 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
28769 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
28772 * gimplify.c (is_oacc_declared): New function.
28773 (oacc_default_clause): Use it to set default flags for acc declared
28774 variables inside parallel regions.
28775 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
28776 declared variables.
28777 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
28778 declare attribute to any decl as necessary.
28780 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
28783 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
28784 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
28785 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
28786 (arm_arch_lpae): This.
28787 * config/arm/arm.c (arm_arch7ve): Rename into ...
28788 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
28789 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
28792 2017-03-22 Martin Liska <mliska@suse.cz>
28795 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
28796 error message instead of an ICE.
28798 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28800 * doc/extend.texi (6.11 Additional Floating Types): Revise.
28802 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
28804 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28806 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28809 2017-03-21 Martin Sebor <msebor@redhat.com>
28811 * doc/extend.texi: Use "cannot" instead of "can't."
28812 * doc/hostconfig.texi: Same.
28813 * doc/install.texi: Same.
28814 * doc/invoke.texi: Same.
28815 * doc/loop.texi: Same.
28816 * doc/md.texi: Same.
28817 * doc/objc.texi: Same.
28818 * doc/rtl.texi: Same.
28819 * doc/tm.texi: Same.
28820 * doc/tm.texi.in: Same.
28821 * doc/trouble.texi: Same.
28823 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
28826 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
28827 (collect_checksum_attributes): Set it.
28828 (die_checksum_ordered): Use it.
28830 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28832 PR tree-optimization/79908
28833 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
28834 change: For a VA_ARG whose LHS has been cast away, use
28835 force_gimple_operand to construct the side effects.
28837 2017-03-21 David Malcolm <dmalcolm@redhat.com>
28839 PR translation/80001
28840 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
28841 more amenable to translation.
28842 (oacc_loop_auto_partitions): Likewise.
28844 2017-03-21 Marek Polacek <polacek@redhat.com>
28845 Martin Sebor <msebor@redhat.com>
28847 PR tree-optimization/80109
28848 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
28849 on INTEGRAL_TYPE_P.
28851 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28852 Segher Boessenkool <segher@kernel.crashing.org>
28855 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
28856 check reg_used_between_p between insn and one of succ or succ2
28857 depending on if succ is artificial insn not inserted into insn
28860 2017-03-21 Martin Liska <mliska@suse.cz>
28862 PR gcov-profile/80081
28863 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
28864 * doc/gcc.texi: Include gcov-dump stuff.
28865 * doc/gcov-dump.texi: New file.
28867 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
28869 PR rtl-optimization/79150
28870 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
28871 conditional jump, if the jump is the last insn of the loop.
28873 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28874 Richard Biener <rguenth@suse.de>
28876 PR tree-optimization/79908
28877 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28878 been cast away, use force_gimple_operand to construct the side
28881 2017-03-21 Martin Liska <mliska@suse.cz>
28883 PR libfortran/79956
28884 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
28887 2017-03-21 Brad Spengler <spender@grsecurity.net>
28890 * plugin.c (htab_hash_plugin): New function.
28891 (add_new_plugin): Use it and adjust.
28892 (parse_plugin_arg_opt): Adjust.
28893 (init_one_plugin): Likewise.
28895 2017-03-21 Richard Biener <rguenther@suse.de>
28897 PR tree-optimization/80032
28898 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
28899 if set force the cleanup to happen unconditionally.
28900 (gimplify_target_expr): Push inserted clobbers with force_uncond
28901 to avoid them being removed by control-dependent DCE.
28903 2017-03-21 Richard Biener <rguenther@suse.de>
28905 PR tree-optimization/80122
28906 * tree-inline.c (copy_bb): Do not expans va-arg packs or
28907 va_arg_pack_len when the inlined call stmt requires pack
28909 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
28911 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28914 * tsan.c (instrument_builtin_call): If the memory model argument
28915 is not a constant, assume it is valid.
28918 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
28921 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28923 PR rtl-optimization/79910
28924 * combine.c (can_combine_p): Do not allow combining an I0 or I1
28925 if its dest is used by an insn before I2 (other than the combined
28926 insns themselves, which are properly handled already).
28928 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28931 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28933 * combine.c (record_used_regs): New static function.
28934 (try_combine): Handle situations where there is an additional
28935 instruction between I2 and I3 which needs to have a LOG_LINK
28939 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28941 * combine.c (try_combine): Delete redundant i1 test. Call
28942 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28944 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28947 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
28948 alternatives 13/14.
28950 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28952 PR tree-optimization/80054
28953 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
28954 the optimization if a PHI or any of its arguments is not dominated
28955 by the candidate's basis. Use gphi* rather than gimple* as
28957 (replace_profitable_candidates): Clean up a gimple* variable that
28958 should be a gphi* variable.
28960 2017-03-20 Martin Sebor <msebor@redhat.com>
28963 * doc/extend.texi (attribute constructor): Document present limitation.
28965 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
28968 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
28969 __POWER9_VECTOR__ #ifdef control, change template definition to
28970 use Power9-specific built-in function.
28971 (vec_any_eq): Likewise.
28972 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
28973 to control outcomes from this test.
28974 (vector_ae_<mode>p): For VEC_F modes, likewise.
28976 2017-03-20 Ian Lance Taylor <iant@google.com>
28978 * config/i386/i386.c (ix86_function_regparm): Save an extra
28979 register for -fsplit-stack with DECL_STATIC_CHAIN.
28981 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28984 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
28985 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
28987 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28989 * config/riscv/riscv.c (riscv_print_operand): Use "fence
28991 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
28994 2017-03-20 Marek Polacek <polacek@redhat.com>
28997 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
28999 2017-03-20 Richard Biener <rguenther@suse.de>
29001 PR tree-optimization/80113
29002 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
29003 allocate extra SSA name for PHI def.
29004 (add_close_phis_to_outer_loops): Likewise.
29005 (add_close_phis_to_merge_points): Likewise.
29006 (copy_loop_close_phi_args): Likewise.
29007 (copy_cond_phi_nodes): Likewise.
29009 2017-03-20 Martin Liska <mliska@suse.cz>
29011 PR middle-end/79753
29012 * tree-chkp.c (chkp_build_returned_bound): Do not build
29013 returned bounds for a LHS that's not a BOUNDED_P type.
29015 2017-03-20 Martin Liska <mliska@suse.cz>
29019 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
29020 COMPLEX_CST and VECTOR_CST.
29022 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29025 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
29026 target operand. A new splitter adds the clobber statement in case
29027 the target operand is dead anyway.
29029 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
29031 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
29032 to age-old versions of binutils and glibc.
29034 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
29036 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
29038 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29040 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
29042 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29044 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
29045 requirement for binutils 2.13.
29047 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29049 * combine.c (try_combine): Delete redundant i1 test. Call
29050 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29052 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
29054 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
29055 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
29057 <riscv64-*-elf>: Re-arrange section
29058 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
29059 <riscv32-*-linux>: Likewise.
29060 <riscv64-*-elf>: Likewise
29061 <riscv64-*-linux>: Likewise.
29063 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
29066 * aarch64.opt(verbose-cost-dump): Fix typo.
29068 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
29071 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
29072 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
29074 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29076 * reload.c (find_reloads): When reloading a nonoffsettable address,
29077 use RELOAD_OTHER for it and its address reloads.
29079 PR rtl-optimization/79910
29080 * combine.c (record_used_regs): New static function.
29081 (try_combine): Handle situations where there is an additional
29082 instruction between I2 and I3 which needs to have a LOG_LINK
29085 2017-03-17 Jeff Law <law@redhat.com>
29087 PR tree-optimization/71437
29088 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
29089 conditional in the hash table first.
29090 (vrp_dom_walker::before_dom_children): Extract condition from
29091 ASSERT_EXPR. Record condition, its inverion and any implied
29092 conditions as well.
29094 2017-03-17 Marek Polacek <polacek@redhat.com>
29095 Markus Trippelsdorf <markus@trippelsdorf.de>
29097 PR tree-optimization/80079
29098 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
29101 2017-03-17 Richard Biener <rguenther@suse.de>
29103 PR middle-end/80075
29104 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
29105 Properly verify the LHS before the RHS possibly claims to be
29107 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
29110 2017-03-17 Martin Jambor <mjambor@suse.cz>
29112 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
29113 (List of -O2 options): Likewise.
29114 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
29117 2017-03-17 Tom de Vries <tom@codesourcery.com>
29119 * gcov-dump.c (print_usage): Print bug_report_url.
29121 2017-03-17 Richard Biener <rguenther@suse.de>
29123 PR middle-end/80050
29124 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
29125 (parser::peek): Likewise.
29127 2017-03-17 Richard Biener <rguenther@suse.de>
29129 PR tree-optimization/80048
29130 * sese.c (free_sese_info): Properly release rename_map and
29131 copied_bb_map elements.
29133 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
29135 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
29136 Add linked-list forward and backlinks. Insert on
29137 construction, remove on destruction.
29138 (class pass_store_merging): Add m_stores_head field.
29139 (pass_store_merging::terminate_and_process_all_chains):
29140 Iterate over m_stores_head list.
29141 (pass_store_merging::terminate_all_aliasing_chains):
29143 (pass_store_merging::execute): Check for debug stmts first.
29144 Push new chains onto the m_stores_head stack.
29146 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29149 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
29150 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
29151 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
29153 2017-03-16 Jeff Law <law@redhat.com>
29155 PR tree-optimization/71437
29156 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
29157 member function. Implementation moved into after_dom_children
29158 member function and into the threader's thread_outgoing_edges
29160 (dom_opt_dom_walker::after_dom_children): Simplify by moving
29161 some code into new thread_outgoing_edges.
29162 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
29163 definition. Simplify marker handling (do it here). Assume we always
29164 have the available expression and the const/copies tables.
29165 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
29167 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
29168 * tree-vrp.c (equiv_stack): No longer file scoped.
29169 (vrp_dom_walker): New class.
29170 (vrp_dom_walker::before_dom_children): New member function.
29171 (vrp_dom_walker::after_dom_children): Likewise.
29172 (identify_jump_threads): Setup domwalker. Use it rather than
29173 walking edges in a random order by hand. Simplify setup/finalization.
29174 (finalize_jump_threads): Remove.
29175 (vrp_finalize): Do not call identify_jump_threads here.
29176 (execute_vrp): Do it here instead and call thread_through_all_blocks
29179 PR tree-optimization/71437
29180 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29182 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29184 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29185 (dom_opt_dom_walker::thread_across_edge): Remove
29186 handle_dominating_asserts argument. All callers changed.
29187 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29188 changes. Remove calls to lhs_of_dominating_assert. Other
29189 uses of handle_dominating_asserts turn into unconditional code
29190 (simplify_control_stmt_condition_1): Likewise.
29191 (simplify_control_stmt_condition): Likewise.
29192 (thread_through_normal_block, thread_across_edge): Likewise.
29193 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29194 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29195 object if it is not an SSA_NAME.
29196 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29197 before calling into the VRP specific simplifiers.
29198 (identify_jump_threads): Remove handle_dominating_asserts
29201 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29204 * tree-diagnostic.c (default_tree_printer): No longer static.
29205 * tree-diagnostic.h (default_tree_printer): New prototype.
29207 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29209 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29210 Change ins into fmov.
29212 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29214 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29215 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29216 Use h_con constraint for operand 1.
29217 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29218 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29220 2017-03-15 Jeff Law <law@redhat.com>
29222 PR tree-optimization/71437
29223 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29224 (record_temporary_equivalences): Use it.
29226 PR tree-optimization/71437
29227 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29228 tree-ssa-scopedtables.
29229 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29230 (record_conditions, record_cond, vuse_eq): Likewise.
29231 (record_edge_info): Adjust to API tweak of record_conditions.
29232 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29233 (record_temporary_equivalences, optimize_stmt): Likewise.
29234 (eliminate_redundant_computations): Likewise.
29235 (record_equivalences_from_stmt): Likewise.
29236 * tree-ssa-scopedtables.c: Include options.h and params.h.
29237 (vuse_eq): New function, moved from tree-ssa-dom.c
29238 (build_and_record_new_cond): Likewise.
29239 (record_conditions): Likewise. Accept vector of conditions rather
29240 than edge_equivalence structure for first argument.
29241 for the first argument.
29242 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29243 from tree-ssa-dom.c.
29244 (avail_exprs_stack::record_cond): Likewise.
29245 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29246 from tree-ssa-dom.c.
29247 (avail_exprs_stack): Add new member functions lookup_avail_expr
29249 (record_conditions): Declare.
29251 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29254 * lra-constraints.c (process_alt_operands): Increase reject for
29255 reloading an input/output operand.
29257 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29260 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29261 insns to convert from signed/unsigned char/short to IEEE 128-bit
29263 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29265 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29268 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29269 subreg of inner mode for values already in registers.
29271 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29273 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29274 iteration reg is used after the loop.
29276 2017-03-14 Martin Sebor <msebor@redhat.com>
29278 PR tree-optimization/79800
29279 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29280 precision in negative-positive range.
29281 (format_floating): Call non-const overload with adjusted precision.
29283 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29286 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29289 2017-03-14 Martin Sebor <msebor@redhat.com>
29291 PR middle-end/80020
29292 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29293 * builtins.def (aligned_alloc): Use it.
29296 * Makefile.in (GTFILES): Add calls.c.
29297 * calls.c: Include "gt-calls.h".
29299 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29301 PR rtl-optimization/79728
29302 * regs.h (struct target_regs): New field
29303 x_contains_allocatable_regs_of_mode.
29304 (contains_allocatable_regs_of_mode): New macro.
29305 * reginfo.c (init_reg_sets_1): Initialize it, and change
29306 contains_reg_of_mode so it includes global regs as well.
29307 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29308 rather than contains_regs_of_mode.
29310 2017-03-14 Martin Liska <mliska@suse.cz>
29312 * doc/invoke.texi: Document options that can't be combined with
29313 -fcheck-pointer-bounds.
29315 2017-03-14 Martin Liska <mliska@suse.cz>
29317 PR middle-end/79831
29318 * doc/invoke.texi (-Wchkp): Document the option.
29320 2017-03-14 Martin Liska <mliska@suse.cz>
29322 * Makefile.in: Install gcov-dump.
29324 2017-03-14 Martin Liska <mliska@suse.cz>
29326 * multiple_target.c (expand_target_clones): Bail out for
29327 an invalid attribute.
29329 2017-03-14 Richard Biener <rguenther@suse.de>
29331 * alias.c (struct alias_set_entry): Pack properly.
29332 * cfgloop.h (struct loop): Likewise.
29333 * cse.c (struct set): Likewise.
29334 * ipa-utils.c (struct searchc_env): Likewise.
29335 * loop-invariant.c (struct invariant): Likewise.
29336 * lra-remat.c (struct cand): Likewise.
29337 * recog.c (struct change_t): Likewise.
29338 * rtl.h (struct address_info): Likewise.
29339 * symbol-summary.h (function_summary): Likewise.
29340 * tree-loop-distribution.c (struct partition): Likewise.
29341 * tree-object-size.c (struct object_size_info): Likewise.
29342 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29343 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29344 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29345 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29346 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29347 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29348 (struct _stmt_vec_info): Likewise.
29350 2017-03-14 Martin Liska <mliska@suse.cz>
29353 * multiple_target.c (create_dispatcher_calls): Check that
29354 a target can create a function dispatcher.
29356 2017-03-14 Martin Liska <mliska@suse.cz>
29359 * multiple_target.c (expand_target_clones): Drop local.local
29360 flag for default implementation.
29362 2017-03-14 Richard Biener <rguenther@suse.de>
29364 PR tree-optimization/80030
29365 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29367 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29369 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29370 gcc_fallthrough() instead of __attribute__((fallthrough));
29372 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29374 * doc/gcc.texi: Remove "up" link to (DIR).
29375 * doc/gccint.texi: Ditto.
29377 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29379 * doc/install.texi (Specific) <avr>: Remove reference to
29382 2017-03-13 Jeff Law <law@redhat.com>
29384 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29385 attribute rather than comments.
29387 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29388 match_scratch operand is highest.
29390 2017-03-13 Martin Liska <mliska@suse.cz>
29392 PR middle-end/78339
29393 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29394 is a CHKP clone, use original declaration.
29396 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29398 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29399 (arc_conditional_register_usage): Use a different allocation order
29400 when optimizing for size.
29401 * common/config/arc/arc-common.c (arc_option_optimization_table):
29402 Section anchors default on when optimizing for size.
29404 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29406 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29408 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29410 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29411 * config/arc/arc.md (cpu_facility): Add cd variant.
29412 (*movqi_insn): Add code density variant.
29413 (*movhi_insn): Likewise.
29414 (*movqi_insn): Likewise.
29415 (*addsi3_mixed): Likewise.
29416 (subsi3_insn): Likewise.
29418 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29420 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29422 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29424 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29425 expressions with MINUS and UNARY ops.
29427 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29430 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29432 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29433 between vec_select and vector argument.
29434 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29435 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29436 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29437 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29438 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29439 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29441 2017-03-13 Richard Biener <rguenther@suse.de>
29444 * params.def (vect-max-peeling-for-alignment): Fix typo.
29446 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29448 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29449 issue that only occurred with binutils below 2.18.
29451 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29453 * doc/install.texi (Specific) <cris-axis-elf>: No longer
29454 refer to binutils 2.11/2.12 minimum.
29456 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29458 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29459 ftp.kernel.org and simplify binutils requirement.
29461 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29463 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29465 (Optimize Options): Ditto. Also remove redundancy.
29467 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29469 PR translation/79848
29470 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
29472 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
29473 to G_ to avoid double translation.
29475 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29477 PR translation/79923
29478 * auto-profile.c (get_combined_location): Convert leading
29479 character of diagnostics to lower case and remove trailing period.
29480 (read_profile): Likewise for various diagnostics.
29481 * config/arm/arm.c (arm_option_override): Remove trailing period
29482 from various diagnostics.
29483 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
29484 (msp430_expand_delay_cycles): Likewise.
29486 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29489 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
29490 full command-line argument, rather than just "str".
29491 (aarch64_validate_march): Likewise.
29492 (aarch64_validate_mtune): Likewise.
29494 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
29496 PR rtl-optimization/78911
29497 * lra-assigns.c (must_not_spill_p): New function.
29498 (spill_for): Use it.
29500 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29502 PR tree-optimization/79981
29503 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
29504 ATOMIC_COMPARE_EXCHANGE ifn result.
29505 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
29506 IFN_ATOMIC_COMPARE_EXCHANGE.
29508 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29511 * opts.c (parse_sanitizer_options): Add missing question mark to
29512 "did you mean" message.
29514 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29516 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
29518 (VMULEUH_UNS): Likewise.
29519 (VMULOUB_UNS): Likewise.
29520 (VMULOUH_UNS): Likewise.
29521 * config/rs6000/rs6000.c (builtin_function_type): Remove
29522 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
29524 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29527 * read-rtl-function.c (function_reader::read_rtx_operand): Update
29528 x with result of extra_parsing_for_operand_code_0.
29529 (function_reader::extra_parsing_for_operand_code_0): Convert
29530 return type from void to rtx, returning x. When reading
29531 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
29532 larger size containing struct block_symbol.
29534 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
29536 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
29537 -mfloat128-hardware without -m64.
29539 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
29542 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
29543 entries to the case statement that marks unsigned arguments to
29544 overloaded functions.
29546 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29548 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
29549 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
29551 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
29554 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
29555 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
29557 2017-03-10 Martin Liska <mliska@suse.cz>
29561 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
29562 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
29565 2017-03-10 Olivier Hainque <hainque@adacore.com>
29567 * tree-switch-conversion (array_value_type): Start by resetting
29568 candidate type to it's main variant.
29570 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29572 PR rtl-optimization/79909
29573 * combine.c (try_combine): Use simplify_replace_rtx on individual
29574 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
29575 of the whole CALL_INSN_FUNCTION_USAGE.
29577 PR tree-optimization/79972
29578 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
29579 get_range_info on SSA_NAMEs. Formatting fixes.
29581 2017-03-10 Richard Biener <rguenther@suse.de>
29582 Jakub Jelinek <jakub@redhat.com>
29584 PR tree-optimization/77975
29585 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
29586 edge to be constant.
29587 (get_val_for): For constant x return it. Formatting fix.
29588 (loop_niter_by_eval): Avoid pointless looping if the next iteration
29589 would use the same bases as the current one.
29591 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29593 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
29594 instead of vec_select for V1TImode.
29595 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
29597 (VSX_LE_128): Add V1TI to this mode iterator.
29598 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
29599 (*vsx_le_perm_store_<mode>): Likewise.
29600 (pre-reload splitter for VSX stores): Likewise.
29601 (post-reload splitter for VSX stores): Likewise.
29602 (*vsx_xxpermdi2_le_<mode>): Likewise.
29603 (*vsx_lxvd2x2_le_<mode>): Likewise.
29604 (*vsx_stxvd2x2_le_<mode>): Likewise.
29606 2017-03-09 Michael Eager <eager@eagercon.com>
29608 Correct failures with --enable-checking=yes,rtl.
29610 * config/microblaze/microblaze.c (microblaze_expand_shift):
29611 Replace GET_CODE test with CONST_INT_P and INTVAL test with
29612 test for const0_rtx.
29613 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
29614 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
29616 2017-03-09 Richard Biener <rguenther@suse.de>
29618 PR tree-optimization/79977
29619 * graphite-scop-detection.c (scop_detection::merge_sese):
29620 Handle the case of extra exits to blocks dominating the entry.
29622 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
29624 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
29627 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
29629 PR rtl-optimization/79949
29630 * lra-constraints.c (process_alt_operands): Check memory when
29631 trying to predict a cycle. Print about the overall increase.
29633 2017-03-09 Richard Biener <rguenther@suse.de>
29635 PR middle-end/79971
29636 * gimple-expr.c (useless_type_conversion_p): Preserve
29637 TYPE_SATURATING for fixed-point types.
29639 2017-03-09 Richard Biener <rguenther@suse.de>
29642 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
29643 alignment of BLKmode params.
29645 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29648 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
29649 (VALL_NO_V2Q): Likewise.
29651 * config/aarch64/aarch64-simd.md
29652 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
29654 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
29655 VALL_NO_V2Q mode iterator.
29656 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
29658 2017-03-09 Martin Liska <mliska@suse.cz>
29660 PR tree-optimization/79631
29661 * tree-chkp-opt.c (chkp_is_constant_addr): Call
29662 tree_int_cst_sign_bit just for INTEGER constants.
29664 2017-03-09 Martin Liska <mliska@suse.cz>
29668 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
29671 2017-03-09 Marek Polacek <polacek@redhat.com>
29674 * tree.c (inchash::add_expr): Handle TREE_VEC.
29676 2017-03-09 Martin Liska <mliska@suse.cz>
29679 (chkp_narrow_size_and_offset): New function.
29680 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
29681 (void chkp_parse_bit_field_ref): New function.
29682 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
29683 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
29685 2017-03-09 Martin Liska <mliska@suse.cz>
29688 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
29689 (chkp_find_bounds_1): Remove gcc_unreachable.
29691 2017-03-09 Jakub Jelinek <jakub@redhat.com>
29694 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
29695 BUILT_IN_SYNC*, determine the access type from the size suffix and
29696 always build a MEM_REF with that type. Handle forgotten
29697 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
29700 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
29701 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
29702 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
29703 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
29704 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
29705 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
29706 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
29707 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
29708 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
29709 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
29710 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
29711 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
29712 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
29713 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
29714 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
29715 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
29716 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
29717 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
29718 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
29719 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
29720 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
29721 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
29722 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
29723 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
29724 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
29725 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
29726 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
29727 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
29728 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
29729 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
29730 definitions outside of __OPTIMIZE__ guarded section.
29733 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
29734 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
29735 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
29736 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
29739 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29741 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
29742 ("vfenez<mode>"): Add missing constraints.
29744 2017-03-08 Martin Sebor <msebor@redhat.com>
29747 * config/nds32/nds32.c (nds32_option_override):
29748 Fix misspelled diagnostic.
29750 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29753 * gimplify.c (gimplify_omp_for): Replace index var in outer
29754 taskloop statement with an artificial variable and add
29755 OMP_CLAUSE_PRIVATE clause for it.
29757 2017-03-08 Richard Biener <rguenther@suse.de>
29759 PR tree-optimization/79955
29760 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
29761 for accesses that are completely outside of the variable.
29763 2017-03-08 Andrew Haley <aph@redhat.com>
29765 PR tree-optimization/79943
29766 * tree-ssa-loop-split.c (compute_new_first_bound): When
29767 calculating the new upper bound, (END-BEG) should be added, not
29770 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29772 * config/avr/avr.md (setmemhi): Make sure match_dup
29773 operand number comes before match_scratch.
29775 2017-03-08 Richard Biener <rguenther@suse.de>
29777 PR tree-optimization/79920
29778 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
29779 with ncopies == 1 to ...
29780 (vect_transform_slp_perm_load): ... here. Properly compute
29781 all element loads by iterating VF times over the group. Do
29782 not handle ncopies (computed in a broken way) in
29783 vect_create_mask_and_perm.
29785 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29788 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
29789 is a uniform vector, use uniform_vector_p return value instead of
29790 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
29792 2017-03-07 Marek Polacek <polacek@redhat.com>
29794 PR middle-end/79809
29795 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
29796 (alloca_call_type): Likewise.
29798 2017-03-07 Martin Liska <mliska@suse.cz>
29800 * gcov.c (process_args): Put comment to correct location.
29802 2017-03-07 Martin Liska <mliska@suse.cz>
29804 PR middle-end/68270
29805 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
29806 Use array_at_struct_end_p instead of DECL_CHAIN (field).
29807 (chkp_narrow_bounds_for_field): Likewise.
29808 (chkp_parse_array_and_component_ref): Pass one more argument to
29811 2017-03-07 Richard Biener <rguenther@suse.de>
29813 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
29816 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
29818 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
29819 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
29821 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29824 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
29825 to end of description.
29826 (PARAM_MAX_STORES_TO_MERGE): Likewise.
29828 2017-03-07 Jakub Jelinek <jakub@redhat.com>
29830 PR rtl-optimization/79901
29831 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
29833 (*avx512f_<code><mode>3<mask_name>): ... this.
29834 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
29835 iterator instead of VI8_AVX2_AVX512BW.
29837 PR rtl-optimization/79901
29838 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
29839 min/max expander, expand it using expand_vec_cond_expr.
29842 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
29845 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29848 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
29849 to void * for PCH reasons.
29850 * dwarf2out.c (output_loc_operands, output_die): Cast
29851 v.val_vec.array to unsigned char *.
29853 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
29856 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
29859 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
29861 PR rtl-optimization/79571
29862 * lra-constraints.c (process_alt_operands): Calculate static
29863 reject and subtract it from overall when only addresses will be
29866 2017-03-06 Julia Koval <julia.koval@intel.com>
29869 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
29870 incoming stack boundary to 128 for 64-bit targets.
29872 2017-03-06 Richard Biener <rguenther@suse.de>
29874 PR tree-optimization/79894
29875 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
29876 to NULL after folding it.
29878 2017-03-06 Richard Biener <rguenther@suse.de>
29880 PR tree-optimization/79824
29881 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
29882 check disabling peeling for gaps.
29884 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
29886 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
29887 attributes): Document gettimeofday.
29889 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29891 * config/s390/s390.c (s390_option_override_internal): Set
29892 PARAM_MIN_VECT_LOOP_BOUND
29894 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29896 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
29897 * config/s390/s390.md: Likewise.
29899 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29902 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
29903 (<avx2_avx512>_perm<mode>): Rename to ...
29904 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
29906 (<avx512>_perm<mode>_mask): Rename to ...
29907 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
29909 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
29910 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
29911 instead of VI8F_256_512.
29912 (avx512f_perm<mode>): New define_expand.
29913 (avx512f_perm<mode>_mask): Likewise.
29914 (avx512f_perm<mode>_1<mask_name>): New define_insn.
29915 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
29917 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29919 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
29920 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
29922 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
29924 2017-03-06 Martin Liska <mliska@suse.cz>
29927 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
29928 when having a SSA NAME w/o VAR_DECL assigned to it.
29930 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29932 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
29933 msa_dpsub_<su>_d): Fix MODE for vec_select.
29935 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29937 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
29939 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
29941 2017-03-06 Richard Biener <rguenther@suse.de>
29943 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
29944 * plugin.c (register_plugin_info): Likewise.
29945 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
29947 2017-03-05 Jakub Jelinek <jakub@redhat.com>
29949 * config/i386/sse.md (sse_storehps, sse_storelps,
29950 avx_<castmode><avxsizesuffix>_<castmode>,
29951 avx512f_<castmode><avxsizesuffix>_<castmode>,
29952 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
29953 in condition that at least one operand is not a MEM.
29955 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29957 PR middle-end/79805
29958 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
29959 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
29961 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
29962 gimple_call_nothrow_p flag based on whether original builtin can throw.
29963 If it can, emit following stmts on the fallthrough edge.
29964 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
29965 don't create new bb if inserting just debug stmts on the edge, try to
29966 insert them on the fallthru bb or just reset debug stmts.
29968 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
29971 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
29972 restore recog_data (including the operand rtxes inside it) around
29973 the call to get_insn_template.
29975 2017-03-03 Martin Sebor <msebor@redhat.com>
29977 PR tree-optimization/79699
29978 * context.c (context::~context): Free MPFR caches to avoid
29979 a memory leak on program exit.
29981 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29983 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
29984 Use wide_int::ulow () instead of .elt (0).
29986 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29988 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
29989 (*pushxf): Limit oF constraint to 32bit targets and add oC
29990 constraint for 64bit targets.
29991 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
29992 (*pushdf): Change rmF constraint to rmC.
29994 2017-03-03 Martin Liska <mliska@suse.cz>
29996 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
29997 Remove unused variable.
29999 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30002 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
30003 is a memory operand, increase num_memory.
30004 (ix86_expand_args_builtin): Likewise.
30006 2017-03-03 Jan Hubicka <jh@suse.cz>
30009 * ipa-devirt.c (maybe_record_node): Properly handle
30010 __cxa_pure_virtual visibility.
30012 2017-03-03 Martin Liska <mliska@suse.cz>
30014 PR tree-optimization/79803
30015 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
30017 (pass_loop_prefetch::execute): Disabled optimization if an
30018 assumption about L1 cache size is not met.
30020 2017-03-03 Martin Liska <mliska@suse.cz>
30022 PR rtl-optimization/79574
30023 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
30024 (hash_scan_set): Likewise.
30025 (dump_hash_table): Likewise.
30026 (hoist_code): Likewise.
30028 2017-03-03 Richard Biener <rguenther@suse.de>
30030 * fixed-value.c (fixed_from_string): Restore use of elt (1)
30031 in place of uhigh ().
30032 (fixed_convert_from_real): Likewise.
30034 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30037 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
30039 2017-03-03 Richard Biener <rguenther@suse.de>
30041 PR middle-end/79818
30042 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
30043 TYPE_OVERFLOW_UNDEFINED check.
30045 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30047 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
30049 (vector_ae_<mode>_p): Likewise.
30050 (vector_nez_<mode>_p): Likewise.
30051 (vector_ne_v2di_p): Likewise.
30052 (vector_ae_v2di_p): Likewise.
30053 (vector_ne_<mode>_p): Likewise.
30054 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
30056 (vsx_tsqrt<mode>2_fe): Likewise.
30058 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
30061 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
30063 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30065 PR rtl-optimization/79780
30066 * cprop.c (one_cprop_pass): When second and further conditional trap
30067 in a single basic block is turned into an unconditional trap, turn it
30068 into a deleted note to avoid RTL verification failures.
30070 2017-03-02 Richard Biener <rguenther@suse.de>
30072 * fold-const.c (const_binop): Use ulow () instead of elt (0).
30074 2017-03-02 Richard Biener <rguenther@suse.de>
30076 PR tree-optimization/79345
30078 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
30079 param and abort the walk, returning -1 if it is hit.
30080 (walk_aliased_vdefs): Take a limit param and pass it on.
30081 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
30082 defaulting to 0 and return a signed int.
30083 * tree-ssa-uninit.c (struct check_defs_data): New struct.
30084 (check_defs): New helper.
30085 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
30086 about uninitialized memory.
30087 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
30088 bogus uninitialized warning.
30089 (fixed_convert_from_real): Likewise.
30091 2017-03-02 Bin Cheng <bin.cheng@arm.com>
30093 PR tree-optimization/66768
30094 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
30095 iv_use if base object can't be determined.
30097 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30099 PR tree-optimization/79345
30100 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
30101 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
30102 (get_pattern_stats): Initialize it.
30103 * genemit.c (gen_expand): Verify match_scratch numbers come after
30104 match_operand/match_dup numbers.
30105 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
30106 match_scratch numbers.
30107 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
30109 * config/s390/s390.md (trunctdsd2): Likewise.
30111 2017-03-02 Richard Biener <rguenther@suse.de>
30113 * wide-int.h (wide_int_storage::operator=): Implement in terms
30116 2017-03-02 Richard Biener <rguenther@suse.de>
30118 PR tree-optimization/79777
30119 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
30120 the to insert expression to sth existing.
30122 2017-03-01 Martin Sebor <msebor@redhat.com>
30124 PR middle-end/79692
30125 * gimple-ssa-sprintf.c
30126 (directive::known_width_and_precision): New function.
30127 (format_integer): Use it.
30128 (get_mpfr_format_length): Consider the full range of precision
30129 when computing %g output with the # flag. Set the likely byte
30130 count to 3 rather than 1 when precision is indeterminate.
30131 (format_floating): Correct the lower bound of precision.
30133 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30135 * doc/invoke.texi: Document default code model for 64-bit Linux.
30137 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30140 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
30141 udiv rather than div since input pattern is unsigned.
30143 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
30145 * config/i386/i386.c (print_reg): Warn for values of
30146 unsupported size in integer register.
30148 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
30151 * config/rs6000/predicates.md (current_file_function_operand): Do
30152 not allow self calls to be local if the function is replaceable.
30154 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
30157 * config/rs6000/altivec.h (vec_ctz and others): Change the
30158 preprocessor macro that controls conditional compilation from
30159 _ARCH_PWR9 to __POWER9_VECTOR__.
30160 (vec_all_ne): Change parameterization of __altivec_scalar_pred
30161 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
30162 control (instead of _ARCH_PWR9 control) so that template
30163 definition uses power9-specific function.
30164 (vec_any_eq): Likewise.
30165 (vec_all_ne): Change macro definition to use a power9-specific
30166 expansion under #ifdef __POWER9_VECTOR__ control (instead of
30167 _ARCH_PWR9 control).
30168 (vec_any_eq) Likewise.
30169 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
30170 expansion for CMPNEF to remove support for xvcmpnesp instruction.
30171 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
30172 support for xvcmpnedp instruction.
30173 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
30174 macro expansion so that Power9 implementation of vec_all_ne does
30175 not use the AltiVec predicate framework.
30176 (VCMPNEH_P): Likewise.
30177 (VCMPNEW_P): Likewise.
30178 (VCMPNED_P): Likewise.
30179 (VCMPNEFP_P): Likewise.
30180 (VCMPNEDP_P): Likewise.
30181 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30182 implementation of vec_any_eq to not use AltiVec predicate
30184 (VCMPAEH_P): Likewise.
30185 (VCMPAEW_P): Likewise.
30186 (VCMPAED_P): Likewise.
30187 (VCMPAEFP_P): Likewise.
30188 (VCMPAEDP_P): Likewise.
30189 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30190 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30191 not use the AltiVec predicate framework.
30192 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30193 of vec_any_eq to not use AltiVec predicate framework.
30194 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30195 support for predefined __POWER9_VECTOR__ macro to indicate that
30196 Power9 instruction selection is enabled.
30197 (altivec_overloaded_builtins): Remove extraneous
30198 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30199 function argument types RS6000_BTI_bool_V16QI and
30200 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30201 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30202 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30203 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30204 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30205 Power9 for implementations of vec_cmpne. Change the signature for
30206 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30207 (representing vec_all_ne) to remove the previously described first
30208 argument of type RS6000_BTI_INTSI, as this was an artifact of
30209 reliance on the AltiVec predicate framework, which is no longer
30210 used in the implementation of these functions. Add
30211 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30212 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30213 since, unlike the AltiVec predicate framework implementation, we
30214 do not share function descriptors between vec_alle and vec_anyeq.
30215 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30216 set of modes that receive special treatment even when
30217 TARGET_P9_VECTOR is true. The special treatment emits code that
30218 does not depend on Power9 instructions.
30219 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30220 define_expand to not rely on AltiVec predicate framework.
30221 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30223 (vector_ne_v2di_p): Change this define_expand to not rely on
30224 AltiVec predicate framework.
30225 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30227 (vector_ne_<mode>_p): Change this define_expand to not rely on
30228 AltiVec predicate framework.
30229 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30231 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30232 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30233 define_insn pattern.
30234 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30235 define_insn pattern because the xvcmpne<VSs>. instruction is not
30237 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30238 instruction is not supported.
30240 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30242 * config/nvptx/nvptx.c: Include intl.h.
30244 2017-03-01 Martin Jambor <mjambor@suse.cz>
30247 * ipa-prop.h (ipa_bits): Removed field known.
30248 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30249 to pointers. Adjusted their comments to warn about their sharing.
30250 (ipcp_transformation_summary): Change bits to a vector of pointers.
30251 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30252 (ipa_get_ipa_bits_for_value): Declare.
30253 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30254 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30255 (ipa_bits_hash_table): Likewise.
30256 (ipa_vr_ggc_hash_traits): Likewise.
30257 (ipa_vr_hash_table): Likewise.
30258 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30259 being pointers and vr_known being removed.
30260 (ipa_set_jf_unknown): Likewise.
30261 (ipa_get_ipa_bits_for_value): New function.
30262 (ipa_set_jfunc_bits): Likewise.
30263 (ipa_get_value_range): New overloaded functions.
30264 (ipa_set_jfunc_vr): Likewise.
30265 (ipa_compute_jump_functions_for_edge): Use the above functions to
30266 construct bits and vr parts of jump functions.
30267 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30268 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30270 (ipcp_grow_transformations_if_necessary): Also allocate
30271 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30273 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30274 them. Fix too long lines.
30275 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30276 vr_known being removed.
30277 (ipa_read_jump_function): Use new setter functions to construct bits
30278 and vr parts of jump functions or set them to NULL.
30279 (write_ipcp_transformation_info): Adjust for bits being pointers.
30280 (read_ipcp_transformation_info): Likewise.
30281 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30283 Include gt-ipa-prop.h.
30284 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30286 (ipcp_store_bits_results): Likewise.
30287 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30288 Do not write to existing jump functions but use a temporary instead.
30290 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30293 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30294 attempt to use its first operand as BIT_FIELD_REF base.
30296 2017-03-01 Richard Biener <rguenther@suse.de>
30298 PR middle-end/79721
30299 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30300 interpolating formula in wrapping arithmetic.
30301 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30303 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30305 PR tree-optimization/79734
30306 * tree-vect-generic.c (expand_vector_condition): Optimize
30307 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30308 Handle VEC_COND_EXPR where comparison has different inner width from
30309 type's inner width.
30311 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30313 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30314 markup, and similar issues. Remove @opindex entries for things
30315 that aren't options. Add missing -mmpy-option entries.
30317 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30319 PR tree-optimization/79737
30320 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30321 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30322 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30323 instead of byte_size. Formatting fix.
30324 (shift_bytes_in_array_right): Formatting fix.
30326 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30329 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30330 condition on optimize for the leaf function test.
30332 2017-02-28 Martin Liska <mliska@suse.cz>
30335 * read-rtl-function.c (function_reader::handle_unknown_directive):
30336 Bail out when one uses -flto.
30338 2017-02-28 Martin Liska <mliska@suse.cz>
30340 * common.opt: Replace space with tabular for options of <number>
30342 * config/i386/i386.opt: Show <number> value for
30343 -mlarge-data-threshold.
30344 * opts.c (print_filtered_help): Do not display number in hexadecimal
30347 2017-02-28 Martin Liska <mliska@suse.cz>
30349 * common.opt: Fix --help=option -Q for options which are of
30352 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30354 * config/i386/i386.c (print_reg): Error out for values
30355 of 8-bit size in invalid integer register.
30357 2017-02-28 Martin Sebor <msebor@redhat.com>
30359 PR tree-optimization/79691
30360 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30362 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30365 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30366 gcc_unreachable with output_operand_lossage.
30368 2017-02-28 Richard Biener <rguenther@suse.de>
30370 PR tree-optimization/79740
30371 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30373 (visit_nary_op): Insert the nary into the hashtable if we
30374 pattern-matched sth.
30375 * tree-ssa-pre.c (eliminate_insert): Robustify.
30377 2017-02-28 Richard Biener <rguenther@suse.de>
30379 PR middle-end/79731
30380 * fold-const.c (decode_field_reference): Reject out-of-bound
30383 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30385 * config/i386/i386.c: Include intl.h.
30386 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30387 instead of just cond ? "..." : "...".
30388 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30389 * coverage.c (read_counts_file): Likewise.
30390 * omp-offload.c: Include intl.h.
30391 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30392 of just cond ? "..." : "...".
30393 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30394 of just cond ? "..." : "...".
30396 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30399 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30401 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30402 'tune for' CPU name.
30403 * config/arm/arm-cpu-data.h: Regenerated.
30405 2017-02-28 Richard Biener <rguenther@suse.de>
30407 PR tree-optimization/79732
30408 * tree-inline.c (expand_call_inline): Do not shadow var.
30410 2017-02-28 Richard Biener <rguenther@suse.de>
30412 PR tree-optimization/79723
30413 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30414 address-space properly.
30416 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30418 * doc/optinfo.texi (Optimization groups): Fix option used for
30420 * doc/invoke.texi (-fopt-info): Document "omp".
30421 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30422 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30423 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30424 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30425 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30427 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30429 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30430 OPTGROUP_OMP with "omp".
30432 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30435 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30436 for arithmetic shift of unsigned V2DI.
30438 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30440 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30441 arc/linux.h headers.
30442 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30443 (LINK_SPEC): Likewise.
30444 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30445 (EXTRA_SPECS): Likewise.
30446 (STARTFILE_SPEC): Likewise.
30447 (ENDFILE_SPEC): Likewise.
30448 (LIB_SPEC): Likewise.
30449 (TARGET_SDATA_DEFAULT): Likewise.
30450 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30451 (MULTILIB_DEFAULTS): Likewise.
30452 (DWARF2_UNWIND_INFO): Likewise.
30453 * config/arc/big.h: New file.
30454 * config/arc/elf.h: Likewise.
30455 * config/arc/linux.h: Likewise.
30456 * config/arc/t-uClibc: Remove.
30458 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30460 PR tree-optimization/77536
30461 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30462 (tree_transform_and_unroll_loop): Use above function to compute the
30463 estimated niter of unrolled loop and use it when scaling profile.
30464 Also use count info rather than frequency if it's non-zero.
30465 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30466 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30467 (vect_transform_loop): Call above function.
30469 2017-02-27 Richard Biener <rguenther@suse.de>
30471 PR tree-optimization/45397
30472 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
30473 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
30474 (visit_nary_op): Add pattern matching for CSEing sign-changed
30475 or truncated operations with wider ones.
30477 2017-02-27 Richard Biener <rguenther@suse.de>
30479 PR tree-optimization/79690
30480 * tree-vect-stmts.c (vectorizable_store): Use vector type
30481 built from the DR with address-space.
30483 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
30485 * doc/invoke.texi (Optimize Options): Refine the description
30486 of asan-use-after-return.
30488 2017-02-25 Alan Modra <amodra@gmail.com>
30490 PR rtl-optimization/79584
30491 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
30492 base, not ad->base_term, the reg within base. Remove assertion
30493 that ad->base == ad->base_term. Replace gen_int_mode using
30494 bogus mode with const0_rtx.
30496 2017-02-25 Jakub Jelinek <jakub@redhat.com>
30498 PR middle-end/79396
30499 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
30500 FMA_EXPR like tcc_binary or tcc_unary.
30502 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
30505 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
30507 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
30508 (output_loc_operands): Handle DW_OP_call_ref and
30509 DW_OP_GNU_variable_value.
30510 (struct variable_value_struct): New type.
30511 (struct variable_value_hasher): Likewise.
30512 (variable_value_hash): New variable.
30513 (string_types): Remove.
30514 (copy_loc_descr): New function.
30515 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
30516 (prepend_loc_descr_to_each): New function.
30517 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
30518 instead of add_loc_descr_to_each if the first argument is single
30519 location list and the second has multiple.
30520 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
30521 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
30522 when looking for variable value which doesn't have other location info.
30523 (loc_list_from_tree): Formatting fix.
30524 (gen_array_type_die): Simplify DW_AT_string_length handling.
30525 (adjust_string_types): Remove.
30526 (gen_subprogram_die): Don't call adjust_string_types nor test/set
30527 string_types. Call resolve_variable_values.
30528 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
30529 (resolve_addr_in_expr): Likewise. Add A argument.
30530 (copy_deref_exprloc): Remove deref argument. Adjust for the
30531 original expression being DW_OP_GNU_variable_value with optionally
30532 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
30533 optionally after it.
30534 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
30535 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
30536 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
30537 (variable_value_hasher::hash, variable_value_hasher::equal): New
30539 (resolve_variable_value_in_expr, resolve_variable_value,
30540 resolve_variable_values, note_variable_value_in_expr,
30541 note_variable_value): New functions.
30542 (dwarf2out_early_finish): Call note_variable_value on all toplevel
30545 2017-02-24 Jakub Jelinek <jakub@redhat.com>
30548 * opts.h (handle_generated_option): Add GENERATED_P argument.
30549 * opts-common.c (handle_option): Adjust function comment.
30550 (handle_generated_option): Add GENERATED_P argument, pass it to
30552 (control_warning_option): Pass false to handle_generated_option
30554 * opts.c (maybe_default_option): Pass true to handle_generated_option
30556 * optc-gen.awk: Likewise.
30558 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30560 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
30561 a REG, look at the REG it is a SUBREG of.
30562 (splitter for cmpeqsi_t): Ditto.
30564 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30566 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
30567 the special USEs with the pattern of the insn, not the insn itself.
30569 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
30572 * doc/invoke.texi: Document -mload-store-pairs.
30574 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30575 Sandra Loosemore <sandra@codesourcery.com>
30577 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
30578 argument isn't a CONST_INT.
30579 (nios2_alternate_compare_const): Assert op is a CONST_INT.
30580 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
30581 (nios2_validate_compare): Bypass alternate compare logic if *op2
30582 is not a CONST_INT.
30583 (ldstwm_operation_p): Return false if first_base is not a REG or
30584 if first_offset is not a CONST_INT.
30586 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30588 * config/cris/cris.md: Use correct operand in a define_peephole2.
30590 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30592 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
30594 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30596 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
30597 this_insn if it is an INSN or JUMP_INSN.
30598 (force_offsettable): Look at base, not at addr.
30599 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
30600 on things that aren't necessarily CONST_INTs.
30602 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
30604 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
30605 -mfpmath=sse is the default also for x86-32 targets with SSE2
30606 instruction set when @option{-ffast-math} is enabled
30608 2017-02-24 Jeff Law <law@redhat.com>
30610 PR rtl-optimizatoin/79286
30611 * ira.c (update_equiv_regs): Drop may_trap_p exception to
30614 2017-02-24 Richard Biener <rguenther@suse.de>
30616 PR tree-optimization/79389
30617 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
30620 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
30622 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
30623 function comment to reflect reality.
30624 (loop_exits_before_overflow): Fix typo in function description.
30626 2017-02-24 Richard Biener <rguenther@suse.de>
30628 PR tree-optimization/79389
30629 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
30630 properly that a threading opportunity exists. Detect conditional
30631 copy/constant propagation opportunities.
30633 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
30635 * config/visium/visium.md (type): Add trap.
30636 (b): New mode attribute.
30637 (*btst): Rename into...
30638 (*btst<mode>): ...this and adjust.
30639 (*cbranchsi4_btst_insn): Rename into...
30640 (*cbranch<mode>4_btst_insn): ...this and adjust.
30641 (trap): New define_insn.
30643 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30645 PR tree-optimization/79389
30646 * ifcvt.c (struct noce_if_info): Add rev_cond field.
30647 (noce_reversed_cond_code): New function.
30648 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
30649 reversed_comparison_code. Formatting fix.
30650 (noce_try_store_flag): Test rev_cond != NULL in addition to
30651 reversed_comparison_code.
30652 (noce_try_store_flag_constants): Likewise.
30653 (noce_try_store_flag_mask): Likewise.
30654 (noce_try_addcc): Use rev_cond if non-NULL instead of
30655 reversed_comparison_code.
30656 (noce_try_cmove_arith): Likewise. Formatting fixes.
30657 (noce_try_minmax, noce_try_abs): Clear rev_cond.
30658 (noce_find_if_block): Initialize rev_cond.
30659 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
30660 instead of false as last argument never attempt to reverse it
30663 2017-02-23 Bin Cheng <bin.cheng@arm.com>
30665 PR tree-optimization/79663
30666 * tree-predcom.c (combine_chains): Process refs in reverse order
30667 only for ZERO length chains, and add explaining comment.
30669 2017-02-23 Jeff Law <law@redhat.com>
30671 PR tree-optimization/79578
30672 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
30673 in call to operand_equal_p.
30675 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
30678 * config/i386/cpuid.h: Fix another undefined behavior.
30680 2017-02-23 Richard Biener <rguenther@suse.de>
30682 PR tree-optimization/79683
30683 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
30684 vector types for data-refs.
30686 2017-02-23 Martin Liska <mliska@suse.cz>
30688 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
30690 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30692 PR middle-end/79665
30693 * internal-fn.c (get_range_pos_neg): Moved to ...
30694 * tree.c (get_range_pos_neg): ... here. No longer static.
30695 * tree.h (get_range_pos_neg): New prototype.
30696 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
30697 are known to be in between 0 and signed maximum inclusive, try to
30698 expand both unsigned and signed divmod and use the cheaper one from
30701 2017-02-22 Jeff Law <law@redhat.com>
30703 PR tree-optimization/79578
30704 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
30705 to compare base operands.
30707 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
30710 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
30711 gpc_reg_operand instead of fpr_reg_operand.
30713 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
30715 * config/mips/mips.c (mips_return_in_memory): Force FP
30716 vector types to be returned in memory for o32 ABI.
30718 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30720 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
30721 instead of DW_TAG_member for static data member declarations and don't
30722 set no_linkage_name for static inline data members.
30723 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
30726 2017-02-22 Martin Liska <mliska@suse.cz>
30728 * doc/invoke.texi: Replace inequality signs with square brackets
30731 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
30734 * lra-constraints.c (simplify_operand_subreg): Handle
30735 WORD_REGISTER_OPERATIONS targets.
30737 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30740 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
30741 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
30742 elimination by swapping fld*.
30744 2017-02-22 Richard Biener <rguenther@suse.de>
30746 PR tree-optimization/79673
30747 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
30748 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
30749 irrelevant address-space qualifiers and avoiding a
30750 ADDR_SPACE_CONVERT_EXPR from fold_convert.
30752 2017-02-22 Richard Biener <rguenther@suse.de>
30754 PR tree-optimization/79666
30755 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
30756 to not symbolically negate if that may introduce undefined
30759 2017-02-22 Martin Liska <mliska@suse.cz>
30762 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
30763 * data-streamer-out.c (streamer_write_gcov_count_stream):
30765 * value-prof.c (stream_out_histogram_value): Make assert more
30766 precise based on type of counter.
30768 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
30771 * config/i386/i386.md (standard_x87sse_constant_load splitter):
30772 Use nonimmediate_operand instead of memory_operand for operand 1.
30773 (float-extend standard_x87sse_constant_load splitter): Ditto.
30775 2017-02-21 Jeff Law <law@redhat.com>
30777 PR tree-optimization/79621
30778 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
30779 blocks with edges to themselves.
30781 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30784 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
30785 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
30786 Use gimple_call_builtin_p.
30789 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
30790 on temporarily removed DEBUG_INSNs.
30792 PR tree-optimization/79649
30793 * tree-loop-distribution.c (classify_partition): Give up on
30794 non-generic address space loads/stores.
30796 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
30798 * doc/loop.texi (Loop manipulation): Remove nonexistent
30799 tree_ssa_loop_version from the documentation.
30800 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
30802 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30805 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
30806 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
30807 * config/rs6000/rs6000.c: Include except.h.
30808 (rs6000_expand_split_stack_prologue): Call
30809 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
30811 2017-02-21 Martin Jambor <mjambor@suse.cz>
30814 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
30815 have been analyzed.
30817 2017-02-21 Martin Jambor <mjambor@suse.cz>
30819 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
30820 for backward compatibility only.
30821 * doc/invoke.texi (Option Summary): Remove all references to
30822 -fipa-cp-alignment.
30824 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
30828 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30830 * lra-constraints.c (curr_insn_transform): Handle
30831 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30833 2017-02-21 Martin Liska <mliska@suse.cz>
30835 * config/i386/i386.opt: Replace -masm-dialect with -masm.
30837 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
30839 PR translation/79638
30840 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
30842 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
30845 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
30846 (arm_function_ok_for_sibcall): Return false for an indirect call by
30847 descriptor if all the argument registers are used.
30848 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
30849 alignment of the function.
30851 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30853 PR tree-optimization/61441
30854 * simplify-rtx.c (simplify_const_unary_operation): For
30855 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
30856 the sNaN unmodified.
30858 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30860 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30861 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30862 instead of SYSTEM_HEADER_DIR.
30864 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
30865 Martin Liška <mliska@suse.cz>
30867 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
30868 Fix typos and grammar, use active voice, and clarify.
30870 2017-02-20 Marek Polacek <polacek@redhat.com>
30872 PR middle-end/79537
30873 * gimplify.c (gimplify_expr): Handle unused *&&L;.
30876 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
30878 2017-02-20 Jakub Jelinek <jakub@redhat.com>
30881 * config/i386/i386.c (ix86_expand_builtin): Handle
30882 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
30883 ix86_builtins_isa[fcode].isa as a requirement of those
30884 flags and any other flag in the bitmask.
30885 (ix86_init_mmx_sse_builtins): Use 0 instead of
30886 ~OPTION_MASK_ISA_64BIT as mask.
30887 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
30888 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
30889 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
30890 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
30892 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30895 * lra-constraints.c (split_reg): Check requested split mode
30896 is supported by the register.
30898 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30900 * lra-constraints.c (simplify_operand_subreg): Remove early
30903 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30906 * lra-constraints.c (curr_insn_transform): Tighten condition
30907 for converting SUBREG reloads from OP_OUT to OP_INOUT.
30909 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30912 * lra-constraints.c (curr_insn_transform): Handle
30913 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30915 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
30918 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
30920 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
30922 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
30925 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
30928 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30930 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
30933 2017-02-18 Jakub Jelinek <jakub@redhat.com>
30935 * final.c (last_columnnum, override_columnnum): New variables.
30936 (final_start_function): Set last_columnnum, pass it to begin_prologue
30937 hook and pass 0 to dwarf2out_begin_prologue.
30938 (final_scan_insn): Update override_columnnum. Pass last_columnnum
30939 to source_line debug hook.
30940 (notice_source_line): Compute last_columnnum and for debug_column_info
30941 return true on column changes.
30942 * debug.h (struct gcc_debug_hooks): Add column argument to
30943 source_line and begin_prologue hooks.
30944 (debug_nothing_int_charstar_int_bool): Remove prototype.
30945 (debug_nothing_int_int_charstar,
30946 debug_nothing_int_int_charstar_int_bool): New prototypes.
30947 (dwarf2out_begin_prologue): Add column argument.
30948 * debug.c (do_nothing_debug_hooks): Adjust source_line and
30949 begin_prologue hooks.
30950 (debug_nothing_int_charstar_int_bool): Remove.
30951 (debug_nothing_int_int_charstar,
30952 debug_nothing_int_int_charstar_int_bool): New functions.
30953 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
30954 through to dwarf2out_source_line.
30955 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
30956 (dwarf2out_source_line): Add column argument, emit it if requested.
30957 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
30959 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30960 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30961 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
30962 through to dwarf2out_begin_prologue.
30963 (vmsdbgout_source_line): Add column argument, pass it through to
30964 dwarf2out_source_line.
30965 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
30966 dbxout_source_line caller.
30967 (dbxout_source_line): Add column argument.
30969 * common.opt (gno-column-info, gcolumn-info): New options.
30970 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
30971 (check_die): Also test for multiple DW_AT_decl_column attributes.
30972 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
30973 DW_AT_decl_column if requested.
30974 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
30976 (gen_variable_die): Likewise.
30977 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
30978 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
30981 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
30982 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
30983 (ix86_handle_option): Handle OPT_m3dnowa.
30984 * doc/invoke.texi (-m3dnowa): Document.
30985 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
30986 -m3dnowa instead of -m3dnow -march=athlon.
30989 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
30990 instead of gcc_assert for K, r and R code checks. Formatting fixes.
30992 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30995 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
30996 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
30997 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
30998 generator for vsx_xxpermdi_<mode>_be.
30999 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
31000 force big-endian semantics.
31001 (vsx_xxpermdi_<mode>_be): New define_expand with same
31002 implementation as previous version of vsx_xxpermdi_<mode>.
31004 2017-02-17 Jakub Jelinek <jakub@redhat.com>
31006 PR tree-optimization/79327
31007 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
31008 variable, its initialization and use.
31010 2017-02-17 Julia Koval <julia.koval@intel.com>
31012 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
31013 (OPTION_MASK_ISA_PKU_UNSET): New.
31014 (ix86_handle_option): Handle -mrdpid.
31015 * config/i386/cpuid.h (bit_RDPID): New.
31016 * config/i386/driver-i386.c (host_detect_local_cpu):
31017 Detect RDPID feature.
31018 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
31019 * config/i386/i386-c.c (ix86_target_macros_internal):
31021 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
31022 (ix86_valid_target_attribute_inner_p): Add "rdpid".
31023 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
31024 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
31025 * config/i386/i386.md (define_insn "rdpid"): New.
31026 * config/i386/i386.opt Add -mrdpid.
31027 * config/i386/immintrin.h (_rdpid_u32): New.
31029 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
31031 PR rtl-optimization/79541
31032 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
31033 instead of transforming it into USE.
31035 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
31037 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
31038 If HONOR_SNANS (SFmode) force the input to a register.
31039 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
31040 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
31041 an frsp or similar insn.
31043 2017-02-17 Martin Liska <mliska@suse.cz>
31045 PR rtl-optimization/79577
31046 * params.def (selsched-max-sched-times): Increase minimum to 1.
31048 2017-02-17 Martin Liska <mliska@suse.cz>
31050 PR rtl-optimization/79574
31051 * gcse.c (want_to_gcse_p): Prevent integer overflow.
31053 2017-02-17 Martin Liska <mliska@suse.cz>
31055 PR tree-optimization/79529
31056 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
31057 ssa_defined_default_def_p to handle cases which are implicitly
31059 * tree-ssa.c (ssa_defined_default_def_p): New function.
31060 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
31061 which are implicitly defined.
31062 * tree-ssa.h (ssa_defined_default_def_p): Declare.
31064 2017-02-17 Richard Biener <rguenther@suse.de>
31066 PR middle-end/79576
31067 * params.def (max-ssa-name-query-depth): Limit to 10.
31069 2017-02-17 Richard Biener <rguenther@suse.de>
31071 PR tree-optimization/79552
31072 * tree-ssa-structalias.c (visit_loadstore): Properly verify
31075 2017-02-17 Richard Biener <rguenther@suse.de>
31078 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
31080 2017-02-17 Marek Polacek <polacek@redhat.com>
31082 PR middle-end/79536
31083 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
31084 (fold_negate_expr): New wrapper.
31086 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
31088 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
31089 Correct terminology and de-emphasize pre-standard behavior.
31091 2017-02-16 Alan Modra <amodra@gmail.com>
31093 PR rtl-optimization/79286
31094 * ira.c (def_dominates_uses): New function.
31095 (update_equiv_regs): Don't create an equivalence for insns that
31096 may trap where the register def does not dominate the use.
31098 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
31100 PR rtl-optimization/78127
31101 * lra.c (lra): Call lra_eliminate before finish the loop after
31104 2017-02-16 Richard Biener <rguenther@suse.de>
31106 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
31108 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
31109 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
31110 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
31111 (isl_val_int_from_wi): New function.
31112 (extract_affine_gmp): Rename to ...
31113 (extract_affine_wi): ... this, take a widest_int.
31114 (extract_affine_int): Just wrap extract_affine_wi.
31115 (add_param_constraints): Use isl_val_int_from_wi.
31116 (add_loop_constraints): Likewise, and extract_affine_wi.
31118 2017-02-15 Jeff Law <law@redhat.com>
31120 PR middle-end/79521
31121 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
31122 ira_init_register_move_cost_if_necessary.
31124 2017-02-15 Martin Sebor <msebor@redhat.com>
31126 PR middle-end/32003
31127 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
31128 removed in a prior commit.
31130 2017-02-15 Bin Cheng <bin.cheng@arm.com>
31132 PR tree-optimization/79347
31133 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
31134 counters during peeling.
31136 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
31138 * Makefile.in (site.exp): Remove "set ISLVER".
31140 2017-02-15 Jakub Jelinek <jakub@redhat.com>
31143 * real.c (real_from_integer): Call real_convert even for decimal.
31145 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
31148 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
31150 2017-02-14 Andrew Pinski <apinski@cavium.com>
31152 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
31153 cores and change the partno/implementer to be correct.
31154 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
31155 the 'B" as the implementer.
31156 * config/aarch64/aarch64-tune.md: Regenerate.
31158 2017-02-14 Carl Love <cel@us.ibm.com>
31160 * config/rs6000/rs6000.c: Add case statement entry to make the
31161 xvcvuxdsp built-in argument unsigned.
31162 * config/rs6000/vsx.md: Fix the source and return operand types so they
31163 match the instruction definitions from the ISA document. Fix typo
31164 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
31167 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
31170 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
31171 member early_clobber_alts.
31172 * lra-lives.c (reg_early_clobber_p): New.
31173 (process_bb_lives): Use it.
31174 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
31175 (debug_operand_data): Initialize early_clobber_alts.
31176 (setup_operand_alternative): Set up early_clobber_alts.
31177 (collect_non_operand_hard_regs): Ditto. Pass early clobber
31178 alternatives to new_insn_reg.
31179 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31181 (lra_update_insn_regno_info): Pass the new arg.
31183 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31185 PR middle-end/79505
31186 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31187 (new_oacc_loop_raw): Don't clear already cleared fields.
31190 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31191 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31192 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31194 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31197 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31199 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31202 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31203 the extra instruction to the right place to store 128-bit constant
31206 2017-02-14 Martin Sebor <msebor@redhat.com>
31208 PR middle-end/79448
31209 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31210 warning for strings of unknown length.
31212 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31214 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31216 2017-02-14 Jeff Law <law@redhat.com>
31219 * ira-costs.c (scan_one_insn): Initialize register move costs
31220 for pseudos seen in USE/CLOBBER insns.
31222 PR tree-optimization/79095
31223 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31224 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31225 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31226 if the operands are known to be not equal, then the resulting range
31228 (intersect_ranges): If the new range is ~[0,0] and the old range is
31229 wide, then prefer ~[0,0].
31230 * tree-vrp.c (overflow_comparison_p_1): New function.
31231 (overflow_comparison_p): New function.
31232 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31233 if NAME is used in an overflow test.
31234 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31235 overflow check that can be expressed as an equality test, then adjust
31236 ops to be that equality test.
31238 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31240 * config/s390/s390-builtin-types.def: Remove flags argument.
31241 * config/s390/s390.c (s390_init_builtins): Likewise.
31243 2017-02-14 Martin Liska <mliska@suse.cz>
31245 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31246 vector. Fix trailing white spaces.
31248 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31250 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31253 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31255 PR rtl-optimization/68664
31256 * config/arm/arm.c (arm_sched_can_speculate_insn):
31257 New function. Declare prototype.
31258 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31260 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31262 PR rtl-optimization/68664
31263 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31265 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31267 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31269 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31270 max skip bytes for function, loop and jump.
31272 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31274 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31275 ABS_EXPR for gimple dump.
31277 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31280 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31282 PR tree-optimization/79408
31283 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31284 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31285 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31286 also if rhs1 is INTEGER_CST.
31288 2017-02-14 Richard Biener <rguenther@suse.de>
31290 PR middle-end/79432
31291 * tree-into-ssa.c (insert_phi_nodes): When the function can
31292 have abnormal edges rewrite SSA names with broken use-def
31293 dominance out of SSA and register them for PHI insertion.
31295 2017-02-13 Martin Sebor <msebor@redhat.com>
31297 PR middle-end/79496
31298 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31299 clearing info.nowrite flag when snprintf size argument is a range.
31301 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31303 * cprop.c (cprop_jump): Add missing space in string literal.
31304 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31305 (get_constraint_for_component_ref): Likewise.
31306 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31307 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31308 * lra-constraints.c (process_alt_operands): Likewise.
31309 * ipa-inline.c (inline_small_functions): Likewise.
31310 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31311 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31312 * trans-mem.c (diagnose_tm_1_op): Likewise.
31313 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31314 (grid_parallel_clauses_gridifiable): Likewise.
31316 * config/nvptx/mkoffload.c (process): Add space in between
31319 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31320 "MOD4_SSE_REGS" and "ALL_REGS".
31322 * spellcheck.c (test_data): Add , in between "foo" and "food".
31324 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31327 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31328 boundary crossing check and subsequent code generation agree.
31330 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31332 * config/aarch64/aarch64.c (has_memory_op): Delete.
31333 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31336 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31338 PR rtl-optimization/79388
31339 PR rtl-optimization/79450
31340 * combine.c (distribute_notes): When removing TEM_INSN for which
31341 corresponding dest has last value recorded, invalidate that last
31344 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31346 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31347 of explicit '@'. Add missing assembly comment marker on branch costs
31350 2017-02-13 Nathan Sidwell <nathan@acm.org>
31352 * gengtype-lex.l (<in_struct>): Add '/'.
31354 2017-02-13 Martin Liska <mliska@suse.cz>
31357 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31359 2017-02-13 Richard Biener <rguenther@suse.de>
31361 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31363 * configure: Re-generate.
31364 * config.in: Likewise.
31365 * graphite-dependences.c: Simplify as if
31366 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31367 * graphite-isl-ast-to-gimple.c: Likewise.
31368 * graphite-optimize-isl.c: Likewise.
31369 * graphite-poly.c: Likewise.
31370 * graphite-sese-to-poly.c: Likewise.
31371 * graphite.h: Likewise.
31372 * toplev.c: Include isl/version.h and use isl_version () for
31373 printing the ISL version.
31374 * doc/install.texi: Update ISL requirement.
31376 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31378 * doc/standards.texi (Standards): Update reference to
31381 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31383 * doc/extend.texi (Named Address Spaces): sourceware.org now
31385 * doc/install.texi (Binaries): Ditto.
31388 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31390 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31391 standard terminology "stringize"/"stringizing" throughout.
31392 * doc/cppinternals.texi: Likewise.
31394 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31396 * doc/extend.texi: Fix some spelling mistakes and typos.
31397 * doc/invoke.texi: Likewise.
31399 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31402 * params.def (inline-min-speedup) Change from 10 to 8.
31404 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31406 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31409 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31412 * ipa-inline-analysis.c (get_minimal_bb): New function.
31413 (record_modified): Use it.
31414 (remap_edge_change_prob): Handle also ancestor functions.
31416 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31418 * doc/contrib.texi (Contributors): Remove broken link into
31419 the Mauve CVS repository.
31421 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31423 PR middle-end/79454
31424 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31425 result computation whenever lhs doesn't have vector mode, not
31426 just when it has BLKmode.
31428 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31430 * doc/makefile.texi (profiledbootstrap): Refer to the
31431 installation instructions only in textual form.
31433 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31436 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31438 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31440 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31441 (Specific): Update mingw-w64 reference.
31443 (Specific): Remove broken link to Renesas RX processor.
31445 2017-02-10 Richard Biener <rguenther@suse.de>
31447 * toplev.c (process_options): Do not mention obsolete graphite
31448 options when printing sorry message about missing graphite support.
31449 Mention -floop-nest-optimize.
31451 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
31453 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
31454 (vtst_p16): Likewise.
31455 (vtstq_p8): Likewise.
31456 (vtstq_p16): Likewise.
31458 (vtstq_p64): Likewise.
31459 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31460 (vset_lane_p64): New.
31461 (vsetq_lane_p64): New.
31463 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31465 PR tree-optimization/79411
31466 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31467 stmt operands are SSA_NAMEs used in abnormal phis.
31468 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
31471 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31474 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
31477 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31479 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
31481 2017-02-09 Jakub Jelinek <jakub@redhat.com>
31483 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
31487 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
31488 not arbitrary TREE_CONSTANT.
31491 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
31492 "omp declare target link" attribute unless is_global_var.
31493 * omp-offload.c (find_link_var_op): Likewise.
31495 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
31496 Chung-Lin Tang <cltang@codesourcery.com>
31498 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
31500 (gimplify_adjust_omp_clauses): Don't delete TILE.
31501 (gimplify_omp_for): Deal with TILE.
31502 * internal-fn.c (expand_GOACC_TILE): New function.
31503 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
31505 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
31506 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
31508 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
31509 avoid DIV for outermost collapse var.
31510 (expand_oacc_for): Insert tile element loop as needed. Adjust.
31511 Remove out of date comments, fix whitespace.
31512 * omp-general.c (omp_extract_for_data): Deal with tiling.
31513 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
31514 adjust OLF_DIM_BASE value.
31515 (struct omp_for_data): Add tiling field.
31516 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
31517 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
31518 for auto loops. Remove default auto determining, moved to
31519 oacc_loop_fixed_partitions.
31520 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
31521 stmts, add e_mask field.
31522 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
31523 (oacc_thread_numbers): Use oacc_dim_call.
31524 (oacc_xform_tile): New.
31525 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
31526 (finish_oacc_loop): Adjust for ifns vector.
31527 (oacc_loop_discover_walk): Append loop abstraction sites to list,
31528 add case for GOACC_TILE fns.
31529 (oacc_loop_xform_loop): Delete.
31530 (oacc_loop_process): Iterate over call list directly, and add
31531 handling for GOACC_TILE fns.
31532 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
31534 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
31535 vector partitioning to outer loops. Assign 2 partitions to loops
31536 when available. Add TILE handling.
31537 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
31538 (execite_oacc_device_lower): Process GOACC_TILE fns,
31539 ignore unknown specs.
31540 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
31541 * tree.c (omp_clause_num_ops): Adjust TILE ops.
31542 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
31544 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
31546 * configure.ac (ACX_BUGURL): Update.
31547 * configure: Regenerate.
31549 2017-02-09 Richard Biener <rguenther@suse.de>
31551 PR tree-optimization/69823
31552 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
31553 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
31555 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31557 * config/arc/arc-c.def: Add __NPS400__ definition.
31558 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
31559 (TARGET_NPS400): Define.
31561 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31563 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
31565 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
31566 pointer, arch_info.
31567 (arc_cpu_types): Fill the arch_info field with a pointer into the
31568 arc_arch_types table.
31569 (arc_selected_cpu): Declare.
31570 * config/arc/arc.c (arc_selected_cpu): Make global.
31571 (arc_selected_arch): Delete.
31572 (arc_base_cpu): Delete.
31573 (arc_override_options): Remove references to deleted variables,
31574 update access to arch information.
31575 (ARC_OPT): Update access to arch information.
31576 (ARC_OPTX): Likewise.
31577 * config/arc/arc.h (arc_base_cpu): Remove declaration.
31578 (TARGET_ARC600): Update access to arch information.
31579 (TARGET_ARC601): Likewise.
31580 (TARGET_ARC700): Likewise.
31581 (TARGET_EM): Likewise.
31582 (TARGET_HS): Likewise.
31583 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
31586 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
31589 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
31590 condition/operands for integer GE/LE/GEU/LEU operations.
31592 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
31594 PR translation/79397
31595 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
31598 2017-02-08 Martin Jambor <mjambor@suse.cz>
31601 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
31602 whether allocation happened.
31603 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
31604 nothing was allocated.
31606 2017-02-08 Jakub Jelinek <jakub@redhat.com>
31608 PR tree-optimization/79408
31609 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
31610 constant, but SSA_NAME with a known integer range, use the minimum
31611 of that range instead of op1 to determine if modulo can be replaced
31612 with its first operand.
31614 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31616 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
31618 2017-02-08 Richard Biener <rguenther@suse.de>
31620 PR tree-optimization/71824
31621 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31622 Check all loops contained in the merged region.
31624 2017-02-07 Andrew Pinski <apinski@cavium.com>
31626 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
31628 2017-02-07 Andrew Pinski <apinski@cavium.com>
31630 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
31631 (thunderxt88): Likewise.
31632 (thunderxt81): Disable LSE and change v8.1 to v8.
31633 (thunderxt83): Likewise.
31635 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31636 Richard Biener <rguenther@suse.de>
31638 PR middle-end/79399
31639 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
31640 type from int to size_t.
31641 * ira-costs.c (struct_costs_size): Change type from int to size_t.
31643 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31645 PR rtl-optimization/79386
31646 * cprop.c (bypass_conditional_jumps): Initialize
31647 bypass_last_basic_block already before splitting bbs after
31648 unconditional traps...
31649 (bypass_conditional_jumps): ... rather than here.
31652 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
31653 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
31654 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
31655 fix -masm=intel patterns.
31657 2017-02-07 Richard Biener <rguenther@suse.de>
31659 PR tree-optimization/79256
31660 PR middle-end/79278
31661 * builtins.c (get_object_alignment_2): Use min_align_of_type
31662 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
31663 and ADJUST_FIELD_ALIGN.
31665 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
31667 * doc/tm.texi: Regenerate.
31668 * stor-layout.c (layout_decl): Adjust.
31669 (update_alignment_for_field): Likewise.
31670 (place_field): Likewise.
31671 (min_align_of_type): Likewise.
31672 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
31673 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
31674 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
31675 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
31676 * config/frv/frv.c (frv_adjust_field_align): Likewise.
31677 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
31678 * config/i386/i386.c (x86_field_alignment): Likewise.
31679 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
31680 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
31681 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
31682 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
31683 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
31684 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
31688 2017-01-30 Richard Biener <rguenther@suse.de>
31690 PR tree-optimization/79256
31691 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31692 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31695 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
31697 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
31698 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
31699 builtins to SImode and emit a zero-extend, if necessary.
31701 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31703 * docs/invoke.texi (RISC-V Options): Alphabetize.
31705 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31707 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
31710 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31712 * config/riscv/riscv.c: New file.
31713 * common/config/riscv/riscv-common.c: Likewise.
31714 * config.gcc: Likewise.
31715 * config/riscv/constraints.md: Likewise.
31716 * config/riscv/elf.h: Likewise.
31717 * config/riscv/generic.md: Likewise.
31718 * config/riscv/linux.h: Likewise.
31719 * config/riscv/multilib-generator: Likewise.
31720 * config/riscv/peephole.md: Likewise.
31721 * config/riscv/pic.md: Likewise.
31722 * config/riscv/predicates.md: Likewise.
31723 * config/riscv/riscv-builtins.c: Likewise.
31724 * config/riscv/riscv-c.c: Likewise.
31725 * config/riscv/riscv-ftypes.def: Likewise.
31726 * config/riscv/riscv-modes.def: Likewise.
31727 * config/riscv/riscv-opts.h: Likewise.
31728 * config/riscv/riscv-protos.h: Likewise.
31729 * config/riscv/riscv.h: Likewise.
31730 * config/riscv/riscv.md: Likewise.
31731 * config/riscv/riscv.opt: Likewise.
31732 * config/riscv/sync.md: Likewise.
31733 * config/riscv/t-elf-multilib: Likewise.
31734 * config/riscv/t-linux: Likewise.
31735 * config/riscv/t-linux-multilib: Likewise.
31736 * config/riscv/t-riscv: Likewise.
31737 * configure.ac: Likewise.
31738 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
31739 Waterman as RISC-V maintainers.
31740 * doc/install.texi: Add RISC-V entries.
31741 * doc/invoke.texi: Add RISC-V options section.
31742 * doc/md.texi: Add RISC-V constraints section.
31743 * configure: Regenerated.
31745 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
31748 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
31749 false values to be constant vectors with all 0 or all 1 bits set.
31750 (vcondu<mode><mode>): Likewise.
31751 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
31753 (fpmask_comparison_operator): Update comment.
31754 (vecint_comparison_operator): New predicate.
31755 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
31756 vector conditionals when the true and false values are constant
31757 vectors with all 0 bits or all 1 bits set.
31759 2017-02-06 Martin Sebor <msebor@redhat.com>
31761 PR tree-optimization/79376
31762 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
31764 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
31766 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
31767 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
31768 to simplify split condition.
31770 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31772 * omp-expand.c (oxpand_omp_atomic_fetch_op,
31773 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
31776 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
31778 PR rtl-optimization/68664
31779 * target.def (can_speculate_insn): New hook.
31780 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
31781 * doc/tm.texi: Regenerate.
31782 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
31783 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
31784 (rs6000_sched_can_speculate_insn): New function.
31786 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31788 PR tree-optimization/79284
31789 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
31790 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
31791 vectorizable_mask_load_store, vectorizable_operation,
31792 vect_is_simple_cond, get_same_sized_vectype): Use it instead
31793 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
31794 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
31795 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
31796 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
31797 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
31798 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
31799 is_gimple_assign (stmt). Replace another such test with
31800 is_gimple_assign (stmt).
31802 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31805 * config/avr/avr.c (rtl-iter.h): Include it.
31806 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
31807 (avr_legitimate_combined_insn): ...and implementation.
31809 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
31811 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
31812 * config/s390/s390.c (s390_const_operand_ok)
31813 (s390_canonicalize_comparison, s390_extract_part)
31814 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
31815 (s390_contiguous_bitmask_p, s390_rtx_costs)
31816 (legitimize_pic_address): Likewise.
31817 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
31818 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
31819 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
31820 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
31821 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
31823 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31825 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
31826 REGNO($0) == REGNO($1).
31828 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31830 * config/s390/linux.h(SIZE_TYPE): Add comment.
31832 2017-02-06 Julian Brown <julian@codesourcery.com>
31833 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31834 Virendra Pathak <virendra.pathak@broadcom.com>
31836 * config/aarch64/aarch64-cores.def: Change the scheduler
31838 * config/aarch64/aarch64.md: Include thunderx2t99.md.
31839 * config/aarch64/thunderx2t99.md: New file.
31841 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31843 * doc/standards.texi (Go Language): Update link to language
31846 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31848 * tree-eh.c (lower_resx): Sanitize profile.
31849 (cleanup_empty_eh_move_lp): Likewise.
31851 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31854 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
31856 * cfgloopmanip.h (loop_version): Update prototype.
31857 * modulo-sched.c (sms_schedule): Update call of loop_version.
31858 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
31859 * tree-parloops.c (gen_parallel_loop): Likewise.
31860 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
31861 * tree-ssa-loop-split.c (split_loop): Likewise.
31862 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
31863 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
31865 2017-02-05 Martin Liska <mliska@suse.cz>
31868 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
31870 (print_operand_address): Initialize a struct to zero.
31872 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31874 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
31875 garbage collector only in textual form.
31877 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31879 * doc/extend.texi (x86 specific memory model extensions for
31880 transactional memory): Simplify a phrase.
31882 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
31885 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
31886 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
31887 (atomic_storedi_1): Likewise.
31889 2017-02-04 Jakub Jelinek <jakub@redhat.com>
31891 PR tree-optimization/79338
31892 * tree-parloops.c (gather_scalar_reductions): Don't call
31893 vect_analyze_loop_form for loop->inner before destroying loop's
31896 2017-02-03 Martin Sebor <msebor@redhat.com>
31898 PR tree-optimization/79327
31899 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
31900 when precision has resulted in leading zeros.
31901 (format_integer): Adjust the likely counter to assume an unknown
31902 argument that may be zero is non-zero.
31904 2017-02-03 Jason Merrill <jason@redhat.com>
31907 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
31908 avoid copying non-taken branch.
31910 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31912 PR tree-optimization/79340
31913 * tree-vect-loop.c (vectorizable_reduction): Release
31914 vec_defs elements after safe_splicing them into other vectors.
31917 PR tree-optimization/79327
31918 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
31919 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
31921 (format_integer): Use wide_int_to_tree instead of build_int_cst
31922 + to_?hwi. If argmin is NULL, just set argmin and argmax to
31923 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
31924 of shortest and longest sequence.
31926 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
31928 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
31929 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
31931 2017-02-03 Walter Lee <walt@tilera.com>
31934 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
31935 after initial stackframe link reg save.
31936 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
31938 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31941 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
31942 wu for stxssp alternative.
31944 2017-02-03 Martin Sebor <msebor@redhat.com>
31946 PR tree-optimization/79352
31947 * gimple-fold.c (get_range_strlen): Add argument.
31948 (get_range_strlen): Change return type to bool.
31949 (get_maxval_strlen): Pass in a dummy argument.
31950 * gimple-fold.h (get_range_strlen): Change return type to bool.
31951 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
31952 * tree.h (array_at_struct_end_p): Add argument.
31953 * tree.c (array_at_struct_end_p): Handle it.
31955 2017-02-03 Martin Liska <mliska@suse.cz>
31958 * multiple_target.c (create_dispatcher_calls): Redirect edge
31959 from a caller of a dispatcher.
31960 (expand_target_clones): Make the clones local.
31961 (ipa_target_clone): Do both target clones and resolvers.
31962 (ipa_dispatcher_calls): Remove the pass.
31963 (pass_dispatcher_calls::gate): Likewise.
31964 (make_pass_dispatcher_calls): Likewise.
31965 * passes.def (pass_target_clone): Put as very first IPA early
31968 2017-02-03 Martin Liska <mliska@suse.cz>
31970 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
31971 in case of a function with ifunc attribute.
31973 2017-02-03 Martin Liska <mliska@suse.cz>
31975 * cgraph.c (cgraph_node::dump): Dump function version info.
31976 * symtab.c (symtab_node::dump_base): Add missing new line.
31978 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31980 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
31981 (ifcombine_ifandif): Use it.
31983 2017-02-03 Martin Liska <mliska@suse.cz>
31985 * doc/invoke.texi: Document default value for
31986 use-after-scope-direct-emission-threshold.
31988 2017-02-03 Martin Liska <mliska@suse.cz>
31990 PR tree-optimization/79339
31991 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
31992 (format_floating): Likewise.
31994 2017-02-03 Martin Liska <mliska@suse.cz>
31997 * ipa-prop.c (ipa_node_params_t::insert): Remove current
31999 (ipa_node_params_t::remove): Likewise.
32000 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
32001 initialization from removed ipa_node_params_t::insert.
32002 (ipa_node_params::~ipa_node_params): Move from removed
32003 ipa_node_params_t::release.
32004 * symbol-summary.h (symbol_summary::m_released): New member.
32005 Do not release a summary twice. Do not allow to call finalizer
32006 for types of a summary that live in GGC memory.
32008 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32010 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
32013 2017-02-02 Martin Sebor <msebor@redhat.com>
32015 PR middle-end/79275
32016 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
32017 (format_string): Tighten up the range of output for non-constant
32018 strings and correct the expected range for wide non-constant strings.
32020 2017-02-02 Martin Sebor <msebor@redhat.com>
32022 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
32024 PR middle-end/32003
32025 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
32027 (-fdump-tree-@var): Add to index and document how to come up
32028 with pass-specific option and dump file names.
32029 (-fdump-passes): Clarify where to look for output.
32031 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32033 PR middle-end/77445
32034 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
32035 statistics of the analyzed path; allow threading for speed when
32036 any of BBs along the path are optimized for speed.
32038 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
32040 PR middle-end/78468
32041 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
32042 settings of the virtual registers.
32045 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
32047 * explow.c (get_dynamic_stack_size): Take known alignment of stack
32048 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
32051 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32053 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
32054 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
32056 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32058 * config/s390/s390.md: Add missing comments with the expanded
32060 * config/s390/vector.md: Likewise.
32061 * config/s390/vx-builtins.md: Likewise.
32063 2017-02-02 Jakub Jelinek <jakub@redhat.com>
32066 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
32067 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
32068 conditions on a single line.
32070 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32072 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32073 __S390_VX__ to __VX__.
32075 2017-02-01 Andrew Pinski <apinski@cavium.com>
32077 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
32078 stmt_info to record_stmt_cost.
32079 (vect_get_known_peeling_cost): Pass stmt_info if known to
32081 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
32082 cpu_vector_cost field into
32083 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
32084 field into vec_int_stmt_cost and vec_fp_stmt_cost.
32085 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
32086 splitting of scalar_stmt_cost and vec_stmt_cost.
32087 (thunderx_vector_cost): Likewise.
32088 (cortexa57_vector_cost): LIkewise.
32089 (exynosm1_vector_cost): Likewise.
32090 (xgene1_vector_cost): Likewise.
32091 (thunderx2t99_vector_cost): Improve after the splitting of the two
32093 (aarch64_builtin_vectorization_cost): Update for the splitting of
32094 scalar_stmt_cost and vec_stmt_cost.
32096 2017-02-01 Torvald Riegel <triegel@redhat.com>
32097 Richard Henderson <rth@redhat.com>
32099 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
32100 conditional on existance of a fast atomic load.
32101 * optabs-query.c (can_atomic_load_p): New function.
32102 * optabs-query.h (can_atomic_load_p): Declare it.
32103 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
32104 no fast atomic load is available for the particular size of access.
32105 (expand_atomic_compare_and_swap): Likewise.
32106 (expand_atomic_load): Likewise.
32107 (expand_atomic_store): Likewise.
32108 (expand_atomic_fetch_op): Likewise.
32109 * testsuite/lib/target-supports.exp
32110 (check_effective_target_sync_int_128): Remove x86 because it provides
32111 no fast atomic load.
32112 (check_effective_target_sync_int_128_runtime): Likewise.
32114 2017-02-01 Richard Biener <rguenther@suse.de>
32116 * graphite.c: Include tree-vectorizer.h for find_loop_location.
32117 (graphite_transform_loops): Provide opt-info for optimized nests.
32118 * tree-parloop.c (parallelize_loops): Provide opt-info for
32119 parallelized loops.
32121 2017-02-01 Richard Biener <rguenther@suse.de>
32123 PR middle-end/79315
32124 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
32125 was not set before.
32127 2017-02-01 Richard Biener <rguenther@suse.de>
32129 PR tree-optimization/71824
32130 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32131 Verify the loops are valid in the merged SESE region.
32132 (scop_detection::can_represent_loop_1): Check analyzing the
32133 evolution of the number of iterations in the region succeeds.
32135 2017-01-31 Ian Lance Taylor <iant@golang.org>
32137 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
32138 REG_ARGS_SIZE note to 32-bit push insns and call insn.
32140 2017-01-31 David Malcolm <dmalcolm@redhat.com>
32142 PR preprocessor/79210
32143 * input.c (get_substring_ranges_for_loc): Replace line_width
32144 assertion with error-handling.
32146 2017-01-31 Richard Biener <rguenther@suse.de>
32148 PR tree-optimization/77318
32149 * graphite-sese-to-poly.c (extract_affine): Fix assert.
32150 (create_pw_aff_from_tree): Take loop parameter.
32151 (add_condition_to_pbb): Pass loop of the condition to
32152 create_pw_aff_from_tree.
32154 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32156 * config/s390/s390.c (s390_asan_shadow_offset): New function.
32157 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
32159 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
32163 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
32165 (convert_int_to_float128): Likewise.
32166 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
32167 (convert_int_to_float128): Likewise.
32168 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
32169 (UNSPEC_IEEE128_CONVERT): Likewise.
32170 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
32171 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
32172 Use local variables for IBM extended format.
32173 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
32174 (fix_trunc<mode>si2_fprs): Likewise.
32175 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
32176 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
32177 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
32178 to know that we can now have integers of all sizes in vector
32180 (fix<uns>_<mode>di2_hw): Likewise.
32181 (float<uns>_<mode>si2_hw): Likewise.
32182 (fix_<mode>si2_hw): Likewise.
32183 (fixuns_<mode>si2_hw): Likewise.
32184 (float<uns>_<mode>di2_hw): Likewise.
32185 (float_<mode>di2_hw): Likewise.
32186 (float_<mode>si2_hw): Likewise.
32187 (floatuns_<mode>di2_hw): Likewise.
32188 (floatuns_<mode>si2_hw): Likewise.
32189 (xscvqp<su>wz_<mode>): Delete, no longer used.
32190 (xscvqp<su>dz_<mode>): Likewise.
32191 (xscv<su>dqp_<mode>): Likewise.
32192 (ieee128_mfvsrd_64bit): Likewise.
32193 (ieee128_mfvsrd_32bit): Likewise.
32194 (ieee128_mfvsrwz): Likewise.
32195 (ieee128_mtvsrw): Likewise.
32196 (ieee128_mtvsrd_64bit): Likewise.
32197 (ieee128_mtvsrd_32bit): Likewise.
32199 2017-01-31 Martin Liska <mliska@suse.cz>
32202 * ipa-prop.c (ipa_free_all_node_params): Call release method
32203 instead of ~sumbol_summary to not to trigger double times
32206 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32208 PR tree-optimization/71691
32209 * bitmap.h (class auto_bitmap): New.
32210 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32211 is_maybe_undefined instead of ssa_undefined_value_p.
32213 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32215 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32216 __S390_ARCH_LEVEL__ to __ARCH__.
32218 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32220 PR tree-optimization/79267
32221 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32222 if should_remove_lhs_p is true.
32224 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32227 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32228 (add_alignment_attribute): New.
32229 (base_type_die): Add alignment attribute.
32230 (subrange_type_die): Likewise.
32231 (modified_type_die): Likewise.
32232 (gen_array_type_die): Likewise.
32233 (gen_descr_array_type_die: Likewise.
32234 (gen_enumeration_type_die): Likewise.
32235 (gen_subprogram_die): Likewise.
32236 (gen_variable_die): Likewise.
32237 (gen_field_die): Likewise.
32238 (gen_ptr_to_mbr_type_die): Likewise.
32239 (gen_struct_or_union_type_die): Likewise.
32240 (gen_subroutine_type_die): Likewise.
32241 (gen_typedef_die): Likewise.
32242 (base_type_cmp): Compare alignment attribute.
32244 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32247 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32248 (setb_unsigned) New pattern for setb with CCUNS.
32249 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32250 subfc./subfe sequence to avoid overflow problems. Generate a
32251 shorter sequence with cmpld/setb for power9.
32252 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32253 for generating subfc. instruction.
32254 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32255 now uses this instruction.
32257 2017-01-30 Ian Lance Taylor <iant@google.com>
32260 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32261 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32263 2017-01-30 Martin Sebor <msebor@redhat.com>
32265 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32266 Move constant to the right of a relational operator.
32267 (get_mpfr_format_length, format_character, format_string): Ditto.
32268 (should_warn_p, maybe_warn): Same.
32270 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32272 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32275 * asan.c (get_translation_unit_decl): Remove function.
32276 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32278 2017-01-30 Martin Liska <mliska@suse.cz>
32280 PR gcov-profile/79259
32281 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32282 -fprofile-generate.
32284 2017-01-30 Martin Liska <mliska@suse.cz>
32287 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32288 Initialize variables with NULL value.
32290 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32293 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32295 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32297 2017-01-30 Richard Biener <rguenther@suse.de>
32299 PR tree-optimization/79276
32300 * tree-vrp.c (process_assert_insertions): Properly adjust common
32301 when removing a duplicate.
32303 2017-01-30 Richard Biener <rguenther@suse.de>
32305 PR tree-optimization/79256
32306 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32307 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32309 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32311 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32314 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32315 ("*r<noxa>sbg_<mode>_sll_bitmask")
32316 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32317 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32318 Use contiguous_bitmask_nowrap_operand.
32320 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32323 * config/rs6000/altivec.h (vec_xl): Revise #define.
32324 (vec_xst): Likewise.
32326 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32328 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32330 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32332 PR rtl-optimization/79194
32333 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32334 traps before call to bypass_conditional_jumps.
32336 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32338 PR tree-optimization/71374
32339 * lra-constraints.c (check_conflict_input_operands): New.
32340 (match_reload): Use it.
32342 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32345 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32346 account to calculate conflict_set.
32348 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32350 PR rtl-optimization/78559
32351 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32352 other_insn in combine.
32354 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32356 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32357 uint16_type_node for BT_UINT16.
32359 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32361 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32362 "RTL Tests" to menu.
32363 (GIMPLE Tests): New node.
32364 (RTL Tests): New node.
32366 2017-01-27 Richard Biener <rguenther@suse.de>
32368 PR tree-optimization/79245
32369 * tree-loop-distribution.c (distribute_loop): Apply cost
32370 modeling also to detected patterns.
32372 2017-01-27 Richard Biener <rguenther@suse.de>
32374 PR tree-optimization/71433
32375 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32376 (compare_assert_loc): New function.
32377 (process_assert_insertions): Sort and optimize assert locations
32378 to remove duplicates and push down identical assertions on
32379 edges to their destination block.
32381 2017-01-27 Richard Biener <rguenther@suse.de>
32383 PR tree-optimization/79244
32384 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32385 out SSA names even if abnormal.
32387 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32389 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32390 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32391 instead of MPFR_RNDN.
32393 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32396 * arm.c (arm_option_override): Don't call build_target_option_node
32397 until after doing all option overrides.
32398 (arm_valid_target_attribute_tree): Likewise.
32400 2017-01-27 Martin Liska <mliska@suse.cz>
32402 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32403 for {cd}tors and C++ {cd}tors.
32405 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32407 * config/s390/s390.md ("*setmem_long_and")
32408 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32410 2017-01-26 Martin Sebor <msebor@redhat.com>
32412 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32415 2017-01-26 Martin Sebor <msebor@redhat.com>
32417 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32418 HAVE_DFmode before using XFmode or DFmode.
32419 (parse_directive): Avoid using the z length modifier to avoid
32420 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32422 PR middle-end/78703
32423 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32424 to accept adjustment as an array.
32425 (get_int_range): New function.
32426 (struct directive): Make width and prec arrays.
32427 (directive::set_width, directive::set_precision): Call get_int_range.
32428 (format_integer, format_floating): Handle width and precision ranges.
32429 (format_string, parse_directive): Same.
32431 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32434 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32435 comdat_type_p set, just clone them, but keep the children in the
32439 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32440 which have direct callers with -fvar-tracking-assignments enabled
32442 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32443 inside of type units.
32445 2017-01-26 Martin Sebor <msebor@redhat.com>
32447 PR middle-end/78703
32448 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32450 (struct format_result): Replace number_chars, number_chars_min,
32451 and number_chars_max with a single member of struct result_range.
32453 (format_result::operator+=): Adjust.
32454 (struct fmtresult): Remove bounded. Handle likely and unlikely
32456 (fmtresult::adjust_for_width_or_precision): New function.
32457 (fmtresult:type_max_digits): New function.
32458 (bytes_remaining): Handle likely and unlikely counters.
32459 (min_bytes_remaining): Remove.
32460 (format_percent): Simplify.
32461 (format_integer, format_floating): Set likely and unlikely counters.
32462 (get_string_length, format_character, format_string): Same.
32463 (format_plain, should_warn_p): New function.
32464 (maybe_warn): Call should_warn_p. Update diagnostic messages
32465 and handle those for all directives, including plain strings.
32466 (format_directive): Handle likely and unlikely counters.
32467 Remove unnecessary quoting from diagnostics. Add an informational
32469 (add_bytes): Remove.
32470 (pass_sprintf_length::compute_format_length): Simplify.
32471 (try_substitute_return_value): Handle likely and unlikely counters.
32473 2017-01-26 Carl Love <cel@us.ibm.com>
32475 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
32476 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
32478 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
32481 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
32482 endianess for subregs into account.
32483 * lra-constraints.c (lra_constraints): Do risky transformations
32484 always on the first iteration.
32485 * lra-lives.c (check_pseudos_live_through_calls): Add arg
32486 last_call_used_reg_set.
32487 (process_bb_lives): Define and use last_call_used_reg_set.
32488 * lra.c (lra): Always continue after lra_constraints on the first
32491 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
32493 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
32495 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
32497 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32499 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
32500 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
32501 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
32502 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
32503 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
32504 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
32505 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
32506 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
32507 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
32509 2017-01-26 Marek Polacek <polacek@redhat.com>
32512 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
32513 for the third operand.
32515 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32517 PR middle-end/79236
32518 * omp-low.c (struct omp_context): Add simt_stmt field.
32519 (scan_omp_for): Return omp_context *.
32520 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
32521 context to the _simt_ SIMD stmt.
32522 (lower_omp_for): For combined SIMD with sibling _simt_
32523 SIMD, make sure to use the same decls in _looptemp_
32524 clauses as in the sibling.
32526 2017-01-26 David Sherwood <david.sherwood@arm.com>
32528 PR middle-end/79212
32529 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
32532 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32535 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
32536 emit fld b; fld a; if possible.
32538 * brig-builtins.def: Update copyright years.
32539 * config/arm/arm_acle_builtins.def: Update copyright years.
32541 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
32544 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
32545 constraint instead of o for the stxsd instruction.
32547 2017-01-25 Carl Love <cel@us.ibm.com>
32549 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
32550 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
32552 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
32554 * doc/invoke.texi (C++ Dialect Options): Fix typo.
32556 2017-01-25 Richard Biener <rguenther@suse.de>
32558 PR tree-optimization/69264
32559 * target.def (vector_alignment_reachable): Improve documentation.
32560 * doc/tm.texi: Regenerate.
32561 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
32563 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
32564 earlier changes with respect to TYPE_USER_ALIGN.
32565 (vector_alignment_reachable_p): Likewise. Improve dumping.
32567 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32570 * config/arm/arm.md (xordi3): Force constant operand into a register
32573 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32575 * doc/invoke.texi (-fstore-merging): Correct default optimization
32576 levels at which it is enabled.
32577 (-O): Move -fstore-merging from list to...
32580 2017-01-25 Richard Biener <rguenther@suse.de>
32583 * omp-expand.c: Include debug.h.
32584 (expand_omp_taskreg): Make sure to generate early debug before
32585 outlining anything from a function.
32586 (expand_omp_target): Likewise.
32587 (grid_expand_target_grid_body): Likewise.
32589 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
32592 * asan.c (get_translation_unit_decl): New function.
32593 (asan_add_global): Extract modules file name from globals
32594 TRANSLATION_UNIT_DECL name.
32596 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
32599 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
32600 for long calls with APCS frame and VFP.
32602 2017-01-24 David Malcolm <dmalcolm@redhat.com>
32604 * cfg.c (original_copy_tables_initialized_p): New function.
32605 * cfg.h (original_copy_tables_initialized_p): New decl.
32606 * cfgrtl.c (relink_block_chain): Guard the call to
32607 free_original_copy_tables with a call to
32608 original_copy_tables_initialized_p.
32609 * cgraph.h (symtab_node::native_rtl_p): New decl.
32610 * cgraphunit.c (symtab_node::native_rtl_p): New function.
32611 (symtab_node::needed_p): Don't assert for early assembly output
32612 for __RTL functions.
32613 (cgraph_node::finalize_function): Set "force_output" for __RTL
32615 (cgraph_node::analyze): Bail out early for __RTL functions.
32616 (analyze_functions): Update assertion to support __RTL functions.
32617 (cgraph_node::expand): Bail out early for __RTL functions.
32618 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
32620 * function.h (struct function): Update comment for field
32622 * gimple-expr.c: Include "tree-pass.h".
32623 (gimple_has_body_p): Return false for __RTL functions.
32624 * Makefile.in (OBJS): Add run-rtl-passes.o.
32625 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
32627 (gcc::pass_manager::get_clean_slate): New accessor.
32628 * passes.c: Include "insn-addr.h".
32629 (should_skip_pass_p): Add logging. Update logic for running
32630 "expand" to be compatible with both __GIMPLE and __RTL. Guard
32631 property-provider override so it is only done for gimple passes.
32633 (skip_pass): New function.
32634 (execute_one_pass): Call skip_pass when skipping passes.
32635 * read-md.c (md_reader::read_char): Support filtering
32636 the input to a subset of line numbers.
32637 (md_reader::md_reader): Initialize fields
32638 m_first_line and m_last_line.
32639 (md_reader::read_file_fragment): New function.
32640 * read-md.h (md_reader::read_file_fragment): New decl.
32641 (md_reader::m_first_line): New field.
32642 (md_reader::m_last_line): New field.
32643 * read-rtl-function.c (function_reader::create_function): Only
32644 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
32645 curr_properties. Set DECL_INITIAL to a dummy block.
32646 (read_rtl_function_body_from_file_range): New function.
32647 * read-rtl-function.h (read_rtl_function_body_from_file_range):
32649 * run-rtl-passes.c: New file.
32650 * run-rtl-passes.h: New file.
32652 2017-01-24 Jeff Law <law@redhat.com>
32654 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
32657 2017-01-24 Bin Cheng <bin.cheng@arm.com>
32659 PR tree-optimization/79159
32660 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
32661 (record_nonwrapping_iv): Improve boundary using above function if no
32662 value range information.
32664 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
32665 Martin Jambor <mjambor@suse.cz>
32667 * brig-builtins.def: New file.
32668 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
32669 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
32670 (DEF_HSAIL_SAT_BUILTIN): Likewise.
32671 (DEF_HSAIL_INTR_BUILTIN): Likewise.
32672 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
32673 * builtin-types.def (BT_INT8): New.
32674 (BT_INT16): Likewise.
32675 (BT_UINT8): Likewise.
32676 (BT_UINT16): Likewise.
32677 (BT_FN_ULONG): Likewise.
32678 (BT_FN_UINT_INT): Likewise.
32679 (BT_FN_UINT_ULONG): Likewise.
32680 (BT_FN_UINT_LONG): Likewise.
32681 (BT_FN_UINT_PTR): Likewise.
32682 (BT_FN_ULONG_PTR): Likewise.
32683 (BT_FN_INT8_FLOAT): Likewise.
32684 (BT_FN_INT16_FLOAT): Likewise.
32685 (BT_FN_UINT32_FLOAT): Likewise.
32686 (BT_FN_UINT16_FLOAT): Likewise.
32687 (BT_FN_UINT8_FLOAT): Likewise.
32688 (BT_FN_UINT64_FLOAT): Likewise.
32689 (BT_FN_UINT16_UINT32): Likewise.
32690 (BT_FN_UINT32_UINT16): Likewise.
32691 (BT_FN_UINT16_UINT16_UINT16): Likewise.
32692 (BT_FN_INT_PTR_INT): Likewise.
32693 (BT_FN_UINT_PTR_UINT): Likewise.
32694 (BT_FN_LONG_PTR_LONG): Likewise.
32695 (BT_FN_ULONG_PTR_ULONG): Likewise.
32696 (BT_FN_VOID_UINT64_UINT64): Likewise.
32697 (BT_FN_UINT8_UINT8_UINT8): Likewise.
32698 (BT_FN_INT8_INT8_INT8): Likewise.
32699 (BT_FN_INT16_INT16_INT16): Likewise.
32700 (BT_FN_INT_INT_INT): Likewise.
32701 (BT_FN_UINT_FLOAT_UINT): Likewise.
32702 (BT_FN_FLOAT_UINT_UINT): Likewise.
32703 (BT_FN_ULONG_UINT_UINT): Likewise.
32704 (BT_FN_ULONG_UINT_PTR): Likewise.
32705 (BT_FN_ULONG_ULONG_ULONG): Likewise.
32706 (BT_FN_UINT_UINT_UINT): Likewise.
32707 (BT_FN_VOID_UINT_PTR): Likewise.
32708 (BT_FN_UINT_UINT_PTR: Likewise.
32709 (BT_FN_UINT32_UINT64_PTR): Likewise.
32710 (BT_FN_INT_INT_UINT_UINT): Likewise.
32711 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
32712 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
32713 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
32714 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
32715 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
32716 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
32717 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
32718 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
32719 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
32720 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
32721 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
32722 * doc/frontends.texi: List BRIG FE.
32723 * doc/install.texi (Testing): Add BRIG tesring requirements.
32724 * doc/invoke.texi (Overall Options): Mention BRIG.
32725 * doc/standards.texi (Standards): Doucment BRIG HSA version.
32727 2017-01-24 Richard Biener <rguenther@suse.de>
32729 PR translation/79208
32730 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
32732 2017-01-24 Martin Jambor <mjambor@suse.cz>
32735 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
32736 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
32737 and known_contexts.
32739 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
32741 PR middle-end/79123
32742 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
32743 casts from signed to unsigned really don't have a range.
32745 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
32747 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
32748 GMP_RNDx for compatiblity.
32750 2017-01-24 Martin Liska <mliska@suse.cz>
32753 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
32754 that would prevent us to call alloca with -1 as argument.
32756 2017-01-24 Jakub Jelinek <jakub@redhat.com>
32758 * dwarf2out.c (output_compilation_unit_header, output_file_names):
32759 Avoid -Wformat-security warning.
32761 2017-01-23 Andrew Pinski <apinski@cavium.com>
32763 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
32766 2017-01-23 Martin Sebor <msebor@redhat.com>
32768 PR middle-end/78703
32769 * gimple-ssa-sprintf.c (warn_level): New global.
32770 (format_integer): Use it here and throughout the rest of the file.
32771 Use the same switch to compute sign as base.
32772 (maybe_warn): New function.
32773 (format_directive): Factor out warnings into maybe_warn.
32774 Add debugging output. Use warn_level.
32775 (add_bytes): Use warn_level.
32776 (pass_sprintf_length::compute_format_length): Add debugging output.
32777 (try_substitute_return_value): Same.
32778 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
32780 PR middle-end/78703
32781 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
32782 (struct fmtresult, format_integer, format_floating): Adjust.
32783 (fmtresult::fmtresult): Set max correctly in two argument ctor.
32784 (get_string_length, format_string,format_directive): Same.
32785 (pass_sprintf_length::compute_format_length): Same.
32786 (try_substitute_return_value): Simplify slightly.
32788 PR middle-end/78703
32789 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
32790 (fmtresult::operator+=): Outlined.
32791 (struct fmtresult): Add ctors.
32792 (struct conversion_spec): Rename...
32793 (struct directive): ...to this. Add and remove data members.
32794 (directive::set_width, directive::set_precision): New functions.
32795 (format_percent): Use fmtresult ctor.
32796 (get_width_and_precision): Remove.
32797 (format_integer): Make naming changes. Avoid computing width and
32799 (format_floating): Same. Adjust indentation.
32800 (format_character, format_none): New functions.
32801 (format_string): Moved character handling to format_character.
32802 (format_directive): Remove arguments, change return type.
32803 (parse_directive): New function.
32804 (pass_sprintf_length::compute_format_length): Move directive
32805 parsing to parse_directive.
32807 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32809 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
32810 (assign_assembler_name_if_needed): ... this.
32811 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
32812 (assign_assembler_name_if_needed): ... this.
32813 (free_lang_data_in_cgraph): Adjust callers.
32814 * cgraphunit.c (cgraph_node::analyze): Likewise.
32815 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
32818 2017-01-23 Richard Biener <rguenther@suse.de>
32820 PR tree-optimization/79088
32821 PR tree-optimization/79188
32822 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
32823 resetting loop bounds after last path deletion. Reset loop
32824 bounds of the target loop, make code match the comments.
32825 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
32826 Make sure loops need no fixups.
32828 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32830 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
32831 exponent support with double type for first argument.
32832 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
32833 type returned by __builtin_vec_extract_sig,
32834 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
32835 functions from "vector int" to "vector unsigned int" or from
32836 "vector long long int" to "vector unsigned long long int".
32837 Changed type returned by __builtin_vec_extract_exp,
32838 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
32839 functions from "vector int" to "vector unsigned int" or from
32840 "vector long long int" to "vector unsigned long long int".
32841 Changed return type of __builtin_vec_test_data_class,
32842 __builtin_vec_test_data_class_sp, and
32843 __builtin_vec_test_data_class_dp from "vector int" to
32844 "vector bool int" or from "vector long long int" to "vector bool
32845 long long int" and changed second argument type from "unsigned
32846 int" to "int". Added new overloaded function forms "vector float
32847 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
32848 "vector float __builtin_vec_insert_exp_sp (vector float, vector
32849 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
32850 double, vector unsigned long long int)" and "vector double
32851 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
32852 long int)". Changed return type of
32853 __builtin_scalar_test_data_class and
32854 __builtin_scalar_test_data_class_sp and
32855 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
32856 int" and changed second argument from "unsigned int" to "int".
32857 Changed type returned by __builtin_scalar_test_neg,
32858 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
32859 from "int" to "bool int". Added new overloaded function form
32860 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
32861 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
32862 exponent double-precision with floating point first argument.
32863 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
32864 documentation of scalar_test_data_class, scalar_test_neg,
32865 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
32866 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
32867 vec_test_data_class built-in functions to reflect refinements in
32868 their type signatures.
32870 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
32872 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
32874 (aarch64_elf_asm_destructor): Likewise.
32876 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
32878 PR rtl-optimization/78634
32879 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
32880 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
32881 * ifcvt.c (noce_try_cmove): Add missing cost check.
32883 PR rtl-optimization/71724
32884 * combine.c (if_then_else_cond): Look for situations where it is
32885 beneficial to undo the work of one of the recursive calls.
32887 2017-01-23 Bin Cheng <bin.cheng@arm.com>
32889 PR tree-optimization/70754
32890 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
32891 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
32892 combined stmt before it if not NULL.
32893 (combine_chains): Process refs reversely and compute dominance point
32896 2017-01-23 Martin Liska <mliska@suse.cz>
32898 PR tree-optimization/79196
32899 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
32900 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
32902 (strlen_optimize_stmt): Call the renamed function.
32904 2017-01-23 Michael Matz <matz@suse.de>
32906 PR tree-optimization/78384
32907 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
32909 2017-01-23 Richard Biener <rguenther@suse.de>
32911 PR tree-optimization/79186
32912 * tree-vrp.c (register_new_assert_for): Make sure we've seen
32913 both incoming edges before moving an assert.
32915 2017-01-23 Martin Jambor <mjambor@suse.cz>
32917 * ipa-prop.c (load_from_param_1): Removed.
32918 (load_from_unmodified_param): Bits from load_from_param_1 put back
32920 (load_from_param): Removed.
32921 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
32922 with stmt. Reverted back to use of load_from_unmodified_param.
32924 2017-01-23 Martin Jambor <mjambor@suse.cz>
32927 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
32928 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
32929 field a pointer to garbage collected vector, mark lattices and
32930 ipcp_orig_node with GTY((skip)).
32931 (ipa_get_param_count): Adjust to descriptors being a pointer.
32932 (ipa_get_param): Likewise.
32933 (ipa_get_type): Likewise.
32934 (ipa_get_param_move_cost): Likewise.
32935 (ipa_set_param_used): Likewise.
32936 (ipa_get_controlled_uses): Likewise.
32937 (ipa_set_controlled_uses): Likewise.
32938 (ipa_is_param_used): Likewise.
32939 (ipa_node_params_t): Move into garbage collector. New methods insert
32941 (ipa_node_params_sum): Annotate wth GTY(()).
32942 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
32944 (ipa_load_from_parm_agg): Adjust declaration.
32945 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
32946 * ipa-profile.c (ipa_profile): Likewise.
32947 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
32948 (ipa_populate_param_decls): Make descriptors parameter garbage
32950 (ipa_dump_param): Adjust to descriptors being a pointer.
32951 (ipa_alloc_node_params): Likewise.
32952 (ipa_initialize_node_params): Likewise.
32953 (load_from_param_1): Make descriptors parameter garbage collected.
32954 (load_from_unmodified_param): Likewise.
32955 (load_from_param): Likewise.
32956 (ipa_load_from_parm_agg): Likewise.
32957 (ipa_node_params::~ipa_node_params): Removed.
32958 (ipa_free_all_node_params): Remove call to delete operator.
32959 (ipa_node_params_t::insert): New.
32960 (ipa_node_params_t::remove): Likewise.
32961 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
32962 copy known_csts and known_contexts vectors.
32963 (ipa_read_node_info): Adjust to descriptors being a pointer.
32964 (ipcp_modif_dom_walker): Make m_descriptors field garbage
32966 (ipcp_transform_function): Make descriptors variable garbage
32969 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
32971 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
32972 * config/i386/avx512dqintrin.h: Ditto.
32973 * config/i386/avx512fintrin.h: Ditto.
32974 * config/i386/i386.c: Handle new builtins.
32975 * config/i386/i386-builtin.def: Add new builtins.
32976 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
32977 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
32979 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32980 Martin Liska <mliska@suse.cz>
32982 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
32983 * asan.c (asan_expand_poison_ifn): Support stores and use
32984 appropriate ASAN report function.
32985 * internal-fn.c (expand_ASAN_POISON_USE): New function.
32986 * internal-fn.def (ASAN_POISON_USE): Declare.
32987 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
32988 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
32989 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
32990 ASAN_POISON calls w/o LHS.
32991 * tree-ssa.c (execute_update_addresses_taken): Create clobber
32992 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
32993 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
32994 * gimplify.c (asan_poison_variables): Add attribute
32995 use_after_scope_memory to variables that really needs to live
32997 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
32998 having the attribute.
33000 2017-01-23 Martin Liska <mliska@suse.cz>
33002 * asan.c (create_asan_shadow_var): New function.
33003 (asan_expand_poison_ifn): Likewise.
33004 * asan.h (asan_expand_poison_ifn): New declaration.
33005 * internal-fn.c (expand_ASAN_POISON): Likewise.
33006 * internal-fn.def (ASAN_POISON): New builtin.
33007 * sanopt.c (pass_sanopt::execute): Expand
33008 asan_expand_poison_ifn.
33009 * tree-inline.c (copy_decl_for_dup_finish): Make function
33011 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
33012 * tree-ssa.c (is_asan_mark_p): New function.
33013 (execute_update_addresses_taken): Rewrite local variables
33014 (identified just by use-after-scope as addressable) into SSA.
33016 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33018 * doc/install.texi (Specific): opensource.apple.com uses https
33019 now. Remove trailing slash.
33021 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33023 * README.Portability: Remove note on an Irix compatibility issue.
33025 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
33027 * gcov.c (INCLUDE_ALGORITHM): Define.
33028 (INCLUDE_VECTOR): Define.
33029 No longer include <vector> and <algorithm> directly.
33031 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
33033 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
33035 * doc/invoke.texi (Code Gen Options): Ditto.
33037 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
33040 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
33042 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
33044 rtl-optimization/79125
33045 * cprop.c (local_cprop_pass): Handle cases where we make an
33046 unconditional trap.
33048 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
33052 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
33053 read from, for big endian.
33055 2017-01-20 Jiong Wang <jiong.wang@arm.com>
33057 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
33058 register pauth builtins for LP64 only.
33060 2017-01-20 Marek Polacek <polacek@redhat.com>
33063 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
33066 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33068 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
33070 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
33071 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
33072 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
33074 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33077 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
33078 in big-endian mode when they are not a single duplicated value.
33080 2017-01-20 Richard Biener <rguenther@suse.de>
33082 * BASE-VER: Bump to 7.0.1.
33084 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33086 * omp-low.c (omplow_simd_context): New struct. Use it...
33087 (lower_rec_simd_input_clauses): ...here and...
33088 (lower_rec_input_clauses): ...here to hold common data. Adjust all
33089 references to idx, lane, max_vf, is_simt.
33091 2017-01-20 Graham Markall <graham.markall@embecosm.com>
33093 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
33096 2017-01-20 Martin Jambor <mjambor@suse.cz>
33098 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
33099 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
33101 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
33102 (GTFILES): Rename hsa.c to hsa-common.c.
33103 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
33104 * hsa-dump.c: Likewise.
33105 * hsa-gen.c: Likewise.
33106 * hsa-regalloc.c: Likewise.
33107 * ipa-hsa.c: Likewise.
33108 * omp-expand.c: Likewise.
33109 * omp-low.c: Likewise.
33110 * toplev.c: Likewise.
33112 2017-01-20 Marek Polacek <polacek@redhat.com>
33115 * doc/invoke.texi: Document -Wduplicated-branches.
33116 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
33117 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
33118 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
33119 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
33120 return 0 only when not OEP_LEXICOGRAPHIC.
33121 (fold_build_cleanup_point_expr): Use the expression
33122 location when building CLEANUP_POINT_EXPR.
33123 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
33124 * tree.c (add_expr): Handle error_mark_node.
33126 2017-01-20 Martin Liska <mliska@suse.cz>
33129 * tree-profile.c (init_ic_make_global_vars): Do not call
33131 (gimple_init_gcov_profiler): Likewise.
33133 2017-01-20 Martin Liska <mliska@suse.cz>
33136 * cgraph.h (maybe_create_reference): Remove argument and
33138 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
33140 * ipa-cp.c (create_specialized_node): Likewise.
33141 * symtab.c (symtab_node::maybe_create_reference): Handle
33142 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
33144 2017-01-20 Martin Liska <mliska@suse.cz>
33146 * read-rtl-function.c (function_reader::create_function): Use
33147 build_decl instread of build_decl_stat.
33149 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
33151 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
33152 * config/i386/avx512dqintrin.h: Ditto.
33153 * config/i386/avx512fintrin.h: Ditto.
33154 * config/i386/i386-builtin-types.def: Add new types.
33155 * config/i386/i386.c: Handle new types.
33156 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
33157 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
33158 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
33159 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
33160 (__builtin_ia32_kshiftridi): New.
33161 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
33163 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
33167 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
33168 define to rs6000_init_stack_protect_guard.
33169 (rs6000_init_stack_protect_guard): New function.
33171 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33172 Yunqiang Su <yunqiang.su@imgtec.com>
33174 * config.gcc (supported_defaults): Add madd4.
33175 (with_madd4): Add validation.
33176 (all_defaults): Add madd4.
33177 * config/mips/mips.opt (mmadd4): New option.
33178 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33180 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33182 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33183 (ISA_HAS_FUSED_MADD4): Likewise.
33184 * doc/invoke.texi (-mmadd4): Document the new option.
33185 * doc/install.texi (--with-madd4): Document the new option.
33187 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33189 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33190 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33191 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33192 (aarch64_init_pauth_hint_builtins): New.
33193 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33194 (aarch64_expand_builtin): Expand new builtins.
33196 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33198 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33199 * combine-stack-adj.c (no_unhandled_cfa): Handle
33200 REG_CFA_TOGGLE_RA_MANGLE.
33201 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33202 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33203 info for return address signing.
33204 (aarch64_expand_epilogue): Likewise.
33206 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33208 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33209 * config/aarch64/aarch64-protos.h
33210 (aarch64_return_address_signing_enabled): New declaration.
33211 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33213 (aarch64_expand_prologue): Sign return address before it's pushed onto
33215 (aarch64_expand_epilogue): Authenticate return address fetched from
33217 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33218 (aarch64_attributes): New function attributes for "sign-return-address".
33219 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33220 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33221 ("*do_return"): Generate combined instructions according to key index.
33222 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33223 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33225 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33226 * config/aarch64/aarch64.opt (msign-return-address=): New.
33227 * doc/extend.texi (AArch64 Function Attributes): Documents
33228 "sign-return-address=".
33229 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33231 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33233 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33234 overall option summary.
33236 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33238 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33239 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33240 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33241 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33243 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33245 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33246 -mpower9-minmax by default for -mcpu=power9.
33247 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33248 128-bit floating point.
33250 2017-01-20 Alan Modra <amodra@gmail.com>
33252 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33253 optimizing for size.
33255 2017-01-20 Alan Modra <amodra@gmail.com>
33258 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33259 for strcmp and strncmp from corresponding builtin decl.
33261 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33263 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33264 instead of i386/rtems-64.h.
33265 * config/i386/rtems-64.h: Remove.
33267 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33271 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33273 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33275 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33277 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33278 Change int to HOST_WIDE_INT.
33279 * config/aarch64/aarch64-protos.h
33280 (aarch64_simd_gen_const_vector_dup): Likewise.
33281 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33283 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33285 * langhooks-def.h (lhd_type_for_size): New decl.
33286 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33287 * langhooks.c (lhd_type_for_size): New function, taken from
33290 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33292 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33293 define_bypass for CR latency.
33294 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33295 (power9-alu2): Add define_bypass for CR latency.
33297 (power9-mul): Update insn latency.
33298 (power9-mul-compare): Update insn latency, bypass latency and remove
33301 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33303 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33305 * config/aarch64/aarch64.md
33306 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33307 aarch64_nopcrelative_literal_loads.
33308 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33310 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33312 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33313 TARGET_LOONGSON_3A.
33314 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33316 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33319 * config.gcc (supported_defaults): Add lxc1-sxc1.
33320 (with_lxc1_sxc1): Add validation.
33321 (all_defaults): Add lxc1-sxc1.
33322 * config/mips/mips.opt (mlxc1-sxc1): New option.
33323 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33325 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33326 __mips_no_lxc1_sxc1.
33327 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33328 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33329 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33331 2017-01-19 Richard Biener <rguenther@suse.de>
33333 PR tree-optimization/72488
33334 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33335 sure to restore SSA info.
33336 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33338 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33340 PR rtl-optimization/79121
33341 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33342 of the inner type when shifting an extended value.
33344 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33347 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33348 interposable aliases.
33350 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33353 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33354 Use the evmergelohi instruction.
33355 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33356 (mov_sitf_e500_subreg8_2_be): Likewise.
33357 (mov_sitf_e500_subreg12_2_le): Likewise.
33358 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33359 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33360 (mov_sitf_e500_subreg8_2_le): Likewise.
33361 (mov_sitf_e500_subreg12_2_be): Likewise.
33363 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33365 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33366 attribute from vecsimple to vecperm.
33367 (altivec_vbpermq2): Likewise.
33369 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33372 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33374 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33375 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33376 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33377 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33378 case where N arg is SIZE_MAX.
33379 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33380 (cmpstrsi): Add pattern.
33382 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33384 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33385 __builtin_vec_revb builtins.
33386 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33387 built-in functions to support generation of the ISA 3.0 XXBR<x>
33388 vector byte reverse instructions.
33389 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33390 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33391 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33392 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33393 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33394 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33395 (P9V_BUILTIN_VEC_REVB): Likewise.
33396 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33397 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33398 (p9_xxbrq_v16qi): Likewise.
33399 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33400 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33401 (p9_xxbrh_v8hi): Likewise.
33402 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33403 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33404 vec_revb built-in functions.
33406 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33408 PR rtl-optimization/78952
33409 * config/i386/i386.md (any_extract): New code iterator.
33410 (*insvqi_2): Use any_extract for source operand.
33411 (*insvqi_3): Use any_shiftrt for source operand.
33413 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33415 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33417 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33419 2017-01-18 Matthias Klose <doko@ubuntu.com>
33421 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33423 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33425 * config/rs6000/altivec.h (vec_bperm): Change #define.
33426 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33427 (altivec_vbpermq2): New define_insn.
33428 (altivec_vbpermd): Likewise.
33429 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33430 function interface.
33431 (VBPERMD): Likewise.
33432 (VBPERM): New polymorphic function interface.
33433 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33434 Add entries for P9V_BUILTIN_VEC_VBPERM.
33435 * doc/extend.texi: Add interfaces for vec_bperm.
33437 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33439 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33440 first letter of error messages.
33441 (s390_resolve_overloaded_builtin): Likewise.
33442 * config/s390/s390.c (s390_expand_builtin): Likewise.
33443 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33444 (s390_valid_target_attribute_inner_p): Likewise.
33445 * config/s390/s390.md ("tabort"): Likewise.
33447 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33449 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33450 (ISA_AVOID_DIV_HILO): New macro.
33451 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33452 (ISA_HAS_DDIV): Likewise.
33454 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33456 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33458 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33460 * doc/invoke.texi (fabi-version): Spelling fix.
33462 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33465 * doc/invoke.texi (fabi-version): Mention mangling fix for
33468 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33471 * doc/invoke.texi (fabi-version): Document discriminator mangling.
33473 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
33476 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
33477 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
33479 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
33481 (stack_protect_test): Ditto.
33482 * config/rs6000/rs6000.opt (mstack-protector-guard=,
33483 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
33485 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
33486 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
33487 -mstack-protector-guard-offset=.
33488 (RS/6000 and PowerPC Options): Ditto.
33490 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33492 * config/i386/i386.h (MASK_CLASS_P): New define.
33493 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
33494 there are no registers from different register sets also when
33495 mask registers are used. Update function comment.
33496 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
33497 to (*k/*r) and (*k/*km) alternatives.
33499 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
33501 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
33502 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
33503 (EH_RETURN_HANDLER_RTX): New define.
33504 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
33505 Force frame pointer in EH return functions.
33506 (aarch64_expand_epilogue): Add barrier for eh_return.
33507 (aarch64_final_eh_return_addr): Remove.
33508 (aarch64_eh_return_handler_rtx): New function.
33509 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
33511 (aarch64_eh_return_handler_rtx): New prototype.
33513 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33515 * config/rs6000/altivec.h (vec_rlmi): New #define.
33516 (vec_vrlnm): Likewise.
33517 (vec_rlnm): Likewise.
33518 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
33519 (UNSPEC_VRLNM): Likewise.
33520 (VIlong): New mode iterator.
33521 (altivec_vrl<VI_char>mi): New define_insn.
33522 (altivec_vrl<VI_char>nm): Likewise.
33523 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
33525 (VRLDNM): Likewise.
33526 (RLNM): New polymorphic function entry.
33527 (VRLWMI): New monomorphic function entry.
33528 (VRLDMI): Likewise.
33529 (RLMI): New polymorphic function entry.
33530 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
33531 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
33532 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
33535 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33538 * dwarf2out.c (field_byte_offset): Restore the
33539 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
33540 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
33541 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
33544 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33547 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
33549 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33552 * dwarf2out.c (add_data_member_location_attribute): For constant
33553 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
33554 instead of DW_AT_data_member_location, DW_AT_bit_offset and
33555 DW_AT_byte_size attributes.
33557 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33559 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
33560 after forcing to constant memory when the code model is medium.
33562 2017-01-17 Julia Koval <julia.koval@intel.com>
33565 * config/i386/avx512fintrin.h
33566 (_mm512_i32gather_ps): Change __addr type to void const*.
33567 (_mm512_mask_i32gather_ps): Ditto.
33568 (_mm512_i32gather_pd): Ditto.
33569 (_mm512_mask_i32gather_pd): Ditto.
33570 (_mm512_i64gather_ps): Ditto.
33571 (_mm512_mask_i64gather_ps): Ditto.
33572 (_mm512_i64gather_pd): Ditto.
33573 (_mm512_mask_i64gather_pd): Ditto.
33574 (_mm512_i32gather_epi32): Ditto.
33575 (_mm512_mask_i32gather_epi32): Ditto.
33576 (_mm512_i32gather_epi64): Ditto.
33577 (_mm512_mask_i32gather_epi64): Ditto.
33578 (_mm512_i64gather_epi32): Ditto.
33579 (_mm512_mask_i64gather_epi32): Ditto.
33580 (_mm512_i64gather_epi64): Ditto.
33581 (_mm512_mask_i64gather_epi64): Ditto.
33582 (_mm512_i32scatter_ps): Change __addr type to void*.
33583 (_mm512_mask_i32scatter_ps): Ditto.
33584 (_mm512_i32scatter_pd): Ditto.
33585 (_mm512_mask_i32scatter_pd): Ditto.
33586 (_mm512_i64scatter_ps): Ditto.
33587 (_mm512_mask_i64scatter_ps): Ditto.
33588 (_mm512_i64scatter_pd): Ditto.
33589 (_mm512_mask_i64scatter_pd): Ditto.
33590 (_mm512_i32scatter_epi32): Ditto.
33591 (_mm512_mask_i32scatter_epi32): Ditto.
33592 (_mm512_i32scatter_epi64): Ditto.
33593 (_mm512_mask_i32scatter_epi64): Ditto.
33594 (_mm512_i64scatter_epi32): Ditto.
33595 (_mm512_mask_i64scatter_epi32): Ditto.
33596 (_mm512_i64scatter_epi64): Ditto.
33597 (_mm512_mask_i64scatter_epi64): Ditto.
33598 * config/i386/avx512pfintrin.h
33599 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
33600 (_mm512_mask_prefetch_i32gather_ps): Ditto.
33601 (_mm512_mask_prefetch_i64gather_pd): Ditto.
33602 (_mm512_mask_prefetch_i64gather_ps): Ditto.
33603 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
33604 (_mm512_prefetch_i32scatter_ps): Ditto.
33605 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
33606 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
33607 (_mm512_prefetch_i64scatter_pd): Ditto.
33608 (_mm512_prefetch_i64scatter_ps): Ditto.
33609 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
33610 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
33611 * config/i386/avx512vlintrin.h
33612 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
33613 (_mm_mmask_i32gather_ps): Ditto.
33614 (_mm256_mmask_i32gather_pd): Ditto.
33615 (_mm_mmask_i32gather_pd): Ditto.
33616 (_mm256_mmask_i64gather_ps): Ditto.
33617 (_mm_mmask_i64gather_ps): Ditto.
33618 (_mm256_mmask_i64gather_pd): Ditto.
33619 (_mm_mmask_i64gather_pd): Ditto.
33620 (_mm256_mmask_i32gather_epi32): Ditto.
33621 (_mm_mmask_i32gather_epi32): Ditto.
33622 (_mm256_mmask_i32gather_epi64): Ditto.
33623 (_mm_mmask_i32gather_epi64): Ditto.
33624 (_mm256_mmask_i64gather_epi32): Ditto.
33625 (_mm_mmask_i64gather_epi32): Ditto.
33626 (_mm256_mmask_i64gather_epi64): Ditto.
33627 (_mm_mmask_i64gather_epi64): Ditto.
33628 (_mm256_i32scatter_ps): Change __addr type to void*.
33629 (_mm256_mask_i32scatter_ps): Ditto.
33630 (_mm_i32scatter_ps): Ditto.
33631 (_mm_mask_i32scatter_ps): Ditto.
33632 (_mm256_i32scatter_pd): Ditto.
33633 (_mm256_mask_i32scatter_pd): Ditto.
33634 (_mm_i32scatter_pd): Ditto.
33635 (_mm_mask_i32scatter_pd): Ditto.
33636 (_mm256_i64scatter_ps): Ditto.
33637 (_mm256_mask_i64scatter_ps): Ditto.
33638 (_mm_i64scatter_ps): Ditto.
33639 (_mm_mask_i64scatter_ps): Ditto.
33640 (_mm256_i64scatter_pd): Ditto.
33641 (_mm256_mask_i64scatter_pd): Ditto.
33642 (_mm_i64scatter_pd): Ditto.
33643 (_mm_mask_i64scatter_pd): Ditto.
33644 (_mm256_i32scatter_epi32): Ditto.
33645 (_mm256_mask_i32scatter_epi32): Ditto.
33646 (_mm_i32scatter_epi32): Ditto.
33647 (_mm_mask_i32scatter_epi32): Ditto.
33648 (_mm256_i32scatter_epi64): Ditto.
33649 (_mm256_mask_i32scatter_epi64): Ditto.
33650 (_mm_i32scatter_epi64): Ditto.
33651 (_mm_mask_i32scatter_epi64): Ditto.
33652 (_mm256_i64scatter_epi32): Ditto.
33653 (_mm256_mask_i64scatter_epi32): Ditto.
33654 (_mm_i64scatter_epi32): Ditto.
33655 (_mm_mask_i64scatter_epi32): Ditto.
33656 (_mm256_i64scatter_epi64): Ditto.
33657 (_mm256_mask_i64scatter_epi64): Ditto.
33658 (_mm_i64scatter_epi64): Ditto.
33659 (_mm_mask_i64scatter_epi64): Ditto.
33660 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
33661 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
33662 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
33663 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
33664 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
33665 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
33666 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
33667 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
33668 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
33669 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
33670 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
33671 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
33672 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
33673 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
33674 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
33675 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
33676 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
33677 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
33678 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
33679 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
33680 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
33681 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
33682 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
33683 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
33684 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
33685 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
33686 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
33687 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
33688 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
33689 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
33690 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
33691 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
33692 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
33693 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
33694 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
33695 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
33696 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
33697 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
33698 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
33699 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
33700 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
33701 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
33702 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
33703 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
33704 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
33705 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
33706 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
33707 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
33708 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
33709 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
33710 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
33711 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
33712 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
33713 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
33714 definitions accordingly.
33716 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
33717 Kuan-Lin Chen <kuanlinchentw@gmail.com>
33720 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
33723 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
33726 * ira-conflicts.c (ira_build_conflicts): Update total conflict
33727 hard regs for inner regno.
33729 2017-01-17 Martin Liska <mliska@suse.cz>
33732 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
33733 assumption and add comment.
33735 2017-01-17 Nathan Sidwell <nathan@acm.org>
33737 * ipa-visibility.c (localize_node): New function, broken out of ...
33738 (function_and_variable_visibility): ... here. Call it.
33740 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33742 PR middle-end/77445
33743 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
33744 correctly set frequency of oudgoing edge.
33745 (duplicate_thread_path): Fix profile updating.
33747 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33750 * configure.ac: Add GCC_BASE_VER.
33751 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
33752 version from BASE-VER file.
33753 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
33754 (gcc.o): Depend on $(BASEVER).
33755 * common.opt (dumpfullversion): New option.
33756 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
33757 * doc/invoke.texi: Document -dumpfullversion.
33758 * doc/install.texi: Document --with-gcc-major-version-only.
33759 * configure: Regenerated.
33761 2017-01-17 Richard Biener <rguenther@suse.de>
33763 PR tree-optimization/71433
33764 * tree-vrp.c (register_new_assert_for): Merge same asserts
33765 on all incoming edges.
33766 (process_assert_insertions_for): Handle insertions at the
33769 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
33771 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
33772 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
33774 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
33777 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
33780 2017-01-17 Alan Modra <amodra@gmail.com>
33783 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
33784 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
33785 symbolic stack limit when pic.
33787 2017-01-16 Martin Sebor <msebor@redhat.com>
33789 PR tree-optimization/78608
33790 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
33792 2017-01-16 Jeff Law <law@redhat.com>
33795 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
33796 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
33797 for several include directories that may be relative to sysroot.
33798 * config/i386/x-mingw32 (gplus_includedir): Define.
33799 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
33800 (native_system_includedir): Likewise.
33801 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
33802 override if TARGET_SYSTEM_ROOT is defined.
33803 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
33805 PR tree-optimization/79090
33806 PR tree-optimization/33562
33807 PR tree-optimization/61912
33808 PR tree-optimization/77485
33809 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
33810 and computed trims into the dump file.
33812 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33814 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
33816 2017-01-16 Jakub Jelinek <jakub@redhat.com>
33819 * gimplify.c (gimplify_init_constructor): If want_value and
33820 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
33824 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
33825 sequence. Formatting fixes.
33826 (doloop_optimize): Formatting fixes.
33829 * gcc.c (debug_level_greater_than_spec_func): New function.
33830 (static_spec_functions): Add debug-level-gt spec function.
33831 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
33833 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33834 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33835 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
33836 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
33837 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
33838 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
33840 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
33842 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
33843 QImode fixups to general and mask registers only.
33845 2017-01-16 Carl Love <cel@us.ibm.com>
33847 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
33848 for built-in functions
33849 vector signed char vec_nabs (vector signed char)
33850 vector signed short vec_nabs (vector signed short)
33851 vector signed int vec_nabs (vector signed int)
33852 vector signed long long vec_nabs (vector signed long long)
33853 vector float vec_nabs (vector float)
33854 vector double vec_nabs (vector double)
33855 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
33857 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
33858 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
33859 * doc/extend.texi: Update the documentation file for the new built-in
33862 2017-01-16 Martin Sebor <msebor@redhat.com>
33864 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
33867 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33869 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
33870 UNSPEC_VSX__XXSPLTD to require special splat handling.
33872 2017-01-16 David Malcolm <dmalcolm@redhat.com>
33875 * system.h: Poison strndup.
33877 2017-01-16 Alan Modra <amodra@gmail.com>
33880 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
33883 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
33885 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
33887 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
33889 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
33890 call recog here. Assert that INSN_CODE (insn) is non-negative.
33892 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33895 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
33897 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
33898 in the currently scheduled RTL fragment.
33900 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33902 PR rtl-optimization/78751
33903 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
33906 2017-01-14 Jeff Law <law@redhat.com>
33908 PR tree-optimization/79090
33909 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
33910 variable length stores.
33911 (compute_trims): Delete dead assignment to *trim_tail.
33912 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
33915 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
33917 PR rtl-optimization/78626
33918 PR rtl-optimization/78727
33919 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
33920 of a block, and split such blocks after everything else is finished.
33922 2017-01-14 Alan Modra <amodra@gmail.com>
33925 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
33926 target legitimate_combined_insn.
33927 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
33928 (rs6000_legitimate_combined_insn): New function.
33929 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
33931 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
33932 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
33933 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
33935 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
33937 * doc/frontends.texi (G++ and GCC): Remove references to Java.
33939 2017-01-13 Jeff Law <law@redhat.com>
33941 PR tree-optimization/33562
33942 PR tree-optimization/61912
33943 PR tree-optimization/77485
33944 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
33946 (delete_dead_assignment): Likewise.
33947 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
33948 statement to delete_dead_call and delete_dead_assignment.
33950 2017-01-13 David Malcolm <dmalcolm@redhat.com>
33953 * substring-locations.c (format_warning_va): Strengthen case 1 so
33954 that both endpoints of the substring must be within the format
33955 range for just the substring to be printed.
33957 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
33959 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
33960 * config/i386/i386.c (ix86_target_string): Add missing options
33961 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
33962 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
33963 flags_other and ix86_target_other to flags2_other. Display unknown
33965 (ix86_valid_target_attribute_inner_p): Add missing options and
33966 reorder options by implied ISAs, as in ix86_target_string.
33968 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33970 * hash-table.h (hash_table::too_empty_p): New function.
33971 (hash_table::expand): Use it.
33972 (hash_table::traverse): Likewise.
33973 (hash_table::empty_slot): Use sizeof (value_type) instead of
33974 sizeof (PTR) to convert bytes to elements. Shrink the table
33975 if the current size is excessive for the current number of
33978 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33980 * ira-costs.c (record_reg_classes): Break from the inner loop
33981 early once alt_fail is known to be true. Update outer loop
33982 handling accordingly.
33984 2017-01-13 Jeff Law <law@redhat.com>
33986 * tree-ssa-dse.c (decrement_count): New function.
33987 (increment_start_addr, maybe_trim_memstar_call): Likewise.
33988 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
33989 when we know the partially dead statement is a mem* function.
33991 PR tree-optimization/61912
33992 PR tree-optimization/77485
33993 * tree-ssa-dse.c: Include expr.h.
33994 (maybe_trim_constructor_store): New function.
33995 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
33997 PR tree-optimization/33562
33998 PR tree-optimization/61912
33999 PR tree-optimization/77485
34000 * doc/invoke.texi: Document new dse-max-object-size param.
34001 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
34002 * tree-ssa-dse.c: Include params.h.
34003 (dse_store_status): New enum.
34004 (initialize_ao_ref_for_dse): New, partially extracted from
34006 (valid_ao_ref_for_dse, normalize_ref): New.
34007 (setup_live_bytes_from_ref, compute_trims): Likewise.
34008 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
34009 (maybe_trim_partially_dead_store): Likewise.
34010 (maybe_trim_complex_store): Likewise.
34011 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
34012 Track what bytes live from the original store. Return tri-state
34013 for dead, partially dead or live.
34014 (dse_dom_walker): Add constructor, destructor and new private members.
34015 (delete_dead_call, delete_dead_assignment): New extracted from
34017 (dse_optimize_stmt): Make a member of dse_dom_walker.
34018 Use initialize_ao_ref_for_dse.
34020 PR tree-optimization/33562
34021 PR tree-optimization/61912
34022 PR tree-optimization/77485
34023 * sbitmap.h (bitmap_count_bits): Prototype.
34024 (bitmap_clear_range, bitmap_set_range): Likewise.
34025 * sbitmap.c (bitmap_clear_range): New function.
34026 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
34028 2017-01-13 Martin Liska <mliska@suse.cz>
34031 * function.c (set_cfun): Add new argument force.
34032 * function.h (set_cfun): Likewise.
34033 * ipa-inline-transform.c (inline_call): Use the function when
34034 strict alising from is dropped for function we inline to.
34036 2017-01-13 Richard Biener <rguenther@suse.de>
34038 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
34039 for dumping GIMPLE INTEGER_CSTs.
34041 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
34043 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
34044 to 201112L since C++17.
34046 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
34049 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
34050 if -fsanitize=kernel-address is present.
34052 2017-01-13 Richard Biener <rguenther@suse.de>
34054 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
34055 as _Literal ( type ) number in case usual suffixes do not
34056 preserve all information.
34058 2017-01-13 Richard Biener <rguenther@suse.de>
34060 PR tree-optimization/77283
34061 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
34062 and ssa-iterators.h.
34063 (is_feasible_trace): Implement a cost model based on joiner
34066 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
34069 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
34070 char or short to __float128/_Float128 directly.
34072 2017-01-12 Martin Sebor <msebor@redhat.com>
34074 to -Wformat-overflow.
34075 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
34076 (min_bytes_remaining): Same.
34077 (get_string_length): Same.
34078 (format_string): Same.
34079 (format_directive): Same.
34081 (pass_sprintf_length::handle_gimple_call): Same.
34083 2017-01-12 Jakub Jelinek <jakub@redhat.com>
34085 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
34086 info.nowrite calls with no lhs that can't throw. Return bool
34087 whether gsi_remove has been called or not.
34088 (pass_sprintf_length::handle_gimple_call): Return bool whether
34089 try_substitute_return_value called gsi_remove. Formatting fix.
34090 (pass_sprintf_length::execute): Don't use gsi_remove if
34091 handle_gimple_call returned true.
34094 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
34095 be removed due to side-effects, don't remove following barrier nor
34096 turn the successor edge into fallthru edge.
34098 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34101 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
34102 element-reversing loads and stores as not swappable.
34104 2017-01-12 Nathan Sidwell <nathan@acm.org>
34105 Nicolai Stange <nicstange@gmail.com>
34107 * combine.c (try_combine): Don't ignore result of overlap checking
34108 loop. Combine overlap & asm check into single loop.
34110 2017-01-12 Richard Biener <rguenther@suse.de>
34112 * tree-pretty-print.c (dump_generic_node): Provide -gimple
34113 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
34115 2017-01-12 Richard Biener <rguenther@suse.de>
34117 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
34118 and TS_TARGET_OPTION directly derive from TS_BASE.
34119 * tree-core.h (tree_optimization_option): Derive from tree_base.
34120 (tree_target_option): Likewise.
34122 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
34124 * config/i386/i386.c (memory_address_length): Increase len
34125 only when rip_relative_addr_p returns false.
34127 2017-01-11 Julia Koval <julia.koval@intel.com>
34129 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
34130 (OPTION_MASK_ISA_SGX_SET): New.
34131 (ix86_handle_option): Handle OPT_msgx.
34132 * config.gcc: Added sgxintrin.h.
34133 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
34134 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
34135 * config/i386/i386.c (ix86_target_string): Add -msgx.
34137 (ix86_option_override_internal): Handle new options.
34138 (ix86_valid_target_attribute_inner_p): Add sgx.
34139 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
34140 * config/i386/i386.opt: Add msgx.
34141 * config/i386/sgxintrin.h: New file.
34142 * config/i386/x86intrin.h: Add sgxintrin.h.
34144 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34147 * fold-const.c (maybe_nonzero_address): Return 1 for function
34149 (tree_single_nonzero_warnv_p): Don't handle function local objects
34153 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
34156 2017-01-11 David Malcolm <dmalcolm@redhat.com>
34159 * opts.c: Include "spellcheck.h"
34160 (struct string_fragment): New struct.
34161 (struct edit_distance_traits<const string_fragment &>): New
34163 (get_closest_sanitizer_option): New function.
34164 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
34166 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34168 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
34170 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
34171 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
34172 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
34173 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
34174 for initial die_offset if dwarf_split_debug_info.
34175 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
34176 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
34178 (output_skeleton_debug_sections): Formatting fix. Use
34179 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34180 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34182 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34184 * config/arm/cortex-a53.md: Add bypasses for
34185 cortex_a53_r2f_cvt.
34186 (cortex_a53_r2f): Only use for transfers.
34187 (cortex_a53_f2r): Likewise.
34188 (cortex_a53_r2f_cvt): Add reservation for conversions.
34189 (cortex_a53_f2r_cvt): Likewise.
34191 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34193 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34194 to all inlined functions, change static to extern.
34196 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34199 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34201 (arm_assemble_integer): Likewise.
34203 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34205 * config.gcc: Use new awk script to check CPU, FPU and architecture
34206 parameters for --with-... options.
34207 * config/arm/parsecpu.awk: New file
34208 * config/arm/arm-cpus.in: New file.
34209 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34211 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34213 * config/arm/t-arm: Update dependency rules.
34214 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34215 of processing .def files.
34216 * config/arm/genopt.sh: Deleted.
34217 * config/arm/gentune.sh: Deleted.
34218 * config/arm/arm-cores.def: Deleted.
34219 * config/arm/arm-arches.def: Deleted.
34220 * config/arm/arm-fpus.def: Deleted.
34221 * config/arm/arm-tune.md: Regenerated.
34222 * config/arm/arm-tables.opt: Regenerated.
34223 * config/arm/arm-cpu.h: New generated file.
34224 * config/arm/arm-cpu-data.h: New generated file.
34225 * config/arm/arm-cpu-cdata.h: New generated file.
34227 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34230 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34232 (input_varpool_node): Unpack dynamically_initialized bit.
34234 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34236 PR rtl-optimization/79032
34237 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34238 the alignment of the adjusted memory reference against that of MODE,
34239 instead of the alignment of the original memory reference.
34241 2017-01-11 Martin Jambor <mjambor@suse.cz>
34243 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34245 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34246 decorated functions.
34248 2017-01-11 Richard Biener <rguenther@suse.de>
34250 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34251 set range/nonnull info for PHI results. Do not set it on
34252 stmts marked for removal.
34254 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34256 * expr.c (store_field): In the bitfield case, fetch the return value
34257 from the registers before applying a single big-endian adjustment.
34258 Always do a final load for a BLKmode value not larger than a word.
34260 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34263 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34264 that we correctly handle column numbers greater than
34265 LINE_MAP_MAX_COLUMN_NUMBER.
34267 2017-01-10 Martin Sebor <msebor@redhat.com>
34269 PR middle-end/78245
34270 * gimple-ssa-sprintf.c (get_destination_size): Call
34271 {init,fini}object_sizes.
34272 * tree-object-size.c (addr_object_size): Adjust.
34273 (pass_through_call): Adjust.
34274 (pass_object_sizes::execute): Adjust.
34275 * tree-object-size.h (fini_object_sizes): Declare.
34277 2017-01-10 Martin Sebor <msebor@redhat.com>
34279 PR tree-optimization/78775
34280 * builtins.c (get_size_range): Move...
34281 * calls.c: ...to here.
34282 (alloc_max_size): Accept zero argument.
34283 (operand_signed_p): Remove.
34284 (maybe_warn_alloc_args_overflow): Call get_size_range.
34285 * calls.h (get_size_range): Declare.
34287 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34289 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34290 from TI's devices.csv file as of September 2016.
34291 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34293 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34295 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34296 * doc/invoke.texi: Likewise.
34297 * doc/md.texi: Likewise.
34298 * doc/objc.texi: Likewise.
34300 2017-01-10 Joshua Conner <joshconner@google.com>
34302 * config/arm/fuchsia-elf.h: New file.
34303 * config/fuchsia.h: New file.
34304 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34305 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34307 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34309 2016-01-10 Richard Biener <rguenther@suse.de>
34311 PR tree-optimization/79034
34312 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34313 Propagate out degenerate PHIs in the joiner.
34315 2017-01-10 Martin Liska <mliska@suse.cz>
34317 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34318 (sort_congruence_classes_by_decl_uid): Likewise.
34319 (sort_congruence_class_groups_by_decl_uid): Likewise.
34320 (sem_item_optimizer::merge_classes): Sort class, groups in these
34321 classes and members in the groups by DECL_UID of declarations.
34322 This would make merge operations stable.
34324 2017-01-10 Martin Liska <mliska@suse.cz>
34326 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34327 usage of m_classes_vec.
34328 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34329 (sem_item_optimizer::get_group_by_hash): Likewise.
34330 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34331 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34332 (sem_item_optimizer::verify_classes): Likewise.
34333 (sem_item_optimizer::process_cong_reduction): Likewise.
34334 (sem_item_optimizer::dump_cong_classes): Likewise.
34335 (sem_item_optimizer::merge_classes): Likewise.
34336 * ipa-icf.h (congruence_class_hash): Rename from
34337 congruence_class_group_hash. Remove declaration of m_classes_vec.
34339 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34341 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34342 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34343 * config.gcc: Add avx512vpopcntdqintrin.h.
34344 * config/i386/avx512vpopcntdqintrin.h: New.
34345 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34346 * config/i386/i386-builtin-types.def: Add new types.
34347 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34348 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34349 __builtin_ia32_vpopcountq_v8di_mask): New.
34350 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34351 __AVX512VPOPCNTDQ__.
34352 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34353 (PTA_AVX512VPOPCNTDQ): Define.
34354 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34355 TARGET_AVX512VPOPCNTDQ_P): Define.
34356 * config/i386/i386.opt: Add mavx512vpopcntdq.
34357 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34358 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34360 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34362 PR middle-end/77484
34363 * predict.def (PRED_CALL): Set to 67.
34365 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34367 * expr.c (store_field): In the bitfield case, if the value comes from
34368 a function call and is of an aggregate type returned in registers, do
34369 not modify the field mode; extract the value in all cases if the mode
34370 is BLKmode and the size is not larger than a word.
34372 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34375 * config/i386/cpuid.h: Fix undefined behavior.
34377 2017-01-04 Jeff Law <law@redhat.com>
34379 PR tree-optimization/79007
34380 PR tree-optimization/67955
34381 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34382 conservative for pt.null when flag_non_call_exceptions is on.
34384 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34386 PR translation/79019
34387 PR translation/79020
34388 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34389 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34390 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34392 * config/avr/avr.opt (maccumulate-args): Likewise.
34393 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34394 * common.opt (freport-bug): Likewise.
34395 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34396 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34397 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34398 translatable string.
34399 * config/i386/i386.c (function_value_32): Likewise.
34400 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34401 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34403 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34404 * common/config/msp430/msp430-common.c (msp430_handle_option):
34406 * symtab.c (symtab_node::verify_base): Likewise.
34407 * opts.c (set_debug_level): Likewise.
34408 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34409 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34410 missing whitespace to translatable strings.
34411 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34412 * config/sh/superh.h: Likewise.
34413 * config/i386/xopintrin.h: Likewise.
34414 * config/i386/znver1.md: Likewise.
34415 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34416 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34417 * double-int.h (struct double_int): Likewise.
34418 * double-int.c (div_and_round_double): Likewise.
34419 * wide-int.cc: Likewise.
34420 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34421 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34422 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34423 (crossjumps_occurred): ... this.
34424 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34427 PR tree-optimization/78899
34428 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34429 returning bool return struct loop *, NULL for failure and the new
34431 (versionable_outer_loop_p): Don't version outer loop if it has
34432 dont_vectorized bit set.
34433 (tree_if_conversion): When versioning outer loop, ensure
34434 tree_if_conversion is performed also on the inner loop of the
34435 non-vectorizable outer loop copy.
34436 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34437 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34438 prevent vectorization of it.
34439 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34440 the outer loop vectorization of the non-scalar version is attempted
34441 before vectorization of the inner loop in scalar version. If
34442 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34443 vectorization of its inner loop.
34444 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34445 has 2 inner loops, rename also on edges from bb whose single pred
34446 is outer_loop->header. Fix typo in function comment.
34448 2017-01-09 Martin Sebor <msebor@redhat.com>
34451 * asan.c (asan_emit_stack_protection): Increase local buffer size
34452 to avoid snprintf truncation warning.
34454 2017-01-09 Andrew Pinski <apinski@cavium.com>
34456 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34457 to reference thunderx2t99 for the tuning structure
34458 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34460 (thunderx2t99_extra_costs): This.
34461 * config/aarch64/aarch64-tune.md: Regenerate.
34462 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34463 (vulcan_addrcost_table): This.
34464 (vulcan_regmove_cost): Rename to ...
34465 (thunderx2t99_regmove_cost): This.
34466 (vulcan_vector_cost): Rename to ...
34467 (thunderx2t99_vector_cost): this.
34468 (vulcan_branch_cost): Rename to ...
34469 (thunderx2t99_branch_cost): This.
34470 (vulcan_tunings): Rename to ...
34471 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
34472 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
34474 2017-01-09 Martin Jambor <mjambor@suse.cz>
34478 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
34479 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
34480 (propagate_vr_accross_jump_function): Use the above function for all
34481 value range computations for pass-through jump functions and type
34482 converasion from explicit value range values.
34483 (ipcp_propagate_stage): Do not attempt to deduce types of formal
34484 parameters from TYPE_ARG_TYPES.
34485 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
34486 (ipa_write_node_info): Stream type of the actual argument.
34487 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
34489 2017-01-09 Martin Liska <mliska@suse.cz>
34492 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
34493 (lookup_compiler): Do not show error message with have_E.
34495 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34497 PR tree-optimization/78938
34498 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
34499 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
34500 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
34501 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
34504 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34506 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
34509 2017-01-09 Richard Biener <rguenther@suse.de>
34511 PR tree-optimization/78997
34512 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
34513 name condition properly.
34515 2017-01-09 Richard Biener <rguenther@suse.de>
34518 * dwarf2out.c (is_cxx): New overload with context.
34519 (is_naming_typedef_decl): Use it.
34521 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34523 * invoke.texi (Option Summary): Correct spacing in option lists
34524 and add line breaks to fix over-long lines.
34526 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34528 PR middle-end/17660
34530 * extend.texi (Common Variable Attributes): Add xref to GCC
34531 Internals manual to explain mode attribute keywords.
34533 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34536 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
34537 and Preprocessor Options.
34538 (Options for Linking): Document -pthread here....
34539 (RS/6000 and PowerPC Options): ...not here.
34540 (Solaris 2 Options): ...or here.
34541 * doc/cppopts.texi: Document -pthread.
34543 2017-01-08 Martin Sebor <msebor@redhat.com>
34545 PR middle-end/77708
34546 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
34547 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
34548 New member functions.
34549 (format_directive): Used them.
34551 (pass_sprintf_length::handle_gimple_call): Same.
34552 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
34553 to avoid truncation for any argument.
34554 (extract_affine_mul): Same.
34555 * tree.c (get_file_function_name): Same.
34557 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34559 PR middle-end/77484
34560 * predict.def (PRED_INDIR_CALL): Set to 86.
34562 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34564 PR preprocessor/54124
34565 * doc/cppopts.texi: Reformat -d subtable to list the full name
34566 of the options. Add cross-reference to the docs for the general
34567 compiler -d options.
34568 * doc/invoke.texi (Developer Options): Add cross-reference to the
34569 preprocessor-specific -d option documentation.
34571 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34573 PR preprocessor/13498
34574 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
34575 redudant material, and reflect new command-line options.
34576 (System Headers): Likewise.
34578 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34580 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
34581 -isystem, and -idirafter. Copy-edit.
34582 * doc/cppopts.texi: Copy-edit. Remove contradiction about
34583 default for -ftrack-macro-expansion. Delete obsolete and
34584 badly-formatted implementation details about -fdebug-cpp output.
34585 * doc/cppwarnopts.texi: Copy-edit.
34587 2017-01-07 David Malcolm <dmalcolm@redhat.com>
34590 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34591 that the transition from a max line width >= 1<<10 to narrower
34592 lines works correctly.
34594 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
34596 * doc/options.texi (PerFunction): New.
34597 * opt-functions.awk (switch_flags): Map both Optimization and
34598 PerFunction to CL_OPTIMIZATION.
34599 * opth-gen.awk: Test for PerFunction flag along with
34601 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
34602 it only when the latter is present. Skip those that don't in
34603 the hash function generator.
34604 * common.opt (fvar-tracking): Mark as PerFunction instead of
34606 (fvar-tracking-assignments): Likewise.
34607 (fvar-tracking-assignments-toggle): Likewise.
34608 (fvar-tracking-uninit): Likewise.
34610 2017-01-07 Jakub Jelinek <jakub@redhat.com>
34612 PR translation/79018
34613 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
34616 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
34619 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
34620 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
34621 TARGET_LONG_JUMP_TABLE_OFFSETS.
34622 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
34623 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
34624 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
34625 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
34626 * config/m68k/m68k.md (tablejump expander): Likewise.
34627 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
34628 TARGET_LONG_JUMP_TABLE_OFFSETS.
34629 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
34630 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
34632 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
34633 David Holsgrove <david.holsgrove@xilinx.com>
34635 * common/config/microblaze/microblaze-common.c
34636 (TARGET_EXCEPT_UNWIND_INFO): Remove.
34637 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
34639 * config/microblaze/microblaze.c (microblaze_must_save_register)
34640 (microblaze_expand_epilogue, microblaze_return_addr): Handle
34642 (microblaze_eh_return): New function.
34643 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
34644 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
34645 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
34646 * config/microblaze/microblaze.md (eh_return): New pattern.
34648 2017-01-06 Jakub Jelinek <jakub@redhat.com>
34650 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
34651 GCC_DIAGNOSTIC_STRINGIFY): Define.
34653 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
34655 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34657 * config/arm/arm.md (<mcrr>): New.
34659 * config/arm/arm.c (arm_arch5te): New.
34660 (arm_option_override): Set arm_arch5te.
34661 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
34663 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
34664 (arm_mcrr_qualifiers): ... this. New.
34665 (MRRC_QUALIFIERS): Define to...
34666 (arm_mrrc_qualifiers): ... this. New.
34667 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
34669 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
34670 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
34671 (MRRCI, mrrc, MRRC): New.
34672 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
34673 VUNSPEC_MRRC2): New.
34675 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34677 * config/arm/arm.md (<mcr>): New.
34679 * config/arm/arm.c (arm_coproc_builtin_available): Add
34680 support for mcr, mrc, mcr2 and mrc2.
34681 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
34682 (arm_mcr_qualifiers): ... this. New.
34683 (MRC_QUALIFIERS): Define to ...
34684 (arm_mrc_qualifiers): ... this. New.
34685 (MCR_QUALIFIERS): Define to ...
34686 (arm_mcr_qualifiers): ... this. New.
34687 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
34689 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
34690 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
34691 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
34692 VUNSPEC_MRC2): New.
34694 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34696 * config/arm/arm.md (*ldc): New.
34700 * config/arm/arm.c (arm_coproc_builtin_available): Add
34701 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
34702 (arm_coproc_ldc_stc_legitimate_address): New.
34703 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
34704 'qualifier_const_pointer'.
34705 (LDC_QUALIFIERS): Define to...
34706 (arm_ldc_qualifiers): ... this. New.
34707 (STC_QUALIFIERS): Define to...
34708 (arm_stc_qualifiers): ... this. New.
34709 * config/arm/arm-protos.h
34710 (arm_coproc_ldc_stc_legitimate_address): New.
34711 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
34712 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
34713 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
34714 stc2, stcl, stc2l): New.
34715 * config/arm/constraints.md (Uz): New.
34716 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
34717 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
34718 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
34719 VUNSPEC_STC2L): New.
34721 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34723 * config/arm/arm.md (<cdp>): New.
34724 * config/arm/arm.c (neon_const_bounds): Rename this ...
34725 (arm_const_bounds): ... this.
34726 (arm_coproc_builtin_available): New.
34727 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
34728 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
34729 (CDP_QUALIFIERS): Define to...
34730 (arm_cdp_qualifiers): ... this. New.
34732 (arm_expand_builtin_args): Add case for 6 arguments.
34733 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
34734 (arm_const_bounds): ... this.
34735 (arm_coproc_builtin_available): New.
34736 * config/arm/arm_acle.h (__arm_cdp): New.
34738 * config/arm/arm_acle_builtins.def (cdp): New.
34740 * config/arm/iterators.md (CDPI,CDP,cdp): New.
34741 * config/arm/neon.md: Rename all 'neon_const_bounds' to
34742 'arm_const_bounds'.
34743 * config/arm/types.md (coproc): New.
34744 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
34745 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
34746 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
34747 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
34749 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34751 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
34752 (UBINOP_QUALIFIERS): New.
34754 (acle_builtin_data): New. Change comment.
34755 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
34756 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
34757 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
34758 arm_acle_builtins.def.
34759 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
34760 (arm_init_acle_builtins): New.
34761 (CRC32_BUILTIN): Remove.
34762 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
34763 crc32cb, crc32ch and crc32cw.
34764 (arm_init_crc32_builtins): Remove.
34765 (arm_init_builtins): Use arm_init_acle_builtins rather
34766 than arm_init_crc32_builtins.
34767 (arm_expand_acle_builtin): New.
34768 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
34769 * config/arm/arm_acle_builtins.def: New.
34771 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34773 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
34774 (arm_builtin_datum): ... this.
34775 (arm_init_neon_builtin): Rename to ...
34776 (arm_init_builtin): ... this. Add a new parameters PREFIX
34777 and USE_SIG_IN_NAME.
34778 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
34779 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
34780 'arm_builtin_datum'.
34781 (arm_init_vfp_builtins): Likewise.
34782 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
34783 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
34784 (arm_expand_neon_args): Rename to ...
34785 (arm_expand_builtin_args): ... this. Rename builtin_arg
34786 enum values and differentiate between ARG_BUILTIN_MEMORY
34787 and ARG_BUILTIN_NEON_MEMORY.
34788 (arm_expand_neon_builtin_1): Rename to ...
34789 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
34790 values, arm_expand_builtin_args and add bool parameter NEON.
34791 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
34792 (arm_expand_vfp_builtin): Likewise.
34793 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
34795 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34797 PR middle-end/77484
34798 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
34799 * predict.c (tree_estimate_probability_bb): Reverse direction of
34800 polymorphic call predictor.
34802 2017-01-06 David Malcolm <dmalcolm@redhat.com>
34804 * passes.c (execute_one_pass): Split out pass-skipping logic into...
34805 (determine_pass_name_match): ...this new function and...
34806 (should_skip_pass_p): ...this new function.
34808 2017-01-06 Nathan Sidwell <nathan@acm.org>
34810 * ipa-visibility.c (function_and_variable_visibility): Reformat
34811 comments and long lines. Remove extrneous if.
34812 * symtab.c (symtab_node::make_decl_local): Fix code format.
34813 (symtab_node::set_section_for_node): Fix comment typo.
34815 2017-01-06 Martin Liska <mliska@suse.cz>
34818 * lra-constraints.c: Rename invariant to lra_invariant.
34819 * predict.c (set_even_probabilities): Initialize e to NULL.
34821 2017-01-05 Martin Sebor <msebor@redhat.com>
34823 PR tree-optimization/78910
34824 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
34825 (format_integer): Correct off-by-one error in the handling
34826 of precision with negative numbers in signed conversions..
34828 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
34830 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
34832 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34834 PR tree-optimization/71016
34835 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
34836 factor_out_conditional_conversion. Formatting fix.
34837 (factor_out_conditional_conversion): Add cond_stmt argument.
34838 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
34839 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
34842 2017-01-05 David Malcolm <dmalcolm@redhat.com>
34844 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
34845 read-rtl-function.o, and selftest-rtl.o.
34846 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
34847 (selftest::aarch64_test_loading_full_dump): New function.
34848 (selftest::aarch64_run_selftests): New function.
34849 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
34850 selftest::aarch64_run_selftests.
34851 * config/i386/i386.c
34852 (selftest::ix86_test_loading_dump_fragment_1): New function.
34853 (selftest::ix86_test_loading_call_insn): New function.
34854 (selftest::ix86_test_loading_full_dump): New function.
34855 (selftest::ix86_test_loading_unspec): New function.
34856 (selftest::ix86_run_selftests): Call the new functions.
34857 * emit-rtl.c (maybe_set_max_label_num): New function.
34858 * emit-rtl.h (maybe_set_max_label_num): New decl.
34859 * function.c (instantiate_decls): Guard call to
34860 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
34861 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
34863 * gensupport.c (gen_reader::gen_reader): Pass "false"
34864 for new "compact" param of rtx_reader.
34865 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
34866 rather than an empty string for NULL strings.
34867 * read-md.c: Potentially include config.h rather than bconfig.h.
34868 Wrap include of errors.h with #ifdef GENERATOR_FILE.
34869 (have_error): New global, copied from errors.c.
34870 (md_reader::read_name): Rename to...
34871 (md_reader::read_name_1): ...this, adding "out_loc" param,
34872 and converting "missing name or number" to returning false, rather
34874 (md_reader::read_name): Reimplement in terms of read_name_1.
34875 (md_reader::read_name_or_nil): New function.
34876 (md_reader::read_string): Handle "(nil)" by returning NULL.
34877 (md_reader::md_reader): Add new param "compact".
34878 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
34879 (md_reader::read_file): New method.
34880 * read-md.h (md_reader::md_reader): Add new param "compact".
34881 (md_reader::read_file): New method.
34882 (md_reader::is_compact): New accessor.
34883 (md_reader::read_name): Convert return type from void to file_location.
34884 (md_reader::read_name_or_nil): New decl.
34885 (md_reader::read_name_1): New decl.
34886 (md_reader::m_compact): New field.
34887 (noop_reader::noop_reader): Pass "false" for new "compact" param
34889 (rtx_reader::rtx_reader): Add new "compact" param.
34890 (rtx_reader::read_rtx_operand): Make virtual and convert return
34891 type from void to rtx.
34892 (rtx_reader::read_until): New decl.
34893 (rtx_reader::handle_any_trailing_information): New virtual function.
34894 (rtx_reader::postprocess): New virtual function.
34895 (rtx_reader::finalize_string): New virtual function.
34896 (rtx_reader::m_in_call_function_usage): New field.
34897 (rtx_reader::m_reuse_rtx_by_id): New field.
34898 * read-rtl-function.c: New file.
34899 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
34900 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
34901 (selftest::verify_three_block_rtl_cfg): New decl.
34902 * read-rtl-function.h: New file.
34903 * read-rtl.c: Potentially include config.h rather than bconfig.h.
34904 For host, include function.h, memmodel.h, and emit-rtl.h.
34905 (one_time_initialization): New function.
34906 (struct compact_insn_name): New struct.
34907 (compact_insn_names): New array.
34908 (find_code): Handle insn codes in compact dumps.
34909 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
34910 (bind_subst_iter_and_attr): Likewise.
34911 (add_condition_to_string): Likewise.
34912 (add_condition_to_rtx): Likewise.
34913 (apply_attribute_uses): Likewise.
34914 (add_current_iterators): Likewise.
34915 (apply_iterators): Likewise.
34916 (initialize_iterators): Guard usage of apply_subst_iterator with
34917 #ifdef GENERATOR_FILE.
34918 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
34919 (md_reader::read_mapping): Likewise.
34920 (add_define_attr_for_define_subst): Likewise.
34921 (add_define_subst_attr): Likewise.
34922 (read_subst_mapping): Likewise.
34923 (check_code_iterator): Likewise.
34924 (rtx_reader::read_rtx): Likewise. Move one-time initialization
34926 (one_time_initialization): New function.
34927 (rtx_reader::read_until): New method.
34928 (read_flags): New function.
34929 (parse_reg_note_name): New function.
34930 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
34931 Handle reuse_rtx ids.
34932 Wrap iterator lookup within #ifdef GENERATOR_FILE.
34933 Add parsing support for RTL dumps, mirroring the special-cases in
34934 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
34935 values, and calling handle_any_trailing_information.
34936 (rtx_reader::read_rtx_operand): Convert return type from void
34937 to rtx, returning return_rtx. Handle case 'e'. Call
34938 finalize_string on XSTR and XTMPL fields.
34939 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
34940 "(nil)" values were omitted. Call the postprocess vfunc on the
34942 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
34943 class ctor. Initialize m_in_call_function_usage. Call
34944 one_time_initialization.
34945 * rtl-tests.c (selftest::test_uncond_jump): Call
34946 set_new_first_and_last_insn.
34947 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
34948 * selftest-rtl.c: New file.
34949 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
34950 (selftest::get_insn_by_uid): New decl.
34951 * selftest-run-tests.c (selftest::run_tests): Call
34952 read_rtl_function_c_tests.
34953 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
34954 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
34957 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
34959 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
34960 operands in a special way. Assert that pos+len <= mode precision.
34962 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34964 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
34965 3 argument Alias with unlimited for the negative form.
34966 (fno-vect-cost-model): Removed.
34968 2017-01-05 Martin Liska <mliska@suse.cz>
34970 * hsa-gen.c (gen_hsa_divmod): New function.
34971 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
34973 2017-01-05 Martin Liska <mliska@suse.cz>
34976 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
34979 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34981 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
34982 small constant length operands.
34984 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34986 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
34987 between loop iterations.
34989 2017-01-05 Martin Liska <mliska@suse.cz>
34992 * gimplify.c (gimplify_decl_expr): Compare to
34993 asan_poisoned_variables instread of checking flags.
34994 (gimplify_target_expr): Likewise.
34995 (gimplify_expr): Likewise.
34996 (gimplify_function_tree): Conditionally initialize
34997 asan_poisoned_variables.
34999 2017-01-04 Jeff Law <law@redhat.com>
35001 PR tree-optimizatin/78812
35002 * rtl.h (contains_mem_rtx_p): Prototype.
35003 * ifcvt.c (containts_mem_rtx_p): Move from here to...
35004 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
35005 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
35006 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
35007 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
35009 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35011 * input.c (assert_char_at_range): Default-initialize actual_range.
35013 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35015 * df-scan.c (df_ref_create_structure): Make regno unsigned,
35016 to match the caller.
35018 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35020 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
35021 insns after final jump in test to emit dummy move.
35023 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35025 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
35026 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
35028 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35030 * multiple_target.c (create_dispatcher_calls): Init e_next.
35031 * tree-ssa-loop-split.c (split_loop): Init border.
35032 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
35035 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
35040 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
35041 (altivec_register_operand): Do not return true if the operand
35042 contains a SUBREG mixing SImode and SFmode.
35043 (vsx_register_operand): Likewise.
35044 (vsx_reg_sfsubreg_ok): New predicate.
35045 (vfloat_operand): Do not return true if the operand contains a
35046 SUBREG mixing SImode and SFmode.
35047 (vint_operand): Likewise.
35048 (vlogical_operand): Likewise.
35049 (gpc_reg_operand): Likewise.
35050 (int_reg_operand): Likewise.
35051 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
35052 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
35053 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
35055 (rs6000_emit_move_si_sf_subreg): New helper function.
35056 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
35057 fixup SUBREGs involving SImode and SFmode.
35058 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
35059 numbers for the new peephole2 optimization.
35060 (peephole2 for SFmode unions): New peephole2 to optimize cases in
35061 the GLIBC math library that do AND/IOR/XOR operations on single
35062 precision floating point.
35063 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
35064 target macros to say whether we need to avoid SUBREGs mixing
35066 (TARGET_ALLOW_SF_SUBREG): Likewise.
35067 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
35068 (UNSPEC_SI_FROM_SF): Likewise.
35069 (iorxor): Change spacing.
35070 (and_ior_xor): New iterator for AND, IOR, and XOR.
35071 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
35072 (movdi_from_sf_zero_ext): Likewise.
35073 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
35074 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
35075 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
35076 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
35077 (fms<mode>4): Likewise.
35078 (fnma<mode>4): Likewise.
35079 (fnms<mode>4): Likewise.
35080 (nfma<mode>4): Likewise.
35081 (nfms<mode>4): Likewise.
35083 2017-01-04 Marek Polacek <polacek@redhat.com>
35086 * doc/invoke.texi: Document -Wpointer-compare.
35088 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35090 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
35093 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
35094 descriptions for -gdwarf-5 and emit them as uleb128 instead of
35097 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
35100 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
35101 documentation of the powerpc_popcntb_ok attribute.
35102 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35103 code to issue warning messages if a requested CPU configuration is
35104 not supported by the binary (assembler and loader) toolchain.
35105 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
35106 made to define a built-in function that has been disabled.
35107 (paired_init_builtins): Add assertion to prevent ICE if attempt is
35108 made to define a built-in function that has been disabled.
35109 (altivec_init_builtins): Add comment explaining why definition
35110 of the DST built-in functions is not preceded by an assertion
35111 check. Add assertions to prevent ICE if attempts are made to
35112 define an altivec predicate or an abs* built-in function that has
35114 (htm_init_builtins): Add comment explaining why definition of the
35115 htm built-in functions is not preceded by an assertion check.
35117 2017-01-04 Jeff Law <law@redhat.com>
35119 PR tree-optimizatin/67955
35120 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
35121 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
35122 the points-to solution does not include pt_null. Use DECL_PT_UID
35125 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
35127 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
35128 Use gen_int_mode instead of gen_lopwart for const_int operands.
35130 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35132 PR tree-optimization/71563
35133 * match.pd: Simplify X << Y into X if Y is known to be 0 or
35134 out of range value - has low bits known to be zero.
35136 2017-01-04 Alan Modra <amodra@gmail.com>
35138 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
35139 * configure: Regenerate.
35140 * config.in: Regenerate.
35142 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35145 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
35146 a substring of the message, but strcmp with the whole message. Ifdef
35147 ENABLE_NLS, translate the message first using dgettext.
35149 2017-01-03 Jeff Law <law@redhat.com>
35151 PR tree-optimizatin/78856
35152 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
35153 (mark_threaded_blocks): Remove code to truncate thread paths that
35154 cross multiple loop headers. Instead invalidate the cached loop
35155 iteration information and handle case of a thread path walking
35156 into an irreducible region.
35158 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
35161 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
35162 assertions. Add support for doing the signbit if the IEEE 128-bit
35163 floating point value is in a GPR.
35164 * config/rs6000/rs6000.md (Fsignbit): Delete.
35165 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
35166 Update the length attribute if the value is in a GPR.
35167 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
35168 the sign or zero extension instruction, since the value is always 0/1.
35169 (signbit<mode>2_dm2): Delete using <Fsignbit>.
35172 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
35173 extracting SImode to a GPR register so that we can generate a
35174 store, limit the vector to be in a traditional Altivec register
35175 for the vextuwrx instruction.
35177 2017-01-03 Ian Lance Taylor <iant@google.com>
35179 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35181 2017-01-03 Martin Sebor <msebor@redhat.com>
35183 PR tree-optimization/78696
35184 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35185 precision. Use MPFR for %f for greater fidelity. Correct handling
35187 (pass_sprintf_length::compute_format_length): Set width and precision
35188 specified by asrerisk to void_node for vararg functions.
35189 (try_substitute_return_value): Adjust dump output.
35191 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35193 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35195 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35197 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35198 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35199 -mlra/-mno-lra was passed to the compiler.
35201 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35203 PR rtl-optimization/65618
35204 * emit-rtl.c (try_split): Move initialization of "before" and
35205 "after" to just before the call to emit_insn_after_setloc.
35207 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35209 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35211 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35213 * dwarf2out.c (gen_enumeration_type_die): When
35214 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35216 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35218 PR tree-optimization/78965
35219 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35220 Change first argument from const call_info & to call_info &. For %n
35221 set info.nowrite to false.
35223 PR middle-end/78901
35224 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35225 possibly throwing calls.
35227 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35228 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35229 and fns handling, rather than in a separate case SSA_NAME.
35231 2017-01-02 Jeff Law <law@redhat.com>
35233 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35234 fixes for first_period and second_period variables.
35236 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35239 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35240 (*insvqi_1): New insn pattern.
35241 (*insvqi_1_mem_rex64): Ditto.
35242 (*insvqi_2): Ditto.
35243 (*insvqi_3): Rename from *insvqi.
35245 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35247 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35249 * doc/cfg.texi (Edges): Remove reference to Java.
35250 (Maintaining the CFG): Ditto.
35252 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35254 PR middle-end/77674
35255 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35256 transparent aliases.
35258 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35260 PR middle-end/77484
35261 * predict.def (PRED_CALL): Update hitrate.
35262 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35263 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35264 into direct/indirect/polymorphic variants.
35266 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35268 Update copyright years.
35270 * gcc.c (process_command): Update copyright notice dates.
35271 * gcov-dump.c (print_version): Ditto.
35272 * gcov.c (print_version): Ditto.
35273 * gcov-tool.c (print_version): Ditto.
35274 * gengtype.c (create_file): Ditto.
35275 * doc/cpp.texi: Bump @copying's copyright year.
35276 * doc/cppinternals.texi: Ditto.
35277 * doc/gcc.texi: Ditto.
35278 * doc/gccint.texi: Ditto.
35279 * doc/gcov.texi: Ditto.
35280 * doc/install.texi: Ditto.
35281 * doc/invoke.texi: Ditto.
35283 Copyright (C) 2017 Free Software Foundation, Inc.
35285 Copying and distribution of this file, with or without modification,
35286 are permitted in any medium without royalty provided the copyright
35287 notice and this notice are preserved.