1 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
6 to poly_int64. Update commentary for positions_needed.large.
7 (read_info_type): Change offset and width from HOST_WIDE_INT
9 (set_usage_bits): Likewise.
10 (canon_address): Return the offset as a poly_int64 rather than
11 a HOST_WIDE_INT. Use strip_offset_and_add.
12 (set_all_positions_unneeded, any_positions_needed_p): Use
13 positions_needed.large to track stores with non-constant widths.
14 (all_positions_needed_p): Likewise. Take the offset and width
15 as poly_int64s rather than ints. Assert that rhs is nonnull.
16 (record_store): Cope with non-constant offsets and widths.
17 Nullify the rhs of an earlier store if we can't tell which bytes
19 (find_shift_sequence): Take the access_size and shift as poly_int64s
21 (get_stored_val): Take the read_offset and read_width as poly_int64s
22 rather than HOST_WIDE_INTs.
23 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
24 non-constant offsets and widths.
26 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
27 Alan Hayward <alan.hayward@arm.com>
28 David Sherwood <david.sherwood@arm.com>
30 * inchash.h (inchash::hash::add_poly_int): New function.
31 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
32 Use poly_int64 rather than HOST_WIDE_INT.
33 (ao_ref::max_size_known_p): New function.
34 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
35 rather than HOST_WIDE_INT.
36 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
37 to temporaries until its interface is adjusted to match.
38 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
39 (aliasing_component_refs_p, decl_refs_may_alias_p)
40 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
41 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
42 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
44 * alias.c (ao_ref_from_mem): Likewise.
45 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
46 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
47 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
48 (maybe_trim_complex_store, maybe_trim_constructor_store)
49 (live_bytes_read, dse_classify_store): Likewise.
50 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
51 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
52 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
53 (vn_reference_lookup_3): Likewise.
54 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
56 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
57 Alan Hayward <alan.hayward@arm.com>
58 David Sherwood <david.sherwood@arm.com>
60 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
61 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
62 instead of ranges_overlap_p.
64 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
65 Alan Hayward <alan.hayward@arm.com>
66 David Sherwood <david.sherwood@arm.com>
68 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
69 sizes as poly_int64s rather than HOST_WIDE_INTs.
71 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
72 Alan Hayward <alan.hayward@arm.com>
73 David Sherwood <david.sherwood@arm.com>
75 * gimple-fold.h (fold_ctor_reference): Take the offset and size
76 as poly_uint64 rather than unsigned HOST_WIDE_INT.
77 * gimple-fold.c (fold_ctor_reference): Likewise.
79 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
80 Alan Hayward <alan.hayward@arm.com>
81 David Sherwood <david.sherwood@arm.com>
83 * target.def (dwarf_poly_indeterminate_value): New hook.
84 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
85 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
86 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
87 * doc/tm.texi: Regenerate.
88 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
89 offset as a poly_int64.
90 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
91 offset as a poly_int64.
92 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
93 Take the offset as a poly_int64.
94 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
95 (frame_pointer_fb_offset): Change to a poly_int64.
96 (int_loc_descriptor): Take the offset as a poly_int64. Use
97 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
98 (based_loc_descr): Take the offset as a poly_int64.
99 Use strip_offset_and_add to handle (plus X (const)).
100 Use new_reg_loc_descr instead of an open-coded version of the
101 previous implementation.
102 (mem_loc_descriptor): Handle CONST_POLY_INT.
103 (compute_frame_pointer_to_fb_displacement): Take the offset as a
104 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
106 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
107 Alan Hayward <alan.hayward@arm.com>
108 David Sherwood <david.sherwood@arm.com>
110 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
111 (gen_rtx_REG_offset): Take the offset as a poly_int64.
112 * inchash.h (inchash::hash::add_poly_hwi): New function.
113 * gengtype.c (main): Register poly_int64.
114 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
115 offset as a poly_int.
116 (reg_attr_hasher::equal): Use must_eq to compare offsets.
117 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
118 offset as a poly_int64.
119 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
120 * print-rtl.c (print_poly_int): New function.
121 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
123 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
125 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
126 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
127 (vt_get_decl_and_offset): Return the offset as a poly_int64.
128 Enforce track_offset_p for parts of a PARALLEL.
129 (vt_add_function_parameter): Use const_offset for the final
130 offset to track. Use get_tracked_reg_offset for the parts
133 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
134 Alan Hayward <alan.hayward@arm.com>
135 David Sherwood <david.sherwood@arm.com>
137 * target.def (truly_noop_truncation): Take poly_uint64s instead of
138 unsigned ints. Change default to hook_bool_puint64_puint64_true.
139 * doc/tm.texi: Regenerate.
140 * hooks.h (hook_bool_uint_uint_true): Delete.
141 (hook_bool_puint64_puint64_true): Declare.
142 * hooks.c (hook_bool_uint_uint_true): Delete.
143 (hook_bool_puint64_puint64_true): New function.
144 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
145 instead of unsigned ints.
146 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
147 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
149 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
150 Alan Hayward <alan.hayward@arm.com>
151 David Sherwood <david.sherwood@arm.com>
153 * optabs.h (expand_operand): Add an int_value field.
154 (create_expand_operand): Add an int_value parameter and use it
155 to initialize the new expand_operand field.
156 (create_integer_operand): Replace with a declaration of a function
157 that accepts poly_int64s. Move the implementation to...
158 * optabs.c (create_integer_operand): ...here.
159 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
160 the mode preserves the value of int_value, instead of calling
161 const_int_operand on the rtx. Use gen_int_mode to generate
164 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
165 Alan Hayward <alan.hayward@arm.com>
166 David Sherwood <david.sherwood@arm.com>
168 * dumpfile.h (dump_dec): Declare.
169 * dumpfile.c (dump_dec): New function.
170 * pretty-print.h (pp_wide_integer): Turn into a function and
171 declare a poly_int version.
172 * pretty-print.c (pp_wide_integer): New function for poly_ints.
174 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
175 Alan Hayward <alan.hayward@arm.com>
176 David Sherwood <david.sherwood@arm.com>
178 * doc/generic.texi (POLY_INT_CST): Document.
179 * tree.def (POLY_INT_CST): New tree code.
180 * treestruct.def (TS_POLY_INT_CST): New tree layout.
181 * tree-core.h (tree_poly_int_cst): New struct.
182 (tree_node): Add a poly_int_cst field.
183 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
184 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
185 instead of a wide_int_ref.
186 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
188 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
189 instead of an unsigned HOST_WIDE_INT.
190 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
191 (ptrdiff_tree_p): Declare.
192 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
193 extern inline implementations if the target doesn't use POLY_INT_CST.
194 (poly_int_tree_p): New function.
195 (wi::unextended_tree): New class.
196 (wi::int_traits <unextended_tree>): New override.
197 (wi::extended_tree): Add a default constructor.
198 (wi::extended_tree::get_tree): New function.
199 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
200 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
201 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
202 (wi::tree_to_poly_wide_ref): New typedefs.
203 (wi::ints_for): Provide overloads for extended_tree and
205 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
206 (wi::to_wide): New functions.
207 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
208 * tree.c (poly_int_cst_hasher): New struct.
209 (poly_int_cst_hash_table): New variable.
210 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
211 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
213 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
214 (init_ttree): Initialize poly_int_cst_hash_table.
215 (build_int_cst, build_int_cst_type, build_invariant_address): Take
216 a poly_int64 instead of a HOST_WIDE_INT.
217 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
218 instead of an unsigned HOST_WIDE_INT.
219 (wide_int_to_tree): Rename to...
220 (wide_int_to_tree_1): ...this.
221 (build_new_poly_int_cst, build_poly_int_cst): New functions.
222 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
223 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
224 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
225 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
226 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
228 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
229 (streamer_read_tree_body): Likewise.
230 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
231 (streamer_write_tree_body): Likewise.
232 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
233 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
234 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
235 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
236 * gimple-expr.h (is_gimple_constant): Likewise.
237 * gimplify.c (maybe_with_size_expr): Likewise.
238 * print-tree.c (print_node): Likewise.
239 * tree-data-ref.c (data_ref_compare_tree): Likewise.
240 * tree-pretty-print.c (dump_generic_node): Likewise.
241 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
242 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
243 * tree-vrp.c (compare_values_warnv): Likewise.
244 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
245 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
246 (force_expr_to_var_cost): Likewise.
247 * tree-ssa-loop.c (for_each_index): Likewise.
248 * fold-const.h (build_invariant_address, size_int_kind): Take a
249 poly_int64 instead of a HOST_WIDE_INT.
250 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
251 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
253 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
254 (int_const_binop_2): New function, split out from...
255 (int_const_binop_1): ...here. Handle POLY_INT_CST.
256 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
257 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
258 * gimple-ssa-strength-reduction.c (slsr_process_add)
259 (slsr_process_mul): Check for INTEGER_CSTs before using them
261 * stor-layout.c (bits_from_bytes): New function.
262 (bit_from_pos): Use it.
263 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
264 by BITS_PER_UNIT to get the TYPE_SIZE.
265 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
266 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
268 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
269 Alan Hayward <alan.hayward@arm.com>
270 David Sherwood <david.sherwood@arm.com>
272 * doc/rtl.texi (const_poly_int): Document. Also document the
273 rtl sharing behavior.
274 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
275 * rtl.h (const_poly_int_def): New struct.
276 (rtx_def::u): Add a cpi field.
277 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
278 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
279 (wi::rtx_to_poly_wide_ref): New typedef
280 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
281 (poly_int_rtx_p): New functions.
282 (trunc_int_for_mode): Declare a poly_int64 version.
283 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
284 (immed_wide_int_const): Take a poly_wide_int_ref rather than
286 (strip_offset): Declare.
287 (strip_offset_and_add): New function.
288 * rtl.def (CONST_POLY_INT): New rtx code.
289 * rtl.c (rtx_size): Handle CONST_POLY_INT.
290 (shared_const_p): Use poly_int_rtx_p.
291 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
293 (gen_int_shift_amount): Likewise.
294 * emit-rtl.c (const_poly_int_hasher): New class.
295 (const_poly_int_htab): New variable.
296 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
297 (const_poly_int_hasher::hash): New function.
298 (const_poly_int_hasher::equal): Likewise.
299 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
300 (immed_wide_int_const): Rename to...
301 (immed_wide_int_const_1): ...this and make static.
302 (immed_wide_int_const): New function, taking a poly_wide_int_ref
303 instead of a wide_int_ref.
304 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
305 (gen_lowpart_common): Handle CONST_POLY_INT.
306 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
307 * cselib.c (cselib_hash_rtx): Likewise.
308 * dwarf2out.c (const_ok_for_output_1): Likewise.
309 * expr.c (convert_modes): Likewise.
310 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
311 * rtlhash.c (add_rtx): Likewise.
312 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
313 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
314 Handle existing CONST_POLY_INT rtxes.
315 * expmed.h (expand_shift): Take a poly_int64 instead of a
317 * expmed.c (expand_shift): Likewise.
318 * rtlanal.c (strip_offset): New function.
319 (commutative_operand_precedence): Give CONST_POLY_INT the same
320 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
322 * rtl-tests.c (const_poly_int_tests): New struct.
323 (rtl_tests_c_tests): Use it.
324 * simplify-rtx.c (simplify_const_unary_operation): Handle
326 (simplify_const_binary_operation): Likewise.
327 (simplify_binary_operation_1): Fold additions of symbolic constants
329 (simplify_subreg): Handle extensions and truncations of
331 (simplify_const_poly_int_tests): New struct.
332 (simplify_rtx_c_tests): Use it.
333 * wide-int.h (storage_ref): Add default constructor.
334 (wide_int_ref_storage): Likewise.
335 (trailing_wide_ints): Use GTY((user)).
336 (trailing_wide_ints::operator[]): Add a const version.
337 (trailing_wide_ints::get_precision): New function.
338 (trailing_wide_ints::extra_size): Likewise.
340 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
341 Alan Hayward <alan.hayward@arm.com>
342 David Sherwood <david.sherwood@arm.com>
344 * emit-rtl.h (gen_int_shift_amount): Declare.
345 * emit-rtl.c (gen_int_shift_amount): New function.
346 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
348 * calls.c (shift_return_value): Likewise.
349 * cse.c (fold_rtx): Likewise.
350 * dse.c (find_shift_sequence): Likewise.
351 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
352 (expand_shift, expand_smod_pow2): Likewise.
353 * lower-subreg.c (shift_cost): Likewise.
354 * optabs.c (expand_superword_shift, expand_doubleword_mult)
355 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
356 (expand_vec_perm_var): Likewise.
357 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
358 (simplify_binary_operation_1): Likewise.
359 * combine.c (try_combine, find_split_point, force_int_to_mode)
360 (simplify_shift_const_1, simplify_shift_const): Likewise.
361 (change_zero_ext): Likewise. Use simplify_gen_binary.
363 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
365 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
367 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
369 * doc/invoke.texi (ARM Options): Document accepted extension options
370 for -march=armv8.3-a.
372 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
375 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
376 or --with-float=softfp, set the default CPU to arm10e.
378 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
380 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
381 * config/visium/predicates.md (const_shift_operand): Likewise.
382 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
383 (visium_legitimize_reload_address): Likewise.
385 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
388 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
390 2017-12-20 Jakub Jelinek <jakub@redhat.com>
393 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
395 2017-12-20 Julia Koval <julia.koval@intel.com>
397 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
398 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
399 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
400 * config.gcc: Include vpclmulqdqintrin.h.
401 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
402 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
404 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
405 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
406 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
407 * config/i386/i386.c (isa2_opts): Add -mcx16.
408 (isa_opts): Add -mpclmulqdq, remove -mcx16.
409 (ix86_option_override_internal): Move mcx16 to flags2.
410 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
411 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
412 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
413 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
414 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
415 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
416 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
417 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
418 * doc/invoke.texi: Add -mvpclmulqdq.
420 2017-12-20 Tom de Vries <tom@codesourcery.com>
423 * config/i386/i386.c (ix86_static_chain): Move DECL_STATIC_CHAIN test ...
424 * calls.c (rtx_for_static_chain): ... here. New function.
425 * calls.h (rtx_for_static_chain): Declare.
426 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
427 instead of targetm.calls.static_chain.
428 * df-scan.c (df_get_entry_block_def_set): Same.
430 2017-12-19 Tom de Vries <tom@codesourcery.com>
432 PR tree-optimization/83493
433 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
436 2017-12-19 Jakub Jelinek <jakub@redhat.com>
438 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
439 inform with hardcoded english plural handling.
441 2017-12-18 Jeff Law <law@redhat.com>
443 PR tree-optimization/83477
444 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
445 a non-virtual PHI, always push a new range.
447 2017-12-19 Martin Sebor <msebor@redhat.com>
450 * builtins.c (compute_objsize): Handle non-constant offsets.
452 2017-12-19 Jakub Jelinek <jakub@redhat.com>
454 PR tree-optimization/83444
455 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
456 character load case, if get_stridx on MEM_REF's operand doesn't
457 look usable, retry with get_addr_stridx.
459 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
462 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
463 (variable_tracking_main_1): Keep markers even when VTA fails.
466 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
467 even if there are markers before it.
468 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
470 2017-12-19 Jakub Jelinek <jakub@redhat.com>
472 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
473 typical order conditions.
474 * sel-sched.c (extract_new_fences_from): Likewise.
475 * config/visium/constraints.md (J, K, L): Likewise.
476 * config/visium/predicates.md (const_shift_operand): Likewise.
477 * config/visium/visium.c (visium_legitimize_address,
478 visium_legitimize_reload_address): Likewise.
479 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
480 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
481 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
482 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
483 SET_DUMP_DETAIL): Likewise.
484 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
485 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
486 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
487 avr_set_core_architecture, avr_set_current_function,
488 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
489 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
490 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
491 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
492 avr_map_decompose, avr_fold_builtin): Likewise.
493 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
494 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
495 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
496 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
497 * config/m32c/m32c.c (m32c_conditional_register_usage,
498 m32c_address_cost): Likewise.
499 * config/m32c/predicates.md (shiftcount_operand,
500 longshiftcount_operand): Likewise.
501 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
502 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
503 can_use_cdx_ldstw): Likewise.
504 * config/nios2/nios2.h (CDX_REG_P): Likewise.
505 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
507 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
508 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
509 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
510 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
511 * config/vax/vax.c (adjacent_operands_p): Likewise.
512 * config/ft32/constraints.md (L, b, KA): Likewise.
513 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
515 * cfgexpand.c (expand_stack_alignment): Likewise.
516 * gcse.c (insert_expr_in_table): Likewise.
517 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
518 * cgraphunit.c (cgraph_node::expand): Likewise.
519 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
520 * emit-rtl.c (add_insn): Likewise.
521 * input.c (dump_location_info): Likewise.
522 * passes.c (NEXT_PASS): Likewise.
523 * read-rtl-function.c (parse_note_insn_name,
524 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
526 * sched-rgn.c (sched_rgn_init): Likewise.
527 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
528 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
529 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
530 * explow.c (eliminate_constant_term): Likewise.
531 * final.c (leaf_renumber_regs_insn): Likewise.
532 * cfgrtl.c (print_rtl_with_bb): Likewise.
533 * genhooks.c (emit_init_macros): Likewise.
534 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
535 * tree-data-ref.c (conflict_fn): Likewise.
536 * selftest.c (assert_streq): Likewise.
537 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
538 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
539 fold_binary_loc, multiple_of_p): Likewise.
540 * reload.c (push_reload, find_equiv_reg): Likewise.
541 * et-forest.c (et_nca, et_below): Likewise.
542 * dbxout.c (dbxout_symbol_location): Likewise.
543 * reorg.c (relax_delay_slots): Likewise.
544 * dojump.c (do_compare_rtx_and_jump): Likewise.
545 * gengtype-parse.c (type): Likewise.
546 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
547 simplify_const_relational_operation): Likewise.
548 * reload1.c (do_output_reload): Likewise.
549 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
550 * gengtype.c (type_for_name): Likewise.
551 * gimple-ssa-sprintf.c (format_directive): Likewise.
553 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
556 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
557 accessing it. Adjust comment.
559 2017-12-19 Jakub Jelinek <jakub@redhat.com>
562 * predict.c (zero_one_minusone): New function.
563 (apply_return_prediction): Avoid return prediction for functions
564 returning only -1, 0 and 1 values, unless they only return -1 and 0
567 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
569 * config/arc/arc.c (legitimate_scaled_address_p): Clean
570 fall-through warning.
571 (arc_compute_frame_size): Remove unused variables.
572 (arc_print_operand): Fix fprintif format.
573 (arc_can_follow_jump): Clean fall-through warning.
575 2017-12-19 Marek Polacek <polacek@redhat.com>
578 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
581 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
583 * config/arc/arc.c (overriderregs): New variable.
584 (arc_override_options): Track fixed/call saved/call options.
585 (arc_conditional_register_usage): Check against overrideregs
586 variable whenever we change register properties.
588 2017-12-19 Nathan Sidwell <nathan@acm.org>
590 * opts.c (finish_options): Don't prefix dump_base_name if it
591 already contains directories.
593 2017-12-19 Martin Liska <mliska@suse.cz>
595 PR rtl-optimization/82675
596 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
597 more element in sbitmap.
599 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
601 * gimplify.c (gimplify_expr): Use error_operand_p.
603 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
606 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
609 2017-12-19 Daniel Cederman <cederman@gaisler.com>
611 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
612 the jump is to a label.
614 2017-12-19 Jakub Jelinek <jakub@redhat.com>
616 PR tree-optimization/83444
617 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
622 * ipa-inline.c (flatten_remove_node_hook): New function.
623 (ipa_inline): Keep only nodes with flatten attribute at the end of
624 the array in the order from ipa_reverse_postorder, only walk that
625 portion of array for flattening, if there is more than one such
626 node, temporarily register a removal hook and ignore removed nodes.
628 PR tree-optimization/80631
629 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
630 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
631 IFN_REDUC_MAX or IFN_REDUC_MIN.
633 2017-12-18 Jeff Law <law@redhat.com>
635 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
636 record symbolic equivalences from backedges in the CFG.
639 2017-11-19 Jeff Law <law@redhat.com>
641 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
642 of degenerates resulting from ignoring an edge.
644 2017-12-18 Martin Sebor <msebor@redhat.com>
647 PR tree-optimization/78450
648 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
649 (handle_builtin_strlen): Call it.
651 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
653 PR rtl-optimization/83424
654 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
656 2017-12-18 Marek Polacek <polacek@redhat.com>
659 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
660 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
663 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
665 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
666 (umaddsidi4, umaddsidi_split): Likewise.
668 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
670 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
673 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
676 * config/s390/s390.c (s390_option_override): Avoid strncpy.
678 2017-12-18 Richard Biener <rguenther@suse.de>
680 PR tree-optimization/81877
681 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
682 (outermost_indep_loop): Adjust.
683 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
684 (can_sm_ref_p): Adjust.
686 2017-12-18 Richard Biener <rguenther@suse.de>
689 * tree.c (array_at_struct_end_p): Return true if the underlying
690 object has space for at least one element in excess of what
691 the array domain specifies.
693 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
695 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
696 example to avoid overfull hbox.
697 * doc/invoke.texi (Option Summary): Add missing @gol.
698 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
701 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
702 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
704 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
705 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
706 (AArch64 Options): Document -mverbose-cost-dump.
707 (ARM Options): Likewise, plus -mflip-thumb.
709 2017-12-17 Martin Sebor <msebor@redhat.com>
712 * gimple-ssa-warn-restrict.c
713 (builtin_memref::offset_out_of_bounds): Correct the handling of
716 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
718 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
720 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
722 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
724 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
726 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
729 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
731 * doc/invoke.texi: Fix some typos.
733 2017-12-16 Martin Sebor <msebor@redhat.com>
735 PR tree-optimization/78918
736 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
737 * builtins.c (check_sizes): Rename...
738 (check_access): ...to this. Rename function arguments for clarity.
739 (check_memop_sizes): Adjust names.
740 (expand_builtin_memchr, expand_builtin_memcpy): Same.
741 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
742 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
743 (check_strncat_sizes, expand_builtin_strncat): Same.
744 (expand_builtin_strncpy, expand_builtin_memset): Same.
745 (expand_builtin_bzero, expand_builtin_memcmp): Same.
746 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
747 (maybe_emit_sprintf_chk_warning): Same.
748 (expand_builtin_strcpy): Adjust.
749 (expand_builtin_stpcpy): Same.
750 (expand_builtin_with_bounds): Detect out-of-bounds accesses
751 in pointer-checking forms of memcpy, memmove, and mempcpy.
752 (gcall_to_tree_minimal, max_object_size): Define new functions.
753 * builtins.h (max_object_size): Declare.
754 * calls.c (alloc_max_size): Call max_object_size instead of
755 hardcoding ssizetype limit.
756 (get_size_range): Handle new argument.
757 * calls.h (get_size_range): Add a new argument.
758 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
759 * doc/invoke.texi (-Wrestrict): Adjust, add example.
760 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
762 (gimple_fold_builtin_memory_chk): Same.
763 (gimple_fold_builtin_stxcpy_chk): New function.
764 * gimple-ssa-warn-restrict.c: New source.
765 * gimple-ssa-warn-restrict.h: New header.
766 * gimple.c (gimple_build_call_from_tree): Propagate location.
767 * passes.def (pass_warn_restrict): Add new pass.
768 * tree-pass.h (make_pass_warn_restrict): Declare.
769 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
771 (handle_builtin_strcat): Same.
772 (strlen_optimize_stmt): Rename...
773 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
774 stpncpy, strncpy, and their checking forms.
776 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
778 PR rtl-optimization/82849
779 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
780 and get_max_loop_iterations_int.
782 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
783 Alan Hayward <alan.hayward@arm.com>
784 David Sherwood <david.sherwood@arm.com>
786 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
787 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
788 size as a poly_uint64.
789 (mode_for_vector, mode_for_int_vector): Take the number of vector
790 elements as a poly_uint64.
791 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
792 size as a poly_uint64.
793 (mode_for_vector, mode_for_int_vector): Take the number of vector
794 elements as a poly_uint64.
796 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
797 Alan Hayward <alan.hayward@arm.com>
798 David Sherwood <david.sherwood@arm.com>
800 * machmode.h (MACRO_MODE): New macro.
801 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
802 * targhooks.c (default_libcall_value, default_secondary_reload)
803 (default_memory_move_cost, default_register_move_cost)
804 (default_class_max_nregs): Likewise.
806 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
807 Alan Hayward <alan.hayward@arm.com>
808 David Sherwood <david.sherwood@arm.com>
810 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
811 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
813 * genautomata.c (main): Likewise.
814 * genconditions.c (write_header): Likewise.
815 * genemit.c (main): Likewise.
816 * genextract.c (print_header): Likewise.
817 * genopinit.c (main): Likewise.
818 * genoutput.c (output_prologue): Likewise.
819 * genpeep.c (main): Likewise.
820 * genpreds.c (write_insn_preds_c): Likewise.
821 * genrecog.c (writer_header): Likewise.
822 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
823 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
824 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
825 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
826 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
827 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
828 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
829 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
830 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
831 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
832 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
833 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
834 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
835 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
836 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
837 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
838 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
839 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
840 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
841 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
842 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
843 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
844 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
845 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
846 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
847 * config/darwin.c (IN_TARGET_CODE): Likewise.
848 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
849 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
850 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
851 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
852 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
853 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
854 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
855 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
856 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
857 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
858 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
859 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
860 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
861 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
862 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
863 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
864 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
865 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
866 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
867 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
868 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
869 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
870 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
871 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
872 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
873 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
874 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
875 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
876 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
877 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
878 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
879 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
880 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
881 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
882 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
883 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
884 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
885 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
886 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
887 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
888 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
889 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
890 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
891 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
892 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
893 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
894 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
895 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
896 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
897 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
898 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
899 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
900 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
901 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
902 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
903 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
904 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
905 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
906 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
907 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
908 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
909 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
910 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
911 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
912 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
913 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
914 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
915 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
916 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
917 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
918 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
919 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
920 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
921 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
922 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
923 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
924 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
925 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
926 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
927 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
928 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
929 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
930 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
931 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
932 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
933 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
934 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
935 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
936 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
937 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
938 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
939 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
940 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
941 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
942 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
943 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
944 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
945 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
946 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
947 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
948 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
949 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
950 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
951 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
952 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
953 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
955 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
956 Alan Hayward <alan.hayward@arm.com>
957 David Sherwood <david.sherwood@arm.com>
959 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
962 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
963 Alan Hayward <alan.hayward@arm.com>
964 David Sherwood <david.sherwood@arm.com>
966 * doc/generic.texi (VEC_SERIES_EXPR): Document.
967 * doc/md.texi (vec_series@var{m}): Document.
968 * tree.def (VEC_SERIES_EXPR): New tree code.
969 * tree.h (build_vec_series): Declare.
970 * tree.c (build_vec_series): New function.
971 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
972 * tree-pretty-print.c (dump_generic_node): Likewise.
973 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
974 * tree-inline.c (estimate_operator_cost): Likewise.
975 * expr.c (expand_expr_real_2): Likewise.
976 * optabs-tree.c (optab_for_tree_code): Likewise.
977 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
978 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
979 * expmed.c (make_tree): Handle VEC_SERIES.
980 * optabs.def (vec_series_optab): New optab.
981 * optabs.h (expand_vec_series_expr): Declare.
982 * optabs.c (expand_vec_series_expr): New function.
983 * tree-vect-generic.c (expand_vector_operations_1): Check that
984 the operands also have vector type.
986 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
987 Alan Hayward <alan.hawyard@arm.com>
988 David Sherwood <david.sherwood@arm.com>
990 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
991 (VEC_COND_EXPR): Add missing @tindex.
992 * doc/md.texi (vec_duplicate@var{m}): Document.
993 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
994 * tree.c (build_vector_from_val): Add stubbed-out handling of
995 variable-length vectors, using VEC_DUPLICATE_EXPR.
996 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
997 * cfgexpand.c (expand_debug_expr): Likewise.
998 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
999 * tree-inline.c (estimate_operator_cost): Likewise.
1000 * tree-pretty-print.c (dump_generic_node): Likewise.
1001 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
1002 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
1003 (test_vec_duplicate_folding): New function.
1004 (fold_const_c_tests): Call it.
1005 * optabs.def (vec_duplicate_optab): New optab.
1006 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
1007 * optabs.h (expand_vector_broadcast): Declare.
1008 * optabs.c (expand_vector_broadcast): Make non-static. Try using
1009 vec_duplicate_optab.
1010 * expr.c (store_constructor): Try using vec_duplicate_optab for
1012 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
1014 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
1017 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
1018 div/mod latencies a bit.
1020 2017-12-15 Jeff Law <law@redhat.com>
1022 PR tree-optimization/36550
1023 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
1024 (mark_threaded_blocks): Rewrite code to avoid block copying when
1025 optimizing for size. Don't pessimize blocks which will be
1026 copied, but all the statements will be dead.
1028 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
1030 PR tree-optimization/81165
1031 * tree-ssa-threadupdate.c (uses_in_bb): New.
1032 (estimate_threading_killed_stmts): New.
1033 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
1034 * tree-ssa-threadedge.c
1035 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
1038 2017-12-15 Jeff Law <law@redhat.com>
1040 PR tree-optimization/83410
1041 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
1042 threads when parallelizing loops.
1044 2017-12-15 Jakub Jelinek <jakub@redhat.com>
1046 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
1048 * config/alpha/alpha.c (vms_attribute_table): Swap
1049 affects_type_identity and handler fields, adjust comments.
1050 * config/mips/mips.c (mips_attribute_table): Likewise.
1051 * config/visium/visium.c (visium_attribute_table): Likewise.
1052 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
1053 * config/microblaze/microblaze.c (microblaze_attribute_table):
1055 * config/spu/spu.c (spu_attribute_table): Likewise.
1056 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
1057 * config/arc/arc.c (arc_attribute_table): Likewise.
1058 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
1059 * config/v850/v850.c (v850_handle_interrupt_attribute,
1060 v850_handle_data_area_attribute): Formatting fixes.
1061 (v850_attribute_table): Swap affects_type_identity and handler
1062 fields, adjust comments.
1063 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
1064 * config/arm/arm.c (arm_attribute_table): Likewise.
1065 * config/avr/avr.c (avr_attribute_table): Likewise.
1066 * config/s390/s390.c (s390_attribute_table): Likewise.
1067 * config/sh/sh.c (sh_attribute_table): Likewise.
1068 * config/i386/i386.c (ix86_handle_cconv_attribute,
1069 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
1070 (ix86_attribute_table): Swap affects_type_identity and handler
1071 fields, adjust comments.
1072 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1073 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
1074 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
1075 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
1076 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
1077 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
1078 * config/rx/rx.c (rx_attribute_table): Likewise.
1079 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
1080 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
1081 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
1082 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
1083 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1084 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
1085 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
1086 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
1087 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
1088 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
1089 * doc/plugins.texi (user_attr): Likewise. Add NULL for
1091 * attribs.c (empty_attribute_table): Swap affects_type_identity and
1093 (register_scoped_attributes, decl_attributes): Formatting fixes.
1095 PR tree-optimization/83269
1096 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
1097 subtraction in arg0's type if type is signed and arg0 is unsigned.
1101 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
1102 convert? on @0 instead of convert. Check type of @1, not @0.
1103 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
1104 convert? on @0 instead of convert. Check type of @1, not @0.
1105 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
1106 only optimize if either both @1 and @2 types are narrower
1107 precision, or both are wider or equal precision, and in the former
1108 case only if both have undefined overflow.
1110 2017-12-15 Richard Biener <rguenther@suse.de>
1113 * internal-fn.def (IFN_NOP): Add.
1114 * internal-fn.c (expand_NOP): Do nothing.
1115 * lto-streamer-in.c (input_function): Instead of removing
1116 sanitizer calls replace them with IFN_NOP calls.
1118 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
1119 Alan Hayward <alan.hayward@arm.com>
1120 David Sherwood <david.sherwood@arm.com>
1122 * dse.c (store_info, read_info_type): Replace begin and end with
1124 (print_range): New function.
1125 (set_all_positions_unneeded, any_positions_needed_p)
1126 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
1128 (record_store): Likewise. Optimize the case in which all positions
1130 (get_stored_val): Replace read_begin and read_end with read_offset
1132 (replace_read): Update call accordingly.
1134 2017-12-15 Bin Cheng <bin.cheng@arm.com>
1136 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
1137 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
1138 (loop_cand::loop_cand): Initialize above members.
1139 (loop_cand::supported_operations): Delete.
1140 (loop_cand::can_interchange_p): Inline above function.
1141 (loop_cand::classify_simple_reduction): Record number of constant
1142 initialized simple reductions.
1143 (should_interchange_loops): New parameters. Check stmt cost of loops
1145 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
1146 Update call to should_interchange_loops.
1147 (should_interchange_loop_nest): Update call to
1148 should_interchange_loops.
1150 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
1153 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
1154 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
1155 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
1157 2017-12-15 Julia Koval <julia.koval@intel.com>
1159 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
1160 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
1161 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
1162 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
1163 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
1165 2017-12-15 Julia Koval <julia.koval@intel.com>
1167 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
1168 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
1169 * config/i386/sse.md (vaesenc_<mode>): New pattern.
1170 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
1171 _mm_aesenc_epi128): New intrinsics.
1173 2017-12-15 Julia Koval <julia.koval@intel.com>
1175 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
1176 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
1177 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
1178 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
1179 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
1181 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1183 * gimple-ssa-strength-reduction.c (analyze_increments):
1184 Distinguish replacement costs for constant strides from those for
1187 2017-12-14 Jakub Jelinek <jakub@redhat.com>
1189 * var-tracking.c (variable_tracking_main_1): Formatting fix.
1191 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
1193 * doc/invoke.texi: Document -Wcast-function-type.
1194 * recog.h (stored_funcptr): Change signature.
1195 * tree-dump.c (dump_node): Avoid warning.
1196 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
1198 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
1201 * gimple-fold.c (get_range_strlen): Add the handling of non-member
1204 2017-12-14 David Malcolm <dmalcolm@redhat.com>
1206 PR tree-optimization/83312
1207 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
1208 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
1209 "val" param, and to cope with arbitrary basic blocks.
1210 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
1211 handle NULL_TREE for "val", dropping "bb" param.
1212 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
1213 drop "bb" param. Handle NULL_TREE for "val".
1214 (find_case_label_for_value): Make "switch_stmt" param const.
1215 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
1217 (vrp_prop::check_all_array_refs): Reimplement as...
1218 (check_array_bounds_dom_walker::before_dom_children): ...this new
1219 vfunc. Replace linear search through BB block list, excluding
1220 those with non-executable in-edges via dominator walk.
1222 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1224 * config/arm/arm.opt (mverbose-cost-dump): New option.
1225 * config/arm/arm.c (arm_rtx_costs): Use it.
1227 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
1230 * reload1.c (emit_input_reload_insns): Skip debug markers.
1232 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
1234 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
1238 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
1239 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
1240 * config/sh/sh.c: Skip debug insns besides notes.
1241 * config/sh/sh.md: Likewise.
1242 * config/sh/sh_treg_combine.cc: Likewise.
1243 * config/sh/sync.md: Likewise.
1245 2017-12-14 Tom de Vries <tom@codesourcery.com>
1247 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
1248 Add item for weak_undefined.
1250 2017-12-14 Richard Biener <rguenther@suse.de>
1252 PR tree-optimization/67842
1253 * sese.h (bb_in_region): Remove #if 0'ed code.
1255 2017-12-14 Richard Biener <rguenther@suse.de>
1257 PR tree-optimization/83326
1258 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
1259 may_be_zero parameter and handle it by not marking the first
1260 peeled copy as not exiting the loop.
1261 (try_peel_loop): Likewise.
1262 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
1263 to handle the case of constant or zero iterations and perform
1264 loop header copying on-the-fly.
1266 2017-12-14 Richard Biener <rguenther@suse.de>
1268 PR tree-optimization/83418
1269 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
1270 Instead of asserting we don't get unfolded comparisons deal with
1273 2017-12-14 Jakub Jelinek <jakub@redhat.com>
1276 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
1277 basic blocks. Assert debug bind insns don't appear outside of bbs,
1278 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
1281 PR tree-optimization/83198
1282 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
1283 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
1284 value if arg is a REAL_CST with incompatible type.
1286 2017-12-14 Sudakshina Das <sudi.das@arm.com>
1287 Bin Cheng <bin.cheng@arm.com>
1290 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
1292 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
1295 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1297 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
1298 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
1300 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
1301 Alan Hayward <alan.hayward@arm.com>
1302 David Sherwood <david.sherwood@arm.com>
1304 * poly-int.h: New file.
1305 * poly-int-types.h: Likewise.
1306 * coretypes.h: Include them.
1307 (POLY_INT_CONVERSION): Define.
1308 * target.def (estimated_poly_value): New hook.
1309 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
1310 * doc/tm.texi: Regenerate.
1311 * doc/poly-int.texi: New file.
1312 * doc/gccint.texi: Include it.
1313 * doc/rtl.texi: Describe restrictions on subreg modes.
1314 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
1315 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
1316 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
1317 * targhooks.h (default_estimated_poly_value): Declare.
1318 * targhooks.c (default_estimated_poly_value): New function.
1319 * target.h (estimated_poly_value): Likewise.
1320 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
1321 (wi::unary_traits): Delete.
1322 (wi::binary_traits::signed_shift_result_type): Define for
1323 offset_int << HOST_WIDE_INT, etc.
1324 (generic_wide_int::operator <<=): Define for all types and use
1325 wi::lshift instead of <<.
1326 (wi::hwi_with_prec): Add a default constructor.
1327 (wi::ints_for): New class.
1328 (operator <<): Define for all wide-int types.
1329 (operator /): New function.
1330 (operator %): Likewise.
1331 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
1332 (ASSERT_MAYBE_NE_AT): New macros.
1334 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
1335 Dominik Vogt <vogt@linux.vnet.ibm.com>
1338 * emit-rtl.c (init_emit): Remove ??? comment.
1339 * explow.c (get_dynamic_stack_size): Take known alignment of stack
1340 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
1341 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
1342 alignment of 3 virtual registers to BITS_PER_WORD.
1344 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
1346 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
1348 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
1349 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
1351 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
1354 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
1356 2017-12-13 Jakub Jelinek <jakub@redhat.com>
1358 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
1359 with the exception of debug begin stmt markers appear before
1363 * final.c (rest_of_handle_final): Call variable_tracking_main only
1364 if !flag_var_tracking.
1366 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
1367 Jakub Jelinek <jakub@redhat.com>
1371 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
1372 labels debug stmts that can only appear after labels.
1374 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
1376 PR rtl-optimization/82398
1377 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
1378 EXPR_USEFULNESS in priority comparison.
1380 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
1382 PR rtl-optimization/83393
1383 * combine.c (move_deaths): If reg_stat points to a too new insn in
1384 last_death, do not use it: find the proper insn instead.
1386 2017-12-12 Jeff Law <law@redhat.com>
1388 PR tree-optimization/83298
1389 PR tree-optimization/83362
1390 PR tree-optimization/83383
1391 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
1392 push_value_range a public interface. Add new argument to
1393 record_ranges_from_stmt.
1394 * gimple-ssa-evrp-analyze.c
1395 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
1396 Update comments. Handle recording temporary equivalences.
1397 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
1398 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
1399 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
1400 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
1401 gimple-ssa-evrp-analyze.h.
1402 (record_temporary_equivalences_from_phis): Add new argument. When
1403 the PHI arg is an SSA_NAME, set the result's range to the range
1405 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
1406 from statements too.
1407 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
1408 Pass it down to children as needed.
1409 (thread_outgoing_edges): Likewise.
1410 (thread_across_edge): Likewise. Push/pop range state as needed.
1411 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
1413 2017-12-12 Julia Koval <julia.koval@intel.com>
1415 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
1416 (PTA_CANNONLAKE): Remove PTA_CLWB.
1418 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1421 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
1422 (vcondv2div2df): Likewise.
1423 (vconduv2dfv2di): Likewise.
1425 2017-12-12 Jakub Jelinek <jakub@redhat.com>
1427 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
1428 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
1429 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
1430 ATTR_MATHFN_FPROUNDING_ERRNO.
1432 2017-12-12 Richard Biener <rguenther@suse.de>
1434 PR tree-optimization/83385
1435 * graphite-scop-detection.c (get_order, order): Remove.
1436 (bb_to_rpo): New global.
1438 (build_scops): Sort pbbs in RPO order.
1440 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
1442 * combine.c (simplify_set): Do not transform subregs to zero_extends
1443 if the destination mode is a vector mode.
1445 2017-12-12 Jakub Jelinek <jakub@redhat.com>
1447 PR tree-optimization/83359
1448 * tree-cfg.h (fold_loop_internal_call): Declare.
1449 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
1450 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
1451 (find_loop_dist_alias): New function.
1452 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
1453 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
1456 PR tree-optimization/80631
1457 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
1458 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
1459 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
1460 hardcoding zero as the value if COND_EXPR is never true. For
1461 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
1462 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
1463 hardcoding MAX_EXPR as the reduction operation.
1464 (is_nonwrapping_integer_induction): Allow negative step.
1465 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
1466 vect_create_epilog_for_reduction, if no value is suitable, don't
1467 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
1469 2017-12-12 Richard Biener <rguenther@suse.de>
1471 PR tree-optimization/81889
1472 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
1473 range info from the non-wrapping IV instead of just the range
1476 2017-12-12 Julia Koval <julia.koval@intel.com>
1478 * config.gcc: Add vaesintrin.h.
1479 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1481 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1482 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1484 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1485 * config/i386/immintrin.h: Include vaesintrin.h.
1486 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1487 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1488 _mm_aesdec_epi128): New intrinsics.
1490 2017-12-12 Julia Koval <julia.koval@intel.com>
1492 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
1493 OPTION_MASK_ISA_VAES_UNSET): New.
1494 (ix86_handle_option): Handle -mvaes.
1495 * config/i386/cpuid.h: Define bit_VAES.
1496 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
1497 * config/i386/i386-c.c (__VAES__): New.
1498 * config/i386/i386.c (ix86_target_string): Add -mvaes.
1499 (ix86_valid_target_attribute_inner_p): Ditto.
1500 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
1501 * config/i386/i386.opt: Add -mvaes.
1502 * doc/invoke.texi: Ditto.
1504 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
1506 * debug.h (gcc_debug_hooks): Add inline_entry.
1507 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
1508 * debug.c (do_nothing_debug_hooks): Likewise.
1509 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
1510 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
1511 (dwarf2_lineno_debug_hooks): Likewise.
1513 * common.opt (gstatement-frontiers): New, setting
1514 debug_nonbind_markers_p.
1515 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
1516 * toplev.c (process_options): Autodetect value for debug statement
1518 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
1519 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
1521 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
1522 markers. Integrate source bind into debug stmt expand loop.
1523 (pass_expand::execute): Check debug marker limit. Avoid deep
1524 TER and expand debug locations for debug bind insns only.
1525 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
1527 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
1528 * final.c (reemit_insn_block_notes): Take current block from
1529 nonbind markers. Declare note where it's first set.
1530 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
1531 begin stmt markers if enabled.
1532 (notice_source_line): Handle nonbind markers. Fail if their
1533 location is unknown or that of builtins.
1534 (rest_of_handle_final): Convert begin stmt markers to notes if
1535 var-tracking didn't run.
1536 (rest_of_clean_state): Skip begin stmt markers.
1537 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
1539 * function.c (allocate_struct_function): Set begin_stmt_markers.
1540 * function.h (struct function): Add debug_marker_count counter
1541 and debug_nonbind_markers flag.
1542 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
1543 * gimple-low.c (lower_function_body): Adjust
1544 debug_nonbind_markers.
1545 (lower_stmt): Drop or skip gimple debug stmts.
1546 (lower_try_catch): Skip debug stmts.
1547 * gimple.c (gimple_build_debug_begin_stmt): New.
1548 (gimple_copy): Increment debug_marker_count if copying one.
1549 * gimple.h (gimple_build_debug_begin_stmt): Declare.
1550 * gimplify.c (rexpr_location): New.
1551 (rexpr_has_location): New.
1552 (warn_switch_unreachable_r): Handle gimple debug stmts.
1553 (shortcut_cond_r): Call expr_location.
1555 (find_goto_label): New.
1556 (shortcut_cond_expr): Call expr_has_location, expr_location, and
1558 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
1560 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
1561 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
1562 (LANG_HOOKS_INITIALIZER): ... this.
1563 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
1564 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
1566 (update_ebb_live_info): Skip debug insn markers.
1567 * lra.c (debug_insn_static_data): Rename to...
1568 (debug_bind_static_data): ... this.
1569 (debug_marker_static_data): New.
1570 (lra_set_insn_recog_data): Select one of the above depending
1572 (lra_update_isn_regno_info): Don't assume debug insns have
1574 (push_insns): Skip debug insns.
1575 * lto-streamer-in.c (input_function): Drop debug stmts
1576 depending on active options. Adjust debug_nonbind_markers.
1577 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
1578 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
1579 begin stmt marker notes.
1580 (print_insn): Likewise.
1581 * recog.c (extract_insn): Recognize rtl for debug markers.
1582 * rtl.def (DEBUG_MARKER): New.
1583 * tree-inline.c: Include params.h.
1584 (remap_gimple_stmt): Handle nonbind markers.
1585 (maybe_move_debug_stmts_to_successors): Likewise.
1586 (copy_debug_stmt): Likewise.
1587 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
1588 markers regardless of no side effects.
1589 (tsi_link_before): Don't update container's side effects when adding
1590 a begin stmt marker.
1591 (tsi_link_after): Likewise.
1592 (expr_first): Skip begin stmt markers.
1593 (expr_last): Likewise.
1594 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
1595 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
1596 Disregard nonbind markers.
1597 * tree.c (make_node_stat): Don't set side effects for begin stmt
1599 (build1_stat): Likewise.
1600 * tree.def (DEBUG_BEGIN_STMT): New.
1601 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
1602 * var-tracking.c (delete_debug_insns): Renamed to...
1603 (delete_vta_debug_insns): ... this.
1604 (reemit_marker_as_note): New.
1605 (vt_initialize): Reemit markers.
1606 (delete_vta_debug_insns): Likewise.
1607 (vt_debug_insns_local): Reemit or delete markers.
1608 (variable_tracking_main_1): Likewise.
1609 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
1610 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
1611 (gimple_debug_nonbind_marker_p): New.
1612 (gimple_build_debug_bind): Adjust.
1613 (gimple_build_debug_begin_stmt): New.
1614 * doc/invoke.texi (max-debug-marker-count): New param.
1615 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
1616 (debug_parameter_ref, debug_marker): New.
1617 (NOTE_INSN_BEGIN_STMT): New.
1618 (DEBUG_INSN): Describe begin stmt markers.
1620 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
1621 without debug insns, we wouldn't, but clean up debug insns
1622 after a control flow insn nevertheless.
1624 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
1625 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
1626 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
1627 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
1628 (rtl_tidy_fallthru_edge): Likewise.
1629 (rtl_verify_fallthru): Likewise.
1630 (rtl_verify_bb_layout): Likewise.
1631 (skip_insns_after_block): Likewise.
1632 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
1633 * dwarf2out.c: Include print-rtl.h.
1634 (dwarf2out_next_real_insn): New.
1635 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
1636 Dump debug binds in asm comments.
1637 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
1638 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
1639 callers to use gsi_start_nondebug_bb instead.
1640 (gsi_after_labels): Skip gimple debug stmts.
1641 (gsi_start_nondebug): New.
1642 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
1643 (proper_loop_form_for_interchange): Adjust.
1644 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
1645 * gimple.h (gimple_seq_last_nondebug_stmt): New.
1646 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
1647 (collect_fallthrough_labels): Likewise.
1648 (should_warn_for_implicit_fallthrough): Likewise.
1649 (warn_implicit_fallthrough_r): Likewise.
1650 (expand_FALLTHROUGH_r): Likewise.
1651 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
1652 (graphite_copy_stmts_from_block): Skip nonbind markers.
1653 * haifa-sched.c (sched_extend_bb): Skip debug insns.
1654 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
1655 * jump.c (clean_barriers): Skip debug insns.
1656 * omp-expand.c (expand_parallel_call): Skip debug insns.
1657 (expand_task_call): Likewise.
1658 (remove_exit_barrier): Likewise.
1659 (expand_omp_taskreg): Likewise.
1660 (expand_omp_for_init_counts): Likewise.
1661 (expand_omp_for_generic): Likewise.
1662 (expand_omp_for_static_nochunk): Likewise.
1663 (expand_omp_for_static_chunk): Likewise.
1664 (expand_omp_simd): Likewise.
1665 (expand_omp_taskloop_for_outer): Likewise.
1666 (expand_omp_taskloop_for_inner): Likewise.
1667 (expand_oacc_for): Likewise.
1668 (expand_omp_sections): Likewise.
1669 (expand_omp_single): Likewise.
1670 (expand_omp_synch): Likewise.
1671 (expand_omp_atomic_load): Likewise.
1672 (expand_omp_atomic_store): Likewise.
1673 (expand_omp_atomic_fetch_op): Likewise.
1674 (expand_omp_atomic_pipeline): Likewise.
1675 (expand_omp_atomic_mutex): Likewise.
1676 (expand_omp_target): Likewise.
1677 (grid_expand_omp_for_loop): Likewise.
1678 (grid_expand_target_grid_body): Likewise.
1679 (build_omp_regions_1): Likewise.
1680 * omp-low.c (check_combined_parallel): Skip debug stmts.
1681 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
1682 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
1683 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
1685 * tree-cfg.c (make_blobs_1): Skip debug stmts.
1686 (make_edges): Likewise.
1687 (cleanup_dead_labels): Likewise.
1688 (gimple_can_merge_blocks_p): Likewise.
1689 (stmt_starts_bb_p): Likewise.
1690 (gimple_block_label): Likewise.
1691 (gimple_redirect_edge_and_branch): Likewise.
1692 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
1694 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
1696 * tree-pretty-print (print_declaration): Omit initializer in slim
1698 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
1700 (eliminate_unnecessary_stmts): Stabilize block removal order.
1701 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
1702 * var-tracking.c (get_first_insn): New.
1703 (vt_emit_notes): Call it.
1704 (vt_initialize): Walk any insns before the first BB.
1705 (delete_debug_insns): Likewise.
1707 * gimple.h (enum gimple_debug_subcode): Add
1708 GIMPLE_DEBUG_BEGIN_STMT.
1709 (gimple_debug_begin_stmt_p): New.
1710 (gimple_debug_nonbind_marker_p): New.
1711 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
1712 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
1713 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
1714 * insn-notes.def (BEGIN_STMT): New.
1715 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
1716 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
1717 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
1718 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
1719 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
1720 (INSN_DEBUG_MARKER_KIND): New.
1721 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
1722 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
1723 (INSN_VAR_LOCATION_PTR): New.
1724 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
1726 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
1727 deps only if debug bind insns are enabled.
1728 (pass_expand::execute): Avoid deep TER and expand
1729 debug locations for debug bind insns only.
1730 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
1731 debug stmts special handling down to debug bind stmts.
1732 * combine.c (try_combine): Narrow debug insns special handling
1733 down to debug bind insns.
1734 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
1735 Narrow debug insns preexisting special handling down to debug
1737 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
1738 handling down to debug bind insns.
1739 * function.c (instantiate_virtual_regs): Skip debug markers,
1740 adjust handling of debug binds.
1741 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
1742 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
1743 * haifa-sched.c (schedule_insn): Narrow special handling of debug
1744 insns to debug bind insns.
1745 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
1746 special handling of debug stmts to debug bind stmts.
1747 * ipa-split.c (split_function): Likewise.
1748 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
1749 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
1751 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
1752 * regrename.c (build_def_use): Likewise.
1753 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
1754 (pass_cprop_hardreg): Narrow special casing of debug insns to
1756 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
1757 * reload1.c (reload): Likewise.
1758 * sese.c (sese_insert_phis_for_liveouts): Narrow special
1759 casing of debug stmts to debug bind stmts.
1760 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
1761 * ssa-iterators.h (num_imm_uses): Likewise.
1762 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
1763 debug stmts to debug bind stmts.
1764 * tree-inline.c (tree_function_versioning): Narrow special casing
1765 of debug stmts to debug bind stmts.
1766 * tree-loop-distribution.c (generate_loops_for_partition):
1767 Narrow special casing of debug stmts to debug bind stmts.
1768 * tree-sra.c (analyze_access_subtree): Narrow special casing
1769 of debug stmts to debug bind stmts.
1770 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
1771 stmts to debug bind stmts.
1772 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
1773 casing of debug stmts to debug bind stmts.
1774 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
1775 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
1776 casing of debug stmts to debug bind stmts.
1777 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
1779 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
1780 debug stmts to debug bind stmts.
1781 (gimple_replace_ssa_lhs): Likewise.
1782 (insert_debug_temp_for_var_def): Likewise.
1783 (insert_debug_temps_for_defs): Likewise.
1784 (reset_debug_uses): Likewise.
1785 * tree-ssanames.c (release_ssa_name_fn): Likewise.
1786 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
1787 (adjust_debug_stmts): Likewise.
1788 (adjust_phi_and_debug_stmts): Likewise.
1789 (vect_do_peeling): Likewise.
1790 * tree-vect-loop.c (vect_transform_loop): Likewise.
1791 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
1792 * var-tracking.c (adjust_mems): Narrow special casing of debug
1793 insns to debug bind insns.
1794 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
1795 (compute_bb_dataflow, vt_find_locations): Likewise.
1796 (vt_expand_loc, emit_notes_for_changes): Likewise.
1797 (vt_init_cfa_base): Likewise.
1798 (vt_emit_notes): Likewise.
1799 (vt_initialize): Likewise.
1800 (vt_finalize): Likewise.
1802 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
1803 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
1804 (next_nonnote_nondebug_insn_bb): New.
1805 (prev_nonnote_nondebug_insn_bb): New.
1806 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
1807 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
1808 (prev_nonnote_nondebug_insn_bb): Declare.
1809 (next_nonnote_nondebug_insn_bb): Declare.
1810 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
1811 * cfgrtl.c (get_last_bb_insn): Likewise.
1812 * lra.c (push_insns): Likewise.
1814 2017-12-11 David Malcolm <dmalcolm@redhat.com>
1817 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
1819 * selftest.c (selftest::test_runner::test_runner): New ctor.
1820 (selftest::test_runner::~test_runner): New dtor.
1821 * selftest.h (class selftest::test_runner): New class.
1823 2017-12-11 Carl Love <cel@us.ibm.com>
1825 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
1826 vec_extract_fp32_from_shortl]): Add #defines.
1827 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
1828 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
1829 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
1830 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
1831 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
1832 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
1833 * doc/extend.texi: Add documentation for the added builtins.
1835 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
1837 PR rtl-optimization/80693
1838 PR rtl-optimization/81019
1839 PR rtl-optimization/81020
1840 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
1841 are not mentioned in i3. Place the REG_UNUSED note on i2,
1842 possibly modified to REG_DEAD, if it did not originate in i3.
1844 2017-12-11 Jakub Jelinek <jakub@redhat.com>
1846 * recog.c (store_data_bypass_p_1): New function.
1847 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
1848 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
1850 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
1852 PR rtl-optimization/83361
1853 * ifcvt.c (if_convert): Call fixup_partitions.
1855 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
1857 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1858 early folding of splat_u{8,16,32}.
1860 2017-12-11 Jakub Jelinek <jakub@redhat.com>
1862 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
1863 output_operand_lossage first argument with capital letter.
1864 (aarch64_override_options): Don't start error and sorry first argument
1865 with capital letter.
1867 2017-12-11 Andi Kleen <ak@linux.intel.com>
1869 PR gcov-profile/83355
1870 * auto-profile.c (string_table::get_index_by_decl): Don't
1871 recurse when abstract origin points to itself.
1873 2017-12-11 Bin Cheng <bin.cheng@arm.com>
1875 PR tree-optimization/83320
1876 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
1877 (prune_datarefs_not_in_loop): Ditto.
1879 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
1881 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
1883 2017-12-10 Jakub Jelinek <jakub@redhat.com>
1885 PR tree-optimization/83337
1886 * gimple-loop-interchange.cc (compute_access_stride): Handle
1887 bitfield DRs properly.
1889 2017-12-09 Jakub Jelinek <jakub@redhat.com>
1891 PR tree-optimization/83338
1892 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
1893 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
1894 vector of unsigned integers to vector of signed integers.
1896 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
1898 PR rtl-optimization/83317
1899 * lra-constraints.c (process_address_1): Add insn code check.
1901 2017-12-08 Michael Matz <matz@suse.de>
1903 Fix PR tree-optimization/83323
1904 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
1905 head-controlled loops and loop BBs.
1906 * common.opt (funroll-and-jam): Remove, instead ...
1907 (floop-unroll-and-jam): ... reuse this option.
1908 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
1909 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
1910 (-floop-unroll-and-jam): ... this option.
1912 2017-12-08 Jakub Jelinek <jakub@redhat.com>
1914 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
1915 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
1916 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
1918 2017-12-08 Richard Biener <rguenther@suse.de>
1921 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
1922 handle accesses outside of zero-sized vars.
1924 2017-12-08 Martin Jambor <mjambor@suse.cz>
1926 PR tree-optimization/83141
1927 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
1928 test for MEM_REFs implicitely changing types with padding. Remove
1930 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
1932 2017-12-08 Jakub Jelinek <jakub@redhat.com>
1934 * config/arc/arc.c (arc_attribute_table): Add exclusions to
1936 * config/avr/avr.c (avr_attribute_table): Likewise.
1937 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
1938 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
1939 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
1940 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
1941 of struct attribute_spec.
1942 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1944 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
1947 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
1949 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
1952 * profile-count.c (profile_count::from_gcov_type): Move from
1953 profile-count.h; handle overflow.
1954 * profile-count.h (profile_count::from_gcov_type): Move offline.
1956 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
1958 PR rtl-optimization/83304
1959 * combine.c (move_deaths): If we do not know where a register died,
1962 2017-12-08 Richard Biener <rguenther@suse.de>
1964 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
1965 Provide -fopt-info-loop feedback when we interchange in a nest.
1967 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1969 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
1970 for armv6 ARM CPU IDs.
1972 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1974 * common/config/arm/arm-common.c: Include <algorithm>.
1975 (INCLUDE_VECTOR): Define.
1976 (compare_opt_names): New function.
1977 (arm_rewrite_selected_arch): Only strip out extensions that can be
1978 expressed through -mfpu. Sort the remaining extensions
1981 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1983 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
1984 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
1985 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
1986 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
1988 2017-06-08 Tristan Gingold <gindold@adacore.com>
1991 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
1992 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
1994 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1996 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
1997 of struct attribute_spec.
1999 2017-12-08 Julia Koval <julia.koval@intel.com>
2001 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
2002 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
2003 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
2004 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
2005 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
2006 _mm_maskz_dpwssds_epi32): Ditto.
2008 2017-12-08 Richard Biener <rguenther@suse.de>
2010 PR tree-optimization/81303
2011 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
2012 conditions try to create a comparison vector type matching
2013 the data vector type.
2014 (vectorizable_condition): Adjust.
2015 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
2016 Leave invariant conditions alone in case we can vectorize those.
2018 2017-12-08 Julia Koval <julia.koval@intel.com>
2020 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
2021 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
2022 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
2023 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
2024 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
2026 2017-12-08 Julia Koval <julia.koval@intel.com>
2028 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
2029 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
2030 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
2031 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
2032 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
2033 _mm_maskz_dpbusds_epi32): New intrinsics.
2035 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
2037 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
2039 (ld<bh>io_signed): Likewise.
2040 (st<bhw>io): Likewise.
2041 * config/nios2/predicates.md (ldstio_memory_operand): Allow
2042 SMALL_INT12 constant integer operand.
2044 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
2046 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
2048 (nios2_large_constant_allowed): ...this. Adjust uses.
2049 (nios2_plus_symbolic_constant_p): Rename to...
2050 (nios2_plus_large_constant_p): ...this. Adjust uses.
2051 (nios2_legitimate_address_p): Correct CONST_INT handling.
2052 (nios2_symbolic_memory_operand_p): Rename to...
2053 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
2054 (nios2_large_constant_p): Check for large constant integers too.
2055 (nios2_split_large_constant): Handle constant integers.
2056 (nios2_split_symbolic_memory_operand): Rename to...
2057 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
2058 (nios2_legitimize_constant_address): Handle constant integers.
2059 (r0rel_constant_p): Handle small constant integers.
2060 (nios2_print_operand_address): Handle r0-relative integer addresses.
2061 * config/nios2/nios2-protos.h: Adjust for renamed functions.
2062 * config/nios2/nios2.md: Adjust for renamed functions.
2064 2017-12-07 Andrew Waterman <andrew@sifive.com>
2066 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
2067 (TARGET_HAVE_SRODATA_SECTION): New define.
2068 (riscv_select_section): New function.
2070 2017-12-08 Joseph Myers <joseph@codesourcery.com>
2071 Alexander Monakov <amonakov@ispras.ru>
2072 Jakub Jelinek <jakub@redhat.com>
2075 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
2077 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2079 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
2080 a bool success value. Don't call output_operand_lossage here.
2081 (aarch64_print_ldpstp_address): Return a bool success value.
2082 (aarch64_print_operand_address): Call output_addr_const if
2083 aarch64_print_address_internal fails.
2084 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
2085 'y'; call output_operand_lossage instead. Call output_operand_lossage
2086 if aarch64_print_ldpstp_address fails.
2088 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2090 * tree-vector-builder.h
2091 (tree_vector_builder::binary_encoded_nelts): Declare.
2092 * tree-vector-builder.c
2093 (tree_vector_builder::binary_encoded_nelts): New function.
2094 * fold-const.c (negate_expr_p): Likewise.
2095 (operand_equal_p, fold_checksum_tree): Likewise.
2096 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
2097 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
2098 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
2099 (uniform_vector_p): Likewise.
2100 * varasm.c (const_hash_1, compare_constant): Likewise.
2101 * tree-ssa-ccp.c: Include tree-vector-builder.h.
2102 (valid_lattice_transition): Operate directly on the VECTOR_CST
2104 * ipa-icf.c: Include tree-vector-builder.h.
2105 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
2106 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
2108 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2110 * tree.c (build_vector): Delete.
2111 * tree.h (build_vector): Make static and move into the self-testing
2114 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2116 * vector-builder.h (vector_builder::derived): New const overload.
2117 (vector_builder::elt): New function.
2118 * tree-vector-builder.h (tree_vector_builder::type): New function.
2119 (tree_vector_builder::apply_step): Declare.
2120 * tree-vector-builder.c (tree_vector_builder::apply_step): New
2122 * gimple-fold.h (tree_vector_builder): Declare.
2123 (gimple_build_vector): Take a tree_vector_builder instead of a
2124 type and vector of elements.
2125 * gimple-fold.c (gimple_build_vector): Likewise.
2126 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
2128 (get_initial_defs_for_reduction): Likewise.
2129 (vectorizable_induction): Likewise.
2131 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2133 * tree-vector-builder.h
2134 (tree_vector_builder::new_binary_operation): Declare.
2135 * tree-vector-builder.c
2136 (tree_vector_builder::new_binary_operation): New function.
2137 * fold-const.c (fold_relational_const): Use it.
2138 (const_binop): Likewise. Check that both input vectors have
2139 the same number of elements, thus excluding things like WIDEN_SUM.
2140 Check whether it is possible to operate directly on the encodings
2143 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2145 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
2146 new_unary_operation, operating only on the encoded elements.
2147 (const_unop): Likewise.
2148 (exact_inverse): Likewise.
2149 (distributes_over_addition_p): New function.
2150 (const_binop): Use tree_vector_builder and new_unary_operation
2151 for combinations of VECTOR_CST and INTEGER_CST. Operate only
2152 on the encoded elements unless the encoding is strided and the
2153 operation does not distribute over addition.
2154 (fold_convert_const): Use tree_vector_builder and
2155 new_unary_operation. Operate only on the encoded elements
2156 for truncating integer conversions, or for non-stepped encodings.
2158 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2160 * config/sparc/sparc.c: Include tree-vector-builder.h.
2161 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
2162 * expmed.c: Include tree-vector-builder.h.
2163 (make_tree): Use tree_vector_builder instead of build_vector.
2164 * fold-const.c: Include tree-vector-builder.h.
2165 (const_binop): Use tree_vector_builder instead of build_vector.
2166 (const_unop): Likewise.
2167 (native_interpret_vector): Likewise.
2168 (fold_vec_perm): Likewise.
2169 (fold_ternary_loc): Likewise.
2170 * gimple-fold.c: Include tree-vector-builder.h.
2171 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
2173 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
2174 (simplify_vector_constructor): Use tree_vector_builder instead
2176 * tree-vect-generic.c: Include tree-vector-builder.h.
2177 (add_rshift): Use tree_vector_builder instead of build_vector.
2178 (expand_vector_divmod): Likewise.
2179 (optimize_vector_constructor): Likewise.
2180 * tree-vect-loop.c: Include tree-vector-builder.h.
2181 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
2182 of build_vector. Explicitly use a stepped encoding for
2184 * tree-vect-slp.c: Include tree-vector-builder.h.
2185 (vect_get_constant_vectors): Use tree_vector_builder instead
2187 (vect_transform_slp_perm_load): Likewise.
2188 (vect_schedule_slp_instance): Likewise.
2189 * tree-vect-stmts.c: Include tree-vector-builder.h.
2190 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
2191 (vect_gen_perm_mask_any): Likewise.
2192 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
2193 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
2195 (build_vector_from_val): Likewise. Explicitly use a duplicate
2198 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
2200 * doc/generic.texi (VECTOR_CST): Describe new representation of
2202 * vector-builder.h: New file.
2203 * tree-vector-builder.h: Likewise.
2204 * tree-vector-builder.c: Likewise.
2205 * Makefile.in (OBJS): Add tree-vector-builder.o.
2206 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
2207 * tree-core.h (tree_base): Add a vector_cst field to the u union.
2208 (tree_vector): Change the number of elements to
2209 vector_cst_encoded_nelts.
2210 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
2211 (VECTOR_CST_ELTS): Delete.
2212 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
2213 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
2214 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
2215 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
2216 (VECTOR_CST_ENCODED_ELT): Likewise.
2217 (vector_cst_encoded_nelts): New function.
2218 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
2219 VECTOR_CST_NELTS_PER_PATTERN as arguments.
2220 (vector_cst_int_elt, vector_cst_elt): Declare.
2221 * tree.c: Include tree-vector-builder.h.
2222 (tree_code_size): Abort if passed VECTOR_CST.
2223 (tree_size): Update for new VECTOR_CST layout.
2224 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
2225 VECTOR_CST_NELTS_PER_PATTERN as arguments.
2226 (build_vector): Use tree_vector_builder.
2227 (vector_cst_int_elt, vector_cst_elt): New functions.
2228 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
2229 encoded elements and then create the vector in the canonical form.
2230 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
2231 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
2232 (tree_c_tests): Call test_vector_cst_patterns.
2233 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
2235 (hash_tree): Likewise.
2236 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
2237 (streamer_write_tree_header): Likewise.
2238 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
2239 (streamer_alloc_tree): Likewise. Update call to make_vector.
2240 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
2242 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2244 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
2245 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
2246 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
2248 2017-12-07 Bin Cheng <bin.cheng@arm.com>
2249 Richard Biener <rguenther@suse.de>
2251 PR tree-optimization/81303
2252 * Makefile.in (gimple-loop-interchange.o): New object file.
2253 * common.opt (floop-interchange): Reuse the option from graphite.
2254 * doc/invoke.texi (-floop-interchange): Ditto. New document for
2255 -floop-interchange and mention it for -O3.
2256 * opts.c (default_options_table): Enable -floop-interchange at -O3.
2257 * gimple-loop-interchange.cc: New file.
2258 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
2259 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
2260 * passes.def (pass_linterchange): New pass.
2261 * timevar.def (TV_LINTERCHANGE): New time var.
2262 * tree-pass.h (make_pass_linterchange): New declaration.
2263 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
2264 interchange. Record IV before/after increment in new parameters.
2265 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
2266 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
2268 (check_reduction_path): ...New function here.
2269 * tree-vectorizer.h (check_reduction_path): New declaration.
2271 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
2274 PR rtl-optimization/80818
2275 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
2276 always early clobbered.
2277 * lra-lives.c (process_bb_lives): Check input hard regs for early
2278 clobbered non-operand hard reg.
2280 2017-12-07 Jakub Jelinek <jakub@redhat.com>
2283 * tree-cfg.c (verify_gimple_assign_binary): Don't require
2284 types_compatible_p, just that TYPE_MODE is the same.
2286 2017-12-07 Martin Sebor <msebor@redhat.com>
2289 * attribs.c (empty_attribute_table): Initialize new member of
2290 struct attribute_spec.
2291 (decl_attributes): Add argument. Handle mutually exclusive
2292 combinations of attributes.
2293 (selftests::test_attribute_exclusions): New function.
2294 (selftests::attribute_c_tests): Ditto.
2295 * attribs.h (decl_attributes): Add default argument.
2296 * selftest.h (attribute_c_tests): Declare.
2297 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
2298 * tree-core.h (attribute_spec::exclusions, exclude): New type and
2300 * doc/extend.texi (Common Function Attributes): Update const and pure.
2301 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
2302 of struct attribute_spec.
2303 * config/arc/arc.c (arc_attribute_table): Same.
2304 * config/arm/arm.c (arm_attribute_table): Same.
2305 * config/avr/avr.c ( avr_attribute_table): Same.
2306 * config/bfin/bfin.c (bfin_attribute_table): Same.
2307 * config/cr16/cr16.c (cr16_attribute_table): Same.
2308 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
2309 * config/h8300/h8300.c (h8300_attribute_table): Same.
2310 * config/i386/i386.c (ix86_attribute_table): Same.
2311 * config/ia64/ia64.c (ia64_attribute_table): Same.
2312 * config/m32c/m32c.c (m32c_attribute_table): Same.
2313 * config/m32r/m32r.c (m32r_attribute_table): Same.
2314 * config/m68k/m68k.c (m68k_attribute_table): Same.
2315 * config/mcore/mcore.c (mcore_attribute_table): Same.
2316 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
2317 * config/mips/mips.c (mips_attribute_table): Same.
2318 * config/msp430/msp430.c (msp430_attribute_table): Same.
2319 * config/nds32/nds32.c (nds32_attribute_table): Same.
2320 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
2321 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
2322 * config/rl78/rl78.c (rl78__attribute_table): Same.
2323 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
2324 * onfig/rx/rx.c (rx_attribute_table): Same.
2325 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
2326 * config/sh/sh.c (sh_attribute_table): Same.
2327 * config/sparc/sparc.c (sparc_attribute_table): Same.
2328 * config/spu/spu.c (spu_attribute_table): Same.
2329 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
2330 * config/v850/v850.c (v850_attribute_table): Same.
2331 * config/visium/visium.c (visium_attribute_table): Same.
2333 2017-12-07 Tamar Christina <tamar.christina@arm.com>
2336 * config/arm/arm.c (INCLUDE_STRING): Define.
2337 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
2338 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
2341 2017-12-07 Michael Matz <matz@suse.de>
2343 Add unroll and jam pass
2345 * gimple-loop-jam.c: New file.
2346 * Makefile.in (OBJS): Add gimple-loop-jam.o.
2347 * common.opt (funroll-and-jam): New option.
2348 * opts.c (default_options_table): Add unroll-and-jam at -O3.
2349 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
2350 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
2351 * passes.def: Add pass_loop_jam.
2352 * timevar.def (TV_LOOP_JAM): Add.
2353 * tree-pass.h (make_pass_loop_jam): Declare.
2354 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
2355 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
2356 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
2357 to flow_loop_tree_node_add.
2358 (duplicate_subloops, copy_loops_to): Append to sibling list.
2359 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
2360 * doc/invoke.texi (-funroll-and-jam): Document new option.
2361 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
2363 2017-12-07 Richard Biener <rguenther@suse.de>
2365 PR tree-optimization/83296
2366 PR tree-optimization/67769
2367 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
2368 flow sensitive info in an unrelated BB.
2369 (value_replacement): Use reset_flow_sensitive_info.
2370 (minmax_replacement): Reset flow sensitive info on the def
2371 we move. Do not reset flow sensitive info in the whole BB
2372 we move the stmt to.
2373 (abs_replacement): Likewise.
2375 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2378 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2379 rs6000_cpu to the given -mcpu=, or to the default processor.
2381 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2383 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
2384 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
2385 instead of rs6000_cpu_attr.
2386 (rs6000_adjust_cost): Ditto.
2387 (is_microcoded_insn): Ditto.
2388 (rs6000_adjust_priority): Ditto.
2389 (rs6000_issue_rate): Ditto.
2390 (rs6000_use_sched_lookahead): Ditto.
2391 (rs6000_use_sched_lookahead_guard): Ditto.
2392 (rs6000_sched_reorder): Ditto.
2393 (force_new_group): Ditto.
2394 * config/rs6000/rs6000.md (cpu attribute): Ditto.
2395 (group_ending_nop): Ditto.
2397 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2399 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
2400 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
2401 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
2402 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
2404 (rs6000_reassociation_width): Ditto.
2405 (rs6000_emit_epilogue): Ditto.
2406 (rs6000_adjust_cost): Ditto.
2407 (is_microcoded_insn): Ditto.
2408 (is_cracked_insn): Ditto.
2409 (rs6000_adjust_priority): Ditto.
2410 (rs6000_sched_reorder): Ditto.
2411 (rs6000_sched_reorder2): Ditto.
2412 (insn_must_be_first_in_group): Ditto.
2413 (insn_must_be_last_in_group): Ditto.
2414 (rs6000_register_move_cost): Ditto.
2415 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
2418 2017-12-07 Julia Koval <julia.koval@intel.com>
2420 * config.gcc: Add vaesintrin.h.
2421 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2423 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2424 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2426 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2427 * config/i386/immintrin.h: Include vaesintrin.h.
2428 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2429 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2430 _mm_aesdec_epi128): New intrinsics.
2432 2017-12-06 David Malcolm <dmalcolm@redhat.com>
2434 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
2435 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
2436 c-family/c-spellcheck.cc.
2437 (best_macro_match::best_macro_match): Likewise.
2439 (struct edit_distance_traits<cpp_hashnode *>): Move to
2440 c-family/c-spellcheck.h.
2441 (class best_macro_match): Likewise.
2443 2017-12-06 Jakub Jelinek <jakub@redhat.com>
2445 PR tree-optimization/83293
2446 * gimple-ssa-strength-reduction.c (insert_initializers): Use
2447 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
2448 might insert into empty bb.
2451 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
2452 simplify for plus with :c added, and pointer_plus without that.
2453 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
2454 with undefined overflow and the conversion is not widening,
2455 perform negation in utype and only convert to type afterwards.
2456 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
2457 simplify for plus with :c added, and pointer_plus without that.
2458 If type is integral with undefined overflow and the conversion is
2459 not widening, perform minus in utype and only convert to type
2460 afterwards. Move the last pointer_diff_expr simplify into the
2463 2017-12-06 Martin Sebor <msebor@redhat.com>
2465 PR tree-optimization/82646
2466 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
2467 strncpy, not maxlen.
2469 2017-12-06 Martin Sebor <msebor@redhat.com>
2471 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
2474 PR tree-optimization/83075
2475 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
2476 strncat/strncpy don't change length of source string.
2478 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
2481 2017-11-29 Martin Aberg <maberg@gaisler.com>
2483 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
2484 to prevent b2bst errata sequence.
2485 (sqrtdf2_fix): Likewise.
2487 2017-12-06 Jakub Jelinek <jakub@redhat.com>
2489 PR tree-optimization/81945
2490 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
2491 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
2492 to dest_cfun has orig_loop_num set, either remap it to the new
2493 loop number if the loop got moved too, or clear it.
2495 2017-12-05 Steve Ellcey <sellcey@cavium.com>
2497 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
2499 (thunderx2t99_nothing): New insn reservation.
2500 (thunderx2t99_mrs): New insn reservation.
2501 (thunderx2t99_multiple): New insn reservation.
2502 (thunderx2t99_alu_basi): Add bfx to reservation.
2503 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
2505 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2508 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
2510 2017-12-05 Bin Cheng <bin.cheng@arm.com>
2512 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
2513 tree-ssa-pre.c::remove_dead_inserted_code.
2514 * tree-ssa-dce.h: New file.
2515 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
2516 (remove_dead_inserted_code): Move and rename to function
2517 tree-ssa-dce.c::simple_dce_from_worklist.
2518 (pass_pre::execute): Update use.
2520 2017-12-05 Richard Biener <rguenther@suse.de>
2522 PR tree-optimization/83277
2523 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
2524 to code-gen liveout vars.
2526 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
2528 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
2529 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
2530 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
2533 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
2535 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
2537 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
2538 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
2541 2017-12-05 Richard Biener <rguenther@suse.de>
2543 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
2544 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
2545 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
2546 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
2548 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2550 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
2552 2017-12-05 Martin Liska <mliska@suse.cz>
2553 Jakub Jelinek <jakub@redhat.com>
2555 * doc/invoke.texi: Document the options.
2556 * flag-types.h (enum sanitize_code): Add
2557 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
2558 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
2559 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
2560 * opts.c: Define new sanitizer options.
2561 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
2562 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
2564 2017-12-05 Julia Koval <julia.koval@intel.com>
2566 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
2567 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
2568 (ix86_handle_option): Handle -mavx512vnni.
2569 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
2570 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
2571 * config/i386/i386-c (__AVX512VNNI__): New.
2572 * config/i386/i386.c (ix86_target_string): Handle new option.
2573 (ix86_valid_target_attribute_inner_p): Handle new option.
2574 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
2575 * config/i386/i386.opt (mavx512vnni): New option.
2577 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
2580 * athlon.md: Disable for generic.
2581 * haswell.md: Enable for generic.
2582 * i386.c (ix86_sched_init_global): Add core hooks for generic.
2583 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
2585 (ix86_adjust_cost): Move generic to haswell path.
2587 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
2589 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
2590 instead of MEM_P in a couple more places. Fix formatting issues.
2592 2017-12-04 Jim Wilson <jimw@sifive.com>
2594 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
2595 instead of GP_REG_LAST-1.
2596 (riscv_adjust_libcall_cfi_prologue): Likewise.
2597 (riscv_adjust_libcall_cri_epilogue): Likewise.
2598 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
2601 2017-12-04 Luis Machado <luis.machado@linaro.org>
2603 * ipa-pure-const.c (check_decl): Add missing newline.
2604 (state_from_flags): Likewise.
2606 2017-12-04 Jeff Law <law@redhat.com>
2608 PR tree-optimizatin/78496
2609 * gimple-ssa-evrp-analyze.h
2610 (evrp_range_analyzer::get_vr_values): Simplify.
2611 * gimple-ssa-evrp-analyze.c: Corresponding changes.
2612 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
2613 and gimple-ssa-evrp-analyze.h.
2614 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
2615 (simplify_stmt_for_jump_threading): Copy a blob of code from
2616 tree-vrp.c to use ranges to simplify statements.
2617 (dom_opt_dom_walker::before_dom_children): Call
2618 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
2619 (dom_opt_dom_walker::after_dom_children): Similarly for
2620 evrp_range_analyzer::leave.
2621 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
2624 * gimple-ssa-evrp-analyze.c
2625 (evrp_range_analyzer::extract_range_from_stmt): Always use
2626 vr_values::update_value_range so preexisting range info is
2627 medged with new range info, even if the new range is VR_VARYING.
2629 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
2631 * combine.c: Adjust comment.
2632 (use_crosses_set_p): Delete.
2633 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
2634 (try_combine): Ditto.
2636 2017-12-04 Richard Biener <rguenther@suse.de>
2638 PR tree-optimization/83255
2639 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
2640 Re-add zero-iteration check.
2642 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
2644 PR rtl-optimization/83245
2645 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
2646 hard registers as earlyclobber, also if not in an asm.
2648 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
2652 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2655 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2656 rs6000_cpu based on cpu_index, not tune_index.
2658 2017-12-04 Richard Biener <rguenther@suse.de>
2660 PR tree-optimization/83238
2661 * graphite-scop-detection.c (scop_detection::merge_sese): Make
2662 code match comment, rejecting invalid SESE regions.
2664 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
2666 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
2667 require base operand is a REG_POINTER prior to reload on targets
2668 with non-equivalent space registers.
2670 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
2672 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
2673 (update_specialized_profile): Fix updating of counts.
2674 (perhaps_add_new_callers): Likewise.
2676 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
2679 * x86-tune.def: Remove obsolette FIXMEs.
2680 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
2681 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
2682 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
2684 (X86_TUNE_PAD_RETURNS): Disable for generic.
2686 2017-12-02 Jakub Jelinek <jakub@redhat.com>
2688 PR tree-optimization/83170
2689 PR tree-optimization/83241
2690 * gimple-ssa-store-merging.c
2691 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
2692 gimple_vuse (ins_stmt) in case it has changed.
2693 (imm_store_chain_info::output_merged_store): Likewise.
2695 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
2699 * tree-cfg.c (pass_warn_function_return::execute): Handle
2700 __builtin_ubsan_handle_missing_return like __builtin_unreachable
2701 with BUILTINS_LOCATION.
2705 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
2706 is BLKmode for vector field with vector raw mode, use TYPE_MODE
2707 instead of DECL_MODE.
2709 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
2710 last argument to rtx pointer.
2711 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
2712 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
2713 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
2714 depending on the chosen ISAs.
2715 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
2716 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
2717 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
2719 * config/i386/sse.md (mov<mode>_internal): Likewise.
2720 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
2722 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2724 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
2726 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2728 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
2729 parameters from prototype.
2730 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
2731 parameters. Don't print a comment.
2732 (emit_fusion_gpr_load): Adjust.
2733 (emit_fusion_load_store): Adjust.
2734 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
2735 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
2736 comment on the second line.
2738 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2741 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2742 rs6000_cpu based on cpu_index, not tune_index.
2744 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2746 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
2747 which_alternative instead of which_alternative + 1.
2748 (output_asm_insn): Print an extra tab if the template is short.
2750 2017-12-01 Jim Wilson <jimw@sifive.com>
2752 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
2754 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
2755 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
2757 * doc/tm.texi: Regenerate.
2759 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
2762 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
2763 whether we can allocate pseudos before trying to fix an address.
2764 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
2765 memory address is indexed or indirect.
2766 (floatuns_<mode>si2_hw2): Likewise.
2768 2017-12-01 Jason Merrill <jason@redhat.com>
2770 * Makefile.in (TAGS): Add c-family/*.cc.
2772 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
2774 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
2775 (call_value_insn): Likewise.
2776 (sibcall_insn): Likewise.
2777 (sibcall_value_insn): Likewise.
2778 (movsi_aarch64): Likewise.
2779 (movdi_aarch64): Likewise.
2780 (add_losym_): Likewise.
2781 (ldr_got_small_): Likewise.
2782 (ldr_got_small_sidi): Likewise.
2783 (ldr_got_small_28k_): Likewise.
2784 (ldr_got_small_28k_sidi): Likewise.
2785 * config/aarch64/aarch64.c (aarch64_print_address_internal):
2786 Move output_addr_const to symbolic case. Add error check.
2788 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2790 * config/s390/predicates.md (plus16_Q_operand): New predicate.
2791 * config/s390/s390.md: Disable MVC merging peephole if it would
2792 disable operand forwarding.
2793 (new peephole2): Split MVCs if it would turn them into up to 2
2796 2017-12-01 Richard Biener <rguenther@suse.de>
2798 PR tree-optimization/83232
2799 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
2800 detection of same access. Instead of breaking the group here
2801 do not consider the duplicate. Add comment explaining real fix.
2803 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
2805 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
2807 2017-12-01 Sudakshina Das <sudi.das@arm.com>
2809 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
2811 2017-12-01 Jakub Jelinek <jakub@redhat.com>
2813 * function.h (struct function): Remove cilk_frame_decl,
2814 is_cilk_function and calls_cilk_spawn fields.
2815 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
2817 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
2818 cilk_elemental field.
2819 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
2820 * target.def: Adjust comment.
2821 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
2822 Don't test cilk_elemental.
2824 PR tree-optimization/83233
2825 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
2826 bswap_stat name for the struct.
2829 * tree.h (SWITCH_BREAK_LABEL_P): Define.
2830 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
2831 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
2832 SWITCH_BREAK_LABEL_P set on the label.
2833 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
2834 added for default case if it was missing and not all cases covered.
2835 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
2836 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
2838 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
2839 to avoid -Wimplicit-fallthrough warning.
2840 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
2841 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
2843 PR tree-optimization/83221
2844 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
2846 (init_reassoc): Formatting fix.
2849 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
2850 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
2854 * tree-cfg.c: Include asan.h.
2855 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
2857 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2859 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
2861 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2863 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
2864 vpcmpeqd instruction.
2866 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2868 * config/i386/i386.c (standard_sse_constant_opcode): Fix
2869 registers type for 128bit mode.
2871 2017-11-30 Jakub Jelinek <jakub@redhat.com>
2873 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
2875 * gengtype-state.c (read_state_pair): Likewise.
2876 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
2877 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
2878 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
2879 * ifcvt.c (noce_try_store_flag_constants): Likewise.
2880 * tree-ssa-ccp.c (ccp_finalize): Likewise.
2881 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
2882 * builtins.c (fold_builtin_3): Likewise.
2883 * graphite-scop-detection.c
2884 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
2885 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
2888 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
2891 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
2892 parameter. Return true if ignore_decl_rtl_set_p is true and other
2893 conditions are satisfied.
2894 * asan.h (asan_protect_global): Add new parameter.
2895 * varasm.c (categorize_decl_for_section): Pass true as second parameter
2896 to asan_protect_global calls.
2898 2017-11-30 Jim Wilson <jimw@sifive.com>
2900 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
2901 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
2902 -mstrict-align, add info on default value. Delete redundant lines for
2903 -mabi. Add missing -mexplicit-relocs docs.
2905 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2907 * config/arc/arc.md (trap): New pattern.
2909 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2911 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
2912 instruction to end into a delay slot.
2913 * config/arc/arc.md (cond_delay_insn): Check if the instruction
2914 can be placed into a delay slot against reg_note.
2916 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2918 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
2919 labels number of usages.
2921 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2923 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
2925 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
2927 2017-11-30 Jakub Jelinek <jakub@redhat.com>
2930 * internal-fn.c (expand_mul_overflow): Optimize unsigned
2931 multiplication by power of 2 constant into two shifts + comparison.
2933 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
2936 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
2938 2017-11-30 Richard Biener <rguenther@suse.de>
2940 PR tree-optimization/83202
2941 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2942 allow_peel argument and guard peeling.
2943 (canonicalize_loop_induction_variables): Likewise.
2944 (canonicalize_induction_variables): Pass false.
2945 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
2946 peeling from cunrolli.
2948 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2950 * combine.c (try_combine): Print a message to dump file whenever
2951 I0, I1, or I2 cannot be combined into I3.
2953 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2955 PR rtl-optimization/83156
2956 PR rtl-optimization/82621
2957 * combine.c (try_combine): Don't split an I2 if one of the dests is
2958 set again before I3. Allow unused dests.
2960 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2962 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
2964 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
2966 PR rtl-optimization/80818
2967 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
2968 recursively. Use insn code for clobber.
2969 (lra_set_insn_recog_data): Pass the new arg to
2970 collect_non_operand_hard_regs.
2971 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
2973 (lra_update_insn_regno_info): Pass insn to
2974 add_regs_to_insn_regno_info.
2976 2017-11-29 Jim Wilson <jimw@sifive.com>
2977 Andrew Waterman <andrew@sifive.com>
2979 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
2980 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
2981 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
2982 use SINGLE_SHIFT_COST.
2983 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
2984 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
2986 2017-11-29 Julia Koval <julia.koval@intel.com>
2988 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
2989 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
2990 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
2991 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
2992 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
2993 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
2994 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
2995 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
2996 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
2997 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
2998 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
2999 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
3000 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
3001 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
3002 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
3003 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
3004 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
3005 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
3006 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
3007 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
3008 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
3009 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
3010 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
3011 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
3012 __builtin_ia32_vpshldv_v2di_mask,
3013 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
3014 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
3015 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
3017 2017-11-29 Julia Koval <julia.koval@intel.com>
3019 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
3020 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
3021 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
3022 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
3023 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
3024 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
3025 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
3026 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
3027 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
3028 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
3029 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
3030 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
3031 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
3032 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
3033 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
3034 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
3035 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
3036 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
3037 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3038 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
3039 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
3041 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3043 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
3044 movsi_pic_gotdata_op instruction as a load for the UT699 errata
3047 2017-11-29 Martin Aberg <maberg@gaisler.com>
3049 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3050 to prevent b2bst errata sequence.
3051 (sqrtdf2_fix): Likewise.
3053 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3055 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
3056 (div_sqrt_insn_p): New function.
3057 (sparc_do_work_around_errata): Insert NOP instructions to
3058 prevent sequences that could trigger the TN-0013 errata for
3059 certain LEON3 processors.
3060 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
3061 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
3062 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
3063 (in_branch_delay): Prevent div and sqrt in delay slot if
3065 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
3067 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3069 * config/sparc/sparc.c (atomic_insn_p): New function.
3070 (sparc_do_work_around_errata): Insert NOP instructions to
3071 prevent sequences that could trigger the TN-0010 errata for
3073 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
3074 instruction referable in atomic_insns_p.
3076 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3078 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
3079 (atomic_compare_and_swap_leon3_1): Likewise.
3082 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3084 * config/sparc/sparc.c (fpop_insn_p): New function.
3085 (sparc_do_work_around_errata): Insert NOP instructions to
3086 prevent sequences that could trigger the TN-0012 errata for
3088 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
3089 * config/sparc/sparc.md (fix_gr712rc): New attribute.
3090 (in_branch_annul_delay): Prevent floating-point instructions
3091 in delay slot of annulled integer branch.
3093 2017-11-29 Richard Biener <rguenther@suse.de>
3095 PR tree-optimization/83202
3096 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
3098 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
3099 nodes vectorized to the same stmts multiple times.
3100 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
3101 (vect_analyze_slp_instance): Adjust.
3102 (scalar_stmts_to_slp_tree_map_t): New typedef.
3103 (vect_schedule_slp_instance): Add a map recording the SLP node
3104 representing the vectorized stmts for a set of scalar stmts.
3105 Avoid code-generating redundancies.
3106 (vect_schedule_slp): Allocate map and pass it down.
3108 2017-11-29 Nathan Sidwell <nathan@acm.org>
3111 * tree.c (build_complex_type): Fix canonicalization. Only fill in
3114 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
3116 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
3117 cases for printing LDP/STP memory addresses.
3118 (aarch64_print_address_internal): Renamed from
3119 aarch64_print_operand_address, added parameter, add Pmode check.
3120 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
3121 (aarch64_print_operand_address): Indirect to
3122 aarch64_print_address_internal.
3123 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
3124 'y' operand output specifier.
3126 2017-11-29 Jakub Jelinek <jakub@redhat.com>
3129 * tree.c (build_simple_mem_ref_loc): Handle
3130 get_addr_base_and_unit_offset returning a MEM_REF.
3133 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
3136 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
3139 2017-11-28 David Malcolm <dmalcolm@redhat.com>
3141 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
3143 (layout::move_to_column): Likewise.
3145 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3148 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
3149 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
3151 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3152 Martin Jambor <mjambor@suse.cz>
3155 * tree.h (expr_type_first_operand_type_p): Declare
3156 * tree.c (expr_type_first_operand_type_p): New function.
3157 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
3158 (ipa_value_from_jfunc): Adjust declaration.
3159 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
3160 Use it as result type for arithmetics, unless it is NULL in which case
3161 be more conservative.
3162 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
3163 ipa_get_jf_pass_through_result.
3164 (propagate_vals_across_pass_through): Likewise.
3165 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
3166 is to propagate_vals_across_pass_through.
3167 (propagate_constants_across_call): Pass PARM_TYPE to
3168 propagate_scalar_across_jump_function.
3169 (find_more_scalar_values_for_callers_subset): Pass parameter type to
3170 ipa_value_from_jfunc.
3171 (cgraph_edge_brings_all_scalars_for_node): Likewise.
3172 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
3173 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
3174 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
3175 target_type, pass it to ipa_value_from_jfunc.
3176 (update_indirect_edges_after_inlining): Pass parameter type to
3177 try_make_edge_direct_simple_call.
3179 2017-11-28 Jeff Law <law@redhat.com>
3181 * gimple-ssa-evrp-analyze.c
3182 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
3183 refine ranges if scev_initialized_p returns true.
3184 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
3186 2017-11-28 Julia Koval <julia.koval@intel.com>
3188 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
3189 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
3190 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
3191 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
3192 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
3193 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
3194 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
3195 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
3196 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
3197 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
3198 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
3199 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
3200 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
3201 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
3202 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
3203 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
3204 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
3205 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
3206 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
3207 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
3208 __builtin_ia32_vpshrd_v2di_mask): New builtins.
3209 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
3211 2017-11-28 Julia Koval <julia.koval@intel.com>
3213 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
3214 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
3215 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
3216 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
3217 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
3218 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
3219 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
3220 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
3221 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
3222 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
3223 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
3224 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
3225 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
3226 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
3227 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
3228 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
3229 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
3230 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
3231 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
3232 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
3233 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
3234 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
3235 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
3236 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
3237 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
3238 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
3239 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
3240 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
3241 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
3242 __builtin_ia32_vpshld_v2di_mask): New builtins.
3243 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3244 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
3246 2017-11-28 Richard Biener <rguenther@suse.de>
3248 PR tree-optimization/80776
3249 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
3251 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
3253 (evrp_range_analyzer::record_ranges_from_incoming_edges):
3254 If the incoming edge is an effective fallthru because the other
3255 edge only reaches a __builtin_unreachable () then record ranges
3256 derived from the controlling condition in SSA info.
3257 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
3258 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
3260 2017-11-28 Olivier Hainque <hainque@adacore.com>
3262 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
3263 on mingw build hosts.
3265 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3268 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
3270 2017-11-28 Richard Biener <rguenther@suse.de>
3273 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
3274 copies generated from memcpy use a character array as reference
3277 2017-11-28 Julia Koval <julia.koval@intel.com>
3278 Sebastian Peryt <sebastian.peryt@intel.com>
3280 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
3281 c-family/c-cilkplus.o, c-family/array-notation-common.o,
3282 cilk-common.o, cilk.h, cilk-common.c): Remove.
3284 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
3285 * builtins.c (is_builtin_name): Remove cilkplus condition.
3286 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
3287 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
3288 cilk-builtins.def, cilkplus.def): Remove.
3289 * cif-code.def (CILK_SPAWN): Remove.
3290 * cilk-builtins.def: Delete.
3291 * cilk-common.c: Ditto.
3293 * cilkplus.def: Ditto.
3294 * config/darwin.h (fcilkplus): Delete.
3295 * cppbuiltin.c: Ditto.
3296 * doc/extend.texi: Remove cilkplus doc.
3297 * doc/generic.texi: Ditto.
3298 * doc/invoke.texi: Ditto.
3299 * doc/passes.texi: Ditto.
3300 * gcc.c (fcilkplus): Remove.
3301 * gengtype.c (cilk.h): Remove.
3302 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
3304 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
3306 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
3308 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
3309 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
3310 cilkplus conditions.
3311 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
3312 inline_read_section): Ditto.
3313 * ipa-inline-analysis.c (cilk.h): Remove.
3314 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
3315 * lto-wrapper.c (merge_and_complain, append_compiler_options,
3316 append_linker_options): Remove condition for fcilkplus.
3317 * lto/lto-lang.c (cilk.h): Remove.
3318 (lto_init): Remove condition for fcilkplus.
3319 * omp-expand.c (expand_cilk_for_call): Delete.
3320 (expand_omp_taskreg, expand_omp_for_static_chunk,
3321 expand_omp_for): Remove cilkplus
3323 (expand_cilk_for): Delete.
3324 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
3325 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
3326 execute_lower_omp, diagnose_sb_0): Ditto.
3327 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
3328 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
3329 * tree-nested.c: Ditto.
3330 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
3331 (dump_generic_node): Ditto.
3332 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
3333 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
3335 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
3337 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3339 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
3340 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
3341 for vector mode and !TARGET_SIMD.
3343 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3345 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
3347 * tree.h (SWITCH_LABELS): Remove.
3348 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
3349 assert SWITCH_BODY is non-NULL.
3350 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
3352 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
3354 PR tree-optimization/80788
3355 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
3356 has TREE_OVERFLOW set, call drop_tree_overflow.
3358 2017-11-28 Richard Biener <rguenther@suse.de>
3360 PR tree-optimization/83158
3361 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
3363 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
3366 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
3367 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
3369 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
3371 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
3372 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
3374 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3377 * varasm.c (compare_constant): Take the mode of the constants into
3378 account when comparing floating point constants.
3380 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
3382 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
3383 from explicit instantiation of debug_helper.
3384 * vec.h (DEFINE_DEBUG_VEC): Ditto.
3386 2017-11-27 Richard Biener <rguenther@suse.de>
3388 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
3391 2017-11-27 Richard Biener <rguenther@suse.de>
3393 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
3394 (type_hash_canon): Also clear int_cst_hash_table entry for
3396 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
3398 2017-11-27 Tamar Christina <tamar.christina@arm.com>
3400 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3401 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3402 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3404 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
3406 * hash-map.h (gt_cleare_cache): Avoid UB.
3408 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
3410 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
3411 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
3412 (decide_unroll_runtime_iterations): Likewise.
3413 (decide_unroll_stupid): Likewise.
3415 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3418 * config/i386/i386.md: Add a loop with incssp.
3420 2017-11-27 Martin Jambor <mjambor@suse.cz>
3422 PR tree-optimization/81248
3423 * tree-sra.c (splice_param_accesses): Remove size check.
3424 (decide_one_param_reduction): Fix size check.
3425 * gimple-pretty-print.c (dump_profile): Silence warning.
3426 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
3428 2017-11-27 Jakub Jelinek <jakub@redhat.com>
3431 * dbxout.c (lastlineno): New variable.
3432 (dbx_debug_hooks): Use dbxout_switch_text_section as
3433 switch_text_section debug hook.
3434 (dbxout_function_end): Switch to current_function_section
3435 rather than function_section. If crtl->has_bb_partition,
3436 output just one N_FUN, depending on in_cold_section_p.
3437 (dbxout_source_line): Remember last lineno in lastlineno.
3438 (dbxout_switch_text_section): New function.
3439 (dbxout_function_decl): Adjust dbxout_block caller.
3440 (dbx_block_with_cold_children): New function.
3441 (dbxout_block): Return true if any LBRAC/RBRAC have been
3442 emitted. Use dbx_block_with_cold_children at depth == 0
3443 in second partition. Add PARENT_BLOCKNUM argument, pass
3444 it optionally adjusted to children. Output LBRAC/RBRAC
3445 around recursive call only if the block is in the current
3446 partition, if not and anything was output, emit empty
3448 * final.c (final_scan_insn): Compute cold_function_name
3449 before calling switch_text_section debug hook. Call
3450 that hook even if dwarf2out_do_frame if not emitting
3454 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
3455 TREE_READONLY decls.
3457 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
3459 PR rtl-optimization/82488
3460 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
3462 2017-11-26 Julia Koval <julia.koval@intel.com>
3464 * config/i386/i386.c (processor_target_table): Add skylake_cost for
3466 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
3469 2017-11-26 Julia Koval <julia.koval@intel.com>
3471 * config/i386/driver-i386.c (host_detect_local_cpu):
3472 Detect skylake-avx512.
3474 2017-11-26 Julia Koval <julia.koval@intel.com>
3476 * config.gcc: Add -march=cannonlake.
3477 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
3478 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
3479 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
3480 (PTA_CANNONLAKE): New.
3481 (processor_target_table): Add cannonlake.
3482 (ix86_option_override_internal): Ditto.
3483 (fold_builtin_cpu): Ditto.
3484 (get_builtin_code_for_version): Handle cannonlake.
3485 (M_INTEL_COREI7_CANNONLAKE): New.
3486 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
3487 * doc/invoke.texi: Add -march=cannonlake.
3489 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
3491 * plugin.c (add_new_plugin): Use platform-specific library extensions.
3492 (try_init_one_plugin): Alternative implementation for MinGW.
3493 * Makefile.in (plugin_implib): New.
3494 (gengtype-lex.c): Fix broken AIX workaround.
3495 * configure: Regenerate.
3496 * doc/plugins.texi: Document support for MinGW.
3498 2017-11-25 Jakub Jelinek <jakub@redhat.com>
3500 PR rtl-optimization/81553
3501 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
3502 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
3503 is a shift where C1 has different mode than the whole shift, use C1's
3504 mode for MULT rather than the shift's mode.
3507 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
3508 builtins not enabled in the currently selected ISA.
3510 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
3512 PR tree-optimization/71026
3513 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
3514 (insert_reciprocals): Change to insert reciprocals before a division
3515 by a square and to insert the square of a reciprocal.
3516 (execute_cse_reciprocals_1): Change to consider division by a square.
3517 (register_division_in): Add importance parameter.
3519 2017-11-24 Richard Biener <rguenther@suse.de>
3521 PR tree-optimization/82402
3522 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
3523 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
3525 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
3527 * match.pd (0-ptr): New transformation.
3529 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
3532 * ipa-inline.c (inline_small_functions): Set current badnes correctly
3533 when skipping checking.
3535 2017-11-24 Richard Biener <rguenther@suse.de>
3537 PR tree-optimization/83128
3538 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
3539 (vn_reference_lookup_3): Likewise.
3541 2017-11-24 Jakub Jelinek <jakub@redhat.com>
3544 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
3545 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
3548 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
3549 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
3550 check for the latter with gimple_call_builtin_p. Do not handle
3551 BUILT_IN_STPNCPY_CHK which is not a pass through call.
3553 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
3555 * config/arm/arm_neon.h: Fix pragma GCC push_options before
3558 2017-11-23 Julia Koval <julia.koval@intel.com>
3560 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
3561 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
3562 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
3563 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
3564 _mm512_maskz_expandloadu_epi16): New intrinsics.
3565 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
3566 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
3567 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
3568 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
3569 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
3570 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
3571 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
3572 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
3573 _mm256_maskz_expandloadu_epi8): New intrinsics.
3574 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
3575 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
3576 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
3577 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
3578 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
3579 * config/i386/sse.md (VI248_VLBW): New iterator.
3580 (expand<mode>_mask, expand<mode>_maskz): New patterns.
3582 2017-11-23 Julia Koval <julia.koval@intel.com>
3584 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
3585 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
3586 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
3587 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
3588 _mm512_mask_compressstoreu_epi16): New.
3589 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
3590 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
3591 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
3592 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
3593 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
3594 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
3595 _mm256_mask_compressstoreu_epi8): New.
3596 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
3597 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
3598 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
3599 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
3600 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
3601 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
3602 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
3603 __builtin_ia32_compresshi128_mask,
3604 __builtin_ia32_compressstoreuqi512_mask,
3605 __builtin_ia32_compressstoreuhi512_mask,
3606 __builtin_ia32_compressstoreuqi256_mask,
3607 __builtin_ia32_compressstoreuqi128_mask,
3608 __builtin_ia32_compressstoreuhi256_mask,
3609 __builtin_ia32_compressstoreuhi128_mask): New builtins.
3610 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
3612 (ix86_expand_special_args_builtin): Handle new types.
3613 (s4fma_expand): Handle new builtin array.
3614 * config/i386/immintrin.h: Include new headers.
3615 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
3616 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
3618 2017-11-23 Jakub Jelinek <jakub@redhat.com>
3620 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
3621 fixes. Declare temp and g variables at the top in order to avoid
3622 {} in most of the cases.
3624 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
3626 * match.pd (ptr-0): New transformation.
3628 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
3630 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
3631 (struct addr_mode_cost_table): New.
3632 (struct tune_params): Add field addr_mode_costs.
3633 * config/arm/arm.c (generic_addr_mode_costs): New.
3634 (arm_slowmul_tune): Initialise addr_mode_costs field.
3635 (arm_fastmul_tune): Likewise.
3636 (arm_strongarm_tune): Likewise.
3637 (arm_xscale_tune): Likewise.
3638 (arm_9e_tune): Likewise.
3639 (arm_marvell_pj4_tune): Likewise.
3640 (arm_v6t2_tune): Likewise.
3641 (arm_cortex_tune): Likewise.
3642 (arm_cortex_a8_tune): Likewise.
3643 (arm_cortex_a7_tune): Likewise.
3644 (arm_cortex_a15_tune): Likewise.
3645 (arm_cortex_a35_tune): Likewise.
3646 (arm_cortex_a53_tune): Likewise.
3647 (arm_cortex_a57_tune): Likewise.
3648 (arm_exynosm1_tune): Likewise.
3649 (arm_xgene1_tune): Likewise.
3650 (arm_cortex_a5_tune): Likewise.
3651 (arm_cortex_a9_tune): Likewise.
3652 (arm_cortex_a12_tune): Likewise.
3653 (arm_cortex_a73_tune): Likewise.
3654 (arm_v7m_tune): Likewise.
3655 (arm_cortex_m7_tune): Likewise.
3656 (arm_v6m_tune): Likewise.
3657 (arm_fa726te_tune): Likewise.
3658 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
3660 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
3662 * config/arm/arm.c (arm_mem_costs): New function.
3663 (arm_rtx_costs_internal): Use arm_mem_costs.
3665 2017-11-23 Mark Wielaard <mark@klomp.org>
3667 * dwarf2out.c (init_sections_and_labels): Use generation to create
3668 unique ranges_section_label and ranges_base_label. Return generation.
3669 (output_rnglists): Add generation argument. Use generation to create
3670 unique ranges labels.
3671 (dwarf2out_finish): Get generation from init_sections_and_labels
3672 and pass generation to output_rnglists.
3674 2017-11-23 Mike Stump <mikestump@comcast.net>
3675 Eric Botcazou <ebotcazou@adacore.com>
3677 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
3678 * cfgloop.h (struct loop): Add unroll field.
3679 * function.h (struct function): Add has_unroll bitfield.
3680 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
3681 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
3682 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
3683 (pass_rtl_unroll_loops::gate): Likewise.
3684 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
3685 for which loop->unroll==1.
3686 (decide_unroll_constant_iterations): Use note for consistency and
3687 take loop->unroll into account. Return early if loop->unroll is set.
3688 Fix thinko in existing test.
3689 (decide_unroll_runtime_iterations): Use note for consistency and
3690 take loop->unroll into account.
3691 (decide_unroll_stupid): Likewise.
3692 * lto-streamer-in.c (input_cfg): Read loop->unroll.
3693 * lto-streamer-out.c (output_cfg): Write loop->unroll.
3694 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
3696 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
3697 (print_loop): Print loop->unroll if set.
3698 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
3699 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
3700 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
3702 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
3703 loop->unroll is set and smaller than the trip count. Otherwise bypass
3704 entirely the heuristics if loop->unroll is set. Remove dead note.
3705 Fix off-by-one bug in other note.
3706 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
3707 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
3709 (tree_unroll_loops_completely): Make static.
3710 (pass_complete_unroll::execute): Use correct type for variable.
3711 (pass_complete_unrolli::execute): Fix formatting.
3712 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
3714 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
3716 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
3717 enable when TARGET_PREFER_AVX128 is set.
3719 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3721 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
3723 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
3724 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
3726 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3728 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
3729 frequencies. Fix estimation of aggregate parameters.
3731 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3733 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
3736 2017-11-23 Tom de Vries <tom@codesourcery.com>
3738 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
3740 2017-11-23 Tom de Vries <tom@codesourcery.com>
3742 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
3745 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
3748 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
3749 sibcall_value_pcrel_fdpic): Use local variable instead of
3751 (calli_tbr_rel): Add missing operand 2.
3752 (call_valuei_tbr_rel): Add missing operand 3.
3754 2017-11-23 Jakub Jelinek <jakub@redhat.com>
3757 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
3758 bitpos/bitsize covering the whole destination, use store_expr only if
3759 the complex mode is the same. Otherwise, use expand_normal and if
3760 it returns CONCAT, subreg each part separately instead of trying to
3761 subreg the whole result.
3763 2017-11-23 Richard Biener <rguenther@suse.de>
3765 PR tree-optimization/23094
3766 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
3767 come here from walking over backedges in the first iteration.
3768 (vn_reference_lookup_3): Skip clobbers that store the same value.
3770 2017-11-23 Richard Biener <rguenther@suse.de>
3772 PR tree-optimization/81403
3773 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
3774 a block we need a leader relative to.
3775 (phi_translate_1): For nary processing require a leader from
3776 get_representative_for given we run expression simplification
3777 using match-and-simplify. Remove previous fix.
3779 2017-11-22 Jeff Law <law@redhat.com>
3781 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
3782 Use new method allocate_value_range rather than accessing the
3783 vrp_value_range_pool data member directly.
3784 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
3785 to use extract_range_from_stmt method to avoid need for
3786 extract_range_from_assignment method.
3787 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
3788 method rather than setting values_propgated data member directly.
3789 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
3790 and values propagated data members and extract_range_from_assignment
3791 method. Reorder private data members to conform to standards.
3792 Add new methods set_lattice_propagation_complete and
3793 allocate_value_range.
3795 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
3797 PR rtl-optimization/83030
3798 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
3799 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
3800 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
3801 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
3802 CROSSING_JUMP_P flag is already set before setting it.
3803 * cfgrtl.c (fixup_partition_crossing): Likewise.
3804 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
3807 2017-11-22 Jakub Jelinek <jakub@redhat.com>
3809 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
3810 Handle the case where both arguments are using gen_const_vec_series.
3812 2017-11-22 David Malcolm <dmalcolm@redhat.com>
3815 * pretty-print.c (pp_format): Move quoting implementation to
3816 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
3817 to pass address of "quote" local.
3818 (pp_begin_quote): New function.
3819 (pp_end_quote): New function.
3820 * pretty-print.h (printer_fn): Convert penultimate param from bool
3822 (pp_begin_quote): New decl.
3823 (pp_end_quote): New decl.
3824 * tree-diagnostic.c (default_tree_printer): Convert penultimate
3825 param from bool to bool *.
3826 * tree-diagnostic.h (default_tree_printer): Likewise.
3828 2017-11-22 Jeff Law <law@redhat.com>
3830 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
3831 Initialize vr_values.
3832 (evrp_range_analyzer::try_find_new_range): Call methods attached to
3833 vr_values via vr_values class instance rather than delegators.
3834 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
3835 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
3836 (evrp_range_analyzer::push_value_range): Likewise.
3837 (evrp_range_analyzer::pop_value_range): Likewise.
3838 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
3839 most delegators. Those remaining are exposed as public interfaces.
3840 Make vr_values a pointer and private.
3841 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
3843 (evrp_range_analyzer::get_vr_value): New method.
3844 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
3845 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
3846 (class evrp_dom_walker): Attach evrp_folder class, initialize
3847 it in the ctor. Remove temporary delegators.
3848 (evrp_dom_walker::before_dom_children): Call methods in attached
3849 evrp_range_analyzer class via class instance pointer. Use
3850 free value_range_constant_singleton to remove need for
3851 op_with_constant_singleton_value delegator method. Do not
3852 create a vrp_prop class instance for every call! Narrow
3853 scope of a couple variables.
3854 (evrp_dom_walker::cleanup): Call methods in attached
3855 evrp_range_analyzer class via class instance pointer.
3856 * vr-values.h (class vr_values): Privatize many methods and
3859 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3861 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
3862 * cfgexpand.c (expand_debug_expr): Remove handling for them.
3863 * expr.c (expand_expr_real_2): Likewise.
3864 * fold-const.c (const_unop): Likewise.
3865 * optabs-tree.c (optab_for_tree_code): Likewise.
3866 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3867 * tree-inline.c (estimate_operator_cost): Likewise.
3868 * tree-pretty-print.c (dump_generic_node): Likewise.
3869 (op_code_prio): Likewise.
3870 (op_symbol_code): Likewise.
3871 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
3872 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
3873 * internal-fn.c (direct_internal_fn_optab): New function.
3874 (direct_internal_fn_array, direct_internal_fn_supported_p
3875 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
3876 * fold-const-call.c (fold_const_reduction): New function.
3877 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
3879 * tree-vect-loop.c: Include internal-fn.h.
3880 (reduction_code_for_scalar_code): Rename to...
3881 (reduction_fn_for_scalar_code): ...this and return an internal
3883 (vect_model_reduction_cost): Take an internal_fn rather than
3885 (vect_create_epilog_for_reduction): Likewise. Build calls rather
3887 (vectorizable_reduction): Use internal functions rather than tree
3888 codes for the reduction operation. Update calls to the functions
3890 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
3891 Use calls to internal functions rather than REDUC tree codes.
3892 * config/aarch64/aarch64-simd.md: Update comment accordingly.
3894 2017-11-22 Olivier Hainque <hainque@adacore.com>
3896 * config/vxworks.c (vxworks_override_options): Pick default
3897 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
3898 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
3899 DWARF_GNAT_ENCODINGS_DEFAULT.
3900 * config/vxworksae.h: Likewise.
3902 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
3904 PR tree-optimization/83104
3905 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
3908 2017-11-22 Marek Polacek <polacek@redhat.com>
3909 H.J. Lu <hongjiu.lu@intel.com>
3910 Jason Merrill <jason@redhat.com>
3915 * calls.c (initialize_argument_information): Call
3916 warn_parameter_passing_abi target hook.
3917 (store_one_arg): Use 0 for empty record size. Don't push 0 size
3918 argument onto stack.
3919 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
3920 * common.opt: Update -fabi-version description.
3921 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
3922 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
3924 (ix86_is_empty_record): New function.
3925 (ix86_warn_parameter_passing_abi): New function.
3926 (TARGET_EMPTY_RECORD_P): Redefine.
3927 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
3928 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
3929 * doc/tm.texi: Regenerated.
3930 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
3931 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
3932 * dwarf2out.c (get_ultimate_context): Move to tree.c.
3933 * explow.c (hard_function_value): Call arg_int_size_in_bytes
3934 instead of int_size_in_bytes.
3935 * expr.c (copy_blkmode_to_reg): Likewise.
3936 * function.c (aggregate_value_p): Return 0 for empty types.
3937 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
3939 (locate_and_pad_parm): Call arg size_in_bytes instead
3941 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
3942 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
3943 * target.def (empty_record_p, warn_parameter_passing_abi): New target
3945 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
3946 (std_gimplify_va_arg_expr): Skip empty records. Call
3947 arg_size_in_bytes instead size_in_bytes.
3948 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
3949 * tree-core.h (tree_type_common): Add empty_flag.
3950 (tree_decl_common): Update comments.
3951 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
3953 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
3954 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
3956 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
3957 * tree.c (default_is_empty_type): New function.
3958 (default_is_empty_record): New function.
3959 (arg_int_size_in_bytes): New function.
3960 (arg_size_in_bytes): New function.
3961 (get_ultimate_context): New function.
3962 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
3963 TRANSLATION_UNIT_WARN_EMPTY_P.
3964 (default_is_empty_record, arg_int_size_in_bytes,
3965 arg_size_in_bytes, get_ultimate_context): Declare.
3967 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3969 * config/arm/arm.c (cmse_clear_registers): New function.
3970 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
3971 code by call to cmse_clear_registers.
3972 (cmse_nonsecure_entry_clear_before_return): Likewise.
3974 2017-11-22 Tamar Christina <tamar.christina@arm.com>
3976 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
3977 (vdot_s32, vdotq_s32): New.
3978 (vdot_lane_u32, vdotq_lane_u32): New.
3979 (vdot_lane_s32, vdotq_lane_s32): New.
3982 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3985 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
3986 for unsigned values with fewer HWIs than the precision.
3987 (test_overflow): New function.
3988 (wide_int_cc_tests): Call it.
3990 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3992 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
3993 has the same mode class as Pmode.
3995 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3997 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
3998 padding_bits_to_clear_ptr.
3999 (cmse_nonsecure_entry_clear_before_return): Likewise.
4001 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4003 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
4004 auto_sbitap instead of integer bitfield to control register needing
4007 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4009 PR tree-optimization/83044
4010 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
4011 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
4012 that need the upper bound. Subtract offset from
4013 get_addr_base_and_unit_offset only if positive and subtract it
4014 before division by eltsize rather than after it.
4017 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
4018 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
4019 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
4023 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
4024 expand_binop with *mul_widen_optab, make sure at least one of the
4025 operands doesn't have VOIDmode.
4028 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
4030 PR rtl-optimization/82044
4031 PR tree-optimization/82042
4032 * dse.c (record_store): Check for overflow.
4033 (check_mem_read_rtx): Properly check for overflow if width == -1, call
4034 add_wild_read instead of clear_rhs_from_active_local_stores on
4035 overflow and log it into dump_file.
4037 2017-11-22 Richard Biener <rguenther@suse.de>
4039 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
4040 fake edges to exit when looking for a place to insert.
4041 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
4043 (insert_into_preds_of_block): Commit edge insertion immediately,
4044 assert that doesn't require new BBs.
4045 (fini_pre): Release expressions.
4046 (pass_pre::execute): Shuffle things around a bit, if the fn
4047 is too large do not compute AVAIL either as this is really the
4050 2017-11-22 Richard Biener <rguenther@suse.de>
4052 PR tree-optimization/83089
4053 * tree-if-conv.c (pass_if_conversion::execute): If anything
4054 changed reset SCEV and free the number of iteration estimates.
4056 2017-11-21 Martin Sebor <msebor@redhat.com>
4058 PR tree-optimization/82945
4059 * calls.h (warn_nonstring_bound): Remove unused function.
4061 2017-11-21 Martin Sebor <msebor@redhat.com>
4063 PR tree-optimization/82945
4064 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
4065 * calls.h (maybe_warn_nonstring_arg): Declare new function.
4066 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
4068 (initialize_argument_information): Call maybe_warn_nonstring_arg.
4069 * calls.h (get_attr_nonstring_decl): Declare new function.
4070 * doc/extend.texi (attribute nonstring): Update.
4071 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
4072 get_attr_nonstring_decl and handle it.
4073 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
4074 detection of nul-termination.
4075 (strlen_to_stridx): Change to a pointer.
4076 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
4077 (pass_strlen::execute): Same.
4079 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
4081 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
4082 for the new option -mprefer-vector-width=[none|128|256|512].
4083 * config/i386/i386.c (ix86_target_string): remove old style options
4084 -mprefer-avx256 and make -mprefer-avx128 as alias.
4085 (ix86_option_override_internal): Apply defaults for the
4086 -mprefer-vector-width=[128|256] option.
4087 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
4088 Implement macros to work with -mprefer-vector-width=.
4089 * config/i386/i386.opt: Implemented option
4090 -mprefer-vector-width=[none|128|256|512].
4091 * doc/invoke.texi: Documentation for
4092 -mprefer-vector-width=[none|128|256|512].
4094 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
4096 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
4098 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
4100 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
4102 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
4104 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
4106 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
4107 * expr.c (expand_expr_real_2): Likewise.
4108 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
4109 fold_binary_loc): Likewise.
4110 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
4111 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
4112 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
4113 MINUS_EXPR transformations.
4114 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
4115 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
4116 * tree-inline.c (estimate_operator_cost): Likewise.
4117 * tree-pretty-print.c (dump_generic_node, op_code_prio,
4118 op_symbol_code): Likewise.
4119 * tree-vect-stmts.c (vectorizable_operation): Likewise.
4120 * vr-values.c (extract_range_from_binary_expr): Likewise.
4121 * varasm.c (initializer_constant_valid_p_1): Likewise.
4122 * tree.def: New tree code POINTER_DIFF_EXPR.
4124 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
4126 * config/i386/i386.md (*bswap<mode>2_movbe): Add
4127 integer suffix to movbe mnemonic.
4128 (*bswaphi2_movbe): Ditto.
4129 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
4131 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4134 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
4135 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
4136 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
4138 2017-11-21 Martin Liska <mliska@suse.cz>
4140 * tree-inline.c (expand_call_inline): Remove not needed
4143 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
4144 Jakub Jelinek <jakub@redhat.com>
4147 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
4148 Remove static keyword from f variable.
4150 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4152 PR tree-optimization/83086
4153 * gimple-ssa-store-merging.c
4154 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
4155 rather than n.base_addr.
4157 2017-11-21 Martin Liska <mliska@suse.cz>
4159 PR rtl-optimization/82044
4160 PR tree-optimization/82042
4161 * dse.c (check_mem_read_rtx): Check for overflow.
4163 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
4165 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
4168 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
4170 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
4171 and its corresponding call arg location note.
4173 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
4174 Andrew Burgess <andrew.burgess@embecosm.com>
4176 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
4178 (arc_return_slot_offset): Likewise.
4179 (arc_eh_return_address_location): New declaration.
4180 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
4181 (MUST_SAVE_REGISTER): Add exception handler case.
4182 (MUST_SAVE_RETURN_ADDR): Likewise.
4183 (arc_frame_pointer_required): Likewise.
4184 (arc_frame_pointer_needed): New function.
4185 (arc_compute_frame_size): Changed.
4186 (arc_expand_prologue): Likewise.
4187 (arc_expand_epilogue): Likewise.
4188 (arc_initial_elimination_offset): Likewise.
4189 (arc_return_slot_offset): Delete.
4190 (arc_eh_return_address_location): New function.
4191 (arc_builtin_setjmp_frame_value): Likewise.
4192 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
4193 (EH_RETURN_STACKADJ_RTX): Define.
4194 (EH_RETURN_HANDLER_RTX): Likewise.
4195 * config/arc/arc.md (eh_return): Delete.
4197 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
4199 * print-tree.h (debug_vec_tree): Remove prototype.
4200 * gdbinit.in (pvt): Remove macro.
4202 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4204 PR tree-optimization/83047
4205 * gimple-ssa-store-merging.c
4206 (imm_store_chain_info::output_merged_store): If the loads with the
4207 same vuse are in different basic blocks, for load_gsi pick a load
4208 location that is dominated by the other loads.
4211 * config/i386/i386.c (ix86_memmodel_check): Start
4212 -Winvalid-memory-model diagnostics with lowercase letter.
4215 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
4216 set in_cold_section_p to the partition containing loc_list->first.
4217 When seeing loc_list->last_before_switch node, update secname and
4218 perform range_across_switch second partition handling only after that.
4221 * run-rtl-passes.c: Include debug.h.
4222 (run_rtl_passes): Call debug_hooks->assembly_start.
4223 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
4227 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
4228 OPTAB_DIRECT in calls to expand_simple_binop.
4230 2017-11-20 David Malcolm <dmalcolm@redhat.com>
4233 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
4235 2017-11-20 David Malcolm <dmalcolm@redhat.com>
4238 * spellcheck.h (best_match::blithely_get_best_candidate): New
4241 2017-11-20 Jakub Jelinek <jakub@redhat.com>
4243 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
4244 with lower case letter.
4246 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
4248 * config/i386/i386.md (bswaphi2): New expander.
4249 (*bswaphi2_movbe): New insn pattern.
4250 (bswaphi -> rorhi pepehole2): New peephole pattern.
4252 2017-11-20 Jeff Law <law@redhat.com>
4254 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
4255 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
4256 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
4257 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
4258 gimple-ssa-evrp-analyze.h.
4260 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
4261 set BB_VISITED here.
4262 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
4264 2017-11-20 Bin Cheng <bin.cheng@arm.com>
4266 * tree-predcom.c: Add general comment on Store-Store chains.
4267 (split_data_refs_to_components): Postpone clearing eliminate_store_p
4269 (get_chain_last_ref_at): Rename into...
4270 (get_chain_last_write_at): ...this.
4271 (get_chain_last_write_before_load): New function.
4272 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
4273 CT_STORE_STORE when write reference is added.
4274 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
4275 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
4276 (initialize_root_vars_store_elim_1): Ditto.
4277 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
4278 definition is created.
4279 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
4280 chain by replacing it with dominant stored value.
4282 2017-11-20 Bin Cheng <bin.cheng@arm.com>
4284 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
4286 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
4288 * vr-values.c (extract_range_from_binary_expr): Use a full range
4291 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
4293 * config/arm/arm.md (R4_REGNUM): Define constant.
4294 (nonsecure_call_internal): Remove r4 clobber.
4295 (nonsecure_call_value_internal): Likewise.
4296 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
4297 clobber and resequence match_operands.
4298 (nonsecure_call_value_reg_thumb1_v5): Likewise.
4299 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
4300 (nonsecure_call_value_reg_thumb2): Likewise.
4302 2017-11-20 Jakub Jelinek <jakub@redhat.com>
4304 PR tree-optimization/78821
4305 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
4306 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
4307 to the address of the base rather than the base itself.
4308 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
4309 (find_bswap_or_nop_finalize): New function.
4310 (find_bswap_or_nop): Use it.
4311 (bswap_replace): Return a tree rather than bool, change first
4312 argument from gimple * to gimple_stmt_iterator, allow inserting
4313 into an empty sequence, allow ins_stmt to be NULL - then emit
4314 all stmts into gsi. Fix up MEM_REF address gimplification.
4315 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
4316 (struct store_immediate_info): Add N and INS_STMT non-static
4318 (store_immediate_info::store_immediate_info): Initialize them
4319 from newly added ctor args.
4320 (merged_store_group::apply_stores): Formatting fixes. Sort by
4322 (stmts_may_clobber_ref_p): For stores call also
4323 refs_anti_dependent_p.
4324 (gather_bswap_load_refs): New function.
4325 (imm_store_chain_info::try_coalesce_bswap): New method.
4326 (imm_store_chain_info::coalesce_immediate_stores): Use it.
4327 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
4328 (imm_store_chain_info::output_merged_store): Fail if number of
4329 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
4330 and NOP_EXPR rhs_code.
4331 (pass_store_merging::process_store): Compute n and ins_stmt, if
4332 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
4333 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
4335 (pass_store_merging::execute): Calculate dominators.
4337 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
4338 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
4339 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
4340 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
4341 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
4342 class pass_optimize_bswap, bswap_replace,
4343 pass_optimize_bswap::execute): Moved to ...
4344 * gimple-ssa-store-merging.c: ... this file.
4345 Include optabs-tree.h.
4346 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
4347 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
4348 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
4349 anonymous namespace, remove static keywords.
4350 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
4351 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
4353 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
4356 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
4358 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
4360 * vec.h (debug_helper): New function.
4361 (DEFINE_DEBUG_VEC): New macro.
4362 * hash-set.h (debug_helper): New function.
4363 (DEFINE_DEBUG_HASH_SET): New macro.
4364 * cfg.c (debug_slim (edge)): New function.
4365 Call DEFINE_DEBUG_VEC for edges.
4366 Call DEFINE_DEBUG_HASH_SET for edges.
4367 * cfghooks.c (debug_slim (basic_block)): New function.
4368 Call DEFINE_DEBUG_VEC for basic blocks.
4369 Call DEFINE_DEBUG_HASH_SET for basic blocks.
4370 * print-tree.c (debug_slim): New function to handle trees.
4371 Call DEFINE_DEBUG_VEC for trees.
4372 Call DEFINE_DEBUG_HASH_SET for trees.
4373 (debug (vec<tree, va_gc>) &): Remove.
4374 (debug (<vec<tree, va_gc>) *): Remove.
4375 * print-rtl.c (debug_slim): New function to handle const_rtx.
4376 Call DEFINE_DEBUG_VEC for rtx_def.
4377 Call DEFINE_DEBUG_VEC for rtx_insn.
4378 Call DEFINE_DEBUG_HASH_SET for rtx_def.
4379 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
4380 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
4381 (debug (vec<rtx_insn *> *ptr): Remove.
4382 (debug_insn_vector): Remove.
4383 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
4385 2017-11-20 Tom de Vries <tom@codesourcery.com>
4387 PR rtl-optimization/82020
4388 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
4389 IF_THEN_ELSE condition.
4391 2017-11-19 Jeff Law <law@redhat.com>
4393 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
4394 of degenerates resulting from ignoring an edge.
4396 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4399 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
4401 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4404 * profile-count.c (profile_count::to_sreal_scale): Fix return value
4405 for uninitialied counts.
4407 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4410 * tree-inline.c (estimate_num_insns): Set to 1 at least.
4412 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4415 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
4417 2017-11-19 Tom de Vries <tom@codesourcery.com>
4419 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
4422 2017-11-19 Tom de Vries <tom@codesourcery.com>
4424 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
4427 2017-11-19 Tom de Vries <tom@codesourcery.com>
4429 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4432 2017-11-19 Tom de Vries <tom@codesourcery.com>
4434 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
4436 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
4438 2017-11-19 Tom de Vries <tom@codesourcery.com>
4440 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
4442 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
4443 semicolon after MCORE_EXPORT_NAME call.
4445 2017-11-19 Tom de Vries <tom@codesourcery.com>
4448 * vmsdbgout.c (vmsdbgout_early_finish): New function.
4449 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
4451 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
4453 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
4454 (cgraph_edge::clone): Cleanup updating of profile.
4455 * ipa-cp.c (update_profiling_info): Likewise.
4456 * ipa-inline-transform.c (inline_transform): Likewise.
4457 * ipa-inline.c (inline_small_functions): Add missing space to dump.
4458 * ipa-split.c (execute_split_functions): Do not split when function
4460 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
4461 * profile-count.c (profile_count::dump): Add global0.
4462 (profile_count::to_cgraph_frequency): Do not ICE when entry is
4464 (profile_count::to_sreal_scale): Likewise.
4465 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
4466 (profile_count::combine_with_ipa_count): New function.
4467 * profile-count.h (profile_guessed_global0adjusted): New.
4468 (profile_count::adjusted_zero): New.
4469 (profile_count::global0adjusted): New.
4470 (profile_count::combine_with_ipa_count): New.
4471 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
4472 correct profile of return block of split functions.
4473 (copy_cfg_body): Remove unused profile_count.
4474 (copy_body): Likewise.
4475 (expand_call_inline): Update.
4476 (tree_function_versioning): Update.
4478 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
4480 * hash-set.h (hash_set::empty): New.
4481 * tree-ssa-threadbackward.h: Delete.
4482 * tree-ssa-threadbackward.c (class thread_jumps): New.
4483 Move max_threaded_paths into class.
4484 (fsm_find_thread_path): Remove arguments that are now in class.
4485 (profitable_jump_thread_path): Rename to...
4486 (thread_jumps::profitable_jump_thread_path): ...this.
4487 (convert_and_register_jump_thread_path): Rename to...
4488 (thread_jumps::convert_and_register_current_path): ...this.
4489 (check_subpath_and_update_thread_path): Rename to...
4490 (thread_jumps::check_subpath_and_update_thread_path): ...this.
4491 (register_jump_thread_path_if_profitable): Rename to...
4492 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
4493 (handle_phi): Rename to...
4494 (thread_jumps::handle_phi): ...this.
4495 (handle_assignment): Rename to...
4496 (thread_jumps::handle_assignment): ...this.
4497 (fsm_find_control_statement_thread_paths): Rename to...
4498 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
4499 (find_jump_threads_backwards): Rename to...
4500 (thread_jumps::find_jump_threads_backwards): ...this.
4501 Initialize path local data.
4502 (pass_thread_jumps::execute): Call find_jump_threads_backwards
4503 from within thread_jumps class.
4504 (pass_early_thread_jumps::execute): Same.
4506 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4508 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
4510 2017-11-17 Steve Ellcey <sellcey@cavium.com>
4512 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
4513 to canonical location.
4515 2017-11-17 Steve Ellcey <sellcey@cavium.com>
4518 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
4520 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
4522 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
4524 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
4525 rcrt1.o%s/grcrt1.o%s for -static-pie.
4527 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4529 * i386.c (ix86_multiplication_cost, ix86_division_cost,
4530 ix86_shift_rotate_cost): Break out from ...
4531 (ix86_rtx_costs): ... here.
4532 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
4535 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4537 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
4538 when entry block was promoted unlikely.
4539 (estimate_bb_frequencies): Increase frequency scale.
4540 * profile-count.h (profile_count): Export precision info.
4542 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4544 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
4545 disturbin profile of entry block.
4547 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4549 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
4552 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4554 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
4555 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
4557 2017-11-17 Jeff Law <law@redhat.com>
4559 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
4560 from evrp_dom_walker class. Various methods moved into new class.
4561 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
4562 (evrp_range_analyzer::enter): New method.
4563 (evrp_range_analyzer::leave): New method.
4564 (evrp_dom_walker): Remove delegators no longer needed by this class.
4565 Replace vr_values data member with evrp_range_analyzer
4567 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
4568 method extracted from evrp_dom_walker::before_dom_children.
4569 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
4570 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
4572 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
4573 Add private copy constructor and move assignment operators.
4574 Privatize methods and class data where trivially possible.
4575 (evrp_dom_walker::cleanup): New function, extracted from
4576 execute_early_vrp. Simplify access to class data.
4578 * vr-values.h (get_output_for_vrp): Prototype.
4579 * vr-values.c (get_output_for_vrp): New function extracted from
4580 vrp_visit_assignment_or_call and extract_range_from_stmt.
4581 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
4583 2017-11-17 Luis Machado <luis.machado@linaro.org>
4585 * config/aarch64/aarch64.c
4586 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
4587 (qdf24xx_tunings) <autoprefetcher_model>: Set to
4588 tune_params::AUTOPREFETCHER_WEAK.
4590 2017-11-17 Tamar Christina <tamar.christina@arm.com>
4593 * config/arm/arm.c (arm_valid_target_attribute_rec):
4594 Parse "arch=" and "+<ext>".
4595 (arm_valid_target_attribute_tree): Re-init global options.
4596 (arm_option_override): Make non-static.
4597 (arm_options_perform_arch_sanity_checks): Make errors fatal.
4598 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
4599 (__ARM_FEATURE_CRC32): Support undef.
4600 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
4601 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
4603 2017-11-17 David Malcolm <dmalcolm@redhat.com>
4605 * gdbinit.in (break-on-diagnostic): New command.
4607 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4609 * config/i386/i386.c (ix86_expand_epilogue): Change simple
4610 return to indirect jump for EH return if control-flow
4611 protection is enabled. Change explicit 'false' argument in
4612 pro_epilogue_adjust_stack with a value of
4614 * config/i386/i386.md (simple_return_indirect_internal):
4615 Remove SImode restriction to support 64-bit.
4617 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
4619 * combine.c (added_notes_insn): New.
4620 (try_combine): Handle added_notes_insn like added_links_insn.
4621 Rewrite return value code.
4622 (distribute_notes): Set added_notes_insn to the earliest insn we added
4625 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
4627 PR rtl-optimization/82621
4628 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
4629 dest of one of those SETs is unused.
4631 2017-11-17 Richard Biener <rguenther@suse.de>
4634 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
4635 * tree-pretty-print.c (dump_generic_node): Handle
4636 annot_expr_parallel_kind.
4637 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
4638 * gimplify.c (gimple_boolify): Likewise.
4640 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4642 * config.gcc (extra_headers): Add cet.h for x86 targets.
4643 * config/i386/cet.h: New file.
4644 * doc/install.texi: Add --enable-cet/--disable-cet.
4646 2017-11-17 Richard Biener <rguenther@suse.de>
4648 PR tree-optimization/83017
4649 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
4650 (gen_parallel_loop): Properly count iterations.
4651 (parallelize_loops): Handle loop->can_be_parallel independent
4652 of flag_loop_parallelize_all. Make static profitability test match
4654 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
4655 * invoke.texi (parloops-min-per-thread): Document.
4657 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
4659 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
4660 upstreaming review comments.
4662 2017-11-17 Tamar Christina <tamar.christina@arm.com>
4664 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
4665 with fast unaligned access.
4666 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
4668 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
4670 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
4671 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
4673 2017-11-17 Richard Biener <rguenther@suse.de>
4675 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
4676 folding of references.
4678 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
4681 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
4682 of replacing call to strncmp with corresponding call to strcmp when
4685 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
4687 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
4688 option prefer-avx256 for skylake-avx512 configuration.
4689 * config/i386/i386.c (ix86_option_override_internal): Ditto.
4690 (get_builtin_code_for_version): Ditto.
4692 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
4693 Monk Chiang <sh.chiang04@gmail.com>
4695 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
4696 (FIXED_REGISTERS): Reserve more register numbers.
4697 (CALL_USED_REGISTERS): Likewise.
4698 (REG_ALLOC_ORDER): Likewise.
4699 (REG_CLASS_CONTENTS): Likewise.
4700 (REGISTER_NAMES): Likewise.
4702 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
4703 Kito Cheng <kito.cheng@gmail.com>
4705 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
4707 * config/nds32/iterators.md: Add vector mode iterators and attributes.
4709 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
4711 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
4712 parameter list for vec_splats.
4714 2017-11-16 Joseph Myers <joseph@codesourcery.com>
4716 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
4718 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
4720 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
4723 * doc/install.texi: Document incompatibility of Perl >=5.6.26
4724 with the required version of automake 1.11.6.
4726 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
4728 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
4730 (DU_C2_3_power9): Correct reservation combinations.
4731 (FP_DIV_power9, VEC_DIV_power9): New.
4732 (power9-alu): Split out rotate/shift...
4733 (power9-rot): ...to here, correct dispatch resource.
4734 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
4736 (power9-fp): Correct latency.
4737 (power9-sdiv): Add div/sqrt resource.
4738 (power9-ddiv): Correct latency, add div/sqrt resource.
4739 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
4740 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
4742 (power9-qpdiv, power9-qpmul): Adjust resource usage.
4744 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
4746 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
4747 switch statement mapping KF built-ins to TF built-ins if we don't
4748 have the proper ISA 3.0 assembler support.
4750 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4752 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
4753 (lower_emutls_function_body): Do not compute it.
4755 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4757 * ipa-split.c (split_bb_info): Turn time to sreal.
4758 (split_point): Likewise.
4759 (dump_split_point): Likewise.
4760 (fine_split_points): Likewise.
4761 (execute_split_functions): Only zero split_bbs; turn time to sreals.
4763 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4765 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
4768 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4770 * predict.c (combine_predictions_for_bb): Preserve zero predicted
4772 (expensive_function_p): Remove useless assert.
4773 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
4775 2017-11-16 Martin Sebor <msebor@redhat.com>
4777 PR tree-optimization/82588
4778 PR tree-optimization/82583
4779 * tree-vrp.c (check_array_ref): Handle flexible array members,
4780 string literals, and inner indices.
4781 (search_for_addr_array): Add detail to diagnostics.
4783 2017-11-16 Nathan Sidwell <nathan@acm.org>
4787 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
4788 (SET_DECL_ASSEMBLER_NAME): Forward to
4789 overwrite_decl_assembler_name.
4790 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
4791 (overwrite_decl_assembler_name): Declare.
4792 * tree.c (overwrite_decl_assembler_name): New.
4793 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
4794 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
4795 (LANG_HOOKS_INITIALIZER): Add it.
4796 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
4797 * langhooks.c (lhd_set_decl_assembler_name): Use
4798 SET_DECL_ASSEMBLER_NAME.
4799 (lhd_overwrite_decl_assembler_name): Default implementation.
4801 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
4802 Jackson Woodruff <jackson.woodruff@arm.com>
4804 PR tree-optimization/71026
4805 * match.pd: Canonicalize constant multiplies in division.
4807 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
4809 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
4810 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
4811 Remove OPT_fomit_frame_pointer entry.
4812 * common/config/arc/arc-common.c: Likewise.
4813 * common/config/arm/arm-common.c: Likewise.
4814 * common/config/avr/avr-common.c: Likewise.
4815 * common/config/c6x/c6x-common.c: Likewise.
4816 * common/config/cr16/cr16-common.c: Likewise.
4817 * common/config/cris/cris-common.c: Likewise.
4818 * common/config/epiphany/epiphany-common.c: Likewise.
4819 * common/config/fr30/fr30-common.c: Likewise.
4820 * common/config/frv/frv-common.c: Likewise.
4821 * common/config/ia64/ia64-common.c: Likewise.
4822 * common/config/iq2000/iq2000-common.c: Likewise.
4823 * common/config/lm32/lm32-common.c: Likewise.
4824 * common/config/m32r/m32r-common.c: Likewise.
4825 * common/config/mcore/mcore-common.c: Likewise.
4826 * common/config/microblaze/microblaze-common.c: Likewise.
4827 * common/config/mips/mips-common.c: Likewise.
4828 * common/config/mmix/mmix-common.c: Likewise.
4829 * common/config/mn10300/mn10300-common.c: Likewise.
4830 * common/config/nios2/nios2-common.c: Likewise.
4831 * common/config/pa/pa-common.c: Likewise.
4832 * common/config/pdp11/pdp11-common.c: Likewise.
4833 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
4834 * common/config/riscv/riscv-common.c: Likewise.
4835 * common/config/rs6000/rs6000-common.c: Likewise.
4836 * common/config/rx/rx-common.c: Likewise.
4837 * common/config/s390/s390-common.c: Likewise.
4838 * common/config/sh/sh-common.c: Likewise.
4839 * common/config/sparc/sparc-common.c: Likewise.
4840 * common/config/tilegx/tilegx-common.c: Likewise.
4841 * common/config/tilepro/tilepro-common.c: Likewise.
4842 * common/config/v850/v850-common.c: Likewise.
4843 * common/config/visium/visium-common.c: Likewise.
4844 * common/config/xstormy16/xstormy16-common.c: Likewise.
4845 * common/config/xtensa/xtensa-common.c: Likewise.
4846 * invoke.texi (-fomit-frame-pointer): Update documentation.
4848 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4850 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
4852 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4854 * cfg.c (scale_bbs_frequencies_int,
4855 cale_bbs_frequencies_gcov_type): Remove.
4856 * cfg.h (scale_bbs_frequencies_int,
4857 cale_bbs_frequencies_gcov_type): Remove.
4859 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4861 * tree-ssa-loop-manip.c
4862 (scale_dominated_blocks_in_loop): Update to profile counts.
4863 (tree_transform_and_unroll_loop): Likewise.
4865 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4867 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
4868 scale_bbs_frequencies_int.
4870 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4872 * final.c (compute_alignments): Use counts rather than frequencies.
4874 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4876 * cfgloopanal.c: Include sreal.h
4877 (average_num_loop_insns): Use counts and sreal for accounting.
4879 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4881 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
4884 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4885 Kito Cheng <kito.cheng@gmail.com>
4887 * config/nds32/constraints.md: Provide more constraints.
4888 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
4889 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
4890 support constraints usage.
4892 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4894 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
4896 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4897 Kito Cheng <kito.cheng@gmail.com>
4899 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
4900 * config/nds32/nds32.opt: Refine the layout.
4901 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
4902 TARGET_EXT_STRING): Support new options.
4903 * config/nds32/nds32.h: Likewise.
4904 * config/nds32/nds32.md: Likewise.
4905 * config/nds32/nds32-predicates.c: Likewise.
4906 * config/nds32/constraints.md: Likewise.
4907 * common/config/nds32/nds32-common.c: Likewise.
4909 2017-11-16 Julia Koval <julia.koval@intel.com>
4912 * config/i386/gfniintrin.h: Add sse check.
4913 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
4915 2017-11-16 Julia Koval <julia.koval@intel.com>
4917 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
4918 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
4919 (ix86_handle_option): Handle -mavx512vbmi2.
4920 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
4921 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4922 * config/i386/i386-c.c (__AVX512VBMI2__): New.
4923 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
4924 (ix86_valid_target_attribute_inner_p): Ditto.
4925 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
4926 * config/i386/i386.opt (mavx512vbmi2): New option.
4927 * doc/invoke.texi: Add new option.
4929 2017-11-16 Julia Koval <julia.koval@intel.com>
4931 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
4932 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
4933 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
4934 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
4935 _mm512_gf2p8mul_epi8): New intrinsics.
4936 * config/i386/i386-builtin-types.def
4937 (V64QI_FTYPE_V64QI_V64QI): New type.
4938 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
4939 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
4940 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
4941 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
4942 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
4943 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4945 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
4947 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
4948 explicitly as a stream of bytes.
4950 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4952 * config/rs6000/altivec.h (vec_xst_be): New #define.
4953 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
4954 and externalize from *altivec_vperm_<mode>_internal.
4955 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
4957 (XL_BE_V8HI): Likewise.
4958 (XL_BE_V4SI): Likewise.
4959 (XL_BE_V4SI): Likewise.
4960 (XL_BE_V2DI): Likewise.
4961 (XL_BE_V4SF): Likewise.
4962 (XL_BE_V2DF): Likewise.
4963 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
4964 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
4965 all array entries with these keys: VSX_BUILTIN_VEC_XL,
4966 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
4967 VSX_BUILTIN_VEC_XST_BE.
4968 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
4969 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
4971 (altivec_init_builtins): Replace conditional calls to def_builtin
4972 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
4973 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
4974 with unconditional calls. Remove calls to def_builtin for
4975 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
4976 __builtin_vec_xst_be.
4977 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
4978 to define_expand, and add alternate RTL generation for P8.
4979 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
4980 vsx_ld_elemrev_v8hi.
4981 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
4982 add alternate RTL generation for P8.
4983 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
4984 vsx_ld_elemrev_v16qi.
4985 (vsx_st_elemrev_v8hi): Convert define_insn
4986 to define_expand, and add alternate RTL generation for P8.
4987 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
4988 vsx_st_elemrev_v8hi.
4989 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
4990 add alternate RTL generation for P8.
4991 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
4992 vsx_st_elemrev_v16qi.
4994 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
4997 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
4998 TARGET_AVX512ER check.
4999 (ix86_option_override_internal): Set MASK_VZEROUPPER if
5000 neither -mzeroupper nor -mno-zeroupper is used and
5001 TARGET_EMIT_VZEROUPPER is set.
5002 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
5003 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
5005 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
5007 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5008 folding of vector compares.
5009 (fold_build_vec_cmp): New helper function.
5010 (fold_compare_helper): New helper function.
5011 (builtin_function_type): Add compare builtins to the list of functions
5012 having unsigned arguments. Cosmetic updates to comment indentation.
5013 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
5014 the not+eq combination.
5016 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5018 PR tree-optimization/82726
5019 PR tree-optimization/70754
5020 * tree-predcom.c (order_drefs_by_pos): New function.
5021 (combine_chains): Move code setting has_max_use_after to...
5022 (try_combine_chains): ...here. New parameter. Sort combined chains
5023 according to position information.
5024 (tree_predictive_commoning_loop): Update call to above function.
5025 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
5027 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5029 PR tree-optimization/82726
5031 2017-01-23 Bin Cheng <bin.cheng@arm.com>
5033 PR tree-optimization/70754
5034 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
5035 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
5036 combined stmt before it if not NULL.
5037 (combine_chains): Process refs reversely and compute dominance point
5041 2017-02-23 Bin Cheng <bin.cheng@arm.com>
5043 PR tree-optimization/79663
5044 * tree-predcom.c (combine_chains): Process refs in reverse order
5045 only for ZERO length chains, and add explaining comment.
5047 2017-11-15 Tamar Christina <tamar.christina@arm.com>
5049 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
5050 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
5051 * doc/invoke.texi (ARM Options): Add armv8.3-a.
5053 2017-11-15 Tamar Christina <tamar.christina@arm.com>
5055 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
5057 2017-11-15 Martin Liska <mliska@suse.cz>
5059 * tree-cfg.c (pass_warn_function_return::execute):
5060 Compare warn_return_type for greater than zero.
5062 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
5066 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
5067 to return true on Xeon and not on Xeon Phi.
5068 (ix86_check_avx256_register): Changed to ...
5069 (ix86_check_avx_upper_register): ... this. Add extra check for
5070 VALID_AVX512F_REG_OR_XI_MODE.
5071 (ix86_avx_u128_mode_needed): Changed
5072 ix86_check_avx256_register to ix86_check_avx_upper_register.
5073 (ix86_check_avx256_stores): Changed to ...
5074 (ix86_check_avx_upper_stores): ... this. Changed
5075 ix86_check_avx256_register to ix86_check_avx_upper_register.
5076 (ix86_avx_u128_mode_after): Changed
5077 avx_reg256_found to avx_upper_reg_found. Changed
5078 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
5079 (ix86_avx_u128_mode_entry): Changed
5080 ix86_check_avx256_register to ix86_check_avx_upper_register.
5081 (ix86_avx_u128_mode_exit): Ditto.
5082 * config/i386/i386.h: (host_detect_local_cpu): New define.
5084 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5086 * config/arm/xgene1.md (xgene1): Split into automatons
5087 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
5088 (xgene1_f_load): Adjust reservations and/or types.
5089 (xgene1_f_store): Likewise.
5090 (xgene1_load_pair): Likewise.
5091 (xgene1_store_pair): Likewise.
5092 (xgene1_fp_load1): Likewise.
5093 (xgene1_load1): Likewise.
5094 (xgene1_store1): Likewise.
5095 (xgene1_move): Likewise.
5096 (xgene1_alu): Likewise.
5097 (xgene1_simd): Likewise.
5098 (xgene1_bfm): Likewise.
5099 (xgene1_neon_load1): Likewise.
5100 (xgene1_neon_store1): Likewise.
5101 (xgene1_neon_logic): Likewise.
5102 (xgene1_neon_st1): Likewise.
5103 (xgene1_neon_ld1r): Likewise.
5104 (xgene1_alu_cond): Added.
5105 (xgene1_shift_reg): Likwise.
5106 (xgene1_bfx): Likewise.
5107 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
5109 2017-11-15 Jakub Jelinek <jakub@redhat.com>
5112 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
5114 (can_widen_mult_without_libcall): New function.
5115 (expand_mul_overflow): If only checking unsigned mul overflow,
5116 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
5117 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
5118 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
5119 (expand_DIVMOD): Formatting fix.
5120 * expmed.h (expand_mult): Add NO_LIBCALL argument.
5121 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
5122 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
5124 PR tree-optimization/82977
5125 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
5126 constructed temporary to strlen_to_stridx.put.
5128 2017-11-15 Martin Liska <mliska@suse.cz>
5130 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
5131 * configure: Regenerate.
5133 2017-11-15 Martin Liska <mliska@suse.cz>
5136 * config/sh/sh-mem.cc: Use proper probability for
5139 2017-11-14 Jeff Law <law@redhat.com>
5141 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
5142 the red zone for stack_clash_protection_final_dynamic_probe targets
5143 when the total dynamic stack size is zero bytes.
5145 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
5146 blocks is post order.
5148 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
5150 * dumpfile.h (TDF_COMPARE_DEBUG): New.
5151 * final.c (rest_of_clean_state): Set it for the
5152 -fcompare-debug dump.
5153 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
5154 class when TDF_COMPARE_DEBUG is set.
5156 * dwarf2out.c (gen_producer_string): Discard
5159 2017-11-15 Joseph Myers <joseph@codesourcery.com>
5162 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
5163 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
5164 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
5165 (__TGMATH_REAL_2_3): Remove macros.
5166 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
5167 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
5169 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
5171 (remquo): Define using __TGMATH_REAL_3.
5173 2017-11-14 Jeff Law <law@redhat.com>
5175 * vr-values.c: New file with contents extracted from tree-vrp.c.
5176 * Makefile.in (OBJS): Add vr-values.o
5177 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
5178 (set_value_range, set_and_canonicalize_value_range): Likewise.
5179 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
5180 (value_range_constant_singleton, symbolic_range_p): Likewise.
5181 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
5182 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
5183 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
5184 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
5185 (find_case_label_range, find_case_label_index): Likewise.
5186 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
5187 (range_int_cst_singleton_p, value_inside_range): Likewise.
5188 (get_single_symbol): Likewise.
5189 (switch_update): Move structure definition here.
5190 (to_remove_edges, to_update_switch_stmts): Provide externs.
5191 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
5192 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
5193 (vrp_val_is_min, set_value_range): Likewise.
5194 (set_and_canonicalize_value_range, copy_value_range): Likewise.
5195 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
5196 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
5197 (range_is_nonnull, range_int_cst_p): Likewwise.
5198 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
5199 (get_single_symbol, operand_less_p): Likewise
5200 (compare_values_warnv, compare_values): Likewise.
5201 (value_inside_range, value_range_constant_singleton): Likewise.
5202 (zero_nonzero_bitgs_from_vr): Likewise.
5203 (extract_range_from_binary_expr_1): Likewise.
5204 (overflow_comparison_p): Likewise.
5205 (to_remove_edges, to_update_switch_stmts): Likewise.
5206 (find_case_label-index, find_case_label_range): Likewise.
5207 (switch_update, set_value_range_to_nonnegative): Remove.
5208 (set_value_range_to_truthvalue): Likewise.
5209 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
5210 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
5211 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
5212 (find_case_label_ranges, test_for_singularity): Likewise.
5213 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
5214 (x_vr_values): Move to its remaining use site.
5216 2017-11-10 Jeff Law <law@redhat.com>
5218 * vr-values.h (VR_INITIALIZER): Move #define here.
5219 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
5220 * Makefile.in (OBJS): Add tree-evrp.o
5221 * tree-vrp.h (assert_info): Move structure definition here.
5222 (set_value_range_to_varying): Prototype.
5223 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
5224 (infer_value_range, register_edge_assert_for): Likewise.
5225 (stmt_interesting_for_vrp): Likewise.
5226 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
5227 (set_value_range_to_varying): No longer static.
5228 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
5229 (infer_value_range, register_edge_assert_for): Likewise.
5231 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
5233 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
5234 generate the XXBRD instruction.
5236 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
5237 (rs6000_builtin_type_compatible): Treat _Float128 and long double
5238 as being compatible if -mabi=ieeelongdouble.
5239 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
5240 to setup float128 built-ins with hardware support.
5241 (BU_FLOAT128_HW_2): Likewise.
5242 (BU_FLOAT128_HW_3): Likewise.
5243 (BU_FLOAT128_HW_VSX_1): Likewise.
5244 (BU_FLOAT128_HW_VSX_2): Likewise.
5245 (scalar_extract_expq): Change float128 built-in functions to
5246 accommodate having both KFmode and TFmode functions. Use the
5247 KFmode variant as the default.
5248 (scalar_extract_sigq): Likewise.
5249 (scalar_test_neg_qp): Likewise.
5250 (scalar_insert_exp_q): Likewise.
5251 (scalar_insert_exp_qp): Likewise.
5252 (scalar_test_data_class_qp): Likewise.
5253 (sqrtf128_round_to_odd): Delete processing the round to odd
5254 built-in functions as special built-in functions, and define them
5255 as float128 built-ins. Use the KFmode variant as the default.
5256 (truncf128_round_to_odd): Likewise.
5257 (addf128_round_to_odd): Likewise.
5258 (subf128_round_to_odd): Likewise.
5259 (mulf128_round_to_odd): Likewise.
5260 (divf128_round_to_odd): Likewise.
5261 (fmaf128_round_to_odd): Likewise.
5262 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
5263 support for KFmode and TFmode xststdcqp calls.
5264 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
5265 point, switch the built-in handlers for the get/set float128
5266 exponent, get float128 mantissa, float128 test built-ins, and the
5267 float128 round to odd built-in functions. Eliminate creating the
5268 float128 round to odd built-in functions as special built-ins.
5269 (rs6000_init_builtins): Eliminate special creation of the float128
5270 round to odd built-in functions.
5271 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
5272 function insns to support both TFmode and KFmode variants.
5273 (xsxsigqp_<mode>): Likewise.
5274 (xsiexpqpf_<mode>): Likewise.
5275 (xsiexpqp_<mode>): Likewise.
5276 (xststdcqp_<mode>): Likewise.
5277 (xststdcnegqp_<mode>): Likewise.
5278 (xststdcqp_<mode>): Likewise.
5280 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5282 * tree-ssa-threadupdate.c (compute_path_counts): Remove
5283 unused path_in_freq_ptr parameter.
5284 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
5286 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5288 * ipa-inline.c (edge_badness): Dump sreal frequency.
5289 (compute_inlined_call_time): Match natural implementaiton ...
5290 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
5291 forgotten division by CGRAPH_FREQ_BASE.
5293 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5295 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
5296 Solaris 11. Update comment.
5297 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
5299 * config/sol2.h (STARTFILE_SPEC): Likewise.
5300 * configure: Regenerate.
5302 2017-11-14 Carl Love <cel@us.ibm.com>
5304 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
5305 le_ and be_ prefixes to swap* variables. Remove
5306 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
5309 2017-11-14 Jason Merrill <jason@redhat.com>
5311 Support GTY((cache)) on hash_map.
5312 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
5313 (ggc_cache_remove): Override it instead of ggc_mx.
5314 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
5315 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
5316 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
5317 (simple_cache_map_traits): Override maybe_mx.
5318 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
5319 (hash_map): Friend gt_cleare_cache.
5320 (gt_cleare_cache): New.
5321 * tree.h (tree_cache_traits): New hash_map traits class.
5322 (tree_cache_map): New typedef.
5324 2017-11-14 Richard Biener <rguenther@suse.de>
5326 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
5327 paramter and handling.
5328 (cleanup_control_flow_bb): Likewise.
5329 (cleanup_control_flow_pre): New helper performing a DFS walk
5330 to call cleanup_control_flow_bb in PRE order.
5331 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
5332 via cleanup_control_flow_pre.
5334 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
5336 * config/aarch64/aarch64-simd.md
5337 (aarch64_simd_bsl<mode>_internal): Remove DImode.
5338 (*aarch64_simd_bsl<mode>_alt): Likewise.
5339 (aarch64_simd_bsldi_internal): New.
5340 (aarch64_simd_bsldi_alt): Likewise.
5342 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5344 * tracer.c (better_p): Do not compare frequencies.
5345 * reg-stack.c (better_edge): Likewise.
5346 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
5349 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5351 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
5352 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
5353 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
5354 * lto-streamer-in.c (input_function): Use update_max_bb_count.
5355 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
5356 * predict.c (maybe_hot_frequency_p): Inline to ...
5357 (maybe_hot_count_p): ... here; rewrite to counts.
5358 (counts_to_freqs): Rename to ...
5359 (update_max_bb_count): ... this one.
5360 (expensive_function_p): Use counts.
5361 (estimate_bb_frequencies): Update.
5362 (rebuild_frequencies): Update.
5363 * predict.h (counts_to_freqs): Rename to ...
5364 (update_max_bb_count): ... this one.
5365 * profile.c (compute_branch_probabilities): Add debug info
5366 * tree-inline.c (expand_call_inline): Update debug info.
5367 (optimize_inline_calls): Use update_max_bb_count..
5368 (tree_function_versioning): Use update_max_bb_count..
5369 * value-prof.c (gimple_value_profile_transformations):
5370 Do not use update_max_bb_count.
5372 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5374 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
5375 always use frequencies.
5377 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5379 * bb-reorder.c: Remove frequencies from comments.
5380 (better_edge_p): Use profile counts.
5381 (find_traces): Dump profile counts.
5382 (rotate_loop): Use profile counts.
5383 (find_traces_1_round): Likewise.
5384 (connect_better_edge_p): Use counts instead of probabilities for
5386 (copy_bb_p): Drop early check for non-0 frequency.
5387 (sanitize_hot_paths): Update comments.
5389 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5391 * ipa-split.c (struct split_point): Add count.
5392 (consider_split): Do not compute incoming frequency; compute incoming
5393 count and store it to split_point.
5394 (split_function): Set count of the call to split part correctly.
5396 2017-11-13 Carl Love <cel@us.ibm.com>
5398 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
5400 2017-11-13 Tom Tromey <tom@tromey.com>
5402 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
5404 2017-11-13 Carl Love <cel@us.ibm.com>
5406 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
5407 Add support for builtins:
5408 unsigned int vec_first_{,miss}_match_{,or_eos}index,
5409 vector {un,}signed {char,int,short},
5410 vector {un,}signed {char,int,short}) arguments.
5411 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
5412 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
5413 Add BU_P9V_AV_2 expansions for the builtins.
5414 * config/rs6000/altivec.h (vec_first_match_index,
5415 vec_first_mismatch_index, vec_first_match_or_eos_index,
5416 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
5417 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
5418 new extern declaration.
5419 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
5420 * config/rs6000/vsx.md (first_match_index_<mode>,
5421 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
5422 first_mismatch_or_eos_index_<mode>): Add define expand.
5423 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
5424 * doc/extend.texi: Update the built-in documenation file for the new
5427 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
5429 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
5430 into the min/max operations for _Float<N> and _Float<N>X types.
5432 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
5435 * dwarf2out.c (do_eh_frame): New static variable.
5436 (dwarf2out_begin_prologue): Set it.
5437 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
5439 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5441 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
5443 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
5445 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
5448 * debug.h (dwarf2out_do_eh_frame): Declare.
5449 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
5450 (dwarf2out_do_frame): Use it.
5451 (dwarf2out_do_cfi_asm): Likewise.
5452 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
5453 (dwarf2out_assembly_start): Likewise.
5454 (dwarf2out_begin_prologue): Fix comment.
5455 * toplev.c (compile_file): Always call dwarf2out_frame_finish
5456 if the target needs either debug or unwind DWARF2 info.
5457 * lto-opts.c (lto_write_options): Do not save -fexceptions,
5458 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
5459 -ftrapping-math, -ftrapv and -fwrapv.
5461 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5463 * cgraph.c (cgraph_edge::sreal_frequency): New function.
5464 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
5465 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
5466 (estimate_edge_size_and_time): Likewise.
5467 (ipa_merge_fn_summary_after_inlining): Likewise.
5468 * ipa-inline.c (cgraph_freq_base_rec): Remove.
5469 (compute_uninlined_call_time): Use sreal_frequency.
5470 (compute_inlined_call_time): Likewise.
5471 (ipa_inline): Do not initialize cgraph_freq_base_rec.
5472 * profile-count.c: Include sreal.h.
5473 (profile_count::to_sreal_scale): New.
5474 * profile-count.h: Forward declare sreal.
5475 (profile_count::to_sreal_scale): Declare.
5477 2017-11-13 Nathan Sidwell <nathan@acm.org>
5479 * diagnostic.c (maybe_line_and_column): New.
5480 (diagnostic_get_location_text): Use it.
5481 (diagnostic_report_current_module): Likewise.
5482 (test_diagnostic_get_location_text): Add tests.
5484 2017-11-13 Luis Machado <luis.machado@linaro.org>
5486 * doc/md.texi (Specifying processor pipeline description): Fix
5487 incorrect latency for the div instruction example.
5489 2017-11-13 Jakub Jelinek <jakub@redhat.com>
5491 PR tree-optimization/78821
5492 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
5493 that bit_not_p is the same.
5494 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
5495 (split_group): Count precisely bit_not_p bits in each statement.
5496 (invert_op): New function.
5497 (imm_store_chain_info::output_merged_store): Use invert_op to
5498 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
5499 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
5501 2017-11-13 Martin Liska <mliska@suse.cz>
5503 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
5504 (struct source_info): Likewise.
5505 (add_branch_counts): Likewise.
5506 (add_line_counts): Likewise.
5507 (function_summary): Likewise.
5508 (output_intermediate_line): Likewise.
5509 (generate_results): Likewise.
5511 2017-11-13 Martin Liska <mliska@suse.cz>
5513 * gcov.c (struct block_info): Remove typedef for block_t.
5514 (struct line_info): Likewise.
5515 (line_info::has_block): Likewise.
5516 (EXIT_BLOCK): Likewise.
5517 (unblock): Likewise.
5518 (circuit): Likewise.
5519 (get_cycles_count): Likewise.
5520 (process_file): Likewise.
5521 (read_graph_file): Likewise.
5522 (solve_flow_graph): Likewise.
5523 (find_exception_blocks): Likewise.
5524 (add_line_counts): Likewise.
5525 (accumulate_line_info): Likewise.
5526 (output_line_details): Likewise.
5528 2017-11-13 Martin Liska <mliska@suse.cz>
5530 * gcov.c (struct arc_info): Remove typedef for arc_t.
5531 (struct line_info): Likewise.
5532 (add_branch_counts): Likewise.
5533 (output_branch_count): Likewise.
5534 (function_info::~function_info): Likewise.
5535 (circuit): Likewise.
5536 (output_intermediate_line): Likewise.
5537 (read_graph_file): Likewise.
5538 (solve_flow_graph): Likewise.
5539 (find_exception_blocks): Likewise.
5540 (add_line_counts): Likewise.
5541 (accumulate_line_info): Likewise.
5542 (output_line_details): Likewise.
5543 (output_function_details): Likewise.
5545 2017-11-13 Martin Liska <mliska@suse.cz>
5547 * gcov.c (struct function_info): Remove typedef for function_t.
5548 (struct source_info): Likewise.
5549 (source_info::get_functions_at_location): Likewise.
5550 (solve_flow_graph): Likewise.
5551 (find_exception_blocks): Likewise.
5552 (add_line_counts): Likewise.
5553 (output_intermediate_file): Likewise.
5554 (process_file): Likewise.
5555 (generate_results): Likewise.
5556 (release_structures): Likewise.
5557 (read_graph_file): Likewise.
5558 (read_count_file): Likewise.
5559 (accumulate_line_counts): Likewise.
5560 (output_lines): Likewise.
5562 2017-11-13 Martin Liska <mliska@suse.cz>
5564 * gcov.c (function_info::function_info): Remove num_counts
5565 and add vector<gcov_type>.
5566 (function_info::~function_info): Use the vector.
5567 (process_file): Likewise.
5568 (read_graph_file): Likewise.
5569 (read_count_file): Likewise.
5570 (solve_flow_graph): Likewise.
5572 2017-11-13 Martin Liska <mliska@suse.cz>
5574 * gcov.c (function_info::is_artificial): New function.
5575 (process_file): Erase all artificial early.
5576 (generate_results): Skip as all artificial are already
5579 2017-11-13 Martin Liska <mliska@suse.cz>
5581 * gcov.c (read_graph_file): Store to global vector of functions.
5582 (read_count_file): Iterate the vector.
5583 (process_file): Likewise.
5584 (generate_results): Likewise.
5585 (release_structures): Likewise.
5587 2017-11-13 Jakub Jelinek <jakub@redhat.com>
5589 PR tree-optimization/82954
5590 * gimple-ssa-store-merging.c
5591 (imm_store_chain_info::coalesce_immediate_stores): If
5592 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
5594 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
5596 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
5597 Upddate call to ENDIAN_LANE_N.
5598 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
5599 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
5600 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
5601 and use aarch64_endian_lane_rtx.
5602 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
5604 2017-11-12 Tom de Vries <tom@codesourcery.com>
5606 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
5608 2017-11-12 Tom de Vries <tom@codesourcery.com>
5610 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
5612 2017-11-12 Tom de Vries <tom@codesourcery.com>
5614 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
5616 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
5617 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
5618 * defaults.h (ASM_OUTPUT_LABELREF): Same.
5620 2017-11-11 Martin Sebor <msebor@redhat.com>
5623 * doc/extend.texi (attribute nonstring): Remove spurious argument.
5626 * prefic.c (translate_name): Replace strncpy with memcpy to
5627 avoid -Wstringop-truncation.
5629 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5631 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
5633 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5635 * predict.c (maybe_hot_frequency_p): Do not use cfun.
5637 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5639 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
5642 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
5644 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
5645 enable generating XXBRH if the value is in a vector register.
5646 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
5647 value is in a vector register.
5648 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
5649 register to register bswap64's instead of doing the GPR sequence
5650 used on previous machines.
5651 (bswapdi2_xxbrd): New insn.
5652 (bswapdi2_reg): Disallow on ISA 3.0.
5653 (register to register bswap64 splitter): Do not split the insn on
5654 ISA 3.0 systems that use XXBRD.
5656 2017-11-10 Martin Sebor <msebor@redhat.com>
5659 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
5661 (find_subframework_file): Same.
5663 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5665 * auto-profile.c (afdo_indirect_call): Drop frequency.
5666 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
5667 (cgraph_node::create_edge): Drop frequency argument.
5668 (cgraph_node::create_indirect_edge): Drop frequency argument.
5669 (cgraph_edge::make_speculative): Drop frequency arguments.
5670 (cgraph_edge::resolve_speculation): Do not update frequencies
5671 (cgraph_edge::dump_edge_flags): Do not dump frequency.
5672 (cgraph_node::dump): Check consistency in IPA mode.
5673 (cgraph_edge::maybe_hot_p): Use IPA counter.
5674 (cgraph_edge::verify_count_and_frequency): Rename to ...
5675 (cgraph_edge::verify_count): ... this one; drop frequency checking.
5676 (cgraph_node::verify_node): Update.
5677 * cgraph.h (struct cgraph_edge): Drop frequency.
5678 (cgraph_edge::frequency): New function.
5679 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
5681 (cgraph_edge::rebuild_edges): Likewise.
5682 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
5683 (duplicate_thunk_for_node): Do not pass frequency.
5684 (cgraph_node::create_clone): Scale only counts.
5685 (cgraph_node::create_virtual_clone): Do not pass frequency.
5686 (cgraph_node::create_edge_including_clones): Do not pass frequency.
5687 (cgraph_node::create_version_clone): Do not pass frequency.
5688 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
5689 (cgraph_node::expand_thunk): Do not pass frequency.
5690 (cgraph_node::create_wrapper): Do not pass frequency.
5691 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
5693 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
5694 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
5695 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
5696 (gather_caller_stats): Use frequency function.
5697 (ipcp_cloning_candidate_p): Use frequency function.
5698 (ipcp_propagate_stage): Use frequency function.
5699 (get_info_about_necessary_edges): Use frequency function.
5700 (update_profiling_info): Update only IPA profile.
5701 (update_specialized_profile): Use frequency functoin.
5702 (perhaps_add_new_callers): Update only IPA profile.
5703 * ipa-devirt.c (ipa_devirt): Use IPA profile.
5704 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
5705 (dump_ipa_call_summary): Use frequency function.
5706 (estimate_edge_size_and_time): Use frequency function.
5707 (ipa_merge_fn_summary_after_inlining): Use frequency function.
5708 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
5709 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
5710 (update_noncloned_counts): ... ths one; scale counts only.
5711 (clone_inlined_nodes): Do not scale frequency.
5712 (inline_call): Do not pass frequency.
5713 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
5714 (compute_inlined_call_time): Use IPA profile.
5715 (want_inline_small_function_p): Use IPA profile.
5716 (want_inline_self_recursive_call_p): Use IPA profile.
5717 (edge_badness): Use IPA profile.
5718 (lookup_recursive_calls): Use IPA profile.
5719 (recursive_inlining): Do not pass frequency.
5720 (resolve_noninline_speculation): Do not update frequency.
5721 (inline_small_functions): Collect max of IPA profile.
5722 (dump_overall_stats): Dump IPA porfile.
5723 (dump_inline_stats): Dump IPA porfile.
5724 (ipa_inline): Collect IPA stats.
5725 * ipa-inline.h (clone_inlined_nodes): Update prototype.
5726 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
5727 (ipa_propagate_frequency): Use frequency function.
5728 (ipa_profile): Cleanup.
5729 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
5730 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
5731 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
5732 (input_node): Do not stream frequency.
5733 (input_edge): Do not stream frequency.
5734 (merge_profile_summaries): Scale only IPA profiles.
5735 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
5736 * predict.c (drop_profile): Do not recompute frequency.
5737 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
5738 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
5739 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
5740 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
5741 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
5742 * tree-inline.c (copy_bb): Do not scale frequency.
5743 (expand_call_inline): Do not scale frequency.
5744 (tree_function_versioning): Do not scale frequency.
5745 * ubsan.c (ubsan_create_edge): Do not pass frequency.
5747 2017-11-10 Julia Koval <julia.koval@intel.com>
5749 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
5750 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
5751 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
5752 (_mm256_maskz_gf2p8affine_epi64_epi8)
5753 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
5754 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
5755 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
5756 (__builtin_ia32_vgf2p8affineqb_v32qi)
5757 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
5758 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
5760 2017-11-10 Tamar Christina <tamar.christina@arm.com>
5764 (arm_option_override): Refactor.
5765 (arm_option_reconfigure_globals): New.
5766 (arm_options_perform_arch_sanity_checks): New.
5767 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
5769 (arm_options_perform_arch_sanity_checks): Likewise
5771 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
5773 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
5774 (power9-qpmul): New.
5775 * rs6000/rs6000.md ("type" attr): Add qmul.
5776 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
5777 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
5778 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
5780 2017-11-10 Martin Sebor <msebor@redhat.com>
5783 * builtins.c (compute_objsize): Handle arrays that
5784 compute_builtin_object_size likes to fail for. Make extern.
5785 * builtins.h (compute_objsize): Declare.
5786 (check_strncpy_sizes): New function.
5787 (expand_builtin_strncpy): Call check_strncpy_sizes.
5788 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
5789 -Wstringop-truncation.
5790 (gimple_fold_builtin_strncat): Same.
5791 * gimple.c (gimple_build_call_from_tree): Set call location.
5792 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
5793 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
5794 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
5795 (handle_builtin_strlen): Use strlen_to_stridx.
5796 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
5798 Use strlen_to_stridx.
5799 (pass_strlen::execute): Release strlen_to_stridx.
5800 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
5801 (-Wstringop-truncation): Document new option.
5803 2017-11-10 Martin Liska <mliska@suse.cz>
5805 PR gcov-profile/82702
5806 * gcov.c (main): Handle intermediate files in a different
5808 (get_gcov_intermediate_filename): New function.
5809 (output_gcov_file): Remove support of intermediate files.
5810 (generate_results): Allocate intermediate file.
5811 (release_structures): Clean-up properly fn_end.
5812 (output_intermediate_file): Start iterating with line 1.
5814 2017-11-10 Jakub Jelinek <jakub@redhat.com>
5816 PR tree-optimization/82929
5817 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
5818 ops_swapped_p non-static data member.
5819 (store_immediate_info::store_immediate_info): Clear it.
5820 (imm_store_chain_info::coalesce_immediate_stores): If swapping
5821 ops set ops_swapped_p.
5822 (count_multiple_uses): Handle ops_swapped_p.
5824 2017-11-10 Martin Liska <mliska@suse.cz>
5826 * coverage.c (coverage_init): Stream information about
5827 support of has_unexecuted_blocks.
5828 * doc/gcov.texi: Document that.
5829 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
5830 * gcov.c (read_graph_file): Likewise.
5831 (output_line_beginning): Fix a small issue with
5834 2017-11-10 Bin Cheng <bin.cheng@arm.com>
5836 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
5837 reference of trivial component.
5839 2017-11-10 Jakub Jelinek <jakub@redhat.com>
5842 * gimple-ssa-store-merging.c
5843 (pass_store_merging::terminate_all_aliasing_chains): For
5844 gimple_store_p stmts also call refs_output_dependent_p.
5846 PR rtl-optimization/82913
5847 * compare-elim.c (try_merge_compare): Punt if def_insn is not
5850 2017-11-09 Jeff Law <law@redhat.com>
5852 * vr-values.h: New file with vr_values class.
5853 * tree-vrp.c: Include vr-values.h
5854 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
5855 data objects into the vr_values class.
5856 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
5857 (get_value_range): Make it a member function within vr_values class.
5858 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
5859 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
5860 (op_with_constant_singleton_value_range): Likewise.
5861 (extract_range_for_var_from_comparison_expr): Likewise.
5862 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
5863 (extract_range_from_binary_expr): Likewise.
5864 (extract_range_from_unary_expr): Likewise.
5865 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
5866 (check_for_binary_op_overflow, extract_range_basic): Likewise.
5867 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
5868 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
5869 (compare_name_with_value, compare_names): Likewise.
5870 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
5871 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
5872 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
5873 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
5874 (extract_range_from_phi_node): Likewise.
5875 (simplify_truth_ops_using_ranges): Likewise.
5876 (simplify_div_or_mod_using_ranges): Likewise.
5877 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
5878 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
5879 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
5880 (simplify_float_conversion_using_ranges): Likewise.
5881 (simplify_internal_call_using_ranges): Likewise.
5882 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
5883 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
5884 poitner via x_vr_values for calls into gimple folder.
5885 (vrp_initialize_lattice): Make this the vr_values ctor.
5886 (vrp_free_lattice): Make this the vr_values dtor.
5887 (set_vr_value): New function.
5888 (class vrp_prop): Add vr_values data member. Add various member
5889 functions as well as member functions that delegate to vr_values.
5890 (check_array_ref): Make a member function within vrp_prop class.
5891 (search_for_addr_array, vrp_initialize): Likewise.
5892 (vrp_finalize): Likewise. Revamp to avoid direct access to
5893 vr_value, values_propagated, etc.
5894 (check_array_bounds): Extract vrp_prop class instance pointer from
5895 walk info structure. Use it to call member functions.
5896 (check_all_array_refs): Make a member function within vrp_prop class.
5897 Smuggle class instance pointer via walk info structure.
5898 (x_vr_values): New local static.
5899 (vrp_valueize): Use x_vr_values to get class instance.
5900 (vr_valueize_1): Likewise.
5901 (class vrp_folder): Add vr_values data member. Add various member
5902 functions as well as member functions that delegate to vr_values.
5903 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
5904 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
5905 class instance from vr_values. Use it to call member functions.
5906 (vrp_dom_walker): Add vr_values data member.
5907 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
5908 instance via x_vr_values.
5909 (identify_jump_threads): Accept vr_values as argument. Store
5910 it into the walker structure.
5911 (evrp_dom_walker): Add vr_values class data member. Add various
5913 (evrp_dom_walker::try_find_new_range): Use vr_values data
5914 member to access the memory allocator.
5915 (evrp_dom_walker::before_dom_children): Store vr_values class
5916 instance into the vrp_folder class.
5917 (evrp_dom_walker::push_value_range): Rework to avoid direct
5918 access to num_vr_values and vr_value.
5919 (evrp_dom_walker::pop_value_range): Likewise.
5920 (execute_early_vrp): Remove call to vrp_initialize_lattice.
5921 Use vr_values to get to dump_all_value_ranges member function.
5922 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
5923 and simplify_cond_using_ranges_2 via vrp_prop class instance.
5924 Pass vr_values class instance down to identify_jump_threads.
5925 Remove call to vrp_free_lattice.
5926 (debug_all_value_ranges): Remove.
5928 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
5929 (vrp_folder): Likewise.
5931 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
5932 Get it from the existing bitmap instead.
5933 (vrp_intersect_ranges_1): Likewise.
5935 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5937 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
5939 (store_immediate_info::store_immediate_info): Add bitnotp argument,
5940 set bit_not_p to it.
5941 (imm_store_chain_info::coalesce_immediate_stores): Break group
5942 if bit_not_p is different.
5943 (count_multiple_uses, split_group,
5944 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
5945 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
5946 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
5947 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
5948 to store_immediate_info ctor.
5950 2017-11-09 Jim Wilson <jimw@sifive.com>
5952 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
5953 (scan_prog_file): Likewise.
5955 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
5957 * bb-reorder.c (max_entry_frequency): Remove.
5958 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
5959 connect_traces, push_to_next_round_p): Remove prototypes.
5960 (find_traces_1_round): Use counts only.
5961 (push_to_next_round_p): Likewise.
5962 (find_traces): Likewise.
5963 (rotate_loop): Likewise.
5964 (find_traces_1_round): Likewise.
5965 (connect_traces): Likewise.
5966 (edge_order): Likewise.
5968 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
5970 * config/arm/arm.c (output_return_instruction): Add comments to
5971 indicate requirement for cmse_nonsecure_entry return to account
5972 for the size of clearing instruction output here.
5973 (thumb_exit): Likewise.
5974 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
5975 return in hardfloat mode.
5977 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
5979 * config/rs6000/rs6000.c (machine_function): Add a bool,
5980 "toc_is_wrapped_separately".
5981 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
5982 if it wasn't explicitly set or unset, we are optimizing for speed, and
5983 doing separate shrink-wrapping.
5984 (rs6000_get_separate_components): Enable the TOC component if
5985 saving the TOC register in the prologue.
5986 (rs6000_components_for_bb): Handle the TOC component.
5987 (rs6000_emit_prologue_components): Store the TOC register where needed.
5988 (rs6000_set_handled_components): Mark TOC as handled, if handled.
5989 (rs6000_emit_prologue): Don't save the TOC if that is already done.
5991 2017-11-09 Martin Jambor <mjambor@suse.cz>
5993 * ipa-param-manipulation.c: New file.
5994 * ipa-param-manipulation.h: Likewise.
5995 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
5996 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
5997 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
5998 (ipa_parm_adjustment): Likewise.
5999 (ipa_parm_adjustment_vec): Likewise.
6000 (ipa_get_vector_of_formal_parms): Moved declaration to
6001 ipa-param-manipulation.h.
6002 (ipa_get_vector_of_formal_parm_types): Likewise.
6003 (ipa_modify_formal_parameters): Likewise.
6004 (ipa_modify_call_arguments): Likewise.
6005 (ipa_combine_adjustments): Likewise.
6006 (ipa_dump_param_adjustments): Likewise.
6007 (ipa_modify_expr): Likewise.
6008 (ipa_get_adjustment_candidate): Likewise.
6009 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
6010 ipa-param-manipulation.c.
6011 (ipa_get_vector_of_formal_parm_types): Likewise.
6012 (ipa_modify_formal_parameters): Likewise.
6013 (ipa_modify_call_arguments): Likewise.
6014 (ipa_modify_expr): Likewise.
6015 (get_ssa_base_param): Likewise.
6016 (ipa_get_adjustment_candidate): Likewise.
6017 (index_in_adjustments_multiple_times_p): Likewise.
6018 (ipa_combine_adjustments): Likewise.
6019 (ipa_dump_param_adjustments): Likewise.
6020 * tree-sra.c: Also include ipa-param-manipulation.h
6021 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
6024 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6025 Alan Hayward <alan.hayward@arm.com>
6026 David Sherwood <david.sherwood@arm.com>
6028 * doc/sourcebuild.texi (vect_masked_store): Document.
6030 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6031 Alan Hayward <alan.hayward@arm.com>
6032 David Sherwood <david.sherwood@arm.com>
6034 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
6036 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6037 Alan Hayward <alan.hayward@arm.com>
6038 David Sherwood <david.sherwood@arm.com>
6040 * doc/sourcebuild.texi (vect_variable_length): Document.
6042 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6043 Alan Hayward <alan.hayward@arm.com>
6044 David Sherwood <david.sherwood@arm.com>
6046 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
6048 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6049 Alan Hayward <alan.hayward@arm.com>
6050 David Sherwood <david.sherwood@arm.com>
6052 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
6054 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6055 Alan Hayward <alan.hayward@arm.com>
6056 David Sherwood <david.sherwood@arm.com>
6058 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
6059 previously undocumented selectors.
6060 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
6062 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6064 * doc/rtl.texi (const_vector): Say that elements can be
6065 const_wide_ints too.
6066 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
6067 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
6068 (gen_vec_duplicate): Use it instead of CONSTANT_P.
6069 * optabs.c (expand_vector_broadcast): Likewise.
6071 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6072 Alan Hayward <alan.hayward@arm.com>
6073 David Sherwood <david.sherwood@arm.com>
6075 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
6076 scaled index even if the unscaled address was invalid.
6077 Don't increase the complexity of using a scale in that case.
6079 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6080 Alan Hayward <alan.hayward@arm.com>
6081 David Sherwood <david.sherwood@arm.com>
6083 * doc/rtl.texi: Rewrite the subreg rules so that they partition
6084 the inner register into REGMODE_NATURAL_SIZE bytes rather than
6085 UNITS_PER_WORD bytes.
6086 * emit-rtl.c (validate_subreg): Divide subregs into blocks
6087 based on REGMODE_NATURAL_SIZE of the inner mode.
6088 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
6089 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
6090 * expmed.c (lowpart_bit_field_p): Divide the value up into
6091 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
6092 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
6093 whether something is likely to occupy more than one register.
6095 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6098 * ipa-inline-transform.c (update_noncloned_frequencies): Use
6099 profile_count::adjust_for_ipa_scaling.
6100 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
6101 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
6103 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
6105 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6107 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
6108 (split_group): Add total_orig and total_new arguments, estimate the
6109 number of statements related to the store group without store merging
6110 and with store merging.
6111 (imm_store_chain_info::output_merged_store): Adjust split_group
6112 callers, punt if estimated number of statements with store merging
6113 is not smaller than estimated number of statements without it.
6115 (handled_load): Remove has_single_use checks.
6116 (pass_store_merging::process_store): Likewise.
6118 2017-11-09 Richard Biener <rguenther@suse.de>
6120 PR tree-optimization/82902
6121 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
6123 2017-11-09 Martin Liska <mliska@suse.cz>
6126 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
6129 2017-11-09 Martin Liska <mliska@suse.cz>
6131 PR tree-optimization/82669
6132 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
6134 2017-11-09 Martin Liska <mliska@suse.cz>
6136 PR gcov-profile/48463
6137 * coverage.c (coverage_begin_function): Output also end locus
6138 of a function and information whether the function is
6140 * gcov-dump.c (tag_function): Parse and print the information.
6141 * gcov.c (INCLUDE_MAP): Add include.
6142 (INCLUDE_SET): Likewise.
6143 (struct line_info): Move earlier in the source file because
6144 of vector<line_info> in function_info structure.
6145 (line_info::line_info): Likewise.
6146 (line_info::has_block): Likewise.
6147 (struct source_info): Add new member index.
6148 (source_info::get_functions_at_location): New function.
6149 (function_info::group_line_p): New function.
6150 (output_intermediate_line): New function.
6151 (output_intermediate_file): Use the mentioned function.
6152 (struct function_start): New.
6153 (struct function_start_pair_hash): Likewise.
6154 (process_file): Add code that identifies group functions.
6155 Assign lines either to global or function scope.
6156 (generate_results): Skip artificial functions.
6157 (find_source): Assign index for each source file.
6158 (read_graph_file): Read new flag artificial and end_line.
6159 (add_line_counts): Assign it either to global of function scope.
6160 (accumulate_line_counts): Isolate core of the function to
6161 accumulate_line_info and call it for both function and global
6163 (accumulate_line_info): New function.
6164 (output_line_beginning): Fix GNU coding style.
6165 (print_source_line): New function.
6166 (output_line_details): Likewise.
6167 (output_function_details): Likewise.
6168 (output_lines): Iterate both source (global) scope and function
6170 (struct function_line_start_cmp): New class.
6171 * doc/gcov.texi: Reflect changes in documentation.
6173 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6176 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
6177 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
6178 and similarly for not instead of neg.
6180 2017-11-08 Andi Kleen <ak@linux.intel.com>
6182 * config/i386/i386.opt: Add -mforce-indirect-call.
6183 * config/i386/predicates.md: Check for flag_force_indirect_call.
6184 * doc/invoke.texi: Document -mforce-indirect-call
6186 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
6188 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
6190 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
6191 config/riscv/riscv.c (predict.h): New include.
6192 (riscv_slow_unaligned_access_p): No longer static.
6193 (riscv_block_move_straight): Add require.
6194 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
6196 2017-11-08 Jakub Jelinek <jakub@redhat.com>
6199 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
6200 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
6201 nonimmediate_operand predicate for operand 1 instead of
6204 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6206 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
6208 * config/aarch64/constraints.md (Uml): New constraint.
6209 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
6212 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6214 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
6215 of two vec_duplicates into a vec_concat.
6217 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6219 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
6220 Simplify vec_merge of vec_duplicate and vec_concat.
6221 * config/aarch64/constraints.md (Utq): New constraint.
6222 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
6225 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6227 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
6228 Simplify vec_merge of vec_duplicate and const_vector.
6229 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
6231 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
6232 mode iterator. Update predicate on operand 1 to
6233 handle non-const_vec constants. Delete constraints.
6234 (*aarch64_combinez_be<mode>): Likewise for operand 2.
6236 2017-11-08 Jakub Jelinek <jakub@redhat.com>
6238 PR tree-optimization/78821
6239 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
6241 (store_operand_info::store_operand_info): Initialize it to false.
6242 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
6243 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
6244 store in the group, and if chain_info is non-NULL, to ignore altogether
6246 (compatible_load_p): Fail if bit_not_p does not match.
6247 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
6248 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
6249 (pass_store_merging::process_store): Adjust
6250 terminate_all_aliasing_chains calls to pass NULL in all current spots,
6251 call terminate_all_aliasing_chains newly when adding a store into
6252 a chain with non-NULL chain_info.
6254 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
6256 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
6258 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
6260 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6262 (aarch64_layout_frame): Initialise emit_frame_chain.
6263 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
6264 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
6266 2017-11-08 Martin Liska <mliska@suse.cz>
6268 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
6269 of gimple_call_internal_p.
6271 2017-11-07 Tom de Vries <tom@codesourcery.com>
6273 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
6275 2017-11-07 Tom de Vries <tom@codesourcery.com>
6277 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6280 2017-11-08 Martin Liska <mliska@suse.cz>
6283 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
6285 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
6287 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
6289 2017-11-07 Jakub Jelinek <jakub@redhat.com>
6292 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
6293 (*cmp<mode>_ccz_1): New insn with $k alternative.
6296 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
6297 RTX_COMM_COMPARE as commutative as well.
6298 (ix86_binary_operator_ok): Formatting fix.
6299 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
6300 *<code><mode>3<mask_name><round_saeonly_name>,
6301 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
6302 *<s>mul<mode>3_highpart<mask_name>,
6303 *vec_widen_umult_even_v16si<mask_name>,
6304 *vec_widen_umult_even_v8si<mask_name>,
6305 *vec_widen_umult_even_v4si<mask_name>,
6306 *vec_widen_smult_even_v16si<mask_name>,
6307 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
6308 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
6309 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
6310 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
6311 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
6312 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
6313 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
6314 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
6315 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
6316 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
6317 ix86_binary_operator_ok. Formatting fixes.
6318 (*<plusminus_insn><mode>3<mask_name><round_name>,
6319 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
6322 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
6324 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
6325 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
6326 eq and ne if TARGET_ISEL.
6327 (cmp): New code_iterator.
6328 (UNS, UNSU_, UNSIK): New code_attrs.
6329 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
6330 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
6332 ("eq<mode>3"): ... this.
6333 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
6335 ("ne<mode>3"): ... this.
6337 2017-11-07 Julia Koval <julia.koval@intel.com>
6340 * common/config/i386/i386-common.c
6341 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
6342 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
6343 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
6344 * config/i386/i386.opt: Ditto.
6345 * config/i386/i386.c (ix86_target_string): Ditto.
6346 (ix86_option_override_internal): Ditto.
6347 (ix86_init_mpx_builtins): Move MPX to args2.
6348 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
6349 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
6350 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
6351 __builtin_ia32_vgf2p8affineinvqb_v32qi,
6352 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
6353 __builtin_ia32_vgf2p8affineinvqb_v16qi,
6354 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
6356 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
6359 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
6360 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
6361 (zero-extendsidi peephole2): Remove peephole.
6363 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
6366 * stor-layout.c: Include attribs.h.
6367 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
6368 explicit lookup of "aligned" attribute.
6370 2017-11-07 Andrew Waterman <andrew@sifive.com>
6372 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
6373 (riscv_expand_block_move): Likewise.
6374 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
6376 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
6377 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
6378 * config/riscv/riscv.c (riscv_block_move_straight): New function.
6379 (riscv_adjust_block_mem): Likewise.
6380 (riscv_block_move_loop): Likewise.
6381 (riscv_expand_block_move): Likewise.
6382 * config/riscv/riscv.md (movmemsi): New pattern.
6384 2017-11-07 Michael Clark <michaeljclark@mac.com>
6386 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
6387 (MUSL_DYNAMIC_LINKER): Likewise.
6389 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
6391 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
6392 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
6395 2017-11-07 Richard Biener <rguenther@suse.de>
6397 * match.pd: Fix build.
6399 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
6400 Jackson Woodruff <jackson.woodruff@arm.com>
6402 PR tree-optimization/71026
6403 * match.pd: Canonicalize negate in division.
6405 2017-11-07 Sudakshina Das <sudi.das@arm.com>
6408 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
6410 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
6412 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
6413 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
6415 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
6417 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
6418 non-scalar integral types.
6419 * match.pd (negate_expr_p): Handle MINUS_EXPR.
6420 (-(A-B), -(~A)): New transformations.
6422 2017-11-07 Tom de Vries <tom@codesourcery.com>
6424 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
6425 semicolon after "do {} while (0)".
6426 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6427 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6428 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6429 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6430 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6432 2017-11-07 Tom de Vries <tom@codesourcery.com>
6434 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
6435 after "do {} while (0)".
6436 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6437 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6438 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6439 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6440 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6442 2017-11-07 Tom de Vries <tom@codesourcery.com>
6445 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
6447 (arm_rtx_costs_internal): Add missing semicolon after
6448 HANDLE_NARROW_SHIFT_ARITH call.
6450 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
6452 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
6453 disable isel if it was not set explicitly.
6455 2017-11-06 James Bowman <james.bowman@ftdichip.com>
6457 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
6458 (add_type_attribute) likewise.
6460 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
6462 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
6463 of struct ix86_frame.
6464 (ix86_initial_elimination_offset): Likewise.
6465 (ix86_expand_split_stack_prologue): Likewise.
6467 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
6469 * tree-vrp.h (enum value_range_type): Update stale comment.
6471 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6472 Alan Hayward <alan.hayward@arm.com>
6473 David Sherwood <david.sherwood@arm.com>
6475 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
6476 (aarch64_expand_vec_perm_const): Take the number of units too.
6477 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
6478 (aarch64_expand_vec_perm_const): Likewise.
6479 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
6480 (vec_perm<mode>): Update accordingly.
6482 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6483 Alan Hayward <alan.hayward@arm.com>
6484 David Sherwood <david.sherwood@arm.com>
6486 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
6487 Take the number of units too.
6488 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
6489 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
6490 but check for a vector mode before rather than after the call.
6491 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
6492 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
6493 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
6494 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
6495 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
6496 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
6497 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
6498 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
6499 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
6500 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
6501 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
6502 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
6503 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
6504 (aarch64_sqdmull2_n<mode>): Update accordingly.
6506 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6507 Alan Hayward <alan.hayward@arm.com>
6508 David Sherwood <david.sherwood@arm.com>
6510 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
6511 the number of units too.
6512 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
6513 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
6514 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
6515 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
6516 (vec_store_lanesxi<mode>): Update accordingly.
6518 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6519 Alan Hayward <alan.hayward@arm.com>
6520 David Sherwood <david.sherwood@arm.com>
6522 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
6523 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
6524 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
6525 of units rather than the mode.
6526 * config/aarch64/iterators.md (nunits): New mode attribute.
6527 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
6528 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
6529 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
6530 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
6531 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
6532 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
6533 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
6534 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
6535 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
6536 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
6537 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
6538 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
6539 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
6540 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
6541 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
6542 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
6543 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
6544 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
6545 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
6546 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
6547 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
6548 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
6549 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
6550 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
6551 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
6552 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
6553 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
6554 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
6555 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
6556 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
6557 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
6558 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
6559 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
6560 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
6561 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
6562 (aarch64_simd_vec_setv2di): Likewise.
6564 2017-11-06 Carl Love <cel@us.ibm.com>
6566 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
6568 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
6569 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
6571 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
6573 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
6574 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
6575 Add power 8 macro expansions.
6576 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
6577 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
6578 power 8 instructions. (VSX_XXBR): Add iterator.
6580 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
6582 * config/arm/arm.md (predicable_short_it): Change default to "no",
6583 improve documentation, remove uses that are identical to the default.
6584 (enabled_for_depr_it): Rename to enabled_for_short_it.
6585 * gcc/config/arm/arm-fixed.md (predicable_short_it):
6586 Remove default uses.
6587 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
6588 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
6589 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
6590 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
6592 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
6595 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
6596 float128 helper macros, which are no longer used after deleting
6597 the old 'q' built-in functions, and moving the round to odd
6598 built-in functions to being special built-in functions.
6599 (BU_FLOAT128_2): Likewise.
6600 (BU_FLOAT128_1_HW): Likewise.
6601 (BU_FLOAT128_2_HW): Likewise.
6602 (BU_FLOAT128_3_HW): Likewise.
6603 (FABSQ): Delete old 'q' built-in functions.
6604 (COPYSIGNQ): Likewise.
6605 (SQRTF128_ODD): Move round to odd built-in functions to be
6606 special built-in functions, so that we can handle
6607 -mabi=ieeelongdouble.
6608 (TRUNCF128_ODD): Likewise.
6609 (ADDF128_ODD): Likewise.
6610 (SUBF128_ODD): Likewise.
6611 (MULF128_ODD): Likewise.
6612 (DIVF128_ODD): Likewise.
6613 (FMAF128_ODD): Likewise.
6614 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
6615 built-in names to 'f128'.
6616 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
6617 old 'q' built-in functions, as the machine independent code for
6618 'f128' built-in functions handles this.
6619 (rs6000_expand_builtin): Add expansion for float128 round to odd
6620 functions, keying off on -mabi=ieeelongdouble of whether to use
6621 the KFmode or TFmode variant.
6622 (rs6000_init_builtins): Initialize the _Float128 round to odd
6624 * doc/extend.texi (PowerPC Built-in Functions): Document the old
6625 _Float128 'q' built-in functions are now mapped into the new
6626 'f128' built-in functions.
6628 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
6630 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
6631 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
6633 2017-11-06 David Malcolm <dmalcolm@redhat.com>
6636 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
6637 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
6638 * toplev.c: Include "ipa-fnsummary.h".
6639 (toplev::finalize): Call ipa_fnsummary_c_finalize.
6641 2017-11-06 Jakub Jelinek <jakub@redhat.com>
6643 PR tree-optimization/82838
6644 * gimple-ssa-store-merging.c
6645 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
6646 on a separate gimple_seq which is then appended to seq.
6648 2017-11-06 Jeff Law <law@redhat.com>
6651 * config/i386/i386.c (PROBE_INTERVAL): Remove.
6652 (get_probe_interval): New functions.
6653 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
6654 (ix86_adjust_stack_and_probe): Likewise.
6655 (output_adjust_stack_and_probe): Likewise.
6656 (ix86_emit_probe_stack_range): Likewise.
6657 (ix86_expand_prologue): Likewise.
6659 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6661 PR tree-optimization/82816
6662 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
6663 if the modes of the two types are the same.
6664 (convert_plusminus_to_widen): Likewise.
6666 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6668 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
6670 (usadv16qi): New define_expand.
6671 (usadv8hi): New define_expand.
6673 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
6676 * ipa-inline-transform.c (update_noncloned_frequencies): Always
6678 (inline_transform): Likewise.
6679 * predict.c (counts_to_freqs): Remove useless conditional.
6680 * profile-count.h (profile_count::apply_scale): Move sanity check.
6681 * tree-inline.c (copy_bb): Always scale.
6682 (copy_cfg_body): Likewise.
6684 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
6687 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
6689 (*cmp_ite0): Add enabled_for_depr_it attribute.
6690 (*cmp_ite1): Likewise.
6692 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
6694 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
6697 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6699 * tree-vrp.c (vrp_int_const_binop): Return true on success and
6700 return the value by pointer.
6701 (extract_range_from_multiplicative_op_1): Update accordingly.
6702 Return as soon as an operation fails.
6704 2017-11-05 Tom de Vries <tom@codesourcery.com>
6707 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
6708 (DEF_SANITIZER_BUILTIN): ... here.
6709 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
6710 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
6712 2017-11-05 Tom de Vries <tom@codesourcery.com>
6715 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
6717 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
6718 ASM_OUTPUT_BEFORE_CASE_LABEL call.
6719 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
6721 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6722 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6723 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6725 2017-11-05 Tom de Vries <tom@codesourcery.com>
6728 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
6731 2017-11-04 Michael Clark <michaeljclark@mac.com>
6733 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
6734 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
6736 (*addsi3_extended): Likewise.
6737 (*addsi3_extended2): Likewise.
6738 (<optab>si3): Likewise.
6739 (<optab>di3): Likewise.
6740 (<optab><mode>3): Likewise.
6741 (<*optabe>si3_internal): Likewise.
6742 (zero_extendqi<SUPERQI:mode>2): Likewise.
6743 (*add<mode>hi3): Likewise.
6744 (*xor<mode>hi3): Likewise.
6745 (<optab>di3): Likewise.
6746 (*<optab>si3_extend): Likewise.
6747 (*sge<u>_<X:mode><GPR:mode>): Likewise.
6748 (*slt<u>_<X:mode><GPR:mode>): Likewise.
6749 (*sle<u>_<X:mode><GPR:mode>): Likewise.
6751 2017-11-04 Andrew Waterman <andrew@sifive.com>
6753 * config/riscv/riscv.c (riscv_option_override): Conditionally set
6754 TARGET_STRICT_ALIGN based upon -mtune argument.
6756 2017-11-04 Andrew Waterman <andrew@sifive.com>
6758 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
6760 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
6762 * config/i386/i386.c (choose_basereg): Use optional scratch
6763 register and add assertion.
6764 (x86_emit_outlined_ms2sysv_save): Use scratch register when
6765 needed, and don't allocate stack.
6766 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
6767 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
6768 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
6770 2017-11-03 Jeff Law <law@redhat.com>
6772 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
6773 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
6774 to probe at the start of a noreturn function.
6776 2017-11-03 Jakub Jelinek <jakub@redhat.com>
6778 PR tree-optimization/78821
6779 * gimple-ssa-store-merging.c: Update the file comment.
6780 (MAX_STORE_ALIAS_CHECKS): Define.
6781 (struct store_operand_info): New type.
6782 (store_operand_info::store_operand_info): New constructor.
6783 (struct store_immediate_info): Add rhs_code and ops data members.
6784 (store_immediate_info::store_immediate_info): Add rhscode, op0r
6785 and op1r arguments to the ctor, initialize corresponding data members.
6786 (struct merged_store_group): Add load_align_base and load_align
6788 (merged_store_group::merged_store_group): Initialize them.
6789 (merged_store_group::do_merge): Update them.
6790 (merged_store_group::apply_stores): Pick the constant for
6791 encode_tree_to_bitpos from one of the two operands, or skip
6792 encode_tree_to_bitpos if neither operand is a constant.
6793 (class pass_store_merging): Add process_store method decl. Remove
6794 bool argument from terminate_all_aliasing_chains method decl.
6795 (pass_store_merging::terminate_all_aliasing_chains): Remove
6796 var_offset_p argument and corresponding handling.
6797 (stmts_may_clobber_ref_p): New function.
6798 (compatible_load_p): New function.
6799 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
6800 if there is overlap and rhs_code is not INTEGER_CST. For
6801 non-overlapping stores terminate group if rhs is not mergeable.
6802 (get_alias_type_for_stmts): Change first argument from
6803 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
6804 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
6805 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
6807 (get_location_for_stmts): Change first argument from
6808 auto_vec<gimple *> & to vec<gimple *> &.
6809 (struct split_store): Remove orig_stmts data member, add orig_stores.
6810 (split_store::split_store): Create orig_stores rather than orig_stmts.
6811 (find_constituent_stmts): Renamed to ...
6812 (find_constituent_stores): ... this. Change second argument from
6813 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
6814 to info structures rather than the statements.
6815 (split_group): Rename ALLOW_UNALIGNED argument to
6816 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
6817 it. Adjust find_constituent_stores caller.
6818 (imm_store_chain_info::output_merged_store): Handle rhs_code other
6819 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
6820 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
6821 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
6822 (mem_valid_for_store_merging): New function.
6823 (handled_load): New function.
6824 (pass_store_merging::process_store): New method.
6825 (pass_store_merging::execute): Use process_store method. Adjust
6826 terminate_all_aliasing_chains caller.
6828 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6830 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
6831 Return true for more constants, symbols and label references.
6832 (aarch64_valid_floating_const): Remove unused function.
6834 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6837 * config/aarch64/aarch64.c (aarch64_layout_frame):
6838 Undo forcing of LR at bottom of frame.
6840 2017-11-03 Jeff Law <law@redhat.com>
6843 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
6844 for int_registers_saved.
6846 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
6847 extracted from tree-ssa-dom.c.
6848 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
6849 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
6850 (record_equivalences_from_incoming_edge): Add additional argument
6851 to single_pred_edge_ignoring_loop_edges call.
6852 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
6853 (uncprop_dom_walker::before_dom_children): Add additional argument
6854 to single_pred_edge_ignoring_loop_edges call.
6855 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
6856 single_pred_edge_ignoring_loop_edges rather than open coding.
6857 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
6859 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
6861 * match.pd (-(-A)): Rewrite.
6863 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6865 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
6866 (rs6000_emit_int_cmove): New declaration.
6867 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
6868 (rs6000_emit_sISEL): Delete.
6869 (rs6000_emit_int_cmove): Make non-static.
6870 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
6871 instead of rs6000_emit_sISEL.
6873 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
6875 * asan.c (create_cond_insert_point): Maintain profile.
6876 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
6878 * basic-block.h (struct basic_block_def): Remove frequency.
6879 (EDGE_FREQUENCY): Use to_frequency
6880 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
6882 (find_traces): Update to use to_frequency.
6883 (find_traces_1_round): Likewise; use only IPA counts.
6884 (bb_to_key): Likewise.
6885 (connect_traces): Use IPA counts only.
6886 (copy_bb_p): Update to use to_frequency.
6887 (fix_up_crossing_landing_pad): Likewise.
6888 (sanitize_hot_paths): Likewise.
6889 * bt-load.c (basic_block_freq): Likewise.
6890 * cfg.c (init_flow): Set count_max to uninitialized.
6891 (check_bb_profile): Remove frequencies; check counts.
6892 (dump_bb_info): Do not dump frequencies.
6893 (update_bb_profile_for_threading): Update counts only.
6894 (scale_bbs_frequencies_int): Likewise.
6895 (MAX_SAFE_MULTIPLIER): Remove.
6896 (scale_bbs_frequencies_gcov_type): Update counts only.
6897 (scale_bbs_frequencies_profile_count): Update counts only.
6898 (scale_bbs_frequencies): Update counts only.
6899 * cfg.h (struct control_flow_graph): Add count-max.
6900 (update_bb_profile_for_threading): Update prototype.
6901 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
6902 (find_many_sub_basic_blocks): Likewise.
6903 * cfgcleanup.c (try_forward_edges): Likewise.
6904 (try_crossjump_to_edge): Likewise.
6905 * cfgexpand.c (expand_gimple_cond): Likewise.
6906 (expand_gimple_tailcall): Likewise.
6907 (construct_init_block): Likewise.
6908 (construct_exit_block): Likewise.
6909 * cfghooks.c (verify_flow_info): Check consistency of counts.
6910 (dump_bb_for_graph): Do not dump frequencies.
6911 (split_block_1): Do not update frequencies.
6912 (split_edge): Do not update frequencies.
6913 (make_forwarder_block): Do not update frequencies.
6914 (duplicate_block): Do not update frequencies.
6915 (account_profile_record): Do not update frequencies.
6916 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
6917 for global heuristics.
6918 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
6919 (expected_loop_iterations_unbounded): Use counts only.
6920 * cfgloopmanip.c (scale_loop_profile): Simplify.
6921 (create_empty_loop_on_edge): Simplify
6923 (duplicate_loop_to_header_edge): Simplify
6924 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
6925 (update_br_prob_note): Take care of removing note when profile
6927 (relink_block_chain): Do not dump frequency.
6928 (rtl_account_profile_record): Use to_frequency.
6929 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
6930 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
6931 (cgraph_update_edges_for_call_stmt_node): Likewise.
6932 (cgraph_edge::verify_count_and_frequency): Update.
6933 (cgraph_node::verify_node): Temporarily disable frequency verification.
6934 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
6935 to_cgraph_frequency.
6936 (cgraph_edge::rebuild_edges): Convert to ipa counts.
6937 * cgraphunit.c (init_lowered_empty_function): Do not initialize
6939 (cgraph_node::expand_thunk): Update profile.
6940 * except.c (dw2_build_landing_pads): Do not update frequency.
6941 * final.c (compute_alignments): Use to_frequency.
6942 (dump_basic_block_info): Do not dump frequency.
6943 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
6944 (dump_gimple_bb_header): Do not dump frequency.
6945 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
6947 * gimple-streamer-in.c (input_bb): Do not stream frequency.
6948 * gimple-streamer-out.c (output_bb): Do not stream frequency.
6949 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
6950 (init_before_recovery): Do not update frequency.
6951 (sched_create_recovery_edges): Do not update frequency.
6952 * hsa-gen.c (convert_switch_statements): Do not update frequency.
6953 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
6954 (ipa_cp_c_finalize): Set max_count to uninitialized.
6955 * ipa-fnsummary.c (get_minimal_bb): Use counts.
6956 (param_change_prob): Use counts.
6957 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
6959 * ipa-split.c (consider_split): Use to_frequency.
6960 (split_function): Use to_frequency.
6961 * ira-build.c (loop_compare_func): Likewise.
6962 (mark_loops_for_removal): Likewise.
6963 (mark_all_loops_for_removal): Likewise.
6964 * loop-doloop.c (doloop_modify): Do not update frequency.
6965 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
6967 * lto-streamer-in.c (input_function): Update count_max.
6968 * omp-expand.c (expand_omp_taskreg): Update count_max.
6969 * omp-simd-clone.c (simd_clone_adjust): Update profile.
6970 * predict.c (maybe_hot_frequency_p): Use to_frequency.
6971 (maybe_hot_count_p): Use ipa counts only.
6972 (maybe_hot_bb_p): Simplify.
6973 (maybe_hot_edge_p): Simplify.
6974 (probably_never_executed): Do not take frequency argument.
6975 (probably_never_executed_bb_p): Do not pass frequency.
6976 (probably_never_executed_edge_p): Likewise.
6977 (combine_predictions_for_bb): Check that profile is nonzero.
6978 (propagate_freq): Do not set frequency.
6979 (drop_profile): Simplify.
6980 (counts_to_freqs): Simplify.
6981 (expensive_function_p): Use to_frequency.
6982 (propagate_unlikely_bbs_forward): Simplify.
6983 (determine_unlikely_bbs): Simplify.
6984 (estimate_bb_frequencies): Add hack to silence graphite issues.
6985 (compute_function_frequency): Use ipa counts.
6986 (pass_profile::execute): Update.
6987 (rebuild_frequencies): Use counts only.
6988 (force_edge_cold): Use counts only.
6989 * profile-count.c (profile_count::dump): Dump new count types.
6990 (profile_count::differs_from_p): Check compatiblity.
6991 (profile_count::to_frequency): New function.
6992 (profile_count::to_cgraph_frequency): New function.
6993 * profile-count.h (struct function): Declare.
6994 (enum profile_quality): Add profile_guessed_local and
6995 profile_guessed_global0.
6996 (class profile_proability): Decrease number of bits to 29;
6997 update from_reg_br_prob_note and to_reg_br_prob_note.
6998 (class profile_count: Update comment; decrease number of bits
6999 to 61. Check compatibility.
7000 (profile_count::compatible_p): New private member function.
7001 (profile_count::ipa_p): New member function.
7002 (profile_count::operator<): Handle global zero correctly.
7003 (profile_count::operator>): Handle global zero correctly.
7004 (profile_count::operator<=): Handle global zero correctly.
7005 (profile_count::operator>=): Handle global zero correctly.
7006 (profile_count::nonzero_p): New member function.
7007 (profile_count::force_nonzero): New member function.
7008 (profile_count::max): New member function.
7009 (profile_count::apply_scale): Handle IPA scalling.
7010 (profile_count::guessed_local): New member function.
7011 (profile_count::global0): New member function.
7012 (profile_count::ipa): New member function.
7013 (profile_count::to_frequency): Declare.
7014 (profile_count::to_cgraph_frequency): Declare.
7015 * profile.c (OVERLAP_BASE): Delete.
7016 (compute_frequency_overlap): Delete.
7017 (compute_branch_probabilities): Do not use compute_frequency_overlap.
7018 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
7019 * sched-ebb.c (rank): Use counts only.
7020 * shrink-wrap.c (handle_simple_exit): Use counts only.
7021 (try_shrink_wrapping): Use counts only.
7022 (place_prologue_for_one_component): Use counts only.
7023 * tracer.c (find_best_predecessor): Use to_frequency.
7024 (find_trace): Use to_frequency.
7025 (tail_duplicate): Use to_frequency.
7026 * trans-mem.c (expand_transaction): Do not update frequency.
7027 * tree-call-cdce.c: Do not update frequency.
7028 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
7029 (gimple_merge_blocks): Likewise.
7030 (gimple_split_edge): Likewise.
7031 (gimple_duplicate_sese_region): Likewise.
7032 (gimple_duplicate_sese_tail): Likewise.
7033 (move_sese_region_to_fn): Likewise.
7034 (gimple_account_profile_record): Likewise.
7035 (insert_cond_bb): Likewise.
7036 * tree-complex.c (expand_complex_div_wide): Likewise.
7037 * tree-eh.c (lower_resx): Update profile.
7038 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
7040 (initialize_cfun): Do not initialize frequencies
7041 (freqs_to_counts): Delete.
7042 (copy_cfg_body): Ignore count parameter.
7043 (copy_body): Update.
7044 (expand_call_inline): Update count_max.
7045 (optimize_inline_calls): Update count_max.
7046 (tree_function_versioning): Update count_max.
7047 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
7048 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
7050 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
7051 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
7052 (try_peel_loop): Likewise.
7053 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
7055 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
7056 (tree_transform_and_unroll_loop): Do not use frequencies
7057 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
7058 Use reliable prediction only.
7059 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
7060 * tree-ssa-sink.c (select_best_block): Use to_frequency.
7061 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
7062 probability scaling.
7063 * tree-ssa-threadupdate.c (create_block_for_threading): Do
7064 not update frequency
7065 (any_remaining_duplicated_blocks): Likewise.
7066 (update_profile): Likewise.
7067 (estimated_freqs_path): Delete.
7068 (freqs_to_counts_path): Delete.
7069 (clear_counts_path): Delete.
7070 (ssa_fix_duplicate_block_edges): Likewise.
7071 (duplicate_thread_path): Likewise.
7072 * tree-switch-conversion.c (gen_inbound_check): Use counts.
7073 * tree-tailcall.c (decrease_profile): Do not update frequency.
7074 (eliminate_tail_call): Likewise.
7075 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
7076 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
7077 (optimize_mask_stores): Likewise.
7078 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
7079 * ubsan.c (ubsan_expand_null_ifn): Update profile.
7080 (ubsan_expand_ptr_ifn): Update profile.
7081 * value-prof.c (gimple_ic): Simplify.
7082 * value-prof.h (gimple_ic): Update prototype.
7083 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
7084 * ipa-inline.c (compute_uninlined_call_time): Be sure that
7086 (want_inline_self_recursive_call_p): Likewise.
7087 (resolve_noninline_speculation): Only cummulate defined counts.
7088 (inline_small_functions): Use nonzero_p.
7089 (ipa_inline): Do not access freed node.
7091 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7093 * config/aarch64/aarch64.c (aarch64_override_options_internal):
7094 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
7096 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
7098 * config/riscv/riscv.c (riscv_legitimize_move): Handle
7099 non-legitimate address.
7101 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7103 * config/rs6000/rs6000.md (*lt0_disi): Delete.
7104 (*lt0_<mode>di, *lt0_<mode>si): New.
7106 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7108 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
7109 TARGET_PAIRED_FLOAT.
7111 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
7112 Jim Wilson <jim.wilson@linaro.org>
7114 * config/aarch64/aarch64-cores.def (saphira): New CPU.
7115 * config/aarch64/aarch64-tune.md: Regenerated.
7116 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
7117 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
7119 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
7121 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
7122 (arc_expand_prologue): Restore blink for millicode.
7123 * config/arc/linux.h (LINK_EH_SPEC): Defined.
7125 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
7128 * config/i386/i386.c (ix86_vector_duplicate_value): Use
7129 gen_vec_duplicate after forcing the scalar into a register.
7131 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
7133 * combine (try_combine): Print the insns input to try_combine to the
7136 2017-11-02 Steve Ellcey <sellcey@cavium.com>
7139 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
7140 Remove second argument from aarch64_process_target_attr call.
7141 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
7143 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
7145 (aarch64_handle_attr_arch): Remove second argument.
7146 (aarch64_handle_attr_cpu): Ditto.
7147 (aarch64_handle_attr_tune): Ditto.
7148 (aarch64_handle_attr_isa_flags): Ditto.
7149 (aarch64_process_one_target_attr): Ditto.
7150 (aarch64_process_target_attr): Ditto.
7151 (aarch64_option_valid_attribute_p): Remove second argument.
7152 on aarch64_process_target_attr call.
7154 2017-11-02 David Malcolm <dmalcolm@redhat.com>
7156 * diagnostic.c: Include "selftest-diagnostic.h".
7157 (selftest::assert_location_text): New function.
7158 (selftest::test_diagnostic_get_location_text): New function.
7159 (selftest::diagnostic_c_tests): Call it.
7161 2017-11-02 David Malcolm <dmalcolm@redhat.com>
7163 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
7164 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
7165 (class selftest::test_diagnostic_context): Move to...
7166 * selftest-diagnostic.c: New file.
7167 * selftest-diagnostic.h: New file.
7169 2017-11-02 James Bowman <james.bowman@ftdichip.com>
7171 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
7172 offset range for FT32B.
7173 * config/ft32/ft32.h: option "mcompress" enables relaxation.
7174 * config/ft32/ft32.md: Add TARGET_NOPM.
7175 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
7176 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
7178 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
7180 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
7182 2017-11-02 Jeff Law <law@redhat.com>
7184 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
7185 virtual keyword on FINAL OVERRIDE members.
7187 * tree-ssa-propagate.h (ssa_propagation_engine): Group
7188 virtuals together. Add virtual destructor.
7189 (substitute_and_fold_engine): Similarly.
7191 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
7193 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
7195 2017-11-02 Richard Biener <rguenther@suse.de>
7197 PR tree-optimization/82795
7198 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
7200 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7202 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
7204 (gcc_GAS_CHECK_FEATURE): Remove.
7205 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
7206 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
7207 * configure: Regenerate.
7209 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
7211 * config/arc/arc.c (hwloop_optimize): Account for empty
7214 2017-11-02 Richard Biener <rguenther@suse.de>
7217 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
7218 Truncate ARRAY_REF index and element size.
7220 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
7222 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
7224 2017-11-01 Jeff Law <law@redhat.com>
7226 * tree-ssa-ccp.c (ccp_folder): New class derived from
7227 substitute_and_fold_engine.
7228 (ccp_folder::get_value): New member function.
7229 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
7230 (ccp_fold_stmt): Remove prototype.
7231 (ccp_finalize): Call substitute_and_fold from the ccp_class.
7232 * tree-ssa-copy.c (copy_folder): New class derived from
7233 substitute_and_fold_engine.
7234 (copy_folder::get_value): Renamed from get_value.
7235 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
7236 * tree-vrp.c (vrp_folder): New class derived from
7237 substitute_and_fold_engine.
7238 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
7239 (vrp_folder::get_value): New member function.
7240 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
7241 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
7242 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
7243 provide a class interface to folder/substitute routines.
7244 (ssa_prop_fold_stmt_fn): Remove typedef.
7245 (ssa_prop_get_value_fn): Likewise.
7246 (subsitute_and_fold): Remove prototype.
7247 (replace_uses_in): Likewise.
7248 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
7249 Renamed from replace_uses_in. Call the virtual member function
7250 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
7251 (substitute_and_fold_dom_walker): Remove initialization of
7252 data member entries for calbacks. Add substitute_and_fold_engine
7253 member and initialize it.
7254 (substitute_and_fold_dom_walker::before_dom_children0: Use the
7255 member functions for get_value, replace_phi_args_in c
7256 replace_uses_in, and fold_stmt calls.
7257 (substitute_and_fold_engine::substitute_and_fold): Renamed from
7258 substitute_and_fold. Remove assert. Update ctor call.
7260 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
7261 (ssa_prop_visit_phi_fn): Likewise.
7262 (class ssa_propagation_engine): New class to provide an interface
7264 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
7266 (ssa_prop_visit_phi): Likewise.
7267 (ssa_propagation_engine::simulate_stmt): Moved into class.
7268 Call visit_phi/visit_stmt from the class rather than via
7269 file scoped static variables.
7270 (ssa_propagation_engine::simulate_block): Moved into class.
7271 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
7272 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
7273 set file scoped statics for the visit_stmt/visit_phi callbacks.
7274 * tree-complex.c (complex_propagate): New class derived from
7275 ssa_propagation_engine.
7276 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
7277 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
7278 (tree_lower_complex): Call ssa_propagate via the complex_propagate
7280 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
7281 ssa_propagation_engine.
7282 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
7283 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
7284 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
7285 * tree-ssa-copy.c (copy_prop): New class derived from
7286 ssa_propagation_engine.
7287 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
7288 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
7289 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
7290 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
7291 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
7292 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
7293 (execute_vrp): Call ssa_propagate from the vrp_prop class.
7295 2017-11-01 Jakub Jelinek <jakub@redhat.com>
7297 PR rtl-optimization/82778
7298 PR rtl-optimization/82597
7299 * compare-elim.c (struct comparison): Add in_a_setter field.
7300 (find_comparison_dom_walker::before_dom_children): Remove killed
7301 bitmap and df_simulate_find_defs call, instead walk the defs.
7302 Compute last_setter and initialize in_a_setter. Merge definitions
7303 with first initialization for a few variables.
7304 (try_validate_parallel): Use insn_invalid_p instead of
7305 recog_memoized. Return insn rather than just the pattern.
7306 (try_merge_compare): Fix up comment. Don't uselessly test if
7307 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
7309 (execute_compare_elim_after_reload): Remove df_chain_add_problem
7312 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7313 Alan Hayward <alan.hayward@arm.com>
7314 David Sherwood <david.sherwood@arm.com>
7316 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
7317 aarch64_hard_regno_nregs to get the number of registers
7320 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7321 Alan Hayward <alan.hayward@arm.com>
7322 David Sherwood <david.sherwood@arm.com>
7324 * config/aarch64/constraints.md (Upl): Rename to...
7326 * config/aarch64/aarch64.md
7327 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
7330 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7331 Alan Hayward <alan.hayward@arm.com>
7332 David Sherwood <david.sherwood@arm.com>
7334 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
7335 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
7338 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7339 Alan Hayward <alan.hayward@arm.com>
7340 David Sherwood <david.sherwood@arm.com>
7342 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
7343 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
7344 (aarch64_evpc_dup): Generate rtl direcly, rather than using
7346 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
7347 of a single element.
7348 * config/aarch64/iterators.md: Add a comment above the permute
7349 unspecs to say that they are generated directly by
7350 aarch64_expand_vec_perm_const.
7351 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
7353 2017-11-01 Nathan Sidwell <nathan@acm.org>
7355 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
7357 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
7359 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
7360 and medany code models, and describe what they do.
7362 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7364 Revert accidental duplicate:
7366 * combine.c (can_change_dest_mode): Reject changes in
7367 REGMODE_NATURAL_SIZE.
7369 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
7371 PR rtl-optimization/64682
7372 PR rtl-optimization/69567
7373 PR rtl-optimization/69737
7374 PR rtl-optimization/82683
7375 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
7376 register mentioned in the note, drop the note, unless it came from I3,
7377 in which case it should go to I3 again.
7379 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7381 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
7382 and return false if not.
7383 (clear_bytes_written_by, live_bytes_read): Update accordingly.
7385 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7387 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
7388 range is known to be empty.
7390 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7391 Alan Hayward <alan.hayward@arm.com>
7392 David Sherwood <david.sherwood@arm.com>
7394 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
7395 and CONST_VECTOR_NUNITS instead of computing the number of units from
7396 the byte sizes of the vector and element.
7397 (simplify_binary_operation_1): Likewise.
7398 (simplify_const_binary_operation): Likewise.
7399 (simplify_ternary_operation): Likewise.
7401 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7402 Alan Hayward <alan.hayward@arm.com>
7403 David Sherwood <david.sherwood@arm.com>
7405 * var-tracking.c (INT_MEM_OFFSET): Replace with...
7406 (int_mem_offset): ...this new function.
7407 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
7408 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
7409 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
7412 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7413 Alan Hayward <alan.hayward@arm.com>
7414 David Sherwood <david.sherwood@arm.com>
7416 * lower-subreg.c (interesting_mode_p): New function.
7417 (compute_costs, find_decomposable_subregs, decompose_register)
7418 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
7419 (resolve_clobber, dump_choices): Use it.
7421 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7422 Alan Hayward <alan.hayward@arm.com>
7423 David Sherwood <david.sherwood@arm.com>
7425 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
7427 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7428 Alan Hayward <alan.hayward@arm.com>
7429 David Sherwood <david.sherwood@arm.com>
7431 * alias.c (find_base_value, find_base_term): Only process integer
7432 truncations. Check the precision rather than the size.
7434 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7435 Alan Hayward <alan.hayward@arm.com>
7436 David Sherwood <david.sherwood@arm.com>
7438 * machmode.h (is_narrower_int_mode): New function
7439 * optabs.c (expand_float, expand_fix): Use it.
7440 * dwarf2out.c (rotate_loc_descriptor): Likewise.
7442 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7443 Alan Hayward <alan.hayward@arm.com>
7444 David Sherwood <david.sherwood@arm.com>
7446 * rtl.h (narrower_subreg_mode): New function.
7447 * ira-color.c (update_costs_from_allocno): Use it.
7449 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7450 Alan Hayward <alan.hayward@arm.com>
7451 David Sherwood <david.sherwood@arm.com>
7453 * optabs-query.h (convert_optab_p): New function, split out from...
7454 (convert_optab_handler): ...here.
7455 (widening_optab_handler): Delete.
7456 (find_widening_optab_handler): Remove permit_non_widening parameter.
7457 (find_widening_optab_handler_and_mode): Likewise. Provide an
7458 override that operates on mode class wrappers.
7459 * optabs-query.c (widening_optab_handler): Delete.
7460 (find_widening_optab_handler_and_mode): Remove permit_non_widening
7461 parameter. Assert that the two modes are the same class and that
7462 the "from" mode is narrower than the "to" mode. Use
7463 convert_optab_handler instead of widening_optab_handler.
7464 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
7465 instead of widening_optab_handler.
7466 * expr.c (expand_expr_real_2): Update calls to
7467 find_widening_optab_handler.
7468 * optabs.c (expand_widen_pattern_expr): Likewise.
7469 (expand_binop_directly): Take the insn_code as a parameter.
7470 (expand_binop): Only call find_widening_optab_handler for
7471 conversion optabs; use optab_handler otherwise. Update calls
7472 to find_widening_optab_handler and expand_binop_directly.
7473 Use convert_optab_handler instead of widening_optab_handler.
7474 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
7475 find_widening_optab_handler and use scalar_mode rather than
7477 (convert_plusminus_to_widen): Likewise.
7479 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7480 Alan Hayward <alan.hayward@arm.com>
7481 David Sherwood <david.sherwood@arm.com>
7483 * machmode.h (fixed_size_mode): New class.
7484 * rtl.h (get_pool_mode): Return fixed_size_mode.
7485 * gengtype.c (main): Add fixed_size_mode.
7486 * target.def (get_raw_result_mode): Return a fixed_size_mode.
7487 (get_raw_arg_mode): Likewise.
7488 * doc/tm.texi: Regenerate.
7489 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
7490 * targhooks.c (default_get_reg_raw_mode): Likewise.
7491 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
7492 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
7493 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
7494 (msp430_get_raw_result_mode): Likewise.
7495 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
7496 * dbxout.c (dbxout_parms): Require fixed-size modes.
7497 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
7498 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
7499 * omp-low.c (lower_oacc_reductions): Likewise.
7500 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
7501 (simplify_subreg): Update accordingly.
7502 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
7503 (force_const_mem): Update accordingly. Return NULL_RTX for modes
7504 that aren't fixed-size.
7505 (get_pool_mode): Return a fixed_size_mode.
7506 (output_constant_pool_2): Take a fixed_size_mode.
7508 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7509 Alan Hayward <alan.hayward@arm.com>
7510 David Sherwood <david.sherwood@arm.com>
7512 * doc/rtl.texi (vec_series): Document.
7513 (const): Say that the operand can be a vec_series.
7514 * rtl.def (VEC_SERIES): New rtx code.
7515 * rtl.h (const_vec_series_p_1): Declare.
7516 (const_vec_series_p): New function.
7517 * emit-rtl.h (gen_const_vec_series): Declare.
7518 (gen_vec_series): Likewise.
7519 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
7520 (gen_vec_series): Likewise.
7521 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
7522 * simplify-rtx.c (simplify_unary_operation): Handle negations
7524 (simplify_binary_operation_series): New function.
7525 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
7526 (test_vector_ops_series): New function.
7527 (test_vector_ops): Call it.
7528 * config/powerpcspe/altivec.md (altivec_lvsl): Use
7529 gen_const_vec_series.
7530 (altivec_lvsr): Likewise.
7531 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
7533 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7534 Alan Hayward <alan.hayward@arm.com>
7535 David Sherwood <david.sherwood@arm.com>
7537 * doc/rtl.texi (const): Update description of address constants.
7538 Say that vector constants are allowed too.
7539 * common.md (E, F): Use CONSTANT_P instead of checking for
7541 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
7542 checking for CONST_VECTOR.
7543 * expmed.c (make_tree): Use build_vector_from_val for a CONST
7545 * expr.c (expand_expr_real_2): Check for vector modes instead
7546 of checking for CONST_VECTOR.
7547 * rtl.h (const_vec_p): New function.
7548 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
7549 (unwrap_const_vec_duplicate): Handle them here too.
7551 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7552 David Malcolm <dmalcolm@redhat.com>
7553 Alan Hayward <alan.hayward@arm.com>
7554 David Sherwood <david.sherwood@arm.com>
7556 * rtl.h (vec_duplicate_p): New function.
7557 * selftest-rtl.c (assert_rtx_eq_at): New function.
7558 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
7559 (assert_rtx_eq_at): Declare.
7560 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
7561 * selftest-run-tests.c (selftest::run_tests): Call it.
7562 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
7563 (simplify_unary_operation_1): Recursively handle vector duplicates.
7564 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
7566 (simplify_subreg): Handle subregs of vector duplicates.
7567 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
7568 (selftest::simplify_rtx_c_tests): New functions.
7570 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7571 Alan Hayward <alan.hayward@arm.com>
7572 David Sherwood <david.sherwood@arm.com>
7574 * emit-rtl.h (gen_const_vec_duplicate): Declare.
7575 (gen_vec_duplicate): Likewise.
7576 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
7578 (gen_const_vector): ...here.
7579 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
7580 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
7581 whose elements are all equal.
7582 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
7583 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
7584 (simplify_relational_operation): Likewise.
7585 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
7587 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
7588 (aarch64_expand_vector_init): Likewise.
7589 * config/arm/arm.c (neon_vdup_constant): Likewise.
7590 (neon_expand_vector_init): Likewise.
7591 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
7592 (arm_block_set_unaligned_vect): Likewise.
7593 (arm_block_set_aligned_vect): Likewise.
7594 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
7595 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
7596 (expand_vec_perm_even_odd_pack): Likewise.
7597 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
7598 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
7599 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
7600 gen_const_vec_duplicate.
7601 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
7602 * config/mips/mips.c (mips_gen_const_int_vector): Use
7603 gen_const_vec_duplicate.
7604 (mips_expand_vector_init): Use CONST0_RTX.
7605 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
7606 (define_split): Use gen_const_vec_duplicate.
7607 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
7608 (define_split): Use gen_const_vec_duplicate.
7609 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
7610 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
7611 * config/spu/spu.c (spu_const): Likewise.
7613 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7614 Alan Hayward <alan.hayward@arm.com>
7615 David Sherwood <david.sherwood@arm.com>
7617 * combine.c (can_change_dest_mode): Reject changes in
7618 REGMODE_NATURAL_SIZE.
7620 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
7622 * configure.ac (--enable-libssp): New.
7623 (gcc_cv_libc_provides_ssp): Check for explicit setting before
7624 trying to determine target-specific default. Adjust indentation.
7625 * configure: Regenerated.
7626 * doc/install.texi (Configuration): Expand --disable-libssp
7629 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
7631 config/i386/i386.c (ix86_expand_epilogue): Correct stack
7634 2017-10-31 Martin Jambor <mjambor@suse.cz>
7637 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
7639 2017-10-31 David Malcolm <dmalcolm@redhat.com>
7641 * auto-profile.c (autofdo_source_profile::read): Use
7642 UNKNOWN_LOCATION rather than 0.
7643 * diagnostic-core.h (warning_at_rich_loc): Rename to...
7644 (warning_at): ...this overload.
7645 (warning_at_rich_loc_n): Rename to...
7646 (warning_n): ...this overload.
7647 (error_at_rich_loc): Rename to...
7648 (error_at): ...this overload.
7649 (pedwarn_at_rich_loc): Rename to...
7650 (pedwarn): ...this overload.
7651 (permerror_at_rich_loc): Rename to...
7652 (permerror): ...this overload.
7653 (inform_at_rich_loc): Rename to...
7654 (inform): ...this overload.
7655 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
7656 (diagnostic_n_impl_richloc): Rename to...
7657 (diagnostic_n_impl): ...this rich_location *-based decl.
7658 (inform_at_rich_loc): Rename to...
7659 (inform): ...this, and add an assertion.
7660 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
7661 (warning_at_rich_loc): Rename to...
7662 (warning_at): ...this, and add an assertion.
7663 (warning_at_rich_loc_n): Rename to...
7664 (warning_n): ...this, and add an assertion.
7665 (warning_n): Update location_t-based implementation for removal of
7666 location_t-based diagnostic_n_impl.
7667 (pedwarn_at_rich_loc): Rename to...
7668 (pedwarn): ...this, and add an assertion.
7669 (permerror_at_rich_loc): Rename to...
7670 (permerror): ...this, and add an assertion.
7671 (error_n): Update for removal of location_t-based diagnostic_n_impl.
7672 (error_at_rich_loc): Rename to...
7673 (error_at): ...this, and add an assertion.
7674 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
7675 (driver::do_spec_on_infiles): Likewise.
7676 * substring-locations.c (format_warning_va): Update for renaming
7677 of inform_at_rich_loc.
7679 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
7681 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
7682 _Float<N>X built-in functions so that the variant without the
7683 "__builtin_" prefix is only enabled for the GNU C and Objective C
7684 languages when they are in non-strict ANSI/ISO mode.
7685 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
7686 * target.def (floatn_builtin_p): Add a target hook to control
7687 whether _Float<N> and _Float<N>X built-in functions without the
7688 "__builtin_" prefix are enabled, and return true for C and
7689 Objective C in the default hook. Include langhooks.h in
7691 * targhooks.h (default_floatn_builtin_p): Likewise.
7692 * targhooks.c (default_floatn_builtin_p): Likewise.
7693 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
7694 floatn_builtin_p target hook.
7695 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
7697 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
7698 Eric Botcazou <ebotcazou@adacore.com>
7700 PR rtl-optimization/81803
7701 * lra-constraints.c (curr_insn_transform): Also reload the whole
7702 register for a strict subreg no wider than a word if this is for
7703 a WORD_REGISTER_OPERATIONS target.
7705 2017-10-31 Jason Merrill <jason@redhat.com>
7707 * gdbinit.in: Skip over inlines from timevar.h.
7709 2017-10-31 Martin Liska <mliska@suse.cz>
7711 * doc/gcov.texi: Document new option.
7712 * gcov.c (print_usage): Likewise print it.
7713 (process_args): Support the argument.
7714 (format_count): New function.
7715 (format_gcov): Use the function.
7717 2017-10-31 Martin Liska <mliska@suse.cz>
7719 * gcov.c (struct name_map): do not use typedef.
7720 Define operator== and operator<.
7721 (name_search): Remove.
7722 (name_sort): Remove.
7723 (main): Do not allocate names.
7724 (process_file): Add vertical space.
7725 (generate_results): Use std::find.
7726 (release_structures): Do not release memory.
7727 (find_source): Use std::find.
7729 2017-10-31 Martin Liska <mliska@suse.cz>
7731 * gcov.c (struct line_info): Remove it's typedef.
7732 (line_info::line_info): Add proper ctor.
7733 (line_info::has_block): Do not use a typedef.
7734 (struct source_info): Do not use typedef.
7735 (circuit): Likewise.
7736 (get_cycles_count): Likewise.
7737 (output_intermediate_file): Iterate via vector iterator.
7738 (add_line_counts): Use std::vector methods.
7739 (accumulate_line_counts): Likewise.
7740 (output_lines): Likewise.
7742 2017-10-31 Martin Liska <mliska@suse.cz>
7744 * gcov.c (struct source_info): Remove typedef.
7745 (source_info::source_info): Add proper ctor.
7746 (accumulate_line_counts): Use struct, not it's typedef.
7747 (output_gcov_file): Likewise.
7748 (output_lines): Likewise.
7749 (main): Do not allocate an array.
7750 (output_intermediate_file): Use size of vector container.
7751 (process_file): Resize the vector.
7752 (generate_results): Do not preallocate, use newly added vector
7754 (release_structures): Do not release sources.
7755 (find_source): Use vector methods.
7756 (add_line_counts): Do not use typedef.
7758 2017-10-31 Martin Liska <mliska@suse.cz>
7760 * doc/gcov.texi: Document that.
7761 * gcov.c (add_line_counts): Mark lines with a non-executed
7763 (output_line_beginning): Handle such lines.
7764 (output_lines): Pass new argument.
7765 (output_intermediate_file): Print it in intermediate format.
7767 2017-10-31 Martin Liska <mliska@suse.cz>
7769 * color-macros.h: New file.
7770 * diagnostic-color.c: Factor out color related to macros to
7772 * doc/gcov.texi: Document -k option.
7773 * gcov.c (INCLUDE_STRING): Include string.h.
7774 (print_usage): Add -k option.
7775 (process_args): Parse it.
7776 (pad_count_string): New function.
7777 (output_line_beginning): Likewise.
7778 (DEFAULT_LINE_START): New macro.
7779 (output_lines): Support color output.
7781 2017-10-31 Martin Liska <mliska@suse.cz>
7783 PR gcov-profile/82633
7784 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
7785 their interaction with GCOV infrastructure.
7786 * configure.ac: Add -fkeep-{inline,static}-functions to
7788 * configure: Regenerate.
7790 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
7793 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
7795 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
7798 * config/rs6000/rs6000.md (allocate_stack): Force update interval
7799 into a register if it does not fit into an immediate offset field.
7801 2017-10-31 Olivier Hainque <hainque@adacore.com>
7803 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
7805 2017-10-31 Julia Koval <julia.koval@intel.com>
7807 * config.gcc: Add gfniintrin.h.
7808 * config/i386/gfniintrin.h: New.
7809 * config/i386/i386-builtin-types.def
7810 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7811 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7812 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7813 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7814 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7815 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
7816 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
7817 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
7818 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
7819 V64QI_FTYPE_V64QI_V64QI_INT): New types.
7820 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
7821 * config/i386/immintrin.h: Include gfniintrin.h.
7822 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
7824 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
7826 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
7828 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
7830 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
7831 (arm_ashldi3_1bit): Remove pattern.
7832 (ashrdi3): Remove shift by 1 expansion.
7833 (arm_ashrdi3_1bit): Remove pattern.
7834 (lshrdi3): Remove shift by 1 expansion.
7835 (arm_lshrdi3_1bit): Remove pattern.
7836 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
7837 cost of ashldi3 by 1.
7838 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
7839 (<shift>di3_neon): Likewise.
7841 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7843 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
7844 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
7845 and (*aarch64_simd_mov<VQ:mode>).
7846 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
7847 pattern alternative.
7848 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
7849 attributes to match pattern alternative.
7851 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
7853 * config.gcc (powerpc*-*-*): Add emmintrin.h.
7854 * config/rs6000/emmintrin.h: New file.
7855 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
7857 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
7859 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
7860 * (movdi_vfp_cortexa8): Remove pattern.
7862 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7864 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
7866 (Specific, alpha*-dec-osf5.1): Remove.
7867 (Specific, mips-sgi-irix5): Remove.
7868 (Specific, mips-sgi-irix6): Remove.
7870 2017-10-30 Jakub Jelinek <jakub@redhat.com>
7873 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
7874 arguments to clear_bit_region_be.
7876 2017-10-30 Jim Wilson <wilson@tuliptree.org>
7878 * gimplify.c: Include memmodel.h.
7880 2017-10-30 Martin Jambor <mjambor@suse.cz>
7882 * omp-grid.c (grid_attempt_target_gridification): Also insert a
7883 condition whether loop should be executed at all.
7885 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
7887 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7888 gimple folding of vec_madd() intrinsics.
7889 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
7890 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
7891 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
7893 2017-10-30 Richard Biener <rguenther@suse.de>
7895 PR tree-optimization/82762
7897 2017-10-23 Richard Biener <rguenther@suse.de>
7899 PR tree-optimization/82129
7901 2017-08-01 Richard Biener <rguenther@suse.de>
7903 PR tree-optimization/81181
7904 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7905 (compute_antic): ... end of iteration here.
7907 2017-10-30 Joseph Myers <joseph@codesourcery.com>
7909 * doc/invoke.texi (C Dialect Options): Document -std=c17,
7910 -std=iso9899:2017 and -std=gnu17.
7911 * doc/standards.texi (C Language): Document C17 support.
7912 * doc/cpp.texi (Overview): Mention -std=c17.
7913 (Standard Predefined Macros): Document C11 and C17 values of
7914 __STDC_VERSION__. Do not refer to C99 support as incomplete.
7915 * doc/extend.texi (Inline): Do not list individual options for
7916 standards newer than C99.
7917 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
7919 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
7922 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
7924 * asan.c (asan_finish_file): Align asan globals array by shadow
7927 2017-10-30 Jakub Jelinek <jakub@redhat.com>
7930 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
7931 (struct store_immediate_info): Add bitregion_start and bitregion_end
7933 (store_immediate_info::store_immediate_info): Add brs and bre
7934 arguments and initialize bitregion_{start,end} from those.
7935 (struct merged_store_group): Add bitregion_start, bitregion_end,
7936 align_base and mask fields. Drop unnecessary struct keyword from
7937 struct store_immediate_info. Add do_merge method.
7938 (clear_bit_region_be): Use memset instead of loop storing zeros.
7939 (merged_store_group::do_merge): New method.
7940 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
7941 stores as long as the surrounding bitregions have no gaps.
7942 (merged_store_group::merge_overlapping): Use do_merge.
7943 (merged_store_group::apply_stores): Test that bitregion_{start,end}
7944 is byte aligned, rather than requiring that start and width are
7945 byte aligned. Drop unnecessary struct keyword from
7946 struct store_immediate_info. Allocate and populate also mask array.
7947 Make start of the arrays relative to bitregion_start rather than
7948 start and size them according to bitregion_{end,start} difference.
7949 (struct imm_store_chain_info): Drop unnecessary struct keyword from
7950 struct store_immediate_info.
7951 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
7952 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
7953 struct keyword from struct store_immediate_info.
7954 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
7955 between stores as long as the surrounding bitregions have no gaps.
7957 (struct split_store): Add orig non-static data member.
7958 (split_store::split_store): Initialize orig to false.
7959 (find_constituent_stmts): Return store_immediate_info *, non-NULL
7960 if there is exactly a single original stmt. Change stmts argument
7961 to pointer from reference, if NULL, don't push anything to it. Add
7962 first argument, use it to optimize skipping over orig stmts that
7963 are known to be before bitpos already. Simplify.
7964 (split_group): Return unsigned int count how many stores are or
7965 would be needed rather than a bool. Add allow_unaligned argument.
7966 Change split_stores argument from reference to pointer, if NULL,
7967 only do a dry run computing how many stores would be produced.
7968 Rewritten algorithm to use both alignment and misalign if
7969 !allow_unaligned and handle bitfield stores with gaps.
7970 (imm_store_chain_info::output_merged_store): Set start_byte_pos
7971 from bitregion_start instead of start. Compute allow_unaligned
7972 here, if true, do 2 split_group dry runs to compute which one
7973 produces fewer stores and prefer aligned if equal. Punt if
7974 new count is bigger or equal than original before emitting any
7975 statements, rather than during that. Remove no longer needed
7976 new_ssa_names tracking. Replace num_stmts with
7977 split_stores.length (). Use 32-bit stack allocated entries
7978 in split_stores auto_vec. Try to reuse original store lhs/rhs1
7979 if possible. Handle bitfields with gaps.
7980 (pass_store_merging::execute): Ignore bitsize == 0 stores.
7981 Compute bitregion_{start,end} for the stores and construct
7982 store_immediate_info with that. Formatting fixes.
7984 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
7987 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
7988 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
7990 2017-10-29 Jim Wilson <wilson@tuliptree.org>
7992 * gimplify.c: Include tm_p.h.
7994 * common.opt (gcoff): Re-add as ignored option.
7995 (gcoff1, gcoff2, gcoff3): Likewise.
7997 * Makefile.in (OBJS): Delete sdbout.o.
7998 (GTFILES): Delete $(srcdir)/sdbout.c.
7999 * debug.h: Delete sdb_debug_hooks.
8000 * final.c: Delete sdbout.h include.
8001 (final_scan_insn): Delete SDB_DEBUG check.
8002 (rest_of_clean_state): Likewise.
8003 * output.h: Delete sdb_begin_function_line.
8006 * toplev.c: Delete sdbout.h include.
8007 (process_options): Delete SDB_DEBUG check.
8008 * tree-core.h (tree_type_common): Delete pointer field of
8010 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
8011 TYPE_SYMTAB_POINTER.
8012 * tree.h (TYPE_SYMTAB_POINTER): Delete.
8013 (TYPE_SYMTAB_IS_POINTER): Delete.
8014 (TYPE_SYMTAB_IS_DIE): Renumber.
8015 * xcoffout.c: Refer to former sdbout.c file.
8016 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
8018 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
8019 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
8020 (Debugging Options): Delete -gcoff.
8021 (-gstabs): Delete SDB reference.
8023 (-gcoff@var{level}): Delete.
8024 * doc/passes.texi (Debugging information output): Delete SDB and
8025 sdbout.c references.
8026 * doc/tm.texi: Regenerate.
8027 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
8028 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
8030 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
8031 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
8032 (SDB_DEBUGGING_INFO): Delete.
8033 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
8034 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
8035 * target.def (output_source_filename): Delete COFF reference.
8037 * common.opt (gcoff): Delete.
8038 (gxcoff+): Update Negative chain.
8039 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
8041 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
8042 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
8043 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
8045 (expand_function_start): Change sdb reference to past tense.
8046 (expand_function_end): Change sdb reference to past tense.
8047 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
8048 * opts.c (debug_type_names): Delete coff entry.
8049 (common_handle_option): Delete OPT_gcoff case.
8050 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
8052 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
8053 * config/cris/cris.h: Delete SDB reference in comment.
8054 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
8055 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
8056 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
8057 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
8059 (ix86_expand_prologue): Likewise.
8060 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
8061 * config/ia64/ia64.h: Likewise.
8062 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
8063 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
8065 * config/mmix/mmix.h: Likewise.
8066 * config/nds32/nds32.c: Likewise.
8067 * config/stormy/storym16.h: Likewise.
8068 * config/visium/visium.h: Likewise.
8069 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
8071 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
8073 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
8074 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
8075 FRAME_POINTER_REGNUM point at high end of local var area.
8077 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
8079 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
8080 Move comment around. Do not reset best_edge for a copiable
8081 destination if the copy would cause a partition change.
8082 (better_edge_p): Remove redundant check.
8084 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
8086 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
8088 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
8090 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
8091 for math functions that have _Float<N> and _Float<N>X variants.
8092 (mathfn_built_in_2): Add support for math functions that have
8093 _Float<N> and _Float<N>X variants.
8094 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
8095 (expand_builtin_mathfn_ternary): Add support for fma with
8096 _Float<N> and _Float<N>X variants.
8097 (expand_builtin): Likewise.
8098 (fold_builtin_3): Likewise.
8099 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
8100 create math function _Float<N> and _Float<N>X variants as external
8102 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
8103 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
8104 the __builtin_ prefix and if not strict ansi, without the prefix.
8105 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
8106 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
8107 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
8108 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
8109 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
8110 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
8111 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
8112 function signatures for fma _Float<N> and _Float<N>X variants.
8113 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
8114 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
8115 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
8116 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
8117 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
8118 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
8119 * gencfn-macros.c (print_case_cfn): Add support for math functions
8120 that have _Float<N> and _Float<N>X variants.
8121 (print_define_operator_list): Likewise.
8122 (fltfn_suffixes): Likewise.
8124 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
8125 for math functions that have _Float<N> and _Float<N>X variants.
8126 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
8127 and _Float<N>X variants.
8128 (COPYSIGN): Likewise.
8131 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
8132 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
8134 (integer_valued_read_call_p): Likewise.
8135 * fold-const-call.c (fold_const_call_ss): Likewise.
8136 (fold_const_call_sss): Add support for copysign, fmin, and fmax
8137 _Float<N> and _Float<N>X variants.
8138 (fold_const_call_ssss): Add support for fma _Float<N> and
8139 _Float<N>X variants.
8140 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
8141 support for copysign and fma _Float<N> and _Float<N>X variants.
8142 (backprop::process_builtin_call_use): Likewise.
8143 * tree-call-cdce.c (can_test_argument_range); Add support for
8144 sqrt _Float<N> and _Float<N>X variants.
8145 (edom_only_function): Likewise.
8146 (get_no_error_domain): Likewise.
8147 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
8148 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
8149 copysign _Float<N> and _Float<N>X variants.
8150 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
8151 handled by machine independent code.
8152 (FMAF128): Likewise.
8153 * doc/cpp.texi (Common Predefined Macros): Document defining
8154 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
8155 fma _Float<N> and _Float<N>X variants.
8157 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
8160 * config/i386/i386-modes.def (CCFPU): Remove definition.
8161 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
8162 (ix86_cc_modes_compatible): Ditto.
8163 (ix86_expand_carry_flag_compare): Ditto.
8164 (ix86_expand_int_movcc): Ditto.
8165 (ix86_expand_int_addcc): Ditto.
8166 (ix86_reverse_condition): Ditto.
8167 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
8168 Return true/false for unordered/ordered fp comparisons.
8169 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
8170 (ix86_prepare_fp_compare_args): Update for rename.
8171 (ix86_expand_fp_compare): Update for rename. Generate unordered
8172 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
8173 (ix86_expand_sse_compare_and_jump): Ditto.
8174 * config/i386/predicates.md (fcmov_comparison_operator):
8175 Remove CCFPU mode handling.
8176 (ix86_comparison_operator): Ditto.
8177 (ix86_carry_flag_operator): Ditto.
8178 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
8179 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
8180 (*cmpu<mode>_cc_i387): Ditto.
8181 (FPCMP): Remove mode iterator.
8182 (unord): Remove mode attribute.
8183 (unord_subst): New define_subst transformation
8184 (unord): New define_subst attribute.
8186 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
8187 (*cmpi<unord>xf_i387): Ditto.
8188 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
8189 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
8190 using unord_subst transformation.
8191 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
8192 (round_saeonly): Also handle CCFP mode.
8193 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
8194 Remove UNSPEC_SAHF unspec handling.
8196 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
8198 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
8200 2017-10-27 Jeff Law <law@redhat.com>
8202 * gimple-ssa-sprintf.c: Include domwalk.h.
8203 (class sprintf_dom_walker): New class, derived from dom_walker.
8204 (sprintf_dom_walker::before_dom_children): New function.
8205 (struct call_info): Moved into sprintf_dom_walker class
8206 (compute_formath_length, handle_gimple_call): Likewise.
8207 (sprintf_length::execute): Call the dominator walker rather
8208 than walking the statements.
8210 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
8211 gimple statement locations.
8212 (check_array_bounds): Corresponding changes. Get the statement's
8213 location directly from wi->stmt.
8215 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
8218 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
8220 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
8222 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
8223 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
8225 2017-10-27 Jakub Jelinek <jakub@redhat.com>
8228 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
8229 * config/i386/i386.c (maybe_get_pool_constant): Removed.
8230 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
8231 maybe_get_pool_constant.
8232 * config/i386/predicates.md (zero_extended_scalar_load_operand):
8235 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8237 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
8238 2.26 caveat. Update gas and gld versions.
8239 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
8242 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8244 * cgraph.h (set_malloc_flag): Declare.
8245 * cgraph.c (set_malloc_flag_1): New function.
8246 (set_malloc_flag): Likewise.
8247 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
8248 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
8250 (read_ipa_call_summary): Add support for reading is_return_callee.
8251 (write_ipa_call_summary): Stream is_return_callee.
8252 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
8253 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
8254 ipa-prop.h, ipa-fnsummary.h.
8255 (pure_const_names): Change to static.
8256 (malloc_state_e): Define.
8257 (malloc_state_names): Define.
8258 (funct_state_d): Add field malloc_state.
8259 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
8260 (check_retval_uses): New function.
8261 (malloc_candidate_p): Likewise.
8262 (analyze_function): Add support for malloc attribute.
8263 (pure_const_write_summary): Stream malloc_state.
8264 (pure_const_read_summary): Add support for reading malloc_state.
8265 (dump_malloc_lattice): New function.
8266 (propagate_malloc): New function.
8267 (warn_function_malloc): New function.
8268 (ipa_pure_const::execute): Call propagate_malloc and
8269 ipa_free_fn_summary.
8270 (pass_local_pure_const::execute): Add support for malloc attribute.
8271 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
8272 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
8274 2017-10-27 Martin Liska <mliska@suse.cz>
8276 PR gcov-profile/82457
8277 * doc/invoke.texi: Document that one needs a non-strict ISO mode
8278 for fork-like functions to be properly instrumented.
8280 2017-10-27 Richard Biener <rguenther@suse.de>
8283 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
8284 info when we redirected EH.
8286 2017-10-26 Michael Collison <michael.collison@arm.com>
8288 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
8290 (<optab>_trunchf<GPI:mode>2: New pattern.
8291 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
8292 * config/aarch64/iterators.md (wv): New mode attribute.
8293 (vf, VF): New mode attributes.
8294 (vgp, VGP): New mode attributes.
8295 (s): Update attribute with SImode and DImode prefixes.
8297 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
8299 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
8300 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
8301 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
8302 (nios2_option_overide): Initialize it. Don't allow R0-relative
8303 addressing with PIC.
8304 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
8305 (nios2_symbolic_constant_p): Likewise.
8306 (nios2_legitimate_address_p): Likewise.
8307 (nios2_r0rel_section_name_p): New.
8308 (nios2_symbol_ref_in_r0rel_data_p): New.
8309 (nios2_emit_move_sequence): Handle r0rel_constant_p.
8310 (r0rel_constant_p): New.
8311 (nios2_print_operand_address): Handle r0rel_constant_p.
8312 (nios2_cdx_narrow_form_p): Likewise.
8313 * config/nios2/nios2.opt (mr0rel-sec=): New option.
8314 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
8315 (Nios II Options): Document -mr0rel-sec.
8317 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
8319 * config/nios2/nios2.c: Include xregex.h.
8320 (nios2_gprel_sec_regex): New.
8321 (nios2_option_overide): Initialize it. Don't allow GP-relative
8322 addressing with PIC.
8323 (nios2_small_section_name_p): Check for regex match.
8324 * config/nios2/nios2.opt (mgprel-sec=): New option.
8325 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
8326 (Nios II Options): Document -mgprel-sec.
8328 2017-10-26 Jim Wilson <wilson@tuliptree.org>
8330 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
8332 2017-10-26 Tom de Vries <tom@codesourcery.com>
8334 PR tree-optimization/82707
8335 * gimple.c (gimple_copy): Fix unsharing of
8336 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
8338 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
8340 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
8341 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
8342 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
8343 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
8344 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
8345 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
8346 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
8347 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
8348 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
8349 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
8350 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
8351 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
8352 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
8353 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
8354 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
8355 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
8356 _mm512_mask_cmpunord_ps_mask): New intrinsics.
8358 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8360 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
8362 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
8363 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
8364 warning to rs6000.c. Remove the Undocumented flag, since it has
8366 (-mabi=ibmlongdouble): Likewise.
8367 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
8368 already set, set the default format for long double.
8369 (rs6000_debug_reg_global): Print whether long double is IBM or
8371 (rs6000_option_override_internal): Rework setting long double
8372 format. Only warn if the user is changing the long double default
8373 and they did not use -Wno-psabi.
8374 * doc/invoke.texi (PowerPC options): Update the documentation for
8375 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
8377 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8378 Alan Hayward <alan.hayward@arm.com>
8379 David Sherwood <david.sherwood@arm.com>
8381 * rtl.h (wider_subreg_mode): New function.
8382 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
8383 rather than an unsigned int *.
8384 * ira-color.c (regno_max_ref_width): Replace with...
8385 (regno_max_ref_mode): ...this new variable.
8386 (coalesced_pseudo_reg_slot_compare): Update accordingly.
8387 Use wider_subreg_mode.
8388 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
8389 rather than an unsigned int *.
8390 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
8391 (process_alt_operands): Likewise.
8392 (invariant_p): Likewise.
8393 * lra-spills.c (assign_mem_slot): Likewise.
8394 (add_pseudo_to_slot): Likewise.
8395 * lra.c (collect_non_operand_hard_regs): Likewise.
8396 (add_regs_to_insn_regno_info): Likewise.
8397 * reload1.c (regno_max_ref_width): Replace with...
8398 (regno_max_ref_mode): ...this new variable.
8399 (reload): Update accordingly. Update call to
8400 ira_sort_regnos_for_alter_reg.
8401 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
8402 (init_eliminable_invariants): Update to use regno_max_ref_mode.
8403 (scan_paradoxical_subregs): Likewise.
8405 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8407 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
8408 (aarch64_frame): Add emit_frame_chain boolean.
8409 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8410 Move eh_return case to aarch64_layout_frame.
8411 (aarch64_layout_frame): Initialize emit_frame_chain.
8412 (aarch64_expand_prologue): Use emit_frame_chain.
8414 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8416 * config/aarch64/aarch64.c (aarch64_layout_frame):
8417 Ensure LR is always stored at the bottom of the callee-saves.
8418 Remove rarely used frame layout which saves callee-saves at top of
8419 frame, so the store of LR can be used as a valid probe in all cases.
8421 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8423 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
8424 Improve unaligned TImode/TFmode base/offset split.
8426 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8427 Alan Hayward <alan.hayward@arm.com>
8428 David Sherwood <david.sherwood@arm.com>
8430 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
8431 * combine.c (find_single_use_1): Likewise.
8432 (expand_field_assignment): Likewise.
8433 (move_deaths): Likewise.
8434 * lra-constraints.c (simplify_operand_subreg): Likewise.
8435 (curr_insn_transform): Likewise.
8436 * lra.c (collect_non_operand_hard_regs): Likewise.
8437 (add_regs_to_insn_regno_info): Likewise.
8438 * rtlanal.c (reg_referenced_p): Likewise.
8439 (covers_regno_no_parallel_p): Likewise.
8441 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8443 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
8444 Don't print any bits outside the precision of the value.
8445 * wide-int.cc (test_printing): Add some new tests.
8447 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8449 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
8450 supports -xbrace_comment option.
8451 * configure: Regenerate.
8452 * config.in: Regenerate.
8453 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
8454 (ASM_CPU_SPEC): Use it.
8456 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8458 * target.def (static_rtx_alignment): New hook.
8459 * targhooks.h (default_static_rtx_alignment): Declare.
8460 * targhooks.c (default_static_rtx_alignment): New function.
8461 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
8462 * doc/tm.texi: Regenerate.
8463 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
8464 instead of targetm.constant_alignment. Remove call to
8466 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8467 (cris_preferred_mininum_alignment): New function, split out from...
8468 (cris_constant_alignment): ...here.
8469 (cris_static_rtx_alignment): New function.
8470 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
8472 (ix86_constant_alignment): ...here.
8473 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8474 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8475 (mmix_static_rtx_alignment): New function.
8476 * config/spu/spu.c (spu_static_rtx_alignment): New function.
8477 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8479 2017-10-26 Tamar Christina <tamar.christina@arm.com>
8482 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
8483 Add flag_trapping_math and flag_fp_int_builtin_inexact.
8485 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
8487 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
8488 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
8489 mark as a sign-extending load.
8490 (local_pic_load_u): Define.
8492 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
8495 * fold-const.c (operand_equal_for_comparison_p): Also return true
8496 if ARG0 is a simple variant of ARG1 with narrower precision.
8497 (fold_ternary_loc): Always pass unstripped operands to the predicate.
8499 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
8501 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
8503 * i386.h (processor_costs): Add gather_static, gather_per_elt,
8504 scatter_static, scatter_per_elt.
8505 * x86-tune-costs.h: Add new cost entries.
8507 2017-10-25 Richard Biener <rguenther@suse.de>
8509 * tree-ssa-sccvn.h (vn_eliminate): Declare.
8510 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
8511 class pass_fre): Move to ...
8512 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
8513 class pass_fre): ... here and adjust for statistics.
8515 2017-10-25 Jakub Jelinek <jakub@redhat.com>
8518 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
8519 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
8520 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
8521 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
8524 2017-10-25 Richard Biener <rguenther@suse.de>
8526 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
8527 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
8528 eliminate_push_avail, eliminate_insert): Move inside...
8529 (class eliminate_dom_walker): ... this class in preparation
8531 (fini_eliminate): Remove by merging with ...
8532 (eliminate): ... this function. Adjust for class changes.
8533 (pass_pre::execute): Remove fini_eliminate call.
8534 (pass_fre::execute): Likewise.
8536 2017-10-24 Jakub Jelinek <jakub@redhat.com>
8539 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
8540 (VPERMI2, VPERMI2I): New mode iterators.
8541 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
8542 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
8544 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
8545 mode iterator. Remove 3 old define_insn patterns.
8546 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
8547 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
8548 VPERMI2 mode iterator, remove the other two expanders.
8549 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
8550 to use VPERMI2 mode iterator, add another alternative for vpermi2*
8551 instructions, remove the other two patterns.
8552 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
8553 mode iterator, remove the other two patterns.
8554 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
8555 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
8556 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
8557 and adjust argument order accordingly.
8558 (ix86_expand_vec_perm): Adjust caller.
8559 (expand_vec_perm_1): Likewise.
8560 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
8561 (expand_vec_perm_vpermt2_vpshub2): ... this.
8562 (ix86_expand_vec_perm_const_1): Adjust caller.
8563 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
8566 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
8567 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
8568 (vec_shl_<mode>): Remove unused expander.
8569 (avx512bw_<shift_insn><mode>3): New define_insn.
8570 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
8571 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
8573 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
8576 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
8579 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
8581 PR rtl-optimization/82396
8582 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
8583 (autopref_multipass_init): Simplify initialization.
8584 (autopref_rank_data): Simplify sort order.
8585 * gcc/sched-int.h (autopref_multipass_data_): Remove
8586 multi_mem_insn_p, min_offset and max_offset.
8588 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
8591 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8592 Check special value of flag_omit_frame_pointer.
8593 (aarch64_can_eliminate): Likewise.
8594 (aarch64_override_options_after_change_1): Simplify handling of
8595 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
8597 2017-10-24 Richard Biener <rguenther@suse.de>
8599 PR tree-optimization/82697
8600 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
8601 zero for conditional load and unconditional store.
8603 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
8605 * doc/install.texi: Document bootstrap-cet.
8607 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
8610 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
8611 ENDBR instruction at function entrance if function is only
8614 2017-10-24 Jakub Jelinek <jakub@redhat.com>
8617 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
8618 patterns to better describe from which operation the CF is computed.
8619 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
8620 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
8621 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
8622 is 0, use _0 suffixed expanders instead of emitting a comparison
8625 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
8627 * config/i386/i386.md(*movsf_internal, *movdf_internal):
8628 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
8630 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
8633 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
8634 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
8635 * loop-iv.c (iv_get_reaching_def): Likewise.
8636 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
8637 variable is promoted and the partition contains undefined values.
8639 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8641 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
8643 (nios2_address_cost): Define.
8644 (nios2_legitimize_address): Recognize (exp + constant) directly.
8645 (TARGET_ADDRESS_COST): Define.
8647 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8649 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
8650 (nios2_symbolic_memory_operand_p): Declare.
8651 (nios2_split_large_constant): Declare.
8652 (nios2_split_symbolic_memory_operand): Declare.
8653 * config/nios2/nios2.c: Adjust includes.
8654 (nios2_symbolic_constant_allowed): New.
8655 (nios2_symbolic_constant_p): New.
8656 (nios2_plus_symbolic_constant_p): New.
8657 (nios2_valid_addr_expr_p): Recognize addresses involving
8659 (nios2_legitimate_address_p): Likewise, also LO_SUM.
8660 (nios2_symbolic_memory_operand_p): New.
8661 (nios2_large_constant_p): New.
8662 (nios2_split_large_constant): New.
8663 (nios2_split_plus_large_constant): New.
8664 (nios2_split_symbolic_memory_operand): New.
8665 (nios2_legitimize_address): Code refactoring. Handle addresses
8666 involving symbolic constants.
8667 (nios2_emit_move_sequence): Likewise.
8668 (nios2_print_operand): Improve error output.
8669 (nios2_print_operand_address): Handle LO_SUM.
8670 (nios2_cdx_narrow_form_p): Likewise.
8671 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
8672 operands involving symbolic constants.
8673 (movhi_internal, movsi_internal): Likewise.
8674 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
8675 (extendhisi2, extendqi<mode>2): Likewise.
8677 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8679 * tree-pass.h (PROP_rtl_split_insns): Define.
8680 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
8682 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8684 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
8686 2017-10-23 Jakub Jelinek <jakub@redhat.com>
8689 * target.def (const_not_ok_for_debug_p): Default to
8690 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
8691 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
8692 * targhooks.c (default_const_not_ok_for_debug_p): New function.
8693 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
8694 which targetm.const_not_ok_for_debug_p returned true.
8695 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
8697 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
8699 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
8700 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
8701 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
8703 (ix86_const_not_ok_for_debug_p): New function.
8704 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
8705 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
8707 2017-10-23 David Malcolm <dmalcolm@redhat.com>
8710 * system.h: Conditionally include "unique-ptr.h" if
8711 INCLUDE_UNIQUE_PTR is defined.
8712 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
8713 of defining INCLUDE_UNIQUE_PTR before including "system.h".
8715 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
8717 * config/rl78/rl78.md: New define_expand "subdi3".
8719 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
8722 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
8723 DF_REF_INSN if DF_REF_INSN_INFO is false.
8725 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
8727 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
8728 xmm_move instead of sse_move.
8729 (sse_store_index): New function.
8730 (ix86_register_move_cost): Be more sensible about mismatch stall;
8731 model AVX moves correctly; make difference between sse->integer and
8733 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
8734 moves; make difference between SSE and AVX.
8735 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
8736 and zmm_move. Increase size of sse load and store tables;
8737 add unaligned load and store tables; add ssemmx_to_integer.
8738 * x86-tune-costs.h: Update all entries according to real
8739 move latencies from Agner Fog's manual and chip documentation.
8741 2017-10-23 Jakub Jelinek <jakub@redhat.com>
8744 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
8745 * config/i386/constraints.md (Wf): New constraint.
8746 * config/i386/i386.md (UNSPEC_SBB): New unspec.
8747 (cmp<dwi>_doubleword): Removed.
8748 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
8749 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
8750 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
8751 expand with cmp<dwi>_doubleword. For LTU and GEU use
8752 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
8754 * common.opt (gcolumn-info): Enable by default.
8755 * doc/invoke.texi (gcolumn-info): Document new default.
8757 2017-10-23 Richard Biener <rguenther@suse.de>
8759 PR tree-optimization/82672
8760 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
8761 Fold the stmt if we propagated into it.
8763 2017-10-23 Richard Biener <rguenther@suse.de>
8765 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
8766 (bitmap_remove_expr_from_set): ... this. All callers call this
8767 for non-constant values.
8768 (bitmap_set_subtract): Rename to...
8769 (bitmap_set_subtract_expressions): ... this. Adjust and
8771 (bitmap_set_contains_value): Remove superfluous check.
8772 (bitmap_set_replace_value): Inline into single caller ...
8773 (bitmap_value_replace_in_set): ... here and simplify.
8774 (dependent_clean): Merge into ...
8775 (clean): ... this using an overload. Adjust.
8776 (prune_clobbered_mems): Adjust.
8777 (compute_antic_aux): Likewise.
8778 (compute_partial_antic_aux): Likewise.
8780 2017-10-23 Richard Biener <rguenther@suse.de>
8782 PR tree-optimization/82129
8784 2017-08-01 Richard Biener <rguenther@suse.de>
8786 PR tree-optimization/81181
8787 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8788 (compute_antic): ... end of iteration here.
8790 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8792 * target.def (starting_frame_offset): New hook.
8793 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
8794 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
8795 * doc/tm.texi.in: Regenerate.
8796 * hooks.h (hook_hwi_void_0): Declare.
8797 * hooks.c (hook_hwi_void_0): New function.
8798 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
8799 STARTING_FRAME_OFFSET.
8800 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
8801 * reload1.c (reload): Likewise.
8802 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
8803 instead of STARTING_FRAME_OFFSET.
8804 * function.c (try_fit_stack_local): Likewise.
8805 (assign_stack_local_1): Likewise
8806 (instantiate_virtual_regs): Likewise.
8807 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
8808 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
8809 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
8810 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
8811 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
8812 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
8813 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
8814 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
8815 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
8816 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
8817 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
8818 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
8819 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
8820 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
8821 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
8822 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
8823 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
8824 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
8825 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
8826 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
8827 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
8828 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
8829 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
8830 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
8831 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
8832 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
8833 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
8834 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
8835 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
8836 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
8837 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
8838 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
8839 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
8840 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
8841 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
8842 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
8843 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
8844 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
8845 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
8846 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
8847 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
8848 * config/avr/avr.c (avr_starting_frame_offset): Make static and
8849 return a HOST_WIDE_INT.
8850 (avr_builtin_setjmp_frame_value): Use it instead of
8851 STARTING_FRAME_OFFSET.
8852 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8853 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
8854 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
8856 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8857 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
8858 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
8859 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8860 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
8861 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
8862 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8863 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
8864 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
8865 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8866 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
8867 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
8869 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8870 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
8871 * config/mips/mips.c (mips_compute_frame_info): Refer to
8872 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
8873 (mips_starting_frame_offset): New function.
8874 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8875 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
8876 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
8877 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
8878 and return a HOST_WIDE_INT.
8879 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8880 (mmix_initial_elimination_offset): Refer to
8881 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
8882 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
8883 * config/pa/pa.c (pa_starting_frame_offset): New function.
8884 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
8885 (pa_expand_prologue): Likewise.
8886 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8887 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
8888 !FRAME_GROWS_DOWNWARD handling to...
8889 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8890 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
8891 !FRAME_GROWS_DOWNWARD handling to...
8892 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8893 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
8894 !FRAME_GROWS_DOWNWARD handling to...
8895 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8896 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
8898 (rs6000_starting_frame_offset): New function.
8899 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
8900 !FRAME_GROWS_DOWNWARD handling to...
8901 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8902 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
8903 !FRAME_GROWS_DOWNWARD handling to...
8904 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8905 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
8906 !FRAME_GROWS_DOWNWARD handling to...
8907 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8908 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
8909 (rs6000_starting_frame_offset): New function.
8910 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
8911 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
8912 * config/vax/vax.c (vax_starting_frame_offset): New function.
8913 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
8914 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8915 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
8916 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
8917 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8918 * system.h (STARTING_FRAME_OFFSET): Poison.
8920 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8922 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
8923 SCALAR_TYPE_MODE instead of TYPE_MODE.
8925 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8926 Alan Hayward <alan.hayward@arm.com>
8927 David Sherwood <david.sherwood@arm.com>
8929 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
8931 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8932 Alan Hayward <alan.hayward@arm.com>
8933 David Sherwood <david.sherwood@arm.com>
8935 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
8937 2017-10-23 Richard Biener <rguenther@suse.de>
8939 PR tree-optimization/82129
8940 * tree-ssa-pre.c (bitmap_set_and): Remove.
8941 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
8942 canonicalizing expressions in the set to those with lowest
8943 ID rather than taking that from the first edge.
8945 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8947 * combine.c (rtx_equal_for_field_assignment_p): Use
8948 byte_lowpart_offset.
8950 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8951 Alan Hayward <alan.hayward@arm.com>
8952 David Sherwood <david.sherwood@arm.com>
8954 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
8955 to a SUBREG_PROMOTED_VAR.
8957 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8958 Alan Hayward <alan.hayward@arm.com>
8959 David Sherwood <david.sherwood@arm.com>
8961 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
8962 (expand_debug_source_expr): Likewise.
8963 * combine.c (combine_simplify_rtx): Likewise.
8964 * cse.c (fold_rtx): Likewise.
8965 * optabs.c (expand_float): Likewise.
8966 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8967 (simplify_binary_operation_1): Likewise.
8969 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8970 Alan Hayward <alan.hayward@arm.com>
8971 David Sherwood <david.sherwood@arm.com>
8973 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
8974 (record_promoted_value): Likewise.
8975 * expr.c (expand_expr_real_2): Likewise.
8976 * ree.c (update_reg_equal_equiv_notes): Likewise.
8977 (combine_set_extension): Likewise.
8978 * rtlanal.c (low_bitmask_len): Likewise.
8979 * simplify-rtx.c (neg_const_int): Likewise.
8980 (simplify_binary_operation_1): Likewise.
8982 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8983 Alan Hayward <alan.hayward@arm.com>
8984 David Sherwood <david.sherwood@arm.com>
8986 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
8987 * regcprop.c (maybe_mode_change): Likewise.
8988 * reload1.c (alter_reg): Likewise.
8990 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8992 * inchash.h (inchash::hash::add_wide_int): New function.
8993 * lto-streamer-out.c (hash_tree): Use it.
8995 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8997 * inchash.h (inchash::hash::add_wide_int): Rename to...
8998 (inchash::hash::add_hwi): ...this.
8999 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
9000 (polymorphic_call_target_hasher::hash): Likewise.
9001 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
9002 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
9003 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
9004 * lto-streamer-out.c (hash_tree): Likewise.
9005 * optc-save-gen.awk: Likewise.
9006 * tree.c (add_expr): Likewise.
9008 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9011 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
9012 for ordered inequality comparisons even with TARGET_IEEE_FP.
9014 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9017 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
9018 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
9019 Expand with cmp<dwi>_doubleword.
9021 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9023 * extend.texi: Add x86 specific to 'nocf_check' attribute.
9024 List CET intrinsics.
9025 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
9026 specific to -fcf-protection option.
9028 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9030 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
9031 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
9032 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
9033 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
9034 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
9035 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
9036 (extra_objs): Add cet.o for Linux/x86 targets.
9037 (tmake_file): Add i386/t-cet for Linux/x86 targets.
9038 * config/i386/cet.c: New file.
9039 * config/i386/cetintrin.h: Likewise.
9040 * config/i386/t-cet: Likewise.
9041 * config/i386/cpuid.h (bit_SHSTK): New.
9042 (bit_IBT): Likewise.
9043 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
9044 pass IBT and SHSTK bits.
9045 * config/i386/i386-builtin-types.def
9046 (VOID_FTYPE_UNSIGNED_PVOID): New.
9047 (VOID_FTYPE_UINT64_PVOID): Likewise.
9048 * config/i386/i386-builtin.def: Add CET intrinsics.
9049 * config/i386/i386-c.c (ix86_target_macros_internal): Add
9050 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
9051 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
9052 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
9054 * config/i386/i386.c (rest_of_insert_endbranch): New.
9055 (pass_data_insert_endbranch): Likewise.
9056 (pass_insert_endbranch): Likewise.
9057 (make_pass_insert_endbranch): Likewise.
9058 (ix86_notrack_prefixed_insn_p): Likewise.
9059 (ix86_target_string): Add -mibt, -mshstk flags.
9060 (ix86_option_override_internal): Add flag_cf_protection
9062 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
9063 (ix86_print_operand): Add 'notrack' prefix output.
9064 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
9065 (ix86_expand_builtin): Expand CET intrinsics.
9066 (x86_output_mi_thunk): Add 'endbranch' instruction.
9067 * config/i386/i386.h (TARGET_IBT): New.
9068 (TARGET_IBT_P): Likewise.
9069 (TARGET_SHSTK): Likewise.
9070 (TARGET_SHSTK_P): Likewise.
9071 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
9072 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
9073 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
9074 (builtin_setjmp_setup): New pattern.
9075 (builtin_longjmp): Likewise.
9076 (rdssp<mode>): Likewise.
9077 (incssp<mode>): Likewise.
9078 (saveprevssp): Likewise.
9079 (rstorssp): Likewise.
9080 (wrss<mode>): Likewise.
9081 (wruss<mode>): Likewise.
9082 (setssbsy): Likewise.
9083 (clrssbsy): Likewise.
9084 (nop_endbr): Likewise.
9085 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
9087 * config/i386/immintrin.h: Include <cetintrin.h>.
9088 * config/i386/linux-common.h
9089 (file_end_indicate_exec_stack_and_cet): New prototype.
9090 (TARGET_ASM_FILE_END): New.
9092 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
9094 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
9095 latencies instead of having separate table; make difference between
9096 integer and float costs.
9097 * i386.h (processor_costs): Remove scalar_stmt_cost,
9098 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
9099 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
9101 * x86-tune-costs.h: Remove entries which has been removed in
9102 procesor_costs from all tables; make cond_taken_branch_cost
9103 and cond_not_taken_branch_cost COST_N_INSNS based.
9105 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
9107 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
9109 2017-10-20 Jakub Jelinek <jakub@redhat.com>
9111 * config/i386/i386.md (isa): Remove fma_avx512f.
9112 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
9113 <avx512>_fmadd_<mode>_mask3<round_name>,
9114 <avx512>_fmsub_<mode>_mask<round_name>,
9115 <avx512>_fmsub_<mode>_mask3<round_name>,
9116 <avx512>_fnmadd_<mode>_mask<round_name>,
9117 <avx512>_fnmadd_<mode>_mask3<round_name>,
9118 <avx512>_fnmsub_<mode>_mask<round_name>,
9119 <avx512>_fnmsub_<mode>_mask3<round_name>,
9120 <avx512>_fmaddsub_<mode>_mask<round_name>,
9121 <avx512>_fmaddsub_<mode>_mask3<round_name>,
9122 <avx512>_fmsubadd_<mode>_mask<round_name>,
9123 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
9124 (*vec_widen_umult_even_v16si<mask_name>,
9125 *vec_widen_smult_even_v16si<mask_name>): Likewise.
9126 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
9128 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9130 * extend.texi: Add 'nocf_check' documentation.
9131 * gimple.texi: Add second parameter to
9132 gimple_build_call_from_tree.
9133 * invoke.texi: Add -fcf-protection documentation.
9134 * rtl.texi: Add REG_CALL_NOTRACK documenation.
9136 2017-10-20 Richard Biener <rguenther@suse.de>
9138 PR tree-optimization/82473
9139 * tree-vect-loop.c (vectorizable_reduction): Properly get at
9140 the largest input type.
9142 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9144 * c-attribs.c (handle_nocf_check_attribute): New function.
9145 (c_common_attribute_table): Add 'nocf_check' handling.
9146 * gimple-parser.c: Add second argument NULL to
9147 gimple_build_call_from_tree.
9148 * attrib.c (comp_type_attributes): Check nocf_check attribute.
9149 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
9151 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
9152 * common.opt: Add fcf-protection flag.
9153 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
9154 * flag-types.h: Add enum cf_protection_level.
9155 * gimple.c (gimple_build_call_from_tree): Add second parameter.
9156 Add 'nocf_check' attribute propagation to gimple call.
9157 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
9158 (gimple_build_call_from_tree): Update prototype.
9159 (gimple_call_nocf_check_p): New function.
9160 (gimple_call_set_nocf_check): Likewise.
9161 * gimplify.c: Add second argument to gimple_build_call_from_tree.
9162 * ipa-icf.c: Add nocf_check attribute in statement hash.
9163 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
9164 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
9165 * toplev.c (process_options): Add flag_cf_protection handling.
9167 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9169 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
9171 2017-10-20 Richard Biener <rguenther@suse.de>
9173 PR tree-optimization/82603
9174 * tree-if-conv.c (predicate_mem_writes): Make sure to only
9175 remove false predicated stores.
9177 2017-10-20 Richard Biener <rguenther@suse.de>
9179 * graphite-isl-ast-to-gimple.c
9180 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
9181 Remove return value and simplify, dump copied stmt after lhs
9183 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
9184 Reduce dump verbosity.
9185 (gsi_insert_earliest): Likewise.
9186 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
9187 * graphite.c (print_global_statistics): Adjust dumping.
9188 (print_graphite_scop_statistics): Likewise.
9189 (print_graphite_statistics): Do not dump loops here.
9190 (graphite_transform_loops): But here.
9192 2017-10-20 Nicolas Roche <roche@adacore.com>
9194 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
9195 * configure: Regenerate.
9197 2017-10-20 Jakub Jelinek <jakub@redhat.com>
9200 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
9201 functions when optimizing replace GIMPLE_RETURN stmts with
9202 calls to __builtin_unreachable ().
9205 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
9206 for -fsanitize=thread link of executables.
9207 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
9208 link of executables.
9211 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
9213 (<shift_insn><mode>3<mask_name>): Change the last of the 3
9214 define_insns for logical vector shifts to use VI248_AVX512BW
9215 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
9216 condition, useless isa and prefix attributes. Change the first
9217 2 of these define_insns to ...
9218 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
9219 define_insn for avx512vl.
9220 (<shift_insn><mode>3): ... and this, new define_insn without
9221 masking for non-avx512vl.
9224 * config/i386/sse.md (*andnot<mode>3,
9225 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
9226 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
9227 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
9228 not applied use empty suffix even for TARGET_AVX512VL.
9229 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
9230 is applied, supply evex,evex or evex,evex,evex instead of just
9233 2017-10-20 Julia Koval <julia.koval@intel.com>
9235 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
9236 (OPTION_MASK_ISA_GFNI_UNSET): New.
9237 (ix86_handle_option): Handle OPT_mgfni.
9238 * config/i386/cpuid.h (bit_GFNI): New.
9239 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
9240 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
9241 * config/i386/i386.c (ix86_target_string): Add -mgfni.
9242 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
9243 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
9244 * config/i386/i386.opt: Add mgfni.
9246 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
9248 * config/msp430/msp430.c (msp430_option_override): Disable
9249 -fdelete-null-pointer-checks.
9250 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
9252 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9254 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
9255 of x87 and SSE instructions.
9257 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9259 * asan.c (create_cond_insert_point): Do not update edge count.
9260 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
9261 (afdo_propagate_circuit): Likewise.
9262 (afdo_calculate_branch_prob): Likewise.
9263 (afdo_annotate_cfg): Likewise.
9264 * basic-block.h (struct edge_def): Remove count.
9265 (edge_def::count): New accessor.
9266 * bb-reorder.c (rotate_loop): Update.
9267 (find_traces_1_round): Update.
9268 (connect_traces): Update.
9269 (sanitize_hot_paths): Update.
9270 * cfg.c (unchecked_make_edge): Update.
9271 (make_single_succ_edge): Update.
9272 (check_bb_profile): Update.
9273 (dump_edge_info): Update.
9274 (update_bb_profile_for_threading): Update.
9275 (scale_bbs_frequencies_int): Update.
9276 (scale_bbs_frequencies_gcov_type): Update.
9277 (scale_bbs_frequencies_profile_count): Update.
9278 (scale_bbs_frequencies): Update.
9279 * cfganal.c (connect_infinite_loops_to_exit): Update.
9280 * cfgbuild.c (compute_outgoing_frequencies): Update.
9281 (find_many_sub_basic_blocks): Update.
9282 * cfgcleanup.c (try_forward_edges): Update.
9283 (try_crossjump_to_edge): Update
9284 * cfgexpand.c (expand_gimple_cond): Update
9285 (expand_gimple_tailcall): Update
9286 (construct_exit_block): Update
9287 * cfghooks.c (verify_flow_info): Update
9288 (redirect_edge_succ_nodup): Update
9289 (split_edge): Update
9290 (make_forwarder_block): Update
9291 (duplicate_block): Update
9292 (account_profile_record): Update
9293 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
9294 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
9295 * cfgloopmanip.c (scale_loop_profile): Update.
9297 (lv_adjust_loop_entry_edge): Update.
9298 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
9299 (force_nonfallthru_and_redirect): Update.
9300 (purge_dead_edges): Update.
9301 (rtl_flow_call_edges_add): Update.
9302 * cgraphunit.c (init_lowered_empty_function): Update.
9303 (cgraph_node::expand_thunk): Update.
9304 * gimple-pretty-print.c (dump_probability): Update.
9305 (dump_edge_probability): Update.
9306 * gimple-ssa-isolate-paths.c (isolate_path): Update.
9307 * haifa-sched.c (sched_create_recovery_edges): Update.
9308 * hsa-gen.c (convert_switch_statements): Update.
9309 * ifcvt.c (dead_or_predicable): Update.
9310 * ipa-inline-transform.c (inline_transform): Update.
9311 * ipa-split.c (split_function): Update.
9312 * ipa-utils.c (ipa_merge_profiles): Update.
9313 * loop-doloop.c (add_test): Update.
9314 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
9315 * lto-streamer-in.c (input_cfg): Update.
9316 (input_function): Update.
9317 * lto-streamer-out.c (output_cfg): Update.
9318 * modulo-sched.c (sms_schedule): Update.
9319 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
9320 * predict.c (maybe_hot_edge_p): Update.
9321 (unlikely_executed_edge_p): Update.
9322 (probably_never_executed_edge_p): Update.
9323 (dump_prediction): Update.
9324 (drop_profile): Update.
9325 (propagate_unlikely_bbs_forward): Update.
9326 (determine_unlikely_bbs): Update.
9327 (force_edge_cold): Update.
9328 * profile.c (compute_branch_probabilities): Update.
9329 * reg-stack.c (better_edge): Update.
9330 * shrink-wrap.c (handle_simple_exit): Update.
9331 * tracer.c (better_p): Update.
9332 * trans-mem.c (expand_transaction): Update.
9333 (split_bb_make_tm_edge): Update.
9334 * tree-call-cdce.c: Update.
9335 * tree-cfg.c (gimple_find_sub_bbs): Update.
9336 (gimple_split_edge): Update.
9337 (gimple_duplicate_sese_region): Update.
9338 (gimple_duplicate_sese_tail): Update.
9339 (gimple_flow_call_edges_add): Update.
9340 (insert_cond_bb): Update.
9341 (execute_fixup_cfg): Update.
9342 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
9343 * tree-complex.c (expand_complex_div_wide): Update.
9344 * tree-eh.c (lower_resx): Update.
9345 (unsplit_eh): Update.
9346 (cleanup_empty_eh_move_lp): Update.
9347 * tree-inline.c (copy_edges_for_bb): Update.
9348 (freqs_to_counts): Update.
9349 (copy_cfg_body): Update.
9350 * tree-ssa-dce.c (remove_dead_stmt): Update.
9351 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
9352 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
9353 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
9354 (unloop_loops): Update.
9355 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9356 * tree-ssa-loop-split.c (connect_loops): Update.
9357 (split_loop): Update.
9358 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
9359 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
9360 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
9361 * tree-ssa-reassoc.c (branch_fixup): Update.
9362 * tree-ssa-tail-merge.c (replace_block_by): Update.
9363 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
9364 (compute_path_counts): Update.
9365 (update_profile): Update.
9366 (recompute_probabilities): Update.
9367 (update_joiner_offpath_counts): Update.
9368 (estimated_freqs_path): Update.
9369 (freqs_to_counts_path): Update.
9370 (clear_counts_path): Update.
9371 (ssa_fix_duplicate_block_edges): Update.
9372 (duplicate_thread_path): Update.
9373 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
9374 (case_bit_test_cmp): Update.
9375 (collect_switch_conv_info): Update.
9376 (gen_inbound_check): Update.
9377 (do_jump_if_equal): Update.
9378 (emit_cmp_and_jump_insns): Update.
9379 * tree-tailcall.c (decrease_profile): Update.
9380 (eliminate_tail_call): Update.
9381 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
9382 (vect_do_peeling): Update.
9383 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9384 * ubsan.c (ubsan_expand_null_ifn): Update.
9385 (ubsan_expand_ptr_ifn): Update.
9386 * value-prof.c (gimple_divmod_fixed_value): Update.
9387 (gimple_mod_pow2): Update.
9388 (gimple_mod_subtract): Update.
9389 (gimple_ic): Update.
9390 (gimple_stringop_fixed_value): Update.
9392 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
9395 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
9397 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
9399 PR rtl-optimization/82395
9400 * ira-color.c (allocno_priority_compare_func): Fix comparison step
9401 based on non_spilled_static_chain_regno_p.
9403 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
9405 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
9406 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
9407 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
9409 2017-10-19 Martin Sebor <msebor@redhat.com>
9411 PR tree-optimization/82596
9412 * tree.c (array_at_struct_end_p): Handle STRING_CST.
9414 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
9416 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
9417 (get_mem_refs_of_builtin_call): Likewise.
9418 * builtins.c (expand_builtin_apply): Adjust call to
9419 allocate_dynamic_stack_space.
9420 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
9421 the third argument to allocate_dynamic_stack_space, otherwise -1.
9422 (expand_builtin): Deal with all alloca variants.
9423 (is_inexpensive_builtin): Likewise.
9424 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
9425 * calls.c (special_function_p): Deal with all alloca variants.
9426 (initialize_argument_information): Adjust call to
9427 allocate_dynamic_stack_space.
9428 (expand_call): Likewise.
9429 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
9430 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
9431 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
9432 use it for the stack usage computation.
9433 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
9434 * function.c (gimplify_parameters): Call build_alloca_call_expr.
9435 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
9436 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
9437 (in_loop_p): Remove first argument and useless check.
9438 (pass_walloca::execute): Remove useless test and adjust call to above.
9439 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
9440 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
9441 (gimplify_call_expr): Deal with all alloca variants.
9442 * hsa-gen.c (gen_hsa_alloca): Likewise.
9443 (gen_hsa_insns_for_call): Likewise.
9444 * ipa-pure-const.c (special_builtin_state): Likewise.
9445 * tree-chkp.c (chkp_build_returned_bound): Likewise.
9446 * tree-object-size.c (alloc_object_size): Likewise.
9447 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
9448 (call_may_clobber_ref_p_1): Likewise.
9449 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
9450 (ccp_fold_stmt): Likewise.
9451 (optimize_stack_restore): Likewise.
9452 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
9453 (mark_all_reaching_defs_necessary_1): Likewise.
9454 (propagate_necessity): Likewise.
9455 (eliminate_unnecessary_stmts): Likewise.
9456 * tree.c (build_common_builtin_nodes): Build
9457 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
9458 (build_alloca_call_expr): New function.
9459 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
9460 (CASE_BUILT_IN_ALLOCA): Likewise.
9461 (build_alloca_call_expr): Declare.
9462 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
9464 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
9467 * dwarf2out.c (new_die_raw): New static inline function.
9468 (new_die): Use it to create the DIE.
9469 (add_AT_external_die_ref): Likewise.
9470 (clone_die): Likewise.
9471 (clone_as_declaration): Likewise.
9472 (dwarf2out_vms_debug_main_pointer): Likewise.
9473 (base_type_die): Likewise. Remove early return for corner cases.
9474 Do not call add_pubtype on the DIE here.
9475 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
9476 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
9477 typedefs for base types with DW_AT_endianity. Make sure a DIE with
9478 native order exists for base types, attach the DIE manually and call
9479 add_pubtype on it. Do not equate a reverse order DIE to the type.
9481 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
9483 * config/arm/arm.c (align_ok_ldrd_strd): New function.
9484 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
9486 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
9488 2017-10-19 Jakub Jelinek <jakub@redhat.com>
9490 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
9491 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
9492 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
9493 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
9494 * opts.c (sanitizer_opts): Add builtin.
9495 * ubsan.c (instrument_builtin): New function.
9496 (pass_ubsan::execute): Call it.
9497 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
9498 * doc/invoke.texi: Document -fsanitize=builtin.
9500 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
9501 builtins, store max (log2 (align), 0) into uchar field instead of
9502 align into uptr field.
9503 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
9504 store uchar 0 field instead of uptr 0 field.
9505 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
9506 instead of passing one address of struct with 2 locations pass
9507 two addresses of structs with 1 location each.
9508 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
9509 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
9510 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
9511 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
9512 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
9513 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
9514 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
9515 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
9517 2017-10-19 Martin Liska <mliska@suse.cz>
9520 * file-find.c (remove_prefix): Remove.
9521 * file-find.h (remove_prefix): Likewise.
9522 * gcc-ar.c: Remove smartness of lookup.
9524 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
9526 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
9527 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
9528 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
9530 2017-10-19 Jakub Jelinek <jakub@redhat.com>
9533 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
9534 (setcc + and to xor + setcc): New peephole2.
9536 2017-10-19 Tom de Vries <tom@codesourcery.com>
9538 * doc/sourcebuild.texi (Test Directives, Variants of
9539 dg-require-support): Add dg-require-stack-size.
9541 2017-10-19 Martin Liska <mliska@suse.cz>
9544 * gimplify.c (gimplify_decl_expr): Do not instrument variables
9545 that have a large alignment.
9546 (gimplify_target_expr): Likewise.
9548 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
9550 PR rtl-optimization/82602
9551 * ira.c (rtx_moveable_p): Return false for volatile asm.
9553 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
9556 * config/i386/i386-modes.def (CCGZ): New CC mode.
9557 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
9558 * config/i386/predicates.md (ix86_comparison_operator):
9560 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
9561 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
9562 with double-word subtraction.
9563 (put_condition_code): Handle CCGZmode.
9565 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
9567 * wide-int.cc (debug (const wide_int &)): New.
9568 (debug (const wide_int *)): New.
9569 (debug (const widest_int &)): New.
9570 (debug (const widest_int *)): New.
9572 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
9575 * lra-constraints.c (curr_insn_transform): Use non-input operand
9576 instead of output one for matched reload.
9578 2017-10-18 Bin Cheng <bin.cheng@arm.com>
9580 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
9581 (tree-ssa-loop-ivopts.h): New header file.
9582 (struct builtin_info): New fields.
9583 (classify_builtin_1): Compute and record base and offset parts for
9584 memset builtin partition by calling strip_offset.
9585 (offset_cmp, fuse_memset_builtins): New functions.
9586 (finalize_partitions): Fuse adjacent memset partitions by calling
9588 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
9589 Expose the interface.
9590 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
9592 2017-10-18 Bin Cheng <bin.cheng@arm.com>
9594 PR tree-optimization/82574
9595 * tree-loop-distribution.c (find_single_drs): New parameter. Check
9596 that data reference must be executed exactly once per iteration
9597 against the outermost loop in nest.
9598 (classify_partition): Update call to above function.
9600 2017-10-18 Richard Biener <rguenther@suse.de>
9602 PR tree-optimization/82591
9603 * graphite.c (graphite_transform_loops): Move code gen message
9605 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
9606 Here. Handle scop_to_isl_ast failing.
9607 (scop_to_isl_ast): Limit the number of ISL operations.
9609 2017-10-18 Richard Biener <rguenther@suse.de>
9611 * graphite-isl-ast-to-gimple.c
9612 (translate_isl_ast_to_gimple::set_rename): Simplify.
9613 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
9614 (graphite_copy_stmts_from_block): ... here.
9615 (copy_bb_and_scalar_dependences): Simplify.
9616 (add_parameters_to_ivs_params): Canonicalize.
9617 (generate_entry_out_of_ssa_copies): Simplify.
9618 * graphite-sese-to-poly.c (extract_affine_name): Simplify
9619 by passing in ISL dimension.
9620 (parameter_index_in_region_1): Rename to ...
9621 (parameter_index_in_region): ... this.
9622 (extract_affine): Adjust assert, pass down parameter index.
9623 (add_param_constraints): Use range-info when available.
9624 (build_scop_context): Adjust.
9625 * sese.c (new_sese_info): Adjust.
9626 (free_sese_info): Likewise.
9627 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
9628 Remove unused typedefs.
9629 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
9631 2017-10-18 Martin Liska <mliska@suse.cz>
9633 * combine.c (simplify_compare_const): Add gcc_fallthrough.
9635 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9637 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
9638 (s390_sched_init): Do not reset s390_sched_state if we entered the
9639 current basic block via a fallthru edge and all others are unlikely.
9641 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9643 * config/s390/s390.c (NUM_SIDES): New variable.
9644 (LONGRUNNING_THRESHOLD): New variable.
9645 (LATENCY_FACTOR): New variable.
9646 (s390_sched_score): Decrease score for long-running instructions on
9648 (s390_sched_variable_issue): Perform bookkeeping for long-running
9651 2017-10-18 Richard Biener <rguenther@suse.de>
9653 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
9654 Simplify with removal of the parameter rename map.
9655 (set_rename): Likewise.
9656 (should_copy_to_new_region): Likewise.
9657 (graphite_copy_stmts_from_block): Likewise.
9658 (copy_bb_and_scalar_dependences): Remove initialization of
9659 unused copied_bb_map.
9661 (copy_internal_parameters): Likewise.
9662 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
9663 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
9664 Use INTEGRAL_TYPE_P.
9665 (parameter_index_in_region_1): Rename to ...
9666 (assign_parameter_index_in_region): ... this. Assert we have
9667 a parameter we handle.
9668 (scan_tree_for_params): Adjust.
9669 * sese.h (parameter_rename_map_t): Remove.
9670 (struct sese_info_t): Remove unused parameter_rename_map and
9671 copied_bb_map members.
9672 * sese.c (new_sese_info): Adjust.
9673 (free_sese_info): Likewise.
9675 2017-10-18 Martin Liska <mliska@suse.cz>
9678 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
9679 on an abnormal edge.
9681 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
9683 * doc/invoke.texi (ffunction-sections and fdata-sections):
9686 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
9688 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
9689 the use statement can throw internally.
9691 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
9693 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
9694 any RTX present on the RHS of a SET.
9695 * compare-elim.c (try_eliminate_compare): Restore comment.
9697 2017-10-17 Jakub Jelinek <jakub@redhat.com>
9699 * langhooks.h (struct lang_hooks): Document that tree_size langhook
9700 may be also called on tcc_type nodes.
9701 * langhooks.c (lhd_tree_size): Likewise.
9703 2017-10-17 David Malcolm <dmalcolm@redhat.com>
9705 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
9706 format_warning_at_substring.
9707 (maybe_warn): Convert source_range * param to a location_t. Pass
9708 UNKNOWN_LOCATION rather than NULL to fmtwarn.
9709 (format_directive): Remove code to extract source_ranges and
9710 source_range * in favor of just a location_t.
9711 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
9713 * substring-locations.c (format_warning_va): Convert
9714 source_range * param to a location_t.
9715 (format_warning_at_substring): Likewise.
9716 * substring-locations.h (format_warning_va): Likewise.
9717 (format_warning_at_substring): Likewise.
9719 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
9721 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
9723 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
9724 and scatter/gather ops.
9726 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
9727 vec_gather_load and vec_scatter_store.
9728 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
9729 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
9730 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
9731 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
9732 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
9733 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
9735 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
9737 * reg-stack.c (compare_for_stack_reg): Add bool argument.
9738 Detect FTST instruction and handle its register pops. Only pop
9739 second operand if can_pop_second_op is true.
9740 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
9741 set can_pop_second_op to false in the compare_for_stack_reg call.
9743 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
9744 output_fp_compare for stack register operands.
9745 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
9746 instructions here. Do not emit stack register pops here. Assert
9747 that FCOMPP pops next to top stack register. Rewrite function.
9749 2017-10-17 Nathan Sidwell <nathan@acm.org>
9752 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
9753 use DECL_ASSEMBLER_NAME_RAW.
9756 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
9759 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
9760 Wilco Dijkstra <wilco.dijkstra@arm.com>
9762 * builtins.c (expand_builtin_update_setjmp_buf): Add a
9763 converstion to Pmode from the buf_addr.
9765 2017-10-17 Richard Biener <rguenther@suse.de>
9767 * graphite-dependences.c (scop_get_reads_and_writes): Change
9768 output parameters to references.
9770 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
9772 PR 71026/tree-optimization
9773 * fold-const.c (distribute_real_division): Removed.
9774 (fold_binary_loc): Remove calls to distribute_real_divison.
9776 2017-10-17 Richard Biener <rguenther@suse.de>
9778 * graphite-scop-detection.c
9779 (scop_detection::stmt_has_simple_data_refs_p): Always use
9780 the full nest as region.
9781 (try_generate_gimple_bb): Likewise.
9782 * sese.c (scalar_evolution_in_region): Simplify now that
9783 SCEV can handle instantiation in regions.
9784 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
9785 in the non-loop part of a function if requested.
9787 2017-10-17 Richard Biener <rguenther@suse.de>
9789 PR tree-optimization/82563
9790 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
9792 (graphite_regenerate_ast_isl): Call it.
9793 * graphite-scop-detection.c (build_scops): Remove entry edge split.
9795 2017-10-17 Jakub Jelinek <jakub@redhat.com>
9797 PR tree-optimization/82549
9798 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
9799 Formatting fixes. Instead of calling make_bit_field_ref with negative
9802 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
9804 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
9805 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
9806 _mm_maskz_reduce_ss): New.
9807 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
9808 __builtin_ia32_reducess_mask): Ditto..
9809 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
9810 * config/i386/sse.md (reduces<mode>): Renamed to ...
9811 (reduces<mode><mask_scalar_name>): ... this.
9812 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
9814 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
9815 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
9817 2017-10-16 David Malcolm <dmalcolm@redhat.com>
9819 * Makefile.in (OBJS): Add unique-ptr-tests.o.
9820 * selftest-run-tests.c (selftest::run_tests): Call
9821 selftest::unique_ptr_tests_cc_tests.
9822 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
9823 * unique-ptr-tests.cc: New file.
9825 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
9828 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9830 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9831 (make_hard_regno_born, make_hard_regno_dead): Update
9832 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
9834 2017-10-16 Jeff Law <law@redhat.com>
9836 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
9838 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
9840 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
9842 2017-10-16 Olivier Hainque <hainque@adacore.com>
9844 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
9845 with_cpu if we were configured for an e500v2 target cpu name.
9847 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
9849 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
9850 * doc/invoke.texi: Document +nodsp as a valid extension for
9853 2017-10-16 Martin Liska <mliska@suse.cz>
9855 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
9856 (test_set_range): Likewise.
9857 (test_range_functions): Rename to ...
9858 (test_bit_in_range): ... this.
9859 (sbitmap_c_tests): Add new test.
9861 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9863 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
9865 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
9866 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
9868 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9870 * config/aarch64/aarch64-builtins.c
9871 (aarch64_types_quadopu_lane_qualifiers): New.
9872 (TYPES_QUADOPU_LANE): New.
9873 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
9874 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
9875 (aarch64_<sur>dot_laneq<vsi2qi>): New.
9876 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
9877 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
9878 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
9879 (Vdottype, DOTPROD): New.
9880 (sur): Add SDOT and UDOT.
9882 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9884 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
9885 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
9886 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
9888 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
9889 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
9890 Enable TARGET_DOTPROD.
9891 (cortex-a75.cortex-a55): Likewise.
9892 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
9894 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9896 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
9897 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
9899 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
9901 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
9902 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
9903 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
9904 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
9905 * config/arm/types.md (neon_dot, neon_dot_q): New.
9906 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
9908 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9910 * config/arm/arm.h (TARGET_DOTPROD): New.
9911 * config/arm/arm.c (arm_arch_dotprod): New.
9912 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
9913 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
9914 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
9915 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
9916 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
9917 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
9918 * doc/invoke.texi (armv8.2-a): Document dotprod
9920 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
9922 * i386.c (ix86_vec_cost): New function.
9923 (ix86_rtx_costs): Handle vector operations better.
9924 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
9925 * x86-tune-costs.h: Add new costs to all tables.
9927 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
9929 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
9931 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
9932 divsd, sqrtss and sqrtsd
9933 * x86-tune-costs.h: Add new entries to all costs.
9934 (znver1_cost): Fix to match real instruction latencies.
9936 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9937 Michael Collison <michael.collison@arm.com>
9939 * compare-elim.c: Include emit-rtl.h.
9940 (can_merge_compare_into_arith): New function.
9941 (try_validate_parallel): Likewise.
9942 (try_merge_compare): Likewise.
9943 (try_eliminate_compare): Call the above when no previous clobber
9945 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
9948 2017-10-14 Jakub Jelinek <jakub@redhat.com>
9952 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
9953 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
9957 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
9958 to be any operand_equal_p operands. For & (B - 1) require
9959 B to be power of 2. Recognize
9960 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
9962 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9965 * optabs.c (expand_memory_blockage): Fix call of
9966 targetm.have_memory_blockage.
9968 2017-10-14 Jakub Jelinek <jakub@redhat.com>
9971 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
9972 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
9973 objects to extra_objs instead of overwriting it.
9975 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9977 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
9978 Use any_fp_register_operand as operand[3] predicate. Simplify
9979 equality test for operands[2] and operands[4] memory location.
9980 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
9981 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
9982 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
9983 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
9984 any_fp_register_operand as operand[1] predicate. Simplify
9985 equality test for operands[0] and operands[3] memory location.
9986 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
9987 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
9988 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
9990 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9992 * target-insns.def: Add memory_blockage.
9993 * optabs.c (expand_memory_blockage): New function.
9994 (expand_asm_memory_barrier): Rename ...
9995 (expand_asm_memory_blockage): ... to this.
9996 (expand_mem_thread_fence): Call expand_memory_blockage
9997 instead of expand_asm_memory_barrier.
9998 (expand_mem_singnal_fence): Ditto.
9999 (expand_atomic_load): Ditto.
10000 (expand_atomic_store): Ditto.
10001 * doc/md.texi (Standard Pattern Names For Generation):
10002 Document memory_blockage instruction pattern.
10004 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
10006 * config/rl78/rl78.c (rl78_emit_libcall): New function.
10007 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
10008 * config/rl78/rl78.md: New define_expand "adddi3".
10010 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10012 * cfghooks.c (verify_flow_info): Disable check that all probabilities
10015 2017-10-13 Jeff Law <law@redhat.com>
10017 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
10019 2017-10-13 Jakub Jelinek <jakub@redhat.com>
10022 * internal-fn.c (expand_mul_overflow): If both operands have
10023 the same highpart of -1 or 0 and the topmost bit of lowpart
10024 is different, overflow is if res <= 0 rather than res < 0.
10026 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
10028 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
10029 TARGET_P9_VECTOR code for unaligned_load case.
10031 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10033 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
10035 2017-10-13 Nathan Sidwell <nathan@acm.org>
10037 * tree-core.h (tree_contains_struct): Make bool.
10038 * tree.c (tree_contains_struct): Likewise.
10039 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
10040 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
10041 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
10042 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
10043 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
10045 2017-10-13 Richard Biener <rguenther@suse.de>
10047 * graphite-isl-ast-to-gimple.c
10048 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
10049 parameters and dominance check.
10050 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
10051 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
10052 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
10053 Do not update SSA form here or do intermediate IL verification.
10054 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
10055 (graphite_initialize): Remove check on the number of loops in
10056 the function and inline into graphite_transform_loops.
10057 (graphite_finalize): Inline into graphite_transform_loops.
10058 (graphite_transform_loops): Perform SSA update and IL verification
10060 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
10062 2017-10-13 Richard Biener <rguenther@suse.de>
10064 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
10065 graphite_expression_type_precision): Avoid global constructor
10067 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
10068 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
10069 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
10070 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
10071 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
10072 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
10074 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
10077 * config/i386/i386.h (ix86_red_zone_size): New.
10078 * config/i386/i386.md (push peephole2s): Replace
10079 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
10081 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10082 Alan Hayward <alan.hayward@arm.com>
10083 David Sherwood <david.sherwood@arm.com>
10085 * combine.c (can_change_dest_mode): Reject changes in
10086 REGMODE_NATURAL_SIZE.
10088 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10089 Alan Hayward <alan.hayward@arm.com>
10090 David Sherwood <david.sherwood@arm.com>
10092 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
10093 (expand_debug_source_expr): Likewise.
10094 * combine.c (combine_simplify_rtx): Likewise.
10095 * cse.c (fold_rtx): Likewise.
10096 * fwprop.c (canonicalize_address): Likewise.
10097 * targhooks.c (default_shift_truncation_mask): Likewise.
10099 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10100 Alan Hayward <alan.hayward@arm.com>
10101 David Sherwood <david.sherwood@arm.com>
10103 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
10104 (widened_mode): Likewise.
10105 (expand_unop): Likewise.
10106 * ree.c (transform_ifelse): Likewise.
10107 (merge_def_and_ext): Likewise.
10108 (combine_reaching_defs): Likewise.
10109 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10111 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10112 Alan Hayward <alan.hayward@arm.com>
10113 David Sherwood <david.sherwood@arm.com>
10115 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
10116 * combine.c (gen_lowpart_for_combine): Likewise.
10117 * dwarf2out.c (rtl_for_decl_location): Likewise.
10118 * final.c (alter_subreg): Likewise.
10119 * rtlhooks.c (gen_lowpart_general): Likewise.
10120 (gen_lowpart_if_possible): Likewise.
10122 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10123 Alan Hayward <alan.hayward@arm.com>
10124 David Sherwood <david.sherwood@arm.com>
10126 * calls.c (expand_call): Use subreg_lowpart_offset.
10127 * cse.c (cse_insn): Likewise.
10128 * regcprop.c (copy_value): Likewise.
10129 (copyprop_hardreg_forward_1): Likewise.
10131 2017-10-13 Jakub Jelinek <jakub@redhat.com>
10134 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
10135 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
10136 =Q constraints to +Q and into insn condition add check
10137 that operands[0] and operands[1] are equal.
10138 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
10139 =Q constraints to +Q and into insn condition add check
10140 that operands[0] is equal to either operands[1] or operands[2].
10143 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
10144 instead of handling MINUS_EXPR twice (once for each argument),
10145 canonicalize operand order and handle just once, use rtype where
10146 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
10149 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
10150 any values of __C while still being pattern recognizable as a simple
10151 rotate instruction.
10153 2017-10-13 Richard Biener <rguenther@suse.de>
10155 PR tree-optimization/82451
10157 2017-10-02 Richard Biener <rguenther@suse.de>
10159 PR tree-optimization/82355
10160 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10161 a mapping for the enclosing loop but avoid generating one for
10162 the loop tree root.
10163 (copy_bb_and_scalar_dependences): Remove premature codegen
10164 error on PHIs in blocks duplicated into multiple places.
10165 * graphite-scop-detection.c
10166 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10167 in the region use it as loop and nest to analyze the DR in.
10168 (try_generate_gimple_bb): Likewise.
10169 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10170 (add_loop_constraints): For blocks in a loop not in the region
10171 create a dimension with a single iteration.
10172 * sese.h (gbb_loop_at_index): Remove assert.
10174 * cfgloop.c (loop_preheader_edge): For the loop tree root
10175 return the single successor of the entry block.
10176 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10177 Reset the SCEV hashtable and niters.
10178 * graphite-scop-detection.c
10179 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
10180 assert that we only have POLYNOMIAL_CHREC that vary in loops
10181 contained in the region.
10182 (scop_detection::graphite_can_represent_expr): Adjust.
10183 (scop_detection::stmt_has_simple_data_refs_p): For loops
10184 not in the region set loop to NULL. The nest is now the
10185 entry edge to the region.
10186 (try_generate_gimple_bb): Likewise.
10187 * sese.c (scalar_evolution_in_region): Adjust for
10188 instantiate_scev change.
10189 * tree-data-ref.h (graphite_find_data_references_in_stmt):
10190 Make nest parameter the edge into the region.
10191 (create_data_ref): Likewise.
10192 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
10193 entry edge into a region and adjust instantiate_scev calls.
10194 (create_data_ref): Likewise.
10195 (graphite_find_data_references_in_stmt): Likewise.
10196 (find_data_references_in_stmt): Pass the loop preheader edge
10197 from the nest argument.
10198 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
10199 parameter the edge into the region.
10200 (instantiate_parameters): Use the loop preheader edge as entry.
10201 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
10203 (get_instantiated_value_entry): Make instantiate_below parameter
10204 the edge into the region.
10205 (instantiate_scev_name): Likewise. Adjust dominance checks,
10206 when we cannot use loop-based instantiation instantiate by
10207 walking use-def chains.
10208 (instantiate_scev_poly): Adjust.
10209 (instantiate_scev_binary): Likewise.
10210 (instantiate_scev_convert): Likewise.
10211 (instantiate_scev_not): Likewise.
10212 (instantiate_array_ref): Remove.
10213 (instantiate_scev_3): Likewise.
10214 (instantiate_scev_2): Likewise.
10215 (instantiate_scev_1): Likewise.
10216 (instantiate_scev_r): Do not blindly handle N-operand trees.
10217 Do not instantiate array-refs. Handle all constants and invariants.
10218 (instantiate_scev): Make instantiate_below parameter
10219 the edge into the region.
10220 (resolve_mixers): Use the loop preheader edge for the region
10221 parameter to instantiate_scev_r.
10222 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
10224 2017-10-13 Richard Biener <rguenther@suse.de>
10226 PR tree-optimization/82525
10227 * graphite-isl-ast-to-gimple.c
10228 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
10230 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
10231 Fail code generation when we cannot represent the isl integer.
10232 (binary_op_to_tree): Elide modulo operations that are no-ops
10233 in the type we code generate. Remove now superfluous code
10236 2017-10-13 Richard Biener <rguenther@suse.de>
10238 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
10239 (scop_detection::harmful_loop_in_region): Remove premature
10240 IV type restriction.
10241 (scop_detection::graphite_can_represent_scev): We can handle
10242 pointer IVs just fine.
10244 2017-10-13 Alan Modra <amodra@gmail.com>
10246 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
10247 "Clobbers and Scratch Registers". Add paragraph on
10248 alternative to clobbers for scratch registers and OpenBLAS
10251 2017-10-13 Alan Modra <amodra@gmail.com>
10253 * doc/extend.texi (Clobbers): Correct vax example. Delete old
10254 example of a memory input for a string of known length. Move
10255 commentary out of table. Add a number of new examples
10256 covering array memory inputs.
10258 2017-10-12 Martin Liska <mliska@suse.cz>
10260 PR tree-optimization/82493
10261 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
10262 (test_range_functions): New function.
10263 (sbitmap_c_tests): Likewise.
10264 * selftest-run-tests.c (selftest::run_tests): Run new tests.
10265 * selftest.h (sbitmap_c_tests): New function.
10267 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10269 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
10271 * config/rs6000/amo.h: Fix spacing issue.
10273 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10276 * config/i386/i386.md (*ashl<mode>3_mask_1,
10277 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
10278 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
10281 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
10283 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
10284 (profile_probability): Set max_probability
10285 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
10287 * profile-count.c (profile_probability::differs_from_p): Do not
10288 rely on max_probaiblity == 10000
10290 2017-10-12 Jeff Law <law@redhat.com>
10292 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
10295 2017-10-12 Martin Sebor <msebor@redhat.com>
10299 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
10300 (handle_alias_pairs): Call it.
10301 * common.opt (-Wattribute-alias): New option.
10302 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
10303 * doc/invoke.texi (-Wattribute-alias): Document.
10305 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
10308 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
10310 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10312 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10313 (make_hard_regno_born, make_hard_regno_dead): Update
10314 bb_killed_pseudos and bb_gen_pseudos.
10316 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
10318 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
10320 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
10322 * config/alpha/alpha.c (alpha_split_conditional_move):
10323 Use std::swap instead of manually swapping.
10324 (alpha_stdarg_optimize_hook): Ditto.
10325 (alpha_canonicalize_comparison): Ditto.
10327 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10329 * tree-loop-distribution.c (struct builtin_info): New struct.
10330 (struct partition): Refactor fields into struct builtin_info.
10331 (partition_free): Free struct builtin_info.
10332 (build_size_arg_loc, build_addr_arg_loc): Delete.
10333 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
10334 information from struct builtin_info.
10335 (find_single_drs): New function refactored from classify_partition.
10336 Also moved builtin validity checks to this function.
10337 (compute_access_range, alloc_builtin): New functions.
10338 (classify_builtin_st, classify_builtin_ldst): New functions.
10339 (classify_partition): Refactor code into functions find_single_drs,
10340 classify_builtin_st and classify_builtin_ldst.
10341 (distribute_loop): Don't do runtime alias check when distributing
10343 (find_seed_stmts_for_distribution): New function.
10344 (pass_loop_distribution::execute): Refactor code finding seed
10345 stmts into above function. Support distribution for the innermost
10346 two-level loop nest. Adjust dump information.
10348 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10350 * tree-loop-distribution.c: Adjust the general comment.
10351 (NUM_PARTITION_THRESHOLD): New macro.
10352 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
10353 (classify_partition): Skip builtin pattern of loop nest's inner loop.
10354 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
10355 in call to build_partition_graph.
10356 (finalize_partitions): New parameter. Make loop distribution more
10357 conservative by fusing more partitions.
10358 (distribute_loop): Don't do runtime alias check in case of loop nest
10360 (find_seed_stmts_for_distribution): New function.
10361 (prepare_perfect_loop_nest): New function.
10362 (pass_loop_distribution::execute): Refactor code finding seed stmts
10363 and loop nest into above functions. Support loop nest distribution.
10364 Adjust dump information accordingly.
10366 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10368 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
10369 and set PTYPE_SEQUENTIAL for merged partition.
10371 2017-10-12 Richard Biener <rguenther@suse.de>
10373 PR tree-optimization/69728
10375 2017-09-19 Richard Biener <rguenther@suse.de>
10377 PR tree-optimization/69728
10378 * graphite-sese-to-poly.c (schedule_error): New global.
10379 (add_loop_schedule): Handle empty domain by failing the
10381 (build_original_schedule): Handle schedule_error.
10383 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
10384 domain by returning an unchanged schedule.
10386 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10388 * genrecog.c (validate_pattern): For VEC_SELECT verify that
10389 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
10391 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
10393 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
10396 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10399 * expr.c (store_field): Don't optimize away bitsize == 0 store
10400 from CALL_EXPR with addressable return type.
10402 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
10404 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
10405 * config/rs6000/rs6000.md (sel): Delete mode attribute.
10406 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
10407 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
10408 TARGET_ISEL instead of TARGET_ISEL<sel>.
10410 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
10412 * config/rs6000/rs6000.c
10413 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
10415 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
10417 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
10419 (reg_or_cint_operand): Fix comment.
10420 (reg_or_zero_operand): New predicate.
10421 * config/rs6000/rs6000-protos.h (output_isel): Delete.
10422 * config/rs6000/rs6000.c (output_isel): Delete.
10423 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
10424 instead of reg_or_cint_operand. Output instruction directly (not via
10426 (isel_unsigned_<mode>): Ditto.
10427 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
10428 gpc_reg_operand. Add an instruction alternative for this. Output
10429 instruction directly.
10430 (*isel_reversed_unsigned_<mode>): Ditto.
10432 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10434 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
10435 (TARGET_CANONICALIZE_COMPARISON): Define.
10437 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
10440 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
10441 Check whether the dest is REG before adding REG_EQUIV note.
10443 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
10446 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10448 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10449 (make_hard_regno_born, make_hard_regno_dead): Update
10450 bb_killed_pseudos and bb_gen_pseudos.
10452 2017-10-11 Nathan Sidwell <nathan@acm.org>
10454 * incpath.h (enum incpath_kind): Name enum, prefix values.
10455 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
10456 * incpath.c (heads, tails): Use INC_MAX.
10457 (add_env_var_paths, add_standard_paths): Use incpath_kind.
10458 (merge_include_chains, split_quote_chain,
10459 register_include_chains): Update incpath_kind names.
10460 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
10461 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
10463 (add_framework_path, darwin_register_objc_includes): Likewise.
10464 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
10466 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10468 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
10469 Do not use float_operator operator predicate.
10470 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
10471 * config/i386/predicates.md (float_operator): Remove predicate.
10473 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10475 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
10476 (*jccxf_i387): Ditto.
10477 (*jcc<mode>_i387): Ditto.
10478 (*jccu<mode>_i387): Ditto.
10479 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
10480 (*jcc_*_i387 splitters): Remove.
10481 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
10482 * config/i386/i386.c (ix86_split_fp_branch): Remove.
10483 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
10486 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
10488 * profile-count.h (slow_safe_scale_64bit): New function.
10489 (safe_scale_64bit): New inline.
10490 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
10491 * profile-count.c: Include wide-int.h
10492 (slow_safe_scale_64bit): New.
10494 2017-10-11 Nathan Sidwell <nathan@acm.org>
10496 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
10497 HAS_DECL_ASSEMBLER_NAME_P.
10498 * gimple-expr.c (gimple_decl_printable_name: Check
10499 HAS_DECL_ASSEMBLER_NAME_P too.
10500 * ipa-utils.h (type_in_anonymous_namespace_p): Check
10501 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
10502 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
10503 * passes.c (rest_of_decl_compilation): Check
10504 HAS_DECL_ASSEMBLER_NAME_P too.
10505 * recog.c (verify_changes): Likewise.
10506 * tree-pretty-print.c (dump_decl_name): Likewise.
10507 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
10509 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
10510 (SET_DECL_ASSEMBLER_NAME): Use it.
10511 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
10512 (COPY_DECL_ASSEMBLER_NAME): Likewise.
10513 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
10515 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
10517 * config.gcc (i386, x86_64): Add extra objects.
10518 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
10519 (ix86_min_insn_size): Declare.
10520 (ix86_issue_rate): Declare.
10521 (ix86_adjust_cost): Declare.
10522 (ia32_multipass_dfa_lookahead): Declare.
10523 (ix86_macro_fusion_p): Declare.
10524 (ix86_macro_fusion_pair_p): Declare.
10525 (ix86_bd_has_dispatch): Declare.
10526 (ix86_bd_do_dispatch): Declare.
10527 (ix86_core2i7_init_hooks): Declare.
10528 (ix86_atom_sched_reorder): Declare.
10529 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
10530 (COSTS_N_BYTES): Move to x86-tune-costs.h.
10531 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
10532 (rip_relative_addr_p): Rename to ...
10533 (ix86_rip_relative_addr_p): ... this one; export.
10534 (memory_address_length): Update.
10535 (ix86_issue_rate): Move to x86-tune-sched.c.
10536 (ix86_flags_dependent): Move to x86-tune-sched.c.
10537 (ix86_agi_dependent): Move to x86-tune-sched.c.
10538 (exact_dependency_1): Move to x86-tune-sched.c.
10539 (exact_store_load_dependency): Move to x86-tune-sched.c.
10540 (ix86_adjust_cost): Move to x86-tune-sched.c.
10541 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
10542 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
10543 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
10544 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
10545 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
10546 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
10547 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
10548 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
10549 (min_insn_size): Rename to ...
10550 (ix86_min_insn_size): ... this one; export.
10551 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
10552 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
10553 (core2i7_first_cycle_multipass_backtrack): Move to
10554 x86-tune-sched-core.c.
10555 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
10556 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
10557 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
10558 (ix86_avoid_jump_mispredicts): Update.
10559 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
10560 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
10561 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
10562 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
10563 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
10564 (MAX_INSN): Move to ix86-tune-sched-bd.c.
10565 (MAX_IMM): Move to ix86-tune-sched-bd.c.
10566 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
10567 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
10568 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
10569 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
10570 (MAX_STORE): Move to ix86-tune-sched-bd.c.
10571 (BIG): Move to ix86-tune-sched-bd.c.
10572 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
10573 (enum insn_path): Move to ix86-tune-sched-bd.c.
10574 (get_mem_group): Move to ix86-tune-sched-bd.c.
10575 (is_cmp): Move to ix86-tune-sched-bd.c.
10576 (dispatch_violation): Move to ix86-tune-sched-bd.c.
10577 (is_branch): Move to ix86-tune-sched-bd.c.
10578 (is_prefetch): Move to ix86-tune-sched-bd.c.
10579 (init_window): Move to ix86-tune-sched-bd.c.
10580 (allocate_window): Move to ix86-tune-sched-bd.c.
10581 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
10582 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
10583 (process_end_window): Move to ix86-tune-sched-bd.c.
10584 (allocate_next_window): Move to ix86-tune-sched-bd.c.
10585 (find_constant): Move to ix86-tune-sched-bd.c.
10586 (get_num_immediates): Move to ix86-tune-sched-bd.c.
10587 (has_immediate): Move to ix86-tune-sched-bd.c.
10588 (get_insn_path): Move to ix86-tune-sched-bd.c.
10589 (get_insn_group): Move to ix86-tune-sched-bd.c.
10590 (count_num_restricted): Move to ix86-tune-sched-bd.c.
10591 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
10592 (add_insn_window): Move to ix86-tune-sched-bd.c.
10593 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
10594 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
10595 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
10596 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
10597 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
10598 (do_dispatch): Move to ix86-tune-sched-bd.c.
10599 (has_dispatch): Move to ix86-tune-sched-bd.c.
10600 * config/i386/t-i386: Add new object files.
10601 * config/i386/x86-tune-costs.h: New file.
10602 * config/i386/x86-tune-sched-atom.c: New file.
10603 * config/i386/x86-tune-sched-bd.c: New file.
10604 * config/i386/x86-tune-sched-core.c: New file.
10605 * config/i386/x86-tune-sched.c: New file.
10607 2017-10-11 Liu Hao <lh_mouse@126.com>
10609 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
10610 the generic version below instead.
10611 (should_colorize): Recognize Windows consoles as terminals
10613 * pretty-print.c [__MINGW32__] (write_all): New function.
10614 [__MINGW32__] (find_esc_head): Likewise.
10615 [__MINGW32__] (find_esc_terminator): Likewise.
10616 [__MINGW32__] (eat_esc_sequence): Likewise.
10617 [__MINGW32__] (mingw_ansi_fputs): New function that handles
10619 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
10622 2017-10-11 Richard Biener <rguenther@suse.de>
10624 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
10625 Properly call analyze_scalar_evolution with the loop of the stmt.
10627 2017-10-11 Richard Biener <rguenther@suse.de>
10629 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
10630 * tree-core.h (tree_base): Add chrec_var union member.
10631 * tree.h (CHREC_VAR): Remove.
10632 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
10633 * tree-chrec.h (build_polynomial_chrec): Adjust.
10634 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
10635 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
10637 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
10639 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
10640 * match.pd: ... here.
10641 ((T) X == (T) Y): Relax condition.
10643 2017-10-11 Bin Cheng <bin.cheng@arm.com>
10645 PR tree-optimization/82472
10646 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
10648 (break_alias_scc_partitions): Update postorder number.
10650 2017-10-11 Martin Liska <mliska@suse.cz>
10653 * opts.c (parse_no_sanitize_attribute): Do not use error_value
10655 * opts.h (parse_no_sanitize_attribute): Remove last argument.
10657 2017-10-11 Martin Liska <mliska@suse.cz>
10659 * print-rtl.c (print_insn): Move declaration of idbuf
10660 to same scope as name.
10662 2017-10-11 Martin Liska <mliska@suse.cz>
10667 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
10668 volatile arguments.
10670 2017-10-11 Martin Liska <mliska@suse.cz>
10673 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
10674 volatile arguments.
10676 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
10678 * config.gcc (default_gnu_indirect_function): Default to yes for
10679 arm*-*-linux* with glibc.
10681 2017-10-11 Richard Biener <rguenther@suse.de>
10683 * tree-scalar-evolution.c (get_scalar_evolution): Handle
10684 default-defs and types we do not want to analyze.
10685 (interpret_loop_phi): Replace unreachable code with an assert.
10686 (compute_scalar_evolution_in_loop): Remove and inline ...
10687 (analyze_scalar_evolution_1): ... here, replacing condition with
10688 what makes the intent clearer. Remove handling of cases
10689 get_scalar_evolution now handles.
10691 2017-10-10 Jim Wilson <wilson@tuliptree.org>
10693 PR rtl-optimization/81434
10694 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
10695 comment for main loop. In sched_group_found if, also add checks for
10696 pass and min_cost_group.
10698 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
10700 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
10701 (rs6000_insn_cost): New function.
10702 * config/rs6000/rs6000.md (cost): New attribute.
10704 2017-10-10 Jakub Jelinek <jakub@redhat.com>
10705 H.J. Lu <hongjiu.lu@intel.com>
10709 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
10710 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
10711 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
10712 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
10713 to OPTION_MASK_ISA_AVX512VL - builtins that have both
10714 OPTION_MASK_ISA_MMX and some other bit set require both
10715 mmx and the ISAs without the mmx bit.
10716 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
10717 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
10718 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
10719 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
10720 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
10721 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
10722 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
10723 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
10724 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
10725 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
10726 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
10727 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
10728 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
10729 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
10730 Add OPTION_MASK_ISA_MMX.
10732 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
10734 * config.gcc (armv7*-*-freebsd*): New target.
10735 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
10737 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
10739 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
10740 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
10743 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
10745 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
10746 a template parameter.
10747 (WIDE_INT_REF_FOR): Update accordingly.
10748 * tree.h (wi::int_traits <const_tree>): Delete.
10749 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
10750 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
10751 (wi::tree_to_wide_ref): New typedef.
10752 (wi::to_wide): New function.
10753 * calls.c (get_size_range): Use wi::to_wide when operating on
10754 trees as wide_ints.
10755 * cgraph.c (cgraph_node::create_thunk): Likewise.
10756 * config/i386/i386.c (ix86_data_alignment): Likewise.
10757 (ix86_local_alignment): Likewise.
10758 * dbxout.c (stabstr_O): Likewise.
10759 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
10760 * expr.c (const_vector_from_tree): Likewise.
10761 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
10762 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
10763 (fold_negate_expr_1, int_const_binop_1, const_binop)
10764 (fold_convert_const_int_from_real, optimize_bit_field_compare)
10765 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
10766 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
10767 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
10768 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
10769 (fold_not_const, round_up_loc): Likewise.
10770 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
10771 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
10772 (alloca_call_type): Likewise.
10773 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
10774 * godump.c (go_output_typedef): Likewise.
10775 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
10776 * internal-fn.c (get_min_precision): Likewise.
10777 * ipa-cp.c (ipcp_store_vr_results): Likewise.
10778 * ipa-polymorphic-call.c
10779 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
10780 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
10781 (ipa_modify_call_arguments): Likewise.
10782 * match.pd: Likewise.
10783 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
10784 * print-tree.c (print_node_brief, print_node): Likewise.
10785 * stmt.c (expand_case): Likewise.
10786 * stor-layout.c (layout_type): Likewise.
10787 * tree-affine.c (tree_to_aff_combination): Likewise.
10788 * tree-cfg.c (group_case_labels_stmt): Likewise.
10789 * tree-data-ref.c (dr_analyze_indices): Likewise.
10790 (prune_runtime_alias_test_list): Likewise.
10791 * tree-dump.c (dequeue_and_dump): Likewise.
10792 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
10793 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
10794 * tree-pretty-print.c (dump_generic_node): Likewise.
10795 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
10796 (simple_iv_with_niters): Likewise.
10797 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
10798 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
10799 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
10800 * tree-ssa-loop-niter.c (split_to_var_and_offset)
10801 (refine_value_range_using_guard, number_of_iterations_ne_max)
10802 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
10803 (get_cst_init_from_scev, record_nonwrapping_iv)
10804 (scev_var_range_cant_overflow): Likewise.
10805 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
10806 * tree-ssa-pre.c (compute_avail): Likewise.
10807 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
10808 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
10809 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
10810 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
10811 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
10812 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
10813 (dump_case_nodes, try_switch_expansion): Likewise.
10814 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
10815 (vect_do_peeling): Likewise.
10816 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
10817 * tree-vect-stmts.c (vectorizable_load): Likewise.
10818 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
10819 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
10820 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
10821 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
10822 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
10823 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
10824 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
10825 (evrp_dom_walker::before_dom_children): Likewise.
10826 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
10827 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
10828 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
10829 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
10830 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
10831 (get_range_pos_neg): Likewise.
10832 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
10833 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
10834 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
10835 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
10836 * config/avr/avr.c (avr_fold_builtin): Likewise.
10837 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
10838 * config/msp430/msp430.c (msp430_attr): Likewise.
10839 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
10840 * config/powerpcspe/powerpcspe-c.c
10841 (altivec_resolve_overloaded_builtin): Likewise.
10842 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
10843 (rs6000_expand_ternop_builtin): Likewise.
10844 * config/rs6000/rs6000-c.c
10845 (altivec_resolve_overloaded_builtin): Likewise.
10846 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
10847 (rs6000_expand_ternop_builtin): Likewise.
10848 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
10850 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10852 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
10853 when copying loop nest with only one inner loop.
10855 2017-10-10 Richard Biener <rguenther@suse.de>
10857 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
10858 blocks if SCEV is active.
10859 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
10861 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
10862 (scev_initialize): Assert we are not yet initialized.
10864 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10866 * tree-loop-distribution.c (generate_loops_for_partition): Remove
10867 inner loop's exit stmt by making it always exit the loop, otherwise
10868 we would generate an infinite empty loop.
10870 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10872 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
10873 renaming variables in new preheader if it's deleted.
10875 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10877 * tree-loop-distribution.c (struct partition): Remove unused field
10878 loops of the structure.
10879 (partition_alloc, partition_free): Ditto.
10880 (build_rdg_partition_for_vertex): Ditto.
10882 2017-10-09 Jeff Law <law@redhat.com>
10884 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
10885 return type to match prototype and documentation.
10887 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10889 * config/rs6000/rs6000.c (processor_costs): Move to ...
10890 * config/rs6000/rs6000.h: ... here.
10891 (rs6000_cost): Declare.
10893 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
10895 * except.c (setjmp_fn): New global variable.
10896 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
10897 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
10898 if DONT_USE_BUILTIN_SETJMP is defined.
10900 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10902 * target.def (insn_cost): New hook.
10903 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
10904 * doc/tm.texi: Regenerate.
10905 * rtlanal.c (insn_cost): Use the new hook.
10907 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10909 * combine.c (combine_validate_cost): Compute the new insn_cost,
10910 not just pattern_cost.
10911 (try_combine): Adjust comment.
10913 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10915 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
10917 * combine.c (uid_insn_cost): Adjust comment.
10918 (combine_validate_cost): Adjust comment. Use pattern_cost instead
10920 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
10921 * dse.c (find_shift_sequence): Ditto.
10922 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
10923 (bb_valid_for_noce_process_p): Use pattern_cost.
10924 * rtl.h (insn_rtx_cost): Delete.
10925 (pattern_cost): New prototype.
10926 (insn_cost): New prototype.
10927 * rtlanal.c (insn_rtx_cost): Rename to...
10928 (pattern_cost): ... this.
10931 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
10933 * config/i386/i386.md (*jcc_2): Remove insn pattern.
10934 (*jcc<mode>_0_r_i387): Ditto.
10935 (*jccxf_r_i387): Ditto.
10936 (*jcc<mode>_r_i387): Ditto.
10937 (*jccu<mode>_r_i387): Ditto.
10938 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
10939 (*jcc): Rename from *jcc_1.
10941 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10943 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
10944 deferred rescans after the lvx/stvx recombination pre-pass.
10946 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
10948 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
10949 memory operation instruction support.
10950 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
10951 (rs6000-ibm-aix[789]*): Likewise.
10952 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
10953 Document new functions.
10955 2017-10-09 Richard Biener <rguenther@suse.de>
10957 PR tree-optimization/82397
10958 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
10959 equality only for semantically equal trees.
10961 2017-10-09 Richard Biener <rguenther@suse.de>
10963 PR tree-optimization/82449
10964 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
10965 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
10966 allow constant addresses.
10967 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
10970 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10972 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
10975 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10978 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
10981 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10984 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
10986 2017-10-09 Jakub Jelinek <jakub@redhat.com>
10989 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
10990 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
10992 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
10994 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
10995 (WI_BINARY_PREDICATE_RESULT): Likewise.
10996 (wi::binary_traits::operator_result): New type.
10997 (wi::binary_traits::predicate_result): Likewise.
10998 (generic_wide_int::operator~, unary generic_wide_int::operator-)
10999 (generic_wide_int::operator==, generic_wide_int::operator!=)
11000 (generic_wide_int::operator&, generic_wide_int::and_not)
11001 (generic_wide_int::operator|, generic_wide_int::or_not)
11002 (generic_wide_int::operator^, generic_wide_int::operator+
11003 (binary generic_wide_int::operator-, generic_wide_int::operator*):
11005 (operator~, unary operator-, operator==, operator!=, operator&)
11006 (operator|, operator^, operator+, binary operator-, operator*): New
11008 * expr.c (get_inner_reference): Use wi::bit_and_not.
11009 * fold-const.c (fold_binary_loc): Likewise.
11010 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
11011 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
11012 (bit_value_binop): Likewise.
11013 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
11014 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
11015 (extract_range_from_binary_expr_1): Likewise.
11016 (masked_increment): Likewise.
11017 (simplify_bit_ops_using_ranges): Likewise.
11019 2017-10-09 Martin Jambor <mjambor@suse.cz>
11022 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
11023 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
11024 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
11026 (hsa_fixup_mov_insn_type): New function.
11027 (hsa_op_with_type::get_in_type): Use it.
11028 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
11029 immediates in an assert.
11030 (hsa_op_with_type::extend_int_to_32bit): New method.
11031 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
11032 types. Convert to dest type if necessary.
11033 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
11034 (reg_for_gimple_ssa): Pass false as min32int to
11035 hsa_type_for_scalar_tree_type.
11036 (gen_hsa_addr): Fixup type when creating addresable temporary.
11037 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
11038 (gen_hsa_unary_operation): Extend operands and convert to dest type if
11039 necessary. Call hsa_fixup_mov_insn_type.
11040 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
11041 extend operands and convert to dest type if necessary.
11042 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
11043 to dest type if necessary.
11044 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
11045 if conversion nt necessary and size matches.
11046 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
11047 to dest type if necessary.
11048 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
11049 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
11051 (gen_hsa_clrsb): Likewise.
11052 (gen_hsa_ffs): Likewise.
11053 (gen_hsa_divmod): Extend operands and convert to dest type if
11055 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
11057 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
11059 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
11060 Remove empty default arguments. Use a brace block as output
11062 (conditional return): Ditto.
11064 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
11065 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
11067 (group_ending_nop): Ditto.
11068 (doloop_end): Ditto.
11069 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
11070 (splitters for those): Ditto.
11072 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
11074 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
11075 a conditional jump (and the compare for it) so that pc_rtx is the
11077 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
11078 for the deleted and renamed ctr<mode>_internal[234] patterns.
11079 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
11080 Delete second conditional return pattern.
11081 (ctr<mode>_internal2): Delete this second bdnz pattern.
11082 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
11083 (ctr<mode>_internal4): Delete this second bdz pattern.
11085 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
11087 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
11088 (always_initialized_rtx_for_ssa_name_p): New predicate.
11089 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
11090 (finish_out_of_ssa): Free new field of SA.
11091 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
11092 * tree-ssa-coalesce.c: Include tree-ssa.h.
11093 (get_parm_default_def_partitions): Remove extern keyword.
11094 (get_undefined_value_partitions): New function.
11095 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
11096 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
11097 uninitialized bits.
11098 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
11100 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
11102 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
11104 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
11106 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
11107 for targets that preffer 128bit.
11109 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
11111 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
11113 2017-10-08 Olivier Hainque <hainque@adacore.com>
11115 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
11116 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
11118 (thumb_set_return_address): Likewise.
11120 2017-10-08 Olivier Hainque <hainque@adacore.com>
11122 * common/config/arm/arm-common.c (arm_except_unwind_info):
11123 Handle DWARF2_UNWIND_INFO.
11125 2017-10-07 Michael Collison <michael.collison@arm.com>
11127 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
11130 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
11132 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
11133 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
11134 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
11135 defined, force the creation of a new block for a dispatch label.
11137 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
11139 * invoke.texi (Wsuggest-attribute=cold): Document.
11140 * common.opt (Wsuggest-attribute=cold): New
11141 * ipa-pure-const.c (warn_function_cold): New function.
11142 * predict.c (compute_function_frequency): Use it.
11143 * predict.h (warn_function_cold): Declare.
11145 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
11147 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
11150 2017-10-06 Martin Liska <mliska@suse.cz>
11152 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
11153 keyword for member functions.
11154 (struct sanopt_tree_couple): New struct.
11155 (struct sanopt_tree_couple_hash): New function.
11156 (struct sanopt_ctx): Add new hash_map.
11157 (has_dominating_ubsan_ptr_check): New function.
11158 (record_ubsan_ptr_check_stmt): Likewise.
11159 (maybe_optimize_ubsan_ptr_ifn): Likewise.
11160 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
11161 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
11163 2017-10-06 Sudakshina Das <sudi.das@arm.com>
11166 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
11167 aarch64_simd_valid_immediate on CONST_VECTORs.
11168 (aarch64_reg_or_bic_imm): Likewise.
11170 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
11172 PR rtl-optimization/82396
11173 * haifa-sched.c (ready_sort_real): Disable qsort checking.
11175 2017-10-06 Sebastian Pop <sebpop@gmail.com>
11177 * graphite-dependences.c (scop_get_reads): Move code to...
11178 (scop_get_must_writes): Move code to...
11179 (scop_get_may_writes): Move code to...
11180 (scop_get_reads_and_writes): ... here.
11181 (scop_get_dependences): Call scop_get_reads_and_writes.
11183 2017-10-06 Jakub Jelinek <jakub@redhat.com>
11185 PR tree-optimization/82434
11186 * fold-const.h (can_native_encode_type_p,
11187 can_native_encode_string_p): Remove.
11188 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
11189 don't encode anything, just return what would be otherwise returned.
11190 (native_encode_fixed, native_encode_complex, native_encode_vector):
11192 (native_encode_string): Likewise. Inline by hand
11193 can_native_encode_string_p.
11194 (can_native_encode_type_p): Remove.
11195 (can_native_encode_string_p): Remove.
11196 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
11197 STRING_CSTs using can_native_encode_string_p, test all
11198 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
11199 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
11200 argument from native_encode_expr.
11201 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
11202 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
11205 2017-10-06 Richard Biener <rguenther@suse.de>
11207 PR tree-optimization/82397
11208 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
11209 operand_equal_p but rely on data_ref_compare_tree for detecting
11211 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
11212 to match up with dr_group_sort_cmp.
11214 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11217 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
11219 * config/s390/s390-builtin-types.def: Regenerate.
11221 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11224 * config/s390/s390-builtin-types.def: Regenerate.
11225 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
11226 Change flag from B_VXE to B_VX.
11227 (s390_vec_min_dbl): Remove B_VXE flag.
11229 2017-10-06 Richard Biener <rguenther@suse.de>
11231 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
11232 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
11233 translate_isl_ast_to_gimple::is_valid_rename,
11234 translate_isl_ast_to_gimple::get_rename,
11235 translate_isl_ast_to_gimple::get_def_bb_for_const,
11236 translate_isl_ast_to_gimple::get_new_name,
11237 translate_isl_ast_to_gimple::collect_all_ssa_names,
11238 translate_isl_ast_to_gimple::copy_loop_phi_args,
11239 translate_isl_ast_to_gimple::collect_all_ssa_names,
11240 translate_isl_ast_to_gimple::copy_loop_phi_args,
11241 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
11242 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
11243 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
11244 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
11245 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
11246 translate_isl_ast_to_gimple::copy_cond_phi_args,
11247 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
11248 translate_isl_ast_to_gimple::edge_for_new_close_phis,
11249 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
11250 translate_isl_ast_to_gimple::rename_uses,
11251 translate_isl_ast_to_gimple::rename_all_uses): Remove.
11252 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
11253 (set_rename_for_each_def): Likewise.
11254 (graphite_copy_stmts_from_block): Handle debug stmt resetting
11255 here. Handle rewriting SCEV analyzable uses here.
11256 (copy_bb_and_scalar_dependences): Generate code for PHI
11258 (graphite_regenerate_ast_isl): Adjust.
11259 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
11260 (add_write, add_read): New functions.
11261 (build_cross_bb_scalars_def): Use it and simplify.
11262 (build_cross_bb_scalars_use): Likewise.
11263 (graphite_find_cross_bb_scalar_vars): Inline into...
11264 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
11265 simulating out-of-SSA. Compute liveout and add dependencies.
11266 (build_scops): Force an empty entry block.
11267 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
11269 (sese_build_liveouts): Declare.
11270 (sese_trivially_empty_bb_p): Likewise.
11271 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
11272 compute liveout and debug_liveout.
11273 (sese_bad_liveouts_use): Remove.
11274 (sese_reset_debug_liveouts_bb): Likewise.
11275 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
11276 (sese_build_liveouts): Build liveout and debug_liveout and store
11278 (new_sese_info): Adjust.
11279 (free_sese_info): Likewise.
11280 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
11281 do not build liveout here.
11282 (move_sese_in_condition): Adjust region entry.
11283 (scev_analyzable_p): Match up with chrec_apply requirements.
11284 (sese_trivially_empty_bb_p): New.
11285 * tree-into-ssa.c (get_reaching_def): Properly support generating
11286 default-defs for incremental rewrite of anonymous names.
11288 2017-10-06 Richard Biener <rguenther@suse.de>
11290 * graphite-sese-to-poly.c (extract_affine): For casts increasing
11291 precision do not perform modulo reduction.
11293 2017-10-06 Richard Biener <rguenther@suse.de>
11295 PR tree-optimization/82436
11296 * tree-vect-slp.c (vect_supported_load_permutation_p): More
11297 conservatively choose the vectorization factor when checking
11298 whether we can perform the required load permutation.
11299 (vect_transform_slp_perm_load): Assert when we may not fail.
11301 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
11303 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
11304 message for incompatible -msdata=* and -mcall-* options.
11306 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11308 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
11309 rate for post-reload scheduling.
11311 2017-10-05 Tamar Christina <tamar.christina@arm.com>
11313 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
11315 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11317 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
11318 to improve monte carlo in scimark.
11320 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11322 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
11323 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
11324 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
11325 pentium4_cost, nocona_cost): Set reassociation width to 1.
11326 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
11327 width to 2 for fp operations and 1 otherwise.
11328 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
11330 (atom_cost): Set reassociation width to 2.
11331 (slm_cost, generic_cost): Set fp reassociation width
11332 to 2 and 1 otherwise.
11333 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
11334 (core_cost): Set fp reassociation width to 4 and vector to 2.
11335 (ix86_reassociation_width): Rewrite using cost table; special case
11336 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
11337 and TARGET_AVX128_OPTIMAL.
11338 * config/i386/i386.h (processor_costs): Add
11339 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
11340 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
11341 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
11342 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
11343 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
11344 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
11346 2017-10-05 Nathan Sidwell <nathan@acm.org>
11348 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
11350 2017-10-05 Tamar Christina <tamar.christina@arm.com>
11352 * config/arm/arm.c (arm_test_fpu_data): New.
11353 (arm_run_selftests): Call arm_test_fpu_data.
11355 2017-10-04 Nathan Sidwell <nathan@acm.org>
11357 * toplev.c (toplev::main): Remove excess parens on pretty_printer
11359 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
11361 2017-10-04 Sudakshina Das <sudi.das@arm.com>
11363 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
11364 check type for aarch64_simd_valid_immediate.
11365 (aarch64_output_simd_mov_immediate): Update prototype.
11366 (aarch64_simd_valid_immediate): Update prototype.
11367 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
11368 support for ORR-immediate.
11369 (and<mode>3): modified pattern to add support for BIC-immediate.
11370 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
11371 now checks for valid immediate for BIC and ORR based on new enum
11373 (aarch64_output_simd_mov_immediate): Function now used to output
11374 BIC/ORR imm as well based on new enum argument.
11375 * config/aarch64/constraints.md (Do): New vector immediate constraint.
11377 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
11378 (aarch64_reg_or_bic_imm): Likewise.
11380 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11382 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
11383 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
11385 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
11389 PR rtl-optimization/82396
11390 * haifa-sched.c (autopref_multipass_init): Simplify
11392 (autopref_rank_data): Simplify sort order.
11393 * sched-int.h (autopref_multipass_data_): Remove
11394 multi_mem_insn_p, min_offset and max_offset.
11396 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11398 * doc/sourcebuild.texi: Document vect_peeling_profitable.
11400 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11402 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
11403 vect_doubleint_cvt.
11405 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11407 * doc/sourcebuild.texi: Document vect_long_mult.
11409 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
11411 PR tree-optimization/82413
11412 * fold-const.c (build_range_check): Use widest_int when comparing
11413 the maximum ETYPE value with HIGH.
11415 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
11417 PR rtl-optimization/82396
11418 * haifa-sched.c (autopref_multipass_init): Simplify
11420 (autopref_rank_data): Simplify sort order.
11421 * sched-int.h (autopref_multipass_data_): Remove
11422 multi_mem_insn_p, min_offset and max_offset.
11424 2017-10-04 Jakub Jelinek <jakub@redhat.com>
11426 PR tree-optimization/82381
11427 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
11428 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
11431 PR tree-optimization/82374
11432 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
11433 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
11434 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
11435 current_function_decl to the new decl.
11437 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
11439 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
11440 helper macro for IEEE float128 hardware built-in functions.
11441 (SQRTF128_ODD): Add built-in functions with the round-to-odd
11443 (TRUNCF128_ODD): Likewise.
11444 (ADDF128_ODD): Likewise.
11445 (SUBF128_ODD): Likewise.
11446 (MULF128_ODD): Likewise.
11447 (DIVF128_ODD): Likewise.
11448 (FMAF128_ODD): Likewise.
11449 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
11450 UNSPEC_TRUNC_ROUND_TO_ODD.
11451 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
11452 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
11453 floating point round to odd instructions.
11454 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
11455 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
11456 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
11457 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
11458 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
11459 (trunc<mode>sf2_hw): Change the truncate with round to odd
11460 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
11461 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
11462 to odd hardware instructions.
11463 (sub<mode>3_odd): Likewise.
11464 (mul<mode>3_odd): Likewise.
11465 (div<mode>3_odd): Likewise.
11466 (sqrt<mode>2_odd): Likewise.
11467 (fma<mode>4_odd): Likewise.
11468 (fms<mode>4_odd): Likewise.
11469 (nfma<mode>4_odd): Likewise.
11470 (nfms<mode>4_odd): Likewise.
11471 (trunc<mode>df2_odd): Change the truncate with round to odd
11472 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
11474 * doc/extend.texi (PowerPC built-in functions): Update documentation
11475 for existing IEEE float128-bit built-in functions. Add built-in
11476 functions that generate the IEEE 128-bit floating point round to
11479 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
11481 PR rtl-optimization/77729
11482 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
11483 to (X&(C1&~C2))|C2 transformations.
11485 2017-10-03 Martin Jambor <mjambor@suse.cz>
11487 PR tree-optimization/82363
11488 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
11489 mismatch, mark lacc written regardless of racc.
11491 2017-10-03 Jakub Jelinek <jakub@redhat.com>
11493 PR tree-optimization/82381
11494 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
11495 stmt_to_insert nor wheather SSA_NAMEs are default defs.
11496 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
11497 fallthrough into reassoc_stmt_dominates_stmt_p.
11500 * combine.c (combine_instructions): Don't combine in unreachable
11503 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
11506 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
11507 function to not use the have_cpu variable. Do not set cpu_index,
11508 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
11509 or the default cpu.
11510 (rs6000_valid_attribute_p): Remove duplicate initializations of
11511 old_optimize and func_optimize.
11512 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
11513 (rs6000_activate_target_options): Make global.
11514 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
11517 2017-10-02 Jakub Jelinek <jakub@redhat.com>
11519 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
11520 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
11521 Set *poffset to 0 and *psize and *pmax_size to -1 if
11522 *poffset + *psize overflows in HOST_WIDE_INT.
11524 PR tree-optimization/82387
11525 PR tree-optimization/82388
11526 PR tree-optimization/82389
11527 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
11528 instead of live_bytes non-NULL.
11530 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
11533 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
11536 2017-10-02 Richard Biener <rguenther@suse.de>
11538 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
11539 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
11540 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
11542 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
11544 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
11545 requested precision matches the type's.
11546 * calls.c (alloc_max_size): Calculate the new candidate size as
11547 a widest_int and use wi::to_widest when comparing it with the
11548 current candidate size.
11549 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
11550 zero rather than integer_zero_node.
11551 * match.pd: Check for a no-op conversion before using wi::add
11552 rather than after. Use tree_to_uhwi when summing small shift
11553 counts into an unsigned int.
11555 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
11556 Alan Hayward <alan.hayward@arm.com>
11557 David Sherwood <david.sherwood@arm.com>
11560 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
11561 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
11562 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
11563 POINTER_AND_FP_REGS.
11565 2017-10-02 Richard Biener <rguenther@suse.de>
11567 PR tree-optimization/82355
11568 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11569 a mapping for the enclosing loop but avoid generating one for
11570 the loop tree root.
11571 (copy_bb_and_scalar_dependences): Remove premature codegen
11572 error on PHIs in blocks duplicated into multiple places.
11573 * graphite-scop-detection.c
11574 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11575 in the region use it as loop and nest to analyze the DR in.
11576 (try_generate_gimple_bb): Likewise.
11577 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11578 (add_loop_constraints): For blocks in a loop not in the region
11579 create a dimension with a single iteration.
11580 * sese.h (gbb_loop_at_index): Remove assert.
11582 2017-10-01 Kevin Buettner <kevinb@redhat.com>
11584 * omp-expand.c (adjust_context_scope): New function.
11585 (expand_parallel_call): Call adjust_context_scope.
11587 2017-10-01 Jeff Law <law@redhat.com>
11589 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
11590 dom_opt_dom_walker class with direct access to private members.
11591 Add comments. Call test_for_singularity.
11592 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
11593 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
11594 m_dummy_cond anymore.
11595 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
11597 (pass_dominator:execute): Build the dummy_cond here and pass it
11598 to the dom_opt_dom_walker ctor.
11599 (test_for_singularity): New function.
11601 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
11602 Maya Rashish <coypu@sdf.org>
11604 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
11605 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
11606 (alpha*-*-netbsd*) Use nbsd_tm_file.
11607 (arm*-*-netbsdelf*) Likewise.
11608 (i[34567]86-*-netbsdelf*) Likewise.
11609 (x86_64-*-netbsd*) Likewise.
11610 (mips*-*-netbsd*) Likewise.
11611 (powerpc-*-netbsd*) Likewise.
11612 (sh*-*-netbsd*) Likewise.
11613 (sparc-*-netbsdelf*) Likewise.
11614 (sparc64-*-netbsd*) Likewise.
11615 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
11617 (vax-*-netbsdelf*) Likewise.
11618 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
11619 (UINT_FAST8_TYPE) Likewise.
11620 (INT_FAST16_TYPE) Check CHAR_FAST16.
11621 (UINT_FAST16_TYPE) Likewise.
11623 2017-09-30 Jakub Jelinek <jakub@redhat.com>
11626 * config/i386/i386.md
11627 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
11628 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
11629 *divmodsi4_zext_2): New define_insn_and_split.
11630 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
11631 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
11632 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
11633 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
11634 New define_insn_and_split.
11635 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
11636 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
11637 operands[1] having DImode when mode is SImode.
11639 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
11640 always SImode for DIV and MOD in REG_EQUAL notes.
11642 2017-09-29 Yury Gribov <tetra2005@gmail.com>
11644 PR middle-end/82319
11645 * match.pd: Fix handling of NaNs in pattern.
11647 2017-09-29 Jeff Law <law@redhat.com>
11649 * sbitmap.c (bitmap_bit_in_range_p): New function.
11650 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
11651 * tree-ssa-dse.c (live_bytes_read): New function.
11652 (dse_classify_store): Ignore reads of dead bytes.
11654 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
11655 typos and whitespace errors.
11656 * config/i386/predicates.md (address_no_seg_operand): Likewise.
11657 * config/s390/s390.c (s390_emit_prologue): Likewise.
11659 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
11662 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
11663 with a symbol for LRA.
11665 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
11667 PR rtl-optimization/82338
11668 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
11670 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
11672 * genmodes.c (calc_wider_mode): Suppress qsort macro.
11673 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
11674 (qsort_chk): Declare.
11675 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
11676 (qsort_chk): New function.
11678 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11680 PR tree-optimization/82337
11681 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
11682 phi definition if the PHI result appears in an abnormal PHI.
11683 (find_basis_for_base_expr): Don't record a basis if the LHS of the
11684 basis appears in an abnormal PHI.
11686 2017-09-29 Richard Biener <rguenther@suse.de>
11688 * graphite-isl-ast-to-gimple.c
11689 (translate_isl_ast_to_gimple::set_codegen_error): New function.
11690 (binary_op_to_tree): Use it.
11691 (get_rename_from_scev): Likewise.
11692 (copy_loop_phi_nodes): Likewise.
11693 (copy_bb_and_scalar_dependences): Likewise.
11694 (translate_pending_phi_nodes): Likewise.
11696 2017-09-29 Jakub Jelinek <jakub@redhat.com>
11699 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
11700 for movabsq $(i32 << shift), r64.
11702 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
11704 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
11705 index when encoding %esp as %rsp to avoid 0x67 prefix.
11707 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
11709 * config/i386/i386.md (*movsf_internal, *movdf_internal):
11710 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
11712 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
11714 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
11715 Extensions with more than 16 double VFP registers.
11716 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
11717 to_clear_mask and all code related to it. Replace the remaining
11718 entry by a sbitmap and adapt code accordingly.
11720 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
11722 * brig-builtins.def: Change pure attributes to const.
11724 2017-09-28 Joseph Myers <joseph@codesourcery.com>
11726 * config.gcc (default_gnu_indirect_function): Default to yes for
11727 sparc*-*-linux* with glibc.
11729 2017-09-28 Joseph Myers <joseph@codesourcery.com>
11731 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
11732 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
11733 when creating .init_array and .fini_array sections with priority
11736 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
11739 * config/aarch64/aarch64.c
11740 (aarch64_builtin_support_vector_misalignment): Always return false
11741 when misalignment is unknown.
11743 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
11745 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
11746 this function to return false if the definition used by the swap
11747 instruction is artificial, or if the memory address from which the
11748 constant value is loaded is not represented by a base address held
11749 in a register or if the base address register is a frame or stack
11750 pointer. Additionally, return false if the base address of the
11751 loaded constant is a SYMBOL_REF but is not considered to be a
11753 (replace_swapped_load_constant): New function.
11754 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
11755 loaded constant vector with a load of a swapped constant vector.
11757 2017-09-27 Carl Love <cel@us.ibm.com>
11759 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
11760 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
11761 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
11764 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
11766 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
11767 first, always call autopref_rank_data otherwise.
11769 2017-09-27 Richard Biener <rguenther@suse.de>
11771 * graphite-scop-detection.c (find_scop_parameters): Move
11772 loop bound handling ...
11773 (gather_bbs::before_dom_children): ... here, avoiding the need
11774 to build scop_info->loop_nest.
11775 (record_loop_in_sese): Remove.
11776 * sese.h (sese_info_t::loop_nest): Remove.
11777 * sese.c (new_sese_info): Do not allocate loop_nest.
11778 (free_sese_info): Do not free loop_nest.
11780 2017-09-27 Jakub Jelinek <jakub@redhat.com>
11783 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
11784 lhs from calls if the lhs has addressable type.
11786 2017-09-27 Richard Biener <rguenther@suse.de>
11788 * graphite.h (scop::max_alias_set): New member.
11789 * graphite-scop-detection.c: Remove references to non-existing
11790 --param in comments.
11791 (build_alias_sets): Record the maximum alias set used for drs.
11792 (build_scops): Support zero as unlimited for
11793 --param graphite-max-arrays-per-scop.
11794 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
11795 and inline into ...
11796 (build_poly_sr_1): ... here. Compute alias set based on the
11797 maximum alias set used for drs rather than
11798 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
11800 2017-09-27 Richard Biener <rguenther@suse.de>
11802 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
11803 --param loop-block-tile-size=0 to disable tiling.
11805 2017-09-27 Richard Biener <rguenther@suse.de>
11807 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
11808 (graphite-max-nb-scop-params): Document special value zero.
11809 * domwalk.h (dom_walker::STOP): New symbolical constant.
11810 (dom_walker::dom_walker): Add optional parameter for bb to
11812 (dom_walker::~dom_walker): Declare.
11813 (dom_walker::before_dom_children): Document STOP return value.
11814 (dom_walker::m_user_bb_to_rpo): New member.
11815 (dom_walker::m_bb_to_rpo): Likewise.
11816 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
11817 mapping here if not provided by the user.
11818 (dom_walker::~dom_walker): Free bb to RPO mapping if not
11819 provided by the user.
11820 (dom_walker::STOP): Define.
11821 (dom_walker::walk): Do not compute bb to RPO mapping here.
11822 Support STOP return value from before_dom_children to stop
11824 * graphite-optimize-isl.c (optimize_isl): If the schedule
11825 is the same still generate code if -fgraphite-identity
11826 or -floop-parallelize-all are given.
11827 * graphite-scop-detection.c: Include cfganal.h.
11828 (gather_bbs::gather_bbs): Get and pass through bb to RPO
11830 (gather_bbs::before_dom_children): Return STOP for BBs
11832 (build_scops): Compute bb to RPO mapping and pass it to
11833 the domwalk. Treat --param graphite-max-nb-scop-params=0
11834 as not limiting the number of params.
11835 * graphite.c (graphite_initialize): Remove limit on the
11836 number of basic-blocks in a function.
11837 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
11838 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
11839 default value of 10.
11841 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11843 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
11844 Adjust code to eliminate needing to do the shift right 32-bits
11845 operation after XSCVDPSPN.
11847 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11849 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
11850 ((X / Y) != 0 -> X >= Y): Likewise.
11852 2017-09-26 Carl Love <cel@us.ibm.com>
11854 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
11855 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
11856 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
11857 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
11858 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
11859 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
11860 definitions and overloading.
11861 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
11862 statement for P9V_BUILTIN_XST_LEN_R.
11863 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
11864 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
11865 define_expand and define_insn for the instructions and builtins.
11866 * doc/extend.texi: Update the built-in documentation file for the new
11867 built-in functions.
11868 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
11869 define_insn for the instructions
11871 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
11874 * gcc/config/netbsd-protos.h: New file.
11875 * gcc/config/netbsd.c: New file.
11876 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
11877 * gcc/config/t-netbsd: New file.
11878 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
11879 (tmake_file) Add t-netbsd.
11880 (extra_objs) Add netbsd.o.
11882 2017-09-26 Janus Weil <janus@gcc.gnu.org>
11886 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
11888 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11890 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
11891 sign extension from a vector register to a GPR by doing a 32-bit
11892 direct move and then an EXTSW.
11893 (extendsi<mode>2 splitter): Likewise.
11894 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
11895 right or vector extract after doing XSCVDPSPN. Use
11896 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
11898 (movdi_from_sf_zero_ext): Likewise.
11899 (reload_gpr_from_vsxsf): Likewise.
11900 (p8_mfvsrd_4_disf): Delete, no longer used.
11901 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
11902 then needing to move the SFmode to a GPR to use the XSCVDPSP
11903 instruction instead of FRSP and XSCVDPSPN.
11904 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
11905 it is adjacent to the other XSCVSPDP insns.
11906 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
11907 SFmode to be in traditional Altivec registers.
11908 (vsx_xscvdpspn): Eliminate useless alternative constraint.
11909 (vsx_xscvspdpn): Likewise.
11910 (vsx_xscvspdpn_scalar): Likewise.
11912 2017-09-26 Martin Jambor <mjambor@suse.cz>
11914 * tree-sra.c (compare_access_positions): Put integral types first,
11915 stabilize sorting of integral types, remove conditions putting
11916 non-full-precision integers last.
11917 (sort_and_splice_var_accesses): Disable scalarization if a
11918 non-integert would be represented by a non-full-precision integer.
11920 2017-09-26 Joseph Myers <joseph@codesourcery.com>
11922 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
11923 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
11924 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
11925 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
11926 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
11927 conditionals inside the function instead of around it. Call
11928 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
11929 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
11931 2017-09-26 Richard Biener <rguenther@suse.de>
11933 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
11935 (scop_detection::build_scop_breadth): ... this. Removed.
11936 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
11937 (scop_detection::harmful_stmt_in_bb): Likewise.
11938 (scop_detection::graphite_can_represent_stmt): Likewise.
11939 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
11940 (scop_detection::can_represent_loop): Remove recursion, fold in ...
11941 (scop_detection::can_represent_loop_1): ... this. Removed.
11942 (scop_detection::harmful_loop_in_region): Simplify after inlining
11943 the above and remove more quadraticness.
11944 (build_scops): Adjust.
11945 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
11948 2017-09-26 Jakub Jelinek <jakub@redhat.com>
11951 * config/i386/i386.c (ix86_print_operand_address_as): Only test
11952 REGNO (base) == SP_REG if base is a REG.
11954 PR middle-end/35691
11955 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
11956 if it is different SSA_NAME.
11957 (optimize_range_tests_cmp_bitwise): New function.
11958 (optimize_range_tests): Call it.
11960 2017-09-26 Richard Biener <rguenther@suse.de>
11962 PR tree-optimization/82321
11963 * graphite.c (canonicalize_loop_closed_ssa): Properly check
11964 for the def being inside the loop.
11966 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11968 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
11970 * config/s390/s390-builtins.def: Fix constraint on op4.
11972 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11974 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
11975 independent expanders.
11976 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
11977 ("vec_ordered", "vec_unordered"): New expanders.
11979 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11981 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
11984 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11986 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
11987 vec_unpacks_lo_v16qi.
11988 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
11990 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11992 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
11993 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
11994 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
11996 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11998 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
12000 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
12002 ("*vec_slb<mode>"): New insn pattern.
12003 ("vec_shr_<mode>"): New expander.
12004 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
12005 and force the shift count operand to V16QImode.
12006 ("vec_srb<mode>"): Set shift count mode to V16QI.
12008 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12010 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
12011 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
12012 ("vec_widen_smult_hi_<mode>"): New expander definitions.
12014 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
12017 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
12019 2017-09-26 Richard Biener <rguenther@suse.de>
12021 PR tree-optimization/82320
12022 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
12025 2017-09-25 Jeff Law <law@redhat.com>
12027 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
12028 prototype for new argument.
12029 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
12030 mostly extracted from rs6000_emit_allocate_stack.
12031 (rs6000_emit_probe_stack_range_stack_clash): New function.
12032 (rs6000_emit_allocate_stack): Call
12033 rs6000_emit_probe_stack_range_stack_clash as needed.
12034 (rs6000_emit_probe_stack_range): Add additional argument
12035 to call to gen_probe_stack_range{si,di}.
12036 (output_probe_stack_range): New.
12037 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
12038 (output_probe_stack_range_stack_clash): New.
12039 (rs6000_emit_prologue): Emit notes into dump file as requested.
12040 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
12041 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
12042 Add additional operand and pass it to output_probe_stack_range.
12044 2017-09-25 Bin Cheng <bin.cheng@arm.com>
12046 PR tree-optimization/82163
12047 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
12048 (checking_verify_loop_closed_ssa): New parameter.
12049 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
12050 (check_loop_closed_ssa_stmt): Delete.
12051 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
12052 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
12053 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
12056 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
12058 * brig-builtins.def: Treat HSAIL barrier builtins as
12059 setjmp/longjump style functions.
12061 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
12063 * target.def (constant_alignment): New hook.
12064 * defaults.h (CONSTANT_ALIGNMENT): Delete.
12065 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
12066 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
12067 * doc/tm.texi: Regenerate.
12068 * targhooks.h (default_constant_alignment): Declare.
12069 (constant_alignment_word_strings): Likewise.
12070 * targhooks.c (default_constant_alignment): New function.
12071 (constant_alignment_word_strings): Likewise.
12072 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
12073 instead of CONSTANT_ALIGNMENT.
12074 * varasm.c (align_variable, get_variable_align, build_constant_desc)
12075 (force_const_mem): Likewise.
12076 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
12077 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
12078 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
12079 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12080 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
12082 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
12083 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12084 constant_alignment_word_strings.
12085 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
12086 (CONSTANT_ALIGNMENT): Likewise.
12087 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12088 (arm_constant_alignment): New function.
12089 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
12090 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12091 constant_alignment_word_strings.
12092 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
12093 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12094 constant_alignment_word_strings.
12095 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
12096 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12097 (cris_constant_alignment): New function.
12098 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
12099 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12100 (epiphany_constant_alignment): New function.
12101 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
12102 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12103 constant_alignment_word_strings.
12104 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
12105 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12106 constant_alignment_word_strings.
12107 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
12108 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12109 constant_alignment_word_strings.
12110 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
12111 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
12112 * config/i386/i386.c (ix86_constant_alignment): Make static.
12113 Use the same interface as the target hook.
12114 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12115 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
12116 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12117 constant_alignment_word_strings.
12118 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
12119 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
12120 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12121 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
12122 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12123 constant_alignment_word_strings.
12124 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
12125 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12126 constant_alignment_word_strings.
12127 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
12128 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12129 constant_alignment_word_strings.
12130 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
12131 * config/microblaze/microblaze.c (microblaze_constant_alignment):
12133 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12134 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
12135 * config/mips/mips.c (mips_constant_alignment): New function.
12136 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12137 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
12138 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
12139 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12140 (mmix_constant_alignment): Make static. Use the same interface
12141 as the target hook.
12142 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
12143 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12144 constant_alignment_word_strings.
12145 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
12146 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12147 constant_alignment_word_strings.
12148 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
12149 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12150 constant_alignment_word_strings.
12151 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
12152 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12153 (rs6000_constant_alignment): New function.
12154 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
12155 * config/riscv/riscv.c (riscv_constant_alignment): New function.
12156 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12157 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
12158 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12159 (rs6000_constant_alignment): New function.
12160 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
12161 * config/s390/s390.c (s390_constant_alignment): New function.
12162 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12163 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
12164 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12165 constant_alignment_word_strings.
12166 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
12167 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12168 (sparc_constant_alignment): New function.
12169 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
12170 * config/spu/spu.c (spu_constant_alignment): New function.
12171 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12172 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
12173 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12174 constant_alignment_word_strings.
12175 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
12176 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12177 constant_alignment_word_strings.
12178 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
12179 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12180 constant_alignment_word_strings.
12181 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
12182 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12183 (visium_constant_alignment): New function.
12184 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
12185 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12186 (xtensa_constant_alignment): New function.
12187 * system.h (CONSTANT_ALIGNMENT): Poison.
12189 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
12191 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12192 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
12193 (rs6000_builtin_valid_without_lhs): New helper function.
12194 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12195 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
12197 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
12199 * target.h (vec_perm_indices): Use unsigned short rather than
12201 (auto_vec_perm_indices): Likewise.
12202 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
12203 Use unsigned int rather than unsigned char.
12204 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
12206 2017-09-25 Richard Biener <rguenther@suse.de>
12208 * cfgloop.h (sort_sibling_loops): Declare.
12209 * cfgloop.c (sort_sibling_loops_cmp): New helper.
12210 (sort_sibling_loops): New function sorting the sibling loop list
12212 * graphite.c (graphite_transform_loops): Sort sibling loops.
12214 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
12216 * target.def (vec_perm_const_ok): Change sel parameter to
12218 * optabs-query.c (can_vec_perm_p): Update accordingly.
12219 * doc/tm.texi: Regenerate.
12220 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
12221 auto_vec_perm_indices and remove separate nelt field.
12222 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
12223 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
12224 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
12225 (aarch64_expand_vec_perm_const): Update accordingly.
12226 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
12227 to vec_perm_indices.
12228 * config/arm/arm.c (expand_vec_perm_d): Change perm to
12229 auto_vec_perm_indices and remove separate nelt field.
12230 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
12231 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
12232 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
12234 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
12235 to vec_perm_indices.
12236 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
12237 sel to vec_perm_indices.
12238 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
12239 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
12240 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
12242 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
12245 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
12248 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
12249 on the FUNCTION_DECL function context if it has a DIE that is a
12252 2017-09-25 Richard Biener <rguenther@suse.de>
12254 PR tree-optimization/82285
12255 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
12258 2017-09-25 Tom de Vries <tom@codesourcery.com>
12262 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
12265 2017-09-25 Richard Biener <rguenther@suse.de>
12267 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
12268 ISL errors other than isl_error_quota happen. Dump if the
12269 schedule is the same.
12270 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
12271 errors instead of aborting inside ISL.
12273 2017-09-25 Iain Sandoe <iain@codesourcery.com>
12276 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
12277 of libgcc_eh for m64.
12278 * config/i386/darwin64.h: Likewise.
12280 2017-09-25 Richard Biener <rguenther@suse.de>
12282 PR middle-end/82144
12283 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
12284 attribute for incomplete types nor twice for complete ones.
12286 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
12289 * config/i386/i386.c (ix86_print_operand_address_as): Encode
12290 %esp as %rsp to avoid 0x67 prefix if there is no index or base
12293 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
12296 * config/i386/i386.opt (mprefer-avx256): Use
12297 ix86_target_flags variable.
12298 * config/i386/i386.c (ix86_target_string): Move
12299 -mprefer-avx256 to flag2_opts.
12301 2017-09-22 Jakub Jelinek <jakub@redhat.com>
12303 PR middle-end/35691
12304 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
12305 and x != -1 | y != -1 into (x & y) != -1.
12307 2017-09-22 Steve Ellcey <sellcey@cavium.com>
12309 * config.gcc: Add new case statement to set
12310 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
12311 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
12312 s390x-*-linux* case statements. Added aarch64 to the list of
12313 supported architectures.
12315 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12317 PR tree-optimization/82289
12318 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
12319 STMT_VINFO_RELEVANT_P.
12321 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12322 Alan Hayward <alan.hayward@arm.com>
12323 David Sherwood <david.sherwood@arm.com>
12325 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
12326 for VR_RANGE only; don't allow VR_ANTI_RANGE.
12327 (extract_range_from_binary_expr_1): Don't call
12328 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
12330 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12331 Alan Hayward <alan.hayward@arm.com>
12332 David Sherwood <david.sherwood@arm.com>
12334 * target.def (preferred_vector_alignment): New hook.
12335 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
12337 * doc/tm.texi: Regenerate.
12338 * targhooks.h (default_preferred_vector_alignment): Declare.
12339 * targhooks.c (default_preferred_vector_alignment): New function.
12340 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
12342 (DR_TARGET_ALIGNMENT): New macro.
12343 (aligned_access_p): Update commentary.
12344 (vect_known_alignment_in_bytes): New function.
12345 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
12347 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
12348 Calculate the misalignment based on the target alignment rather than
12350 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
12351 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
12352 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
12353 the target alignment, rather than masking the element misalignment
12354 with the number of elements in a vector. Also use the target
12355 alignment when calculating the maximum number of peels.
12356 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
12357 instead of TYPE_ALIGN_UNIT.
12358 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
12359 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
12360 (vect_create_addr_base_for_vector_ref): Update call accordingly.
12361 (vect_create_data_ref_ptr): Likewise.
12362 (vect_setup_realignment): Realign by ANDing with
12363 -DR_TARGET_MISALIGNMENT.
12364 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
12365 the number of peels based on DR_TARGET_ALIGNMENT.
12366 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
12367 with the guaranteed alignment boundary when deciding whether
12369 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
12370 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
12371 (ensure_base_align): Remove stmt_info parameter. Get the
12372 target base alignment from DR_TARGET_ALIGNMENT.
12373 (vectorizable_store): Update call accordingly. Interpret
12374 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
12376 (vectorizable_load): Likewise.
12378 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12379 Alan Hayward <alan.hayward@arm.com>
12380 David Sherwood <david.sherwood@arm.com>
12382 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
12383 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
12384 (vect_enhance_data_refs_alignment): Likewise.
12386 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
12388 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
12389 error. Only quit immediately if parsing is complete.
12390 (BEGIN): Initialize fatal_err and parse_done.
12391 (begin fpu, end fpu): Check number of arguments.
12392 (begin arch, end arch): Likewise.
12393 (begin cpu, end cpu): Likewise.
12394 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
12395 (optalias): Likewise.
12397 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
12399 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
12400 * config/arm/arm-isa.h: Delete. Move definitions to ...
12401 * arm-cpus.in: ... here. Use new feature and fgroup values.
12402 * config/arm/arm.c (arm_option_override): Use lower case for feature
12404 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
12405 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
12406 * config/arm/parsecpu.awk (END): Add new command 'isa'.
12408 (print_isa_bits_for): New function.
12409 (gen_isa): New function.
12410 (gen_comm_data): Use print_isa_bits_for.
12411 (define feature): New keyword.
12412 (define fgroup): New keyword.
12413 * config/arm/t-arm (TM_H): Remove.
12414 (GTM_H): Add arm-isa.h.
12415 (arm-isa.h): Add rule to generate file.
12416 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
12417 case for feature bit names.
12419 2017-09-22 Richard Biener <rguenther@suse.de>
12421 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
12423 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
12424 print of no dependency loops ...
12425 * graphite.c (graphite_transform_loops): ... here.
12426 (canonicalize_loop_closed_ssa_form): Work from inner to outer
12428 (same_close_phi_node, remove_duplicate_close_phi,
12429 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
12430 (canonicalize_loop_closed_ssa): ... here and simplify.
12431 * graphite-optimize-isl.c: Include tree-vectorizer.h.
12432 (optimize_isl): Use dump_printf_loc to tell when we stopped
12433 optimizing because of an ISL timeout.
12435 2017-09-22 Richard Biener <rguenther@suse.de>
12437 PR tree-optimization/82291
12438 * tree-if-conv.c (predicate_mem_writes): Make sure to
12439 remove writes in blocks predicated with false.
12441 2017-09-22 Richard Biener <rguenther@suse.de>
12443 * sese.c: Include cfganal.h.
12444 (if_region_set_false_region): Remove.
12445 (create_if_region_on_edge): Likewise.
12446 (move_sese_in_condition): Re-implement without destroying
12449 2017-09-22 Richard Biener <rguenther@suse.de>
12451 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
12452 Verify both BBs contain loop PHI nodes before dispatching to
12453 copy_loop_phi_args.
12454 (graphite_regenerate_ast_isl): Do not recompute dominators,
12455 do not verify three times. Restructure for clarity.
12456 * graphite-scop-detection.c (same_close_phi_node,
12457 remove_duplicate_close_phi, make_close_phi_nodes_unique,
12458 defined_in_loop_p, canonicalize_loop_closed_ssa,
12459 canonicalize_loop_closed_ssa_form): Simplify, remove excess
12460 checking and SSA rewrite, move to ...
12461 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
12462 (graphite_initialize): Do not pass in ctx, do not reset the
12463 SCEV cache, compute only dominators.
12464 (graphite_transform_loops): Allocate ISL ctx after
12465 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
12466 Maintain post-dominators only around build_scops.
12467 * sese.c (if_region_set_false_region): Make static. Free
12468 and recompute dominators.
12469 (move_sese_in_condition): Assert we don't get called with
12470 post-dominators computed.
12471 * sese.h (if_region_set_false_region): Remove.
12473 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
12475 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
12476 mode attribute for TARGET_AVX512VL.
12478 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
12480 * config/i386/i386.opt (mprefer-avx256): New option.
12481 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
12483 (ix86_preferred_simd_mode): Return 256-bit AVX modes
12484 for TARGET_PREFER_AVX256.
12485 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
12487 2017-09-21 Jeff Law <law@redhat.com>
12489 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
12490 Fix dump output if the only stack space is for pushed registers.
12492 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12494 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
12497 2017-09-21 Martin Sebor <msebor@redhat.com>
12500 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
12501 code (in C++) or code that triggers warnings.
12503 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
12505 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
12507 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
12509 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
12510 * sched-rgn.c: Ditto.
12511 * sel-sched-ir.c: Ditto.
12513 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
12515 * toplev.h (set_random_seed): Adjust return type.
12516 * toplev.c (init_local_tick): Move eager initialization of random_seed
12517 to get_random_seed. Adjust comment.
12518 (init_random_seed): Inline to get_random_seed, delete.
12519 (get_random_seed): Initialize random_seed lazily.
12520 (set_random_seed): Do not return previous value.
12521 (print_switch_value): Do not call get_random_seed.
12523 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
12525 * cgraph.c (delete_function_version): New, broken out from...
12526 (cgraph_node::delete_function_version): ...here. Rename to
12527 cgraph_node::delete_function_version_by_decl. Update all uses.
12528 (cgraph_node::remove): Call delete_function_version.
12530 2017-09-21 Jakub Jelinek <jakub@redhat.com>
12533 * tree-inline.c (expand_call_inline): Emit clobber stmts for
12534 VAR_DECLs to which addressable non-volatile parameters are mapped
12535 and for id->retvar after the return value assignment. Clear
12536 id->retval and id->retbnd after inlining.
12538 2017-09-21 Richard Biener <rguenther@suse.de>
12540 PR tree-optimization/82276
12541 PR tree-optimization/82244
12542 * tree-vrp.c (build_assert_expr_for): Set
12543 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
12545 (remove_range_assertions): Revert earlier change.
12547 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
12550 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
12552 2017-09-21 Richard Biener <rguenther@suse.de>
12554 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12555 Restore valid IL after code generation errors.
12556 * graphite.c (graphite_transform_loops): Diagnose code
12557 generation issues as MSG_MISSED_OPTIMIZATION and continue
12558 with processing SCOPs.
12560 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12561 Alan Hayward <alan.hayward@arm.com>
12562 David Sherwood <david.sherwood@arm.com>
12564 * calls.c (compute_argument_addresses): Use simplify_gen_binary
12565 rather than choosing between plus_constant and gen_rtx_<CODE>.
12566 * expr.c (emit_push_insn): Likewise.
12567 (expand_expr_real_2): Likewise.
12569 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12570 Alan Hayward <alan.hayward@arm.com>
12571 David Sherwood <david.sherwood@arm.com>
12573 * loop-unroll.c (split_iv): Call copy_rtx on the step.
12575 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12576 Alan Hayward <alan.hayward@arm.com>
12577 David Sherwood <david.sherwood@arm.com>
12579 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
12580 calling tree_to_uhwi.
12582 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12583 Alan Hayward <alan.hayward@arm.com>
12584 David Sherwood <david.sherwood@arm.com>
12586 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
12587 INTEGER_CST rather than a negative test for ADDR_EXPR.
12589 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12590 Alan Hayward <alan.hayward@arm.com>
12591 David Sherwood <david.sherwood@arm.com>
12593 * tree-vrp.c (extract_range_from_binary_expr_1): Check
12594 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
12596 2017-09-21 Richard Biener <rguenther@suse.de>
12598 PR tree-optimization/71351
12599 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
12600 graphite_create_new_loop_guard): Remove, fold remaining parts
12602 (translate_isl_ast_node_for): ... here and simplify.
12604 2017-09-21 Jakub Jelinek <jakub@redhat.com>
12607 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
12608 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
12609 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
12610 alternative always use QI mode, for -Os imov (=R,R) alternative
12611 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
12614 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12615 Jeff Law <law@redhat.com>
12617 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
12618 (allocate_stack_space): New function, partially extracted from
12619 s390_emit_prologue.
12620 (s390_emit_prologue): Track offset to most recent stack probe.
12621 Code to allocate space moved into allocate_stack_space.
12622 Dump actions when no stack is allocated.
12623 (s390_prologue_plus_offset): New function.
12624 (s390_emit_stack_probe): Likewise.
12626 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
12628 * common.opt (Wa, Wl, Wp, g, gz=): Add
12630 (gno-column-info): Remove.
12631 (gcolumn-info): Drop RejectNegative.
12632 (gno-): New prefix.
12633 (gno-record-gcc-switches): Remove.
12634 (grecord-gcc-switches): Drop RejectNegative.
12635 (gno-split-dwarf): Remove.
12636 (gsplit-dwarf): Drop RejectNegative.
12637 (gno-strict-dwarf): Remove.
12638 (gstrict-dwarf): Drop RejectNegative.
12639 * config/darwin.opt (gfull, gused): Add RejectNegative.
12640 * dwarf2out.c (gen_producer_string): Drop
12641 gno-record-gcc-switches handler.
12642 * optc-gen.awk: Add g to prefixes with negative forms.
12643 * opts-common.c (remapping_prefix_p): New.
12644 (find_opt): Check it.
12645 (generate_canonical_option): Test g prefix.
12646 (option_map): Add -gno- mapping.
12647 (add_misspelling_candidates): Check remapping_prefix_p.
12649 2017-09-20 Jeff Law <law@redhat.com>
12651 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
12652 thinko in stack clash protection support.
12654 * explow.c (compute_stack_clash_protection_loop_data): Use
12655 CONST_INT_P instead of explicit test. Verify object is a
12656 CONST_INT_P before looking at INTVAL.
12657 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
12658 instead of explicit test.
12660 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
12663 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
12664 address instead of to r1 and r11.
12666 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
12668 * config.gcc: Support "knm".
12669 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
12670 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12672 * config/i386/i386.c (m_KNM): Define.
12673 (processor_target_table): Add "knm".
12675 (ix86_option_override_internal): Add "knm".
12676 (ix86_issue_rate): Add PROCESSOR_KNM.
12677 (ix86_adjust_cost): Ditto.
12678 (ia32_multipass_dfa_lookahead): Ditto.
12679 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
12680 (fold_builtin_cpu): Add M_INTEL_KNM.
12681 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
12682 (processor_type): Add PROCESSOR_KNM.
12683 * config/i386/x86-tune.def: Add m_KNM.
12684 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
12686 2017-09-20 Richard Biener <rguenther@suse.de>
12688 PR tree-optimization/80213
12689 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
12690 are allowed in empty BBs as well.
12691 (canonicalize_loop_closed_ssa): Also look for other complex
12693 (scop_detection::get_sese): Include the loop-closed PHI block
12695 (scop_detection::merge_sese): Remove code adding extra blocks.
12696 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
12697 (build_scops): Assert the final returned scop is invalid.
12699 2017-09-20 Richard Biener <rguenther@suse.de>
12701 PR tree-optimization/82264
12702 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
12704 (vn_phi_lookup): Likewise.
12705 (vn_phi_insert): Likewise.
12707 2017-09-20 Jakub Jelinek <jakub@redhat.com>
12709 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
12710 that fits into uhwi or shwi, add DW_AT_const_value regardless
12711 of early_dwarf without going through RTL, using add_AT_unsigned
12714 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
12715 (DEBUG_LTO_ABBREV_SECTION): Likewise.
12716 (DEBUG_LTO_MACINFO_SECTION): Likewise.
12717 (DEBUG_MACRO_SECTION): Likewise.
12718 (DEBUG_LTO_MACRO_SECTION): Likewise.
12719 (DEBUG_STR_DWO_SECTION): Likewise.
12720 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
12721 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
12722 (DEBUG_LTO_DWO_LINE_SECTION): Define.
12723 (DEBUG_LTO_LINE_STR_SECTION): Define.
12724 (init_sections_and_labels): Initialize debug_line_str_section
12725 variable. Initialize debug_loc_section for -gdwarf-5 to
12726 DEBUG_LOCLISTS_SECTION. Formatting fixes.
12728 2017-09-20 Richard Biener <rguenther@suse.de>
12730 * graphite-sese-to-poly.c (extract_affine): Properly handle
12731 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
12733 2017-09-20 Richard Biener <rguenther@suse.de>
12735 PR tree-optimization/81373
12736 * graphite-scop-detection.c (build_cross_bb_scalars_def):
12737 Force SESE live-out defs to be handled even if they are
12740 2017-09-19 Jeff Law <law@redhat.com>
12742 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
12743 nothing for stack adjustments with REG_STACK_CHECK.
12744 * sched-deps.c (parse_add_or_inc): Reject insns with
12745 REG_STACK_CHECK from dependency breaking.
12746 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
12747 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
12748 * reg-notes.def (STACK_CHECK): New note.
12750 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
12751 (ix86_expand_prologue): Dump stack clash info as needed.
12752 Call ix86_adjust_stack_and_probe_stack_clash as needed.
12754 * function.c (dump_stack_clash_frame_info): New function.
12755 * function.h (dump_stack_clash_frame_info): Prototype.
12756 (enum stack_clash_probes): New enum.
12758 * config/alpha/alpha.c (alpha_expand_prologue): Also check
12759 flag_stack_clash_protection.
12760 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
12761 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
12762 (arm_frame_pointer_required): Likewise.
12763 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
12764 (ia64_expand_prologue): Likewise.
12765 * config/mips/mips.c (mips_expand_prologue): Likewise.
12766 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
12767 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
12768 (sparc_flat_expand_prologue): Likewise.
12769 * config/spu/spu.c (spu_expand_prologue): Likewise.
12771 * explow.c: Include "params.h".
12772 (anti_adjust_stack_and_probe_stack_clash): New function.
12773 (get_stack_check_protect): Likewise.
12774 (compute_stack_clash_protection_loop_data): Likewise.
12775 (emit_stack_clash_protection_loop_start): Likewise.
12776 (emit_stack_clash_protection_loop_end): Likewise.
12777 (allocate_dynamic_stack_space): Use get_stack_check_protect.
12778 Use anti_adjust_stack_and_probe_stack_clash.
12779 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
12780 (emit_stack_clash_protection_loop_start): Likewise.
12781 (emit_stack_clash_protection_loop_end): Likewise.
12782 * rtl.h (get_stack_check_protect): Prototype.
12783 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
12784 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
12785 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
12787 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
12789 * doc/tm.texi: Rebuilt.
12790 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
12791 get_stack_check_protect.
12792 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
12793 * config/arm/arm.c (arm_expand_prologue): Likewise.
12794 (arm_frame_pointer_required): Likewise.
12795 * config/i386/i386.c (ix86_expand_prologue): Likewise.
12796 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
12797 * config/mips/mips.c (mips_expand_prologue): Likewise.
12798 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
12799 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
12800 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
12801 (sparc_flat_expand_prologue): Likewise.
12803 * common.opt (-fstack-clash-protection): New option.
12804 * flag-types.h (enum stack_check_type): Note difference between
12805 -fstack-check= and -fstack-clash-protection.
12806 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
12807 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
12808 * toplev.c (process_options): Issue warnings/errors for cases
12809 not handled with -fstack-clash-protection.
12810 * doc/invoke.texi (-fstack-clash-protection): Document new option.
12811 (-fstack-check): Note additional problem with -fstack-check=generic.
12812 Note that -fstack-check is primarily for Ada and refer users
12813 to -fstack-clash-protection for stack-clash-protection.
12814 Document new params for stack clash protection.
12816 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
12818 * config/i386/i386.c (ix86_split_long_move): Do not handle
12819 address used for LEA in a special way.
12821 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
12823 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
12825 2017-09-19 Martin Sebor <msebor@redhat.com>
12828 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
12829 of incompatible types.
12831 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
12833 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12834 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
12835 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12836 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
12838 2017-09-19 Richard Biener <rguenther@suse.de>
12840 PR tree-optimization/82244
12841 * tree-vrp.c (remove_range_assertions): Do not propagate
12842 a constant to abnormals but replace the assert with a copy.
12844 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
12846 PR rtl-optimization/57878
12847 PR rtl-optimization/68988
12848 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
12849 avoidance test involving non_reload_pseudos. Move frequency test
12850 below the general fragmentation avoidance test.
12852 2017-09-19 Richard Biener <rguenther@suse.de>
12854 PR tree-optimization/69728
12855 * graphite-sese-to-poly.c (schedule_error): New global.
12856 (add_loop_schedule): Handle empty domain by failing the
12858 (build_original_schedule): Handle schedule_error.
12860 2017-09-19 Richard Biener <rguenther@suse.de>
12862 * graphite-scop-detection.c (scop_detection::can_represent_loop):
12863 Do not iterate to sibling loops but only to siblings of inner
12866 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
12869 * config/m68k/m68k.md (moveq feeding equality comparison): Check
12870 that the registers are different.
12872 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
12874 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
12875 to processor_model and "amdfam17h" to arch_names_table.
12876 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
12878 2017-09-18 Jakub Jelinek <jakub@redhat.com>
12881 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
12883 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12884 Alan Hayward <alan.hayward@arm.com>
12885 David Sherwood <david.sherwood@arm.com>
12887 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
12889 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
12891 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
12892 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
12893 vect_schedule_slp_instance.
12894 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
12895 Update call to vect_slp_analyze_node_operations. Simplify return
12897 (vect_slp_analyze_bb_1): Update call accordingly.
12898 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
12899 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
12900 (vect_schedule_slp): Update call accordingly.
12902 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12903 Alan Hayward <alan.hayward@arm.com>
12904 David Sherwood <david.sherwood@arm.com>
12906 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
12907 with types that aren't in fact scalar.
12909 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12911 * tree-vect-slp.c (vect_record_max_nunits): New function,
12913 (vect_build_slp_tree_1): ...here.
12914 (vect_build_slp_tree_2): Call it for phis too.
12916 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12918 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
12919 to vect_get_vec_def_for_operand when getting the mask operand.
12921 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12922 Alan Hayward <alan.hayward@arm.com>
12923 David Sherwood <david.sherwood@arm.com>
12925 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
12928 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12929 Alan Hayward <alan.hayward@arm.com>
12930 David Sherwood <david.sherwood@arm.com>
12932 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
12933 calculation for vector booleans.
12935 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12936 Alan Hayward <alan.hayward@arm.com>
12937 David Sherwood <david.sherwood@arm.com>
12939 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
12941 (vect_transform_stmt): ...here.
12942 (vect_analyze_stmt): Use it instead of calling
12943 vectorizable_live_operation directly.
12945 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
12947 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
12948 non-SIMT targets in acc vector loops.
12950 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
12952 * configure.ac: Add arc and check if assembler supports gdwarf2.
12953 * configure: Regenerate.
12955 2017-09-18 Richard Biener <rguenther@suse.de>
12957 PR tree-optimization/82220
12958 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
12959 epilogue niters from the min_profitable_iters compute.
12961 2017-09-18 Jakub Jelinek <jakub@redhat.com>
12964 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
12965 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
12966 (ix86_init_pic_reg): Revert 2017-09-01 changes.
12968 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
12971 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
12972 switching to a new text section.
12974 2017-09-18 Richard Biener <rguenther@suse.de>
12976 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
12978 (build_alias_set): Reject aliases with no access function.
12980 2017-09-18 Richard Biener <rguenther@suse.de>
12982 PR tree-optimization/79622
12983 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
12985 (build_cross_bb_scalars_use): Likewise.
12987 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
12989 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
12991 2017-09-18 Alan Modra <amodra@gmail.com>
12994 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
12995 stack_pointer_rtx for count 0. Update comments. Break up
12996 large rtl expression.
12998 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
13000 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
13001 Increase to 20 bytes.
13002 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
13003 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
13004 or avx version of the stub.
13006 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
13009 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
13010 compute the minimum stack alignment. Also update preferred stack
13011 boundary for leaf functions.
13013 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
13015 PR tree-optimization/82228
13016 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
13019 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13021 * common/config/nds32/nds32-common.c
13022 (nds32_option_optimization_table): Refine formatting.
13023 (nds32_option_optimization_table): Use -fsched-pressure and
13024 -fomit-frame-pointer for specific optimization level.
13026 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13028 * config/nds32/nds32.c: Refine formatting and comments.
13029 * config/nds32/nds32.h: Likewise.
13030 * config/nds32/nds32.md: Likewise.
13031 * config/nds32/nds32-cost.c: Likewise.
13032 * config/nds32/nds32-isr.c: Likewise.
13033 * config/nds32/nds32-md-auxiliary.c: Likewise.
13034 * config/nds32/nds32-multiple.md: Likewise.
13035 * config/nds32/nds32-predicates.c: Likewise.
13037 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
13038 Jakub Jelinek <jakub@redhat.com>
13040 Add support for -std=c++2a.
13041 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
13043 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
13045 2017-09-15 Steve Ellcey <sellcey@cavium.com>
13048 * doc/extend.texi (Common Function Attributes): Add
13049 references to ARM, AArch64, and S/390 specific attributes.
13050 (Function Specific Option Pragmas): Add AArch64 and S/390
13051 to list of back ends that support the target pragma.
13053 2017-09-15 Nathan Sidwell <nathan@acm.org>
13055 * doc/standards.texi: Fix C++17 description. Update URLs for
13058 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
13060 * common.opt (Wcast-align=strict): New warning option.
13061 * doc/invoke.texi: Document -Wcast-align=strict.
13063 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
13065 * cgraph.h (cgraph_thunk_info): Add comments.
13066 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
13067 assert for VIRTUAL_* arguments stricter.
13069 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
13071 PR tree-optimization/71026
13072 * match.pd: Move RDIV patterns from fold-const.c
13073 * fold-const.c (distribute_real_division): Removed.
13074 (fold_binary_loc): Remove calls to distribute_real_divison.
13076 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13078 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
13079 c++1z and gnu++1z as deprecated. Change other references to
13080 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
13081 Change -Wc++1z-compat to -Wc++17-compat.
13082 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
13083 * dwarf2out.c (highest_c_language): Handle C++17.
13084 (gen_compile_unit_die): Likewise.
13086 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13088 PR rtl-optimization/82192
13089 * combine.c (make_extraction): Don't look through non-paradoxical
13090 SUBREGs or TRUNCATE if pos + len is or might be bigger than
13093 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13094 Alan Hayward <alan.hayward@arm.com>
13095 David Sherwood <david.sherwood@arm.com>
13097 * target.def (function_arg_offset): New hook.
13098 * targhooks.h (default_function_arg_offset): Declare.
13099 * targhooks.c (default_function_arg_offset): New function.
13100 * function.c (locate_and_pad_parm): Use
13101 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
13102 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
13103 (TARGET_FUNCTION_ARG_OFFSET): ...this.
13104 * doc/tm.texi: Regenerate.
13105 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
13106 * config/spu/spu.c (spu_function_arg_offset): New function.
13107 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
13108 * system.h (FUNCTION_ARG_OFFSET): Poison.
13110 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13111 Alan Hayard <alan.hayward@arm.com>
13112 David Sherwood <david.sherwood@arm.com>
13114 * target.def (truly_noop_truncation): New hook.
13115 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
13116 than TRULY_NOOP_TRUNCATION.
13117 * hooks.h (hook_bool_uint_uint_true): Declare.
13118 * hooks.c (hook_bool_uint_uint_true): New function.
13119 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
13120 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
13121 * doc/tm.texi: Regenerate.
13122 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
13123 rather than TRULY_NOOP_TRUNCATION in comments.
13124 (simplify_comparison): Likewise.
13125 (record_truncated_value): Likewise.
13126 * expmed.c (extract_bit_field_1): Likewise.
13127 (extract_split_bit_field): Likewise.
13128 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
13129 instead of TRULY_NOOP_TRUNCATION.
13130 * function.c (assign_parm_setup_block): Likewise.
13131 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
13132 * rtlhooks.c: Include target.h.
13133 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
13134 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
13135 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
13136 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
13137 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
13138 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
13139 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
13140 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
13141 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
13142 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
13143 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
13144 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
13145 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
13146 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
13147 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
13148 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
13149 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
13150 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
13151 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
13152 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
13153 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
13154 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
13155 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
13156 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
13157 * config/mips/mips.c (mips_truly_noop_truncation): New function.
13158 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13159 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13160 rather than TRULY_NOOP_TRUNCATION in comments.
13161 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
13162 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
13163 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
13164 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
13165 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
13166 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
13167 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
13168 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
13169 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
13170 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
13171 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
13172 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13173 rather than TRULY_NOOP_TRUNCATION in comments.
13174 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
13175 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
13176 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
13177 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
13178 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
13179 TRULY_NOOP_TRUNCATION condition.
13180 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
13181 (TRULY_NOOP_TRUNCATION): Delete.
13182 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
13183 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
13184 * config/spu/spu.c (spu_truly_noop_truncation): New function.
13185 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13186 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
13187 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
13188 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
13189 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13190 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13191 rather than TRULY_NOOP_TRUNCATION in comments.
13192 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
13193 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
13194 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
13195 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
13196 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
13197 * system.h (TRULY_NOOP_TRUNCATION): Poison.
13199 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
13202 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
13203 (*cmp_ior): Likewise.
13204 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
13205 (*ior_scc_scc_cmp): Likewise.
13206 (*and_scc_scc): Likewise.
13207 (*and_scc_scc_cmp): Likewise.
13209 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13210 Alan Hayard <alan.hayward@arm.com>
13211 David Sherwood <david.sherwood@arm.com>
13213 * target.def (can_change_mode_class): New hook.
13214 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
13215 (hard_regno_nregs): Likewise.
13216 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
13217 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
13218 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
13219 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
13220 (LOAD_EXTEND_OP): Update accordingly.
13221 * doc/tm.texi: Regenerate.
13222 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
13223 CANNOT_CHANGE_MODE_CLASS.
13224 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
13225 (REG_CAN_CHANGE_MODE_P): ...this new macro.
13226 * combine.c (simplify_set): Update accordingly.
13227 * emit-rtl.c (validate_subreg): Likewise.
13228 * recog.c (general_operand): Likewise.
13229 * regcprop.c (mode_change_ok): Likewise.
13230 * reload1.c (choose_reload_regs): Likewise.
13231 (inherit_piecemeal_p): Likewise.
13232 * rtlanal.c (simplify_subreg_regno): Likewise.
13233 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
13234 instead of CANNOT_CHANGE_MODE_CLASS.
13235 (reload_cse_simplify_operands): Likewise.
13236 * reload.c (push_reload): Use targetm.can_change_mode_class
13237 instead of CANNOT_CHANGE_MODE_CLASS.
13238 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
13239 REG_CANNOT_CHANGE_MODE_P.
13240 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13241 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
13242 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13243 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13244 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13245 (arm_can_change_mode_class): New function.
13246 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
13247 than CANNOT_CHANGE_MODE_CLASS in comments.
13248 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13249 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
13250 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
13251 (ix86_can_change_mode_class): ...this new function, inverting the
13252 sense of the return value.
13253 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13254 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13255 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13256 (ia64_can_change_mode_class): New function.
13257 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13258 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
13259 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
13260 (m32c_can_change_mode_class): ...this new function, inverting the
13261 sense of the return value.
13262 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13263 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13264 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
13265 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
13266 (mips_can_change_mode_class): ...this new function, inverting the
13267 sense of the return value.
13268 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13269 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13270 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13271 (msp430_can_change_mode_class): New function.
13272 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13273 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
13274 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13275 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13276 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13277 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
13278 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13279 (pa_cannot_change_mode_class): Replace with...
13280 (pa_can_change_mode_class): ...this new function, inverting the
13281 sense of the return value.
13282 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
13283 than CANNOT_CHANGE_MODE_CLASS in comments.
13284 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13285 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
13286 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13287 (pdp11_cannot_change_mode_class): Replace with...
13288 (pdp11_can_change_mode_class): ...this new function, inverting the
13289 sense of the return value.
13290 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13291 * config/powerpcspe/powerpcspe-protos.h
13292 (rs6000_cannot_change_mode_class_ptr): Delete.
13293 * config/powerpcspe/powerpcspe.c
13294 (rs6000_cannot_change_mode_class_ptr): Delete.
13295 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13296 (rs6000_option_override_internal): Assign to
13297 targetm.can_change_mode_class instead of
13298 rs6000_cannot_change_mode_class_ptr.
13299 (rs6000_cannot_change_mode_class): Replace with...
13300 (rs6000_can_change_mode_class): ...this new function, inverting the
13301 sense of the return value.
13302 (rs6000_debug_cannot_change_mode_class): Replace with...
13303 (rs6000_debug_can_change_mode_class): ...this new function.
13304 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13305 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
13306 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13307 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13308 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
13310 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
13311 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13312 (rs6000_option_override_internal): Assign to
13313 targetm.can_change_mode_class instead of
13314 rs6000_cannot_change_mode_class_ptr.
13315 (rs6000_cannot_change_mode_class): Replace with...
13316 (rs6000_can_change_mode_class): ...this new function, inverting the
13317 sense of the return value.
13318 (rs6000_debug_cannot_change_mode_class): Replace with...
13319 (rs6000_debug_can_change_mode_class): ...this new function.
13320 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13321 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
13322 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
13323 (s390_can_change_mode_class): ...this new function, inverting the
13324 sense of the return value.
13325 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13326 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13327 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
13328 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13329 (sh_cannot_change_mode_class): Replace with...
13330 (sh_can_change_mode_class): ...this new function, inverting the
13331 sense of the return value.
13332 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13333 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13334 (sparc_can_change_mode_class): New function.
13335 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13336 * config/spu/spu.c (spu_can_change_mode_class): New function.
13337 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13338 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13339 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13340 (visium_can_change_mode_class): New function.
13341 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
13343 2017-09-15 Richard Biener <rguenther@suse.de>
13345 PR tree-optimization/82217
13346 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
13347 but not undefined case.
13349 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13352 * postreload.c (reload_cse_simplify_operands): Skip
13353 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
13355 2017-09-15 Richard Biener <rguenther@suse.de>
13357 PR tree-optimization/68823
13358 * graphite-scop-detection.c (build_alias_set): If we have a
13359 possible dependence check whether we can handle them by just
13360 looking at the DRs DR_ACCESS_FNs.
13361 (build_scops): If build_alias_set fails, fail the SCOP.
13363 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
13365 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
13366 to support float128 built-in functions that require the ISA 3.0
13368 (BU_FLOAT128_3_HW): Likewise.
13369 (SQRTF128): Add support for the IEEE 128-bit square root and fma
13370 built-in functions.
13371 (FMAF128): Likewise.
13373 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
13374 support for built-in functions that need the ISA 3.0 IEEE 128-bit
13375 floating point instructions.
13376 (rs6000_invalid_builtin): Likewise.
13377 (rs6000_builtin_mask_names): Likewise.
13378 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
13379 (RS6000_BTM_FLOAT128_HW): Likewise.
13380 (RS6000_BTM_COMMON): Likewise.
13381 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
13383 * doc/extend.texi (RS/6000 built-in functions): Document the
13384 IEEE 128-bit floating point square root and fused multiply-add
13385 built-in functions.
13387 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
13389 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
13390 reg (r2) isn't in the set of registers defined in the prologue.
13392 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13393 Alan Hayward <alan.hayward@arm.com>
13394 David Sherwood <david.sherwood@arm.com>
13396 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
13397 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
13398 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
13399 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
13400 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
13402 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
13403 max_vectorization_factor.
13404 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
13406 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13407 Alan Hayward <alan.hayward@arm.com>
13408 David Sherwood <david.sherwood@arm.com>
13410 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
13411 (vect_worthwhile_without_simd_p): Declare.
13412 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
13413 (vectorizable_reduction): Use it.
13414 * tree-vect-stmts.c (vectorizable_shift): Likewise.
13415 (vectorizable_operation): Likewise.
13417 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13418 Alan Hayward <alan.hayward@arm.com>
13419 David Sherwood <david.sherwood@arm.com>
13421 * tree-vectorizer.h (vect_get_num_copies): New function.
13422 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
13423 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13424 (vectorizable_induction): Likewise.
13425 (vectorizable_live_operation): Likewise.
13426 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13427 (vectorizable_bswap): Likewise.
13428 (vectorizable_call): Likewise.
13429 (vectorizable_conversion): Likewise.
13430 (vectorizable_assignment): Likewise.
13431 (vectorizable_shift): Likewise.
13432 (vectorizable_operation): Likewise.
13433 (vectorizable_store): Likewise.
13434 (vectorizable_load): Likewise.
13435 (vectorizable_condition): Likewise.
13436 (vectorizable_comparison): Likewise.
13437 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
13439 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13440 Alan Hayward <alan.hayward@arm.com>
13441 David Sherwood <david.sherwood@arm.com>
13443 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
13444 of vect_init_vector.
13446 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13447 Alan Hayward <alan.hayward@arm.com>
13448 David Sherwood <david.sherwood@arm.com>
13450 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
13451 an inline wrapper that provides a location.
13452 (gimple_build_vector): Likewise.
13453 * gimple-fold.c (gimple_build_vector_from_val): New function.
13454 (gimple_build_vector): Likewise.
13455 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
13456 functions to build the initial value. Always return a gimple value.
13457 (get_initial_defs_for_reduction): Likewise. Only compute
13459 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
13460 vect_init_vector on the results from get_initial_def(s)_for_reduction.
13461 (vectorizable_induction): Use gimple_build_vector rather than
13464 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13465 Alan Hayward <alan.hayward@arm.com>
13466 David Sherwood <david.sherwood@arm.com>
13468 * target.h (vec_perm_indices): New typedef.
13469 (auto_vec_perm_indices): Likewise.
13470 * optabs-query.h: Include target.h
13471 (can_vec_perm_p): Take a vec_perm_indices *.
13472 * optabs-query.c (can_vec_perm_p): Likewise.
13473 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
13474 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
13475 * tree-vect-generic.c (lower_vec_perm): Likewise.
13476 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
13477 (vect_grouped_load_supported): Likewise.
13478 (vect_shift_permute_load_chain): Likewise.
13479 (vect_permute_store_chain): Use auto_vec_perm_indices.
13480 (vect_permute_load_chain): Likewise.
13481 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
13482 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
13483 Update uses of can_vec_perm_p.
13484 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
13485 mode with a number of elements. Take a vec_perm_indices *.
13486 (vect_create_epilog_for_reduction): Update accordingly.
13487 Use auto_vec_perm_indices.
13488 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
13489 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
13490 (vect_transform_slp_perm_load): Likewise.
13491 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
13492 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
13493 (vect_gen_perm_mask_checked): Likewise.
13494 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
13495 (vect_gen_perm_mask_checked): Likewise.
13496 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
13497 (vectorizable_store): Likewise.
13498 (vectorizable_load): Likewise.
13499 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
13500 (vectorizable_bswap): Likewise.
13502 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13503 Alan Hayward <alan.hayward@arm.com>
13504 David Sherwood <david.sherwood@arm.com>
13506 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
13507 * tree.c (build_vector): Likewise.
13508 (build_vector_from_ctor): Update accordingly.
13509 (build_vector_from_val): Likewise.
13510 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
13511 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
13512 * tree-vect-generic.c (add_rshift): Likewise.
13513 (expand_vector_divmod): Likewise.
13514 (optimize_vector_constructor): Likewise.
13515 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
13516 (vect_transform_slp_perm_load): Likewise.
13517 (vect_schedule_slp_instance): Likewise.
13518 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
13519 (vectorizable_call): Likewise.
13520 (vect_gen_perm_mask_any): Likewise. Add elements in order.
13521 * expmed.c (make_tree): Likewise.
13522 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
13523 a vector passed to build_vector.
13524 (fold_convert_const): Likewise.
13525 (exact_inverse): Likewise.
13526 (fold_ternary_loc): Likewise.
13527 (fold_relational_const): Likewise.
13528 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
13529 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
13530 (const_unop): Likewise. Store the reduction accumulator in a
13531 variable rather than an array.
13532 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
13533 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
13534 the new vector, rather than constructing it after the input arrays.
13535 (native_interpret_vector): Use auto_vec<tree> when building
13536 a vector passed to build_vector. Add elements in order.
13537 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
13538 auto_vec<tree> when building a vector passed to build_vector.
13539 (vect_create_epilog_for_reduction): Likewise.
13540 (vectorizable_induction): Likewise.
13541 (get_initial_def_for_reduction): Likewise. Fix indentation of
13543 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
13545 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
13546 passed to build_vector.
13548 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13549 Alan Hayward <alan.hayward@arm.com>
13550 David Sherwood <david.sherwood@arm.com>
13552 * tree-core.h (tree_base::u): Add an "nelts" field.
13553 (tree_vector): Use VECTOR_CST_NELTS as the length.
13554 * tree.c (tree_size): Likewise.
13555 (make_vector): Initialize VECTOR_CST_NELTS.
13556 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
13557 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
13558 TYPE_VECTOR_SUBPARTS.
13559 * expr.c (const_vector_mask_from_tree): Consistently use "units"
13560 as the number of units, setting it from VECTOR_CST_NELTS.
13561 (const_vector_from_tree): Likewise.
13562 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
13563 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
13564 (fold_negate_expr_1): Likewise.
13565 (fold_convert_const): Likewise.
13566 (const_binop): Likewise. Differentiate the number of output and
13568 (const_unop): Likewise.
13569 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
13570 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
13571 in cases that did the opposite.
13573 2017-09-14 Richard Biener <rguenther@suse.de>
13575 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
13578 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
13580 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
13582 2017-09-14 Jakub Jelinek <jakub@redhat.com>
13585 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
13586 if and where to split a bb, except for splitting before debug insn
13587 sequences followed by non-label real insn. Delete debug insns
13588 in between basic blocks.
13590 * combine.c (make_compound_operation_int): Formatting fixes.
13592 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
13593 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13594 * config/netbsd.h (LINK_EH_SPEC): Likewise.
13595 * config/sol2.h (LINK_EH_SPEC): Likewise.
13596 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13597 * config/s390/linux.h (LINK_SPEC): Likewise.
13598 * config/freebsd.h (LINK_EH_SPEC): Likewise.
13599 * config/openbsd.h (LINK_EH_SPEC): Likewise.
13600 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13601 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
13602 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
13603 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13604 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
13605 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
13607 2017-09-13 Jakub Jelinek <jakub@redhat.com>
13609 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
13611 (ENDFILE_LINUX_SPEC): Likewise.
13612 (LINK_EH_SPEC): Likewise.
13613 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
13614 (LINK_OS_LINUX_SPEC32): Likewise.
13615 (LINK_OS_LINUX_SPEC64): Likewise.
13616 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
13617 (LINK_OS_LINUX_SPEC): Likewise.
13619 2017-09-13 Martin Liska <mliska@suse.cz>
13621 PR middle-end/82154
13622 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
13623 CASE_HIGH is NULL_TREE.
13625 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
13626 Alan Hayward <alan.hayward@arm.com>
13627 David Sherwood <david.sherwood@arm.com>
13629 * target.def (secondary_memory_needed): New hook.
13630 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
13631 instead of SECONDARY_MEMORY_NEEDED.
13632 (secondary_memory_needed_mode): Likewise.
13633 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
13634 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
13635 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
13636 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
13637 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
13638 * doc/tm.texi: Regenerate.
13639 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
13640 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
13641 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13642 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
13643 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
13644 * config/i386/i386.c (inline_secondary_memory_needed): Put the
13645 mode argument first and change the reg_class arguments to reg_class_t.
13646 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
13647 Make static. Update the call to inline_secondary_memory_needed.
13648 (ix86_register_move_cost): Update the call to
13649 inline_secondary_memory_needed.
13650 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13651 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
13653 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
13654 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
13656 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
13657 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
13658 * config/mips/mips.c (mips_secondary_memory_needed): Make static
13659 and match hook interface. Add comment from mips.h.
13660 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13661 * config/mmix/mmix.md (truncdfsf2): Refer to
13662 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
13664 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
13665 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
13666 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13667 (pa_secondary_memory_needed): New function.
13668 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
13669 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
13670 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13671 (pdp11_secondary_memory_needed): Make static and match hook interface.
13672 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
13673 * config/powerpcspe/powerpcspe-protos.h
13674 (rs6000_secondary_memory_needed_ptr): Delete.
13675 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
13677 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13678 (rs6000_option_override_internal): Assign to
13679 targetm.secondary_memory_needed rather than
13680 rs6000_secondary_memory_needed_ptr.
13681 (rs6000_secondary_memory_needed): Match hook interface.
13682 (rs6000_debug_secondary_memory_needed): Likewise.
13683 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
13684 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
13685 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
13686 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13687 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
13688 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
13690 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
13691 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13692 (rs6000_option_override_internal): Assign to
13693 targetm.secondary_memory_needed rather than
13694 rs6000_secondary_memory_needed_ptr.
13695 (rs6000_secondary_memory_needed): Match hook interface.
13696 (rs6000_debug_secondary_memory_needed): Likewise.
13697 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
13698 * config/s390/s390.c (s390_secondary_memory_needed): New function.
13699 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13700 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
13701 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13702 (sparc_secondary_memory_needed): New function.
13703 * lra-constraints.c (check_and_process_move): Refer to
13704 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
13706 (curr_insn_transform): Likewise.
13707 (process_alt_operands): Use targetm.secondary_memory_needed
13708 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13709 (check_secondary_memory_needed_p): Likewise.
13710 (choose_split_class): Likewise.
13711 * reload.c: Unconditionally include code that was previously
13712 conditional on SECONDARY_MEMORY_NEEDED.
13713 (push_secondary_reload): Use targetm.secondary_memory_needed
13714 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13715 (push_reload): Likewise.
13716 * reload1.c: Unconditionally include code that was previously
13717 conditional on SECONDARY_MEMORY_NEEDED.
13718 (choose_reload_regs): Use targetm.secondary_memory_needed
13719 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13720 (gen_reload): Likewise.
13721 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
13723 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
13724 Alan Hayward <alan.hayward@arm.com>
13725 David Sherwood <david.sherwood@arm.com>
13727 * target.def (secondary_memory_needed_mode): New hook:
13728 * targhooks.c (default_secondary_memory_needed_mode): Declare.
13729 * targhooks.h (default_secondary_memory_needed_mode): New function.
13730 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
13731 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
13732 * doc/tm.texi: Regenerate.
13733 * lra-constraints.c (check_and_process_move): Use
13734 targetm.secondary_memory_needed_mode instead of
13735 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
13736 (curr_insn_transform): Likewise.
13737 * reload.c (get_secondary_mem): Likewise.
13738 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13739 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
13741 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13742 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13743 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
13744 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13745 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
13747 * config/powerpcspe/powerpcspe-protos.h
13748 (rs6000_secondary_memory_needed_mode): Delete.
13749 * config/powerpcspe/powerpcspe.c
13750 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13751 (rs6000_secondary_memory_needed_mode): Make static.
13752 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13753 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
13755 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
13757 (rs6000_secondary_memory_needed_mode): Make static.
13758 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13759 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
13760 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13761 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13762 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
13764 (sparc_secondary_memory_needed_mode): New function.
13765 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
13767 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
13769 * config/aarch64/constraints.md (Umq): New constraint.
13770 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
13772 (mov<mode>): Update condition.
13774 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13776 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
13777 when bitposition is the same.
13779 2017-09-13 Richard Biener <rguenther@suse.de>
13781 * dwarf2out.c (output_die_symbol): Remove.
13782 (output_die): Do not output a DIEs symbol.
13784 2017-09-13 Richard Biener <rguenther@suse.de>
13786 PR middle-end/82128
13787 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
13788 default-def to avoid breaking iterator update with the weird
13789 interaction with cgraph_update_edges_for_call_stmt_node.
13791 2017-09-13 Richard Biener <rguenther@suse.de>
13793 * tree-cfg.c (verify_gimple_assign_binary): Add verification
13794 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
13795 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
13796 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
13798 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
13800 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
13801 Disable pc relative literal load irrespective of
13802 TARGET_FIX_ERR_A53_84341 for default.
13804 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
13806 * config/sparc/sparc.c (output_return): Output the source location of
13807 the insn in the delay slot, if any.
13808 (output_sibcall): Likewise.
13810 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
13813 * common.opt (-static-pie): New alias.
13814 (shared): Negate static-pie.
13815 (-no-pie): Update help text.
13817 (static-pie): New option.
13818 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
13819 -static-pie support.
13820 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
13821 (LINK_EH_SPEC): Likewise.
13822 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13823 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
13824 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
13825 * gcc.c (LINK_COMMAND_SPEC): Likewise.
13826 (init_gcc_specs): Likewise.
13827 (init_spec): Likewise.
13828 (display_help): Update help message for -pie.
13829 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
13832 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
13834 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
13835 (movdi_aarch64): Likewise.
13836 (movti_aarch64): Likewise.
13838 2017-09-12 Simon Wright <simon@pushface.org>
13841 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
13842 calculation of the minor version, always output as 0.
13844 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13847 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13848 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
13849 on it early, rather than manual conversion late. For
13850 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
13851 instead of performing manual conversion.
13853 2017-09-12 Carl Love <cel@us.ibm.com>
13855 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
13856 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
13858 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13859 VMULOSW): Add definitions.
13860 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13861 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13862 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13863 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13864 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13866 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
13868 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
13870 (movti_aarch64): Likewise.
13871 (movdf_aarch64): Likewise.
13872 (movtf_aarch64): Likewise.
13873 (load_pairdi): Likewise.
13874 (store_pairdi): Likewise.
13875 (load_pairdf): Likewise.
13876 (store_pairdf): Likewise.
13877 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
13878 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
13879 (ldr_got_small_<mode>): Likewise.
13880 (ldr_got_small_28k_<mode>): Likewise.
13881 (ldr_got_tiny): Likewise.
13882 * config/aarch64/iterators.md (ldst_sz): New.
13883 (ldpstp_sz): Likewise.
13884 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
13886 (thunderx_load): Split load_8 to load_16.
13887 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
13889 (thunderx2t99_storepair_basic): Split store_8 to store_16.
13890 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
13891 (xgene1_store_pair): Split store_8 to store_16.
13892 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
13893 (falkor_st_0_st_sd): Split store_8 to store_16.
13895 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
13897 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
13898 and store1/2/3/4 to store_4/8/12/16.
13899 * config/aarch64/aarch64.md: Update for rename.
13900 * config/arm/arm.md: Likewise.: Likewise.
13901 * config/arm/arm.c: Likewise.
13902 * config/arm/thumb1.md: Likewise.
13903 * config/arm/thumb2.md: Likewise.
13904 * config/arm/vfp.md: Likewise.
13905 * config/arm/arm-generic.md: Likewise.
13906 * config/arm/arm1020e.md: Likewise.
13907 * config/arm/arm1026ejs.md: Likewise.
13908 * config/arm/arm1136jfs.md: Likewise.
13909 * config/arm/arm926ejs.md: Likewise.
13910 * config/arm/cortex-a15.md: Likewise.
13911 * config/arm/cortex-a17.md: Likewise.
13912 * config/arm/cortex-a5.md: Likewise.
13913 * config/arm/cortex-a53.md: Likewise.
13914 * config/arm/cortex-a57.md: Likewise.
13915 * config/arm/cortex-a7.md: Likewise.
13916 * config/arm/cortex-a8.md: Likewise.
13917 * config/arm/cortex-a9.md: Likewise.
13918 * config/arm/cortex-m4.md: Likewise.
13919 * config/arm/cortex-m7.md: Likewise.
13920 * config/arm/cortex-r4.md: Likewise.
13921 * config/arm/exynos-m1.md: Likewise.
13922 * config/arm/fa526.md: Likewise.
13923 * config/arm/fa606te.md: Likewise.
13924 * config/arm/fa626te.md: Likewise.
13925 * config/arm/fa726te.md: Likewise.
13926 * config/arm/fmp626.md: Likewise.
13927 * config/arm/iwmmxt.md: Likewise.
13928 * config/arm/ldmstm.md: Likewise.
13929 * config/arm/marvell-pj4.md: Likewise.
13930 * config/arm/xgene1.md: Likewise.
13931 * config/aarch64/thunderx.md: Likewise.
13932 * config/aarch64/thunderx2t99.md: Likewise.
13933 * config/aarch64/falkor.md: Likewise.
13935 2017-09-12 Martin Liska <mliska@suse.cz>
13937 * attribs.c (private_lookup_attribute): New function.
13938 * attribs.h (private_lookup_attribute): Declared here.
13939 (lookup_attribute): Called from this place.
13941 2017-09-12 Richard Biener <rguenther@suse.de>
13943 PR tree-optimization/82157
13944 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
13945 stmts with side-effects.
13947 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13948 Alan Hayward <alan.hayward@arm.com>
13949 David Sherwood <david.sherwood@arm.com>
13951 * target.def (hard_regno_nregs): New hook.
13952 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
13953 * targhooks.h (default_hard_regno_nregs): Declare.
13954 * targhooks.c (default_hard_regno_nregs): New function.
13955 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
13956 (TARGET_HARD_REGNO_NREGS): ...this hook.
13957 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
13958 (CLASS_MAX_NREGS): Likewise.
13959 * doc/tm.texi: Regenerate.
13960 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
13961 instead of HARD_REGNO_NREGS.
13962 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
13963 HARD_REGNO_NREGS in the comment.
13964 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
13965 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
13966 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
13967 Return an unsigned int.
13968 (TARGET_HARD_REGNO_NREGS): Redefine.
13969 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
13970 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
13971 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
13972 (arc_hard_regno_nregs): New function.
13973 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
13974 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
13975 (arm_hard_regno_nregs): New function.
13976 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
13977 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
13978 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
13979 (TARGET_HARD_REGNO_NREGS): Redefine.
13980 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
13981 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
13982 (HARD_REGNO_NREGS): Delete.
13983 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
13984 (cr16_hard_regno_nregs): New function.
13985 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
13986 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
13987 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
13988 (cris_hard_regno_nregs): New function.
13989 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
13990 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
13991 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
13992 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
13993 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
13994 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
13995 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
13996 (frv_hard_regno_nregs): Make static. Take and return an
13998 (frv_class_max_nregs): Remove outdated copy of documentation.
13999 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
14000 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
14001 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
14002 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
14003 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
14004 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
14005 (TARGET_HARD_REGNO_NREGS): Redefine.
14006 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
14007 (CLASS_MAX_NREGS): Update comment.
14008 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
14009 (ia64_hard_regno_nregs): New function.
14010 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
14011 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
14012 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
14013 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
14014 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
14016 (m32c_hard_regno_nregs): Likewise. Make static.
14017 (TARGET_HARD_REGNO_NREGS): Redefine.
14018 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
14019 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
14020 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
14021 (m68k_hard_regno_nregs): New function.
14022 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
14023 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
14024 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
14025 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
14026 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
14027 Take and return an unsigned int.
14028 (TARGET_HARD_REGNO_NREGS): Redefine.
14029 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
14030 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
14031 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
14032 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
14033 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
14034 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
14035 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
14036 (msp430_hard_regno_nregs): Make static. Take and return an
14038 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
14039 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
14040 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
14041 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
14042 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
14043 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
14044 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
14045 (TARGET_HARD_REGNO_NREGS): Redefine.
14046 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
14047 (PA_HARD_REGNO_NREGS): ...this.
14048 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
14049 (PA_HARD_REGNO_NREGS): ...this.
14050 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
14051 (pa_hard_regno_nregs): New function.
14052 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
14053 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
14054 (pdp11_hard_regno_nregs): New function.
14055 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
14056 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
14057 (rs6000_hard_regno_nregs_hook): New function.
14058 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
14059 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
14060 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
14061 Take and return an unsigned int. Move earlier in file.
14062 (TARGET_HARD_REGNO_NREGS): Redefine.
14063 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
14064 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
14065 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
14066 (rl78_hard_regno_nregs): Make static. Take and return an
14068 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
14069 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
14070 (rs6000_hard_regno_nregs_hook): New function.
14071 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
14072 * config/rx/rx.c (rx_hard_regno_nregs): New function.
14073 (TARGET_HARD_REGNO_NREGS): Redefine.
14074 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
14075 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
14076 instead of HARD_REGNO_NREGS.
14077 (s390_hard_regno_nregs): New function.
14078 (s390_hard_regno_mode_ok): Add comment from s390.h.
14079 (TARGET_HARD_REGNO_NREGS): Redefine.
14080 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
14081 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
14082 (sh_hard_regno_nregs): New function.
14083 (sh_pass_in_reg_p): Use it.
14084 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
14085 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14086 (sparc_hard_regno_nregs): New function.
14087 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
14088 * config/spu/spu.c (spu_hard_regno_nregs): New function.
14089 (spu_function_arg_advance): Use it, supplying a valid register number.
14090 (TARGET_HARD_REGNO_NREGS): Redefine.
14091 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
14092 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
14093 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
14094 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
14095 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
14096 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
14097 (CLASS_MAX_NREGS): Remove copy of old documentation.
14098 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
14099 (visium_hard_regno_nregs): New function.
14100 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
14101 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
14102 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
14103 xtensa_hard_regno_nregs): New function.
14104 * system.h (HARD_REGNO_NREGS): Poison.
14106 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14108 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
14109 hard_regno_nregs instead of HARD_REGNO_NREGS.
14110 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
14111 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
14112 (c6x_expand_epilogue): Likewise.
14113 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
14114 (frv_read_iacc_argument): Likewise.
14115 * config/sh/sh.c: Include regs.h.
14116 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
14117 (regs_used): Likewise.
14118 (output_stack_adjust): Likewise.
14119 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
14120 * expmed.c: Include regs.h.
14121 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
14122 * ree.c: Include regs.h.
14123 (combine_reaching_defs): Use hard_regno_nregs instead of
14125 (add_removable_extension): Likewise.
14127 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14129 * regs.h (hard_regno_nregs): Turn into a function.
14130 (end_hard_regno): Update accordingly.
14131 * caller-save.c (setup_save_areas): Likewise.
14132 (save_call_clobbered_regs): Likewise.
14133 (replace_reg_with_saved_mem): Likewise.
14134 (insert_restore): Likewise.
14135 (insert_save): Likewise.
14136 * combine.c (can_change_dest_mode): Likewise.
14137 (move_deaths): Likewise.
14138 (distribute_notes): Likewise.
14139 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
14140 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
14141 (rs6000_split_multireg_move): Likewise.
14142 (rs6000_register_move_cost): Likewise.
14143 (rs6000_memory_move_cost): Likewise.
14144 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
14145 (rs6000_split_multireg_move): Likewise.
14146 (rs6000_register_move_cost): Likewise.
14147 (rs6000_memory_move_cost): Likewise.
14148 * cselib.c (cselib_reset_table): Likewise.
14149 (cselib_lookup_1): Likewise.
14150 * emit-rtl.c (set_mode_and_regno): Likewise.
14151 * function.c (aggregate_value_p): Likewise.
14152 * ira-color.c (setup_profitable_hard_regs): Likewise.
14153 (check_hard_reg_p): Likewise.
14154 (calculate_saved_nregs): Likewise.
14155 (assign_hard_reg): Likewise.
14156 (improve_allocation): Likewise.
14157 (calculate_spill_cost): Likewise.
14158 * ira-emit.c (modify_move_list): Likewise.
14159 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
14160 (ira_hard_reg_in_set_p): Likewise.
14161 * ira.c (setup_reg_mode_hard_regset): Likewise.
14162 (clarify_prohibited_class_mode_regs): Likewise.
14163 (check_allocation): Likewise.
14164 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14165 (lra_setup_reg_renumber): Likewise.
14166 (setup_try_hard_regno_pseudos): Likewise.
14167 (spill_for): Likewise.
14168 (assign_hard_regno): Likewise.
14169 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
14170 * lra-constraints.c (in_class_p): Likewise.
14171 (lra_constraint_offset): Likewise.
14172 (simplify_operand_subreg): Likewise.
14173 (lra_constraints): Likewise.
14174 (split_reg): Likewise.
14175 (split_if_necessary): Likewise.
14176 (invariant_p): Likewise.
14177 (inherit_in_ebb): Likewise.
14178 * lra-lives.c (process_bb_lives): Likewise.
14179 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
14180 (get_hard_regs): Likewise.
14181 (do_remat): Likewise.
14182 * lra-spills.c (assign_spill_hard_regs): Likewise.
14183 * mode-switching.c (create_pre_exit): Likewise.
14184 * postreload.c (reload_combine_recognize_pattern): Likewise.
14185 * recog.c (peep2_find_free_register): Likewise.
14186 * regcprop.c (kill_value_regno): Likewise.
14187 (set_value_regno): Likewise.
14188 (copy_value): Likewise.
14189 (maybe_mode_change): Likewise.
14190 (find_oldest_value_reg): Likewise.
14191 (copyprop_hardreg_forward_1): Likewise.
14192 * regrename.c (check_new_reg_p): Likewise.
14193 (regrename_do_replace): Likewise.
14194 * reload.c (push_reload): Likewise.
14195 (combine_reloads): Likewise.
14196 (find_dummy_reload): Likewise.
14197 (operands_match_p): Likewise.
14198 (find_reloads): Likewise.
14199 (find_equiv_reg): Likewise.
14200 (reload_adjust_reg_for_mode): Likewise.
14201 * reload1.c (count_pseudo): Likewise.
14202 (count_spilled_pseudo): Likewise.
14203 (find_reg): Likewise.
14204 (clear_reload_reg_in_use): Likewise.
14205 (free_for_value_p): Likewise.
14206 (allocate_reload_reg): Likewise.
14207 (choose_reload_regs): Likewise.
14208 (reload_adjust_reg_for_temp): Likewise.
14209 (emit_reload_insns): Likewise.
14210 (delete_output_reload): Likewise.
14211 * rtlanal.c (subreg_get_info): Likewise.
14212 * sched-deps.c (sched_analyze_reg): Likewise.
14213 * sel-sched.c (init_regs_for_mode): Likewise.
14214 (mark_unavailable_hard_regs): Likewise.
14215 (choose_best_reg_1): Likewise.
14216 (verify_target_availability): Likewise.
14217 * valtrack.c (dead_debug_insert_temp): Likewise.
14218 * var-tracking.c (track_loc_p): Likewise.
14219 (emit_note_insn_var_location): Likewise.
14220 * varasm.c (make_decl_rtl): Likewise.
14221 * reginfo.c (choose_hard_reg_mode): Likewise.
14222 (init_reg_modes_target): Refer directly to
14223 this_target_regs->x_hard_regno_nregs.
14225 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14227 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
14228 instead of hard_regno_nregs.
14230 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14232 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
14233 end_hard_regno instead of hard_regno_nregs.
14234 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
14235 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
14236 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
14237 * ira-color.c (improve_allocation): Likewise.
14238 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14239 * lra-lives.c (mark_regno_live): Likewise.
14240 (mark_regno_dead): Likewise.
14241 * lra-remat.c (operand_to_remat): Likewise.
14242 * lra.c (collect_non_operand_hard_regs): Likewise.
14243 * postreload.c (reload_combine_note_store): Likewise.
14244 (move2add_valid_value_p): Likewise.
14245 * reload.c (regno_clobbered_p): Likewise.
14247 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14249 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
14251 * config/v850/v850.c (v850_reorg): Likewise.
14252 * reload.c (refers_to_regno_for_reload_p): Likewise.
14253 (find_equiv_reg): Likewise.
14254 * reload1.c (reload_reg_reaches_end_p): Likewise.
14256 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14258 * caller-save.c (add_used_regs): Use REG_NREGS instead of
14260 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
14261 * config/arm/arm.c (output_move_neon): Likewise.
14262 (arm_attr_length_move_neon): Likewise.
14263 (neon_split_vcombine): Likewise.
14264 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
14265 (c6x_mark_reg_written): Likewise.
14266 (c6x_dwarf_register_span): Likewise.
14267 * config/i386/i386.c (ix86_save_reg): Likewise.
14268 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
14269 (rws_access_reg): Likewise.
14270 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
14271 * mode-switching.c (create_pre_exit): Likewise.
14272 * ree.c (combine_reaching_defs): Likewise.
14273 (add_removable_extension): Likewise.
14274 * regcprop.c (find_oldest_value_reg): Likewise.
14275 (copyprop_hardreg_forward_1): Likewise.
14276 * reload.c (reload_inner_reg_of_subreg): Likewise.
14277 (push_reload): Likewise.
14278 (combine_reloads): Likewise.
14279 (find_dummy_reload): Likewise.
14280 (reload_adjust_reg_for_mode): Likewise.
14281 * reload1.c (find_reload_regs): Likewise.
14282 (forget_old_reloads_1): Likewise.
14283 (reload_reg_free_for_value_p): Likewise.
14284 (reload_adjust_reg_for_temp): Likewise.
14285 (emit_reload_insns): Likewise.
14286 (delete_output_reload): Likewise.
14287 * sel-sched.c (choose_best_reg_1): Likewise.
14288 (choose_best_pseudo_reg): Likewise.
14290 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14291 Alan Hayward <alan.hayward@arm.com>
14292 David Sherwood <david.sherwood@arm.com>
14294 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
14295 * target.def (slow_unaligned_access): New hook.
14296 * targhooks.h (default_slow_unaligned_access): Declare.
14297 * targhooks.c (default_slow_unaligned_access): New function.
14298 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
14299 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
14300 * doc/tm.texi: Regenerate.
14301 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
14302 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
14303 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
14305 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
14306 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
14308 (rs6000_slow_unaligned_access): New function.
14309 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
14310 (expand_block_compare): Likewise.
14311 (expand_strn_compare): Likewise.
14312 (rs6000_rtx_costs): Likewise.
14313 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
14314 (riscv_slow_unaligned_access): Likewise.
14315 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
14316 (riscv_slow_unaligned_access_p): ...this and make static.
14317 (riscv_option_override): Update accordingly.
14318 (riscv_slow_unaligned_access): New function.
14319 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
14320 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
14321 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
14322 (rs6000_slow_unaligned_access): New function.
14323 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
14324 (rs6000_rtx_costs): Likewise.
14325 * config/rs6000/rs6000-string.c (expand_block_compare)
14326 (expand_strn_compare): Use targetm.slow_unaligned_access instead
14327 of SLOW_UNALIGNED_ACCESS.
14328 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
14329 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
14330 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
14331 of SLOW_UNALIGNED_ACCESS.
14332 * expmed.c (simple_mem_bitfield_p): Likewise.
14333 * expr.c (alignment_for_piecewise_move): Likewise.
14334 (emit_group_load_1): Likewise.
14335 (emit_group_store): Likewise.
14336 (copy_blkmode_from_reg): Likewise.
14337 (emit_push_insn): Likewise.
14338 (expand_assignment): Likewise.
14339 (store_field): Likewise.
14340 (expand_expr_real_1): Likewise.
14341 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
14342 * lra-constraints.c (simplify_operand_subreg): Likewise.
14343 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
14344 * gimple-ssa-store-merging.c: Likewise in block comment at start
14346 * tree-ssa-strlen.c: Include target.h.
14347 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
14348 of SLOW_UNALIGNED_ACCESS.
14349 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
14351 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14353 PR rtl-optimization/82185
14354 * expmed.c (emit_store_flag_int): Only test tem if it has been
14357 2017-09-12 Richard Biener <rguenther@suse.de>
14359 PR middle-end/82149
14360 * match.pd ((FTYPE) N CMP CST): Fix typo.
14362 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
14364 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
14366 (mips_near_type_p): Add 'short_call' attribute as a synonym
14368 * doc/extend.texi (short_call): Document new function attribute.
14370 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14373 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
14374 assertion check that in the condition.
14375 (get_atomic_generic_size): Likewise. Before testing if parameter
14376 has pointer type, if it has array type, call for C++
14377 default_conversion to perform array-to-pointer conversion.
14379 2017-09-12 Richard Biener <rguenther@suse.de>
14381 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
14382 for operations we cannot scalarize.
14384 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
14386 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
14387 vectors heap vectors. Clean up comments.
14388 Make visited_bbs a reference.
14389 (profitable_jump_thread_path): Make GC
14390 vectors heap vectors. Clean up comments.
14392 (convert_and_register_jump_thread_path): Make GC vectors heap
14394 (check_subpath_and_update_thread_path): Same. Clean up comments.
14395 Make visited_bbs a reference.
14396 (handle_phi): Abstract common code to to
14397 register_jump_thread_path_if_profitable.
14398 Rename VAR_BB to DEF_BB.
14400 Make GC vectors heap vectors.
14401 Make visited_bbs a reference.
14402 (handle_assignment): Same.
14403 (register_jump_thread_path_if_profitable): New.
14404 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
14406 Make GC vectors heap vectors. Clean up comments.
14407 Make visited_bbs a reference.
14408 (find_jump_threads_backwards): Make visited_bbs live in the stack.
14409 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
14412 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
14415 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
14416 words of E_DImode object are reachable by xtensa_uimm8x4 access.
14418 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
14420 Revert r251800 and r251799.
14422 2017-09-11 Martin Jambor <mjambor@suse.cz>
14425 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
14426 arguments in advance.
14427 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
14428 use it to find predecessor edges.
14429 (naive_outof_ssa): Collect vector of predecessors.
14431 2017-09-08 Jason Merrill <jason@redhat.com>
14433 PR c++/70029 - ICE with ref-qualifier and -flto
14434 * langhooks.h (struct lang_hooks_for_types): Add
14435 copy_lang_qualifiers.
14436 * attribs.c (build_type_attribute_qual_variant): Use it.
14437 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
14439 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
14440 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
14442 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
14445 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
14446 (*mulsi3_sp64): New instruction.
14447 (mulsi3): New expander.
14449 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
14451 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
14453 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14455 * sancov.c: Include memmodel.h.
14457 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
14460 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
14463 2017-09-07 Carl Love <cel@us.ibm.com>
14465 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
14466 the sldi instruction.
14468 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
14470 * sancov.c: Include tm_p.h.
14472 2017-09-07 Jakub Jelinek <jakub@redhat.com>
14475 * output.h (switch_to_other_text_partition): New declaration.
14476 * varasm.c (switch_to_other_text_partition): New function.
14477 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
14478 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
14479 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
14480 to the other text partition before emitting LCL label and switch back
14481 after emitting the word after it.
14483 2017-09-07 Richard Biener <rguenther@suse.de>
14485 * passes.def (pass_split_crit_edges): Remove instance before PRE.
14486 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
14487 critical edges here, after loop init.
14488 (pass_data_pre): Remove PROP_no_crit_edges flags.
14489 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
14490 for valueization of call args to avoid leaking VN_TOP.
14491 (visit_use): Assert we do not visit default defs.
14492 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
14493 Use error_mark_node to more easily detect leaking VN_TOP.
14494 All default-defs are varying, not VN_TOP. Mark them visited.
14495 (run_scc_vn): Make code match comment.
14497 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14499 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
14500 OPTION_MASK_FLOAT128_KEYWORD.
14501 (POWERPC_MASKS): Likewise.
14502 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
14503 support for the -mfloat128-type option, and make -mfloat128
14504 default on PowerPC Linux systems. Define or undefine
14505 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
14506 Define __float128 to be __ieee128 if IEEE 128-bit support is
14507 enabled, or undefine it.
14508 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
14509 Delete defining __FLOAT128_TYPE__.
14510 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
14511 -mfloat128-type option and make -mfloat128 default on PowerPC
14513 (TARGET_FLOAT128_TYPE): Likewise.
14514 (-mfloat128-type): Likewise.
14515 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14516 Delete the -mfloat128-type option and make -mfloat128 default on
14517 PowerPC Linux systems. Always use __ieee128 to be the keyword for
14518 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
14519 128-bit floating point is enabled. Change tests from using
14520 -mfloat128-type to -mfloat128.
14521 (rs6000_mangle_type): Use the correct mangling for the __float128
14522 type even if normal long double is restricted to 64-bits.
14523 (floatn_mode): Enable the _Float128 type by default on VSX Linux
14525 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
14526 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
14527 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
14529 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
14530 documentation for -mfloat128.
14532 2017-09-06 Olivier Hainque <hainque@adacore.com>
14534 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
14536 2017-09-06 Wish Wu <wishwu007@gmail.com>
14537 Jakub Jelinek <jakub@redhat.com>
14539 * asan.c (initialize_sanitizer_builtins): Add
14540 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
14541 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
14542 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
14543 BT_FN_VOID_UINT64_PTR variables.
14544 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
14545 (BT_FN_VOID_UINT16_UINT16): Likewise.
14546 (BT_FN_VOID_UINT32_UINT32): Likewise.
14547 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
14548 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
14549 (BT_FN_VOID_UINT64_PTR): Likewise.
14550 * common.opt (flag_sanitize_coverage): New variable.
14551 (fsanitize-coverage=trace-pc): Remove.
14552 (fsanitize-coverage=): Add.
14553 * flag-types.h (enum sanitize_coverage_code): New enum.
14554 * fold-const.c (fold_range_test): Disable non-short-circuit
14555 optimization if flag_sanitize_coverage.
14556 (fold_truth_andor): Likewise.
14557 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
14558 * opts.c (COVERAGE_SANITIZER_OPT): Define.
14559 (coverage_sanitizer_opts): New array.
14560 (get_closest_sanitizer_option): Add OPTS argument, handle also
14561 OPT_fsanitize_coverage_.
14562 (parse_sanitizer_options): Adjusted to also handle
14563 OPT_fsanitize_coverage_.
14564 (common_handle_option): Add OPT_fsanitize_coverage_.
14565 * sancov.c (instrument_comparison, instrument_switch): New function.
14566 (sancov_pass): Add trace-cmp support.
14567 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
14568 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
14569 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
14570 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
14571 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
14572 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
14573 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
14574 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
14575 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
14576 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
14578 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
14580 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14581 error. Only quit immediately if parsing is complete.
14582 (BEGIN): Initialize fatal_err and parse_done.
14583 (begin fpu, end fpu): Check number of arguments.
14584 (begin arch, end arch): Likewise.
14585 (begin cpu, end cpu): Likewise.
14586 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14587 (optalias): Likewise.
14589 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
14591 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14592 * config/arm/arm-isa.h: Delete. Move definitions to ...
14593 * arm-cpus.in: ... here. Use new feature and fgroup values.
14594 * config/arm/arm.c (arm_option_override): Use lower case for feature
14596 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14597 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14598 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14600 (print_isa_bits_for): New function.
14601 (gen_isa): New function.
14602 (gen_comm_data): Use print_isa_bits_for.
14603 (define feature): New keyword.
14604 (define fgroup): New keyword.
14605 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
14606 (arm-isa.h): Add rule to generate file.
14607 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14608 case for feature bit names.
14610 2017-09-06 Richard Biener <rguenther@suse.de>
14612 * tree-ssa-pre.c (NECESSARY): Remove.
14613 (create_expression_by_pieces): Do not touch pass-local flags.
14614 (insert_into_preds_of_block): Likewise.
14615 (do_pre_regular_insertion): Likewise.
14616 (eliminate_insert): Likewise.
14617 (eliminate_dom_walker::before_dom_children): Likewise.
14618 (fini_eliminate): Do not look at inserted_exprs.
14619 (mark_operand_necessary): Remove.
14620 (remove_dead_inserted_code): Replace with simple work-list
14621 algorithm based on inserted_exprs and SSA uses.
14622 (pass_pre::execute): Re-order fini_eliminate and
14623 remove_dead_inserted_code.
14625 2017-09-06 Olivier Hainque <hainque@adacore.com>
14627 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
14628 for VxWorks 7. Adjust surrounding comments.
14630 2017-09-06 Richard Biener <rguenther@suse.de>
14632 * gimple-ssa-strength-reduction.c
14633 (find_candidates_dom_walker::before_dom_children): Also allow
14636 2017-09-06 Richard Biener <rguenther@suse.de>
14638 PR tree-optimization/82108
14639 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
14640 for gap in the non-permutation SLP case.
14642 2017-09-06 Martin Jambor <mjambor@suse.cz>
14644 PR tree-optimization/82078
14645 * tree-sra.c (sort_and_splice_var_accesses): Move call to
14646 add_access_to_work_queue...
14647 (build_accesses_from_assign): ...here.
14648 (propagate_all_subaccesses): Make sure racc is the group
14649 representative, if there is one.
14651 2017-09-06 Jakub Jelinek <jakub@redhat.com>
14653 PR middle-end/82095
14654 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
14657 2017-09-06 Richard Biener <rguenther@suse.de>
14659 * gimple-ssa-strength-reduction.c
14660 (find_candidates_dom_walker::before_doom_children): Use a
14661 type and not a mode check.
14663 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
14666 * config/arm/predicates.md (arm_general_adddi_operand): Create new
14668 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
14670 2017-09-05 Jeff Law <law@redhat.com>
14672 PR tree-optimization/64910
14673 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
14674 cases where we have 3 or more operands.
14676 2017-09-05 Jakub Jelinek <jakub@redhat.com>
14678 PR middle-end/81768
14679 * omp-low.c (lower_omp_for): Recompute tree invariant if
14680 gimple_omp_for_initial/final is ADDR_EXPR.
14682 PR middle-end/81768
14683 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
14684 into gimple val before gimplification fo the COND_EXPR.
14686 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
14688 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
14689 REGION_COPY argument.
14690 (thread_through_all_blocks): Remove unused argument to
14691 duplicate_thread_path.
14693 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14694 Alan Hayward <alan.hayward@arm.com>
14695 David Sherwood <david.sherwood@arm.com>
14697 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
14698 Take a scalar_mode rather than a machine_mode.
14699 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14700 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
14701 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14702 (aarch64_gen_adjusted_ldpstp): Likewise.
14703 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
14705 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14706 Alan Hayward <alan.hayward@arm.com>
14707 David Sherwood <david.sherwood@arm.com>
14709 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
14710 Take a scalar_int_mode instead of a machine_mode.
14711 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
14712 (aarch64_output_scalar_simd_mov_immediate): Likewise.
14713 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
14714 (aarch64_simd_attr_length_rglist): Delete.
14715 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
14716 a scalar_int_mode instead of a machine_mode.
14717 (aarch64_add_offset): Likewise.
14718 (aarch64_internal_mov_immediate): Likewise
14719 (aarch64_add_constant_internal): Likewise.
14720 (aarch64_add_constant): Likewise.
14721 (aarch64_movw_imm): Likewise.
14722 (aarch64_rtx_arith_op_extract_p): Likewise.
14723 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
14724 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
14725 Remove assert that the mode isn't a vector.
14726 (aarch64_output_scalar_simd_mov_immediate): Likewise.
14727 (aarch64_expand_mov_immediate): Update calls after above changes.
14728 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
14729 (aarch64_and_bitmask_imm): Check for scalar integer modes.
14730 (aarch64_move_imm): Likewise.
14731 (aarch64_can_const_movi_rtx_p): Likewise.
14732 (aarch64_strip_extend): Likewise.
14733 (aarch64_extr_rtx_p): Likewise.
14734 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
14735 a CONST_INT when the mode parameter is VOIDmode.
14736 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
14738 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14740 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
14741 * stor-layout.c (bitwise_mode_for_mode): Likewise.
14742 (bitwise_type_for_mode): Update accordingly.
14744 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14746 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
14747 * stor-layout.c (mode_for_size_tree): Likewise.
14748 (mode_for_array): Update accordingly.
14749 (layout_decl): Likewise.
14750 (compute_record_mode): Likewise. Only set the mode once.
14752 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14754 * target.def (get_mask_mode): Change return type to opt_mode.
14756 * doc/tm.texi: Regenerate.
14757 * targhooks.h (default_get_mask_mode): Return an opt_mode.
14758 * targhooks.c (default_get_mask_mode): Likewise.
14759 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
14760 * optabs-query.c (can_vec_mask_load_store_p): Update use of
14761 targetm.get_mask_mode.
14762 * tree.c (build_truth_vector_type): Likewise.
14764 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14766 * machmode.h (mode_for_vector): Return an opt_mode.
14767 * stor-layout.c (mode_for_vector): Likewise.
14768 (mode_for_int_vector): Update accordingly.
14769 (layout_type): Likewise.
14770 * config/i386/i386.c (emit_memmov): Likewise.
14771 (ix86_expand_set_or_movmem): Likewise.
14772 (ix86_expand_vector_init): Likewise.
14773 (ix86_get_mask_mode): Likewise.
14774 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
14776 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
14777 * expmed.c (extract_bit_field_1): Likewise.
14778 * expr.c (expand_expr_real_2): Likewise.
14779 * optabs-query.c (can_vec_perm_p): Likewise.
14780 (can_vec_mask_load_store_p): Likewise.
14781 * optabs.c (expand_vec_perm): Likewise.
14782 * targhooks.c (default_get_mask_mode): Likewise.
14783 * tree-vect-stmts.c (vectorizable_store): Likewise.
14784 (vectorizable_load): Likewise.
14785 (get_vectype_for_scalar_type_and_size): Likewise.
14787 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14789 * machmode.h (mode_for_int_vector): New function.
14790 * stor-layout.c (mode_for_int_vector): Likewise.
14791 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
14792 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
14793 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
14794 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
14795 (s390_expand_vcond): Likewise.
14797 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14799 * machmode.h (opt_machine_mode): New type.
14800 (opt_mode<T>): Allow construction from anything that can be
14802 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
14803 (mode_for_size): Return an opt_machine_mode.
14804 * stor-layout.c (mode_for_size): Likewise.
14805 (mode_for_size_tree): Update call accordingly.
14806 (bitwise_mode_for_mode): Likewise.
14807 (make_fract_type): Likewise.
14808 (make_accum_type): Likewise.
14809 * caller-save.c (replace_reg_with_saved_mem): Update call
14811 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14812 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14813 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14814 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14815 * expmed.c (extract_bit_field_1): Likewise.
14816 * reload.c (get_secondary_mem): Likewise.
14817 * varasm.c (assemble_integer): Likewise.
14818 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
14821 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14823 * machmode.h (decimal_float_mode_for_size): New function.
14824 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
14825 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
14826 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
14827 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
14828 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
14829 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
14831 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14833 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
14834 (get_builtin_sync_mode): Likewise.
14835 (expand_ifn_atomic_compare_exchange): Likewise.
14836 (expand_builtin_atomic_clear): Likewise.
14837 (expand_builtin_atomic_test_and_set): Likewise.
14838 (fold_builtin_atomic_always_lock_free): Likewise.
14839 * calls.c (compute_argument_addresses): Likewise.
14840 (emit_library_call_value_1): Likewise.
14841 (store_one_arg): Likewise.
14842 * combine.c (combine_instructions): Likewise.
14843 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
14844 * config/arm/arm.c (arm_function_value): Likewise.
14845 (aapcs_allocate_return_reg): Likewise.
14846 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
14847 * config/i386/i386.c (construct_container): Likewise.
14848 (ix86_gimplify_va_arg): Likewise.
14849 (ix86_expand_sse_cmp): Likewise.
14850 (emit_memmov): Likewise.
14851 (emit_memset): Likewise.
14852 (expand_small_movmem_or_setmem): Likewise.
14853 (ix86_expand_pextr): Likewise.
14854 (ix86_expand_pinsr): Likewise.
14855 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
14856 * config/microblaze/microblaze.c (microblaze_block_move_straight):
14858 * config/mips/mips.c (mips_function_value_1) Likewise.
14859 (mips_block_move_straight): Likewise.
14860 (mips_expand_ins_as_unaligned_store): Likewise.
14861 * config/powerpcspe/powerpcspe.c
14862 (rs6000_darwin64_record_arg_advance_flush): Likewise.
14863 (rs6000_darwin64_record_arg_flush): Likewise.
14864 * config/rs6000/rs6000.c
14865 (rs6000_darwin64_record_arg_advance_flush): Likewise.
14866 (rs6000_darwin64_record_arg_flush): Likewise.
14867 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
14868 (sparc_function_value_1): Likewise.
14869 * config/spu/spu.c (adjust_operand): Likewise.
14870 (spu_emit_branch_or_set): Likewise.
14871 (arith_immediate_p): Likewise.
14872 * emit-rtl.c (gen_lowpart_common): Likewise.
14873 * expr.c (expand_expr_real_1): Likewise.
14874 * function.c (assign_parm_setup_block): Likewise.
14875 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
14876 * reload1.c (alter_reg): Likewise.
14877 * stor-layout.c (mode_for_vector): Likewise.
14878 (layout_type): Likewise.
14880 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14882 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
14883 (spu_convert_move): Likewise.
14884 * lower-subreg.c (resolve_simple_move): Likewise.
14886 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14889 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
14890 define_insn to a define_expand.
14891 (altivec_vsum2sws_direct): New define_insn.
14892 (altivec_vsumsws): Convert from a define_insn to a define_expand.
14894 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
14896 * config/arm/arm.c (arm_option_params_internal): Improve setting of
14899 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
14904 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
14905 realign stack if stack alignment needed is less than incoming
14908 2017-09-05 Marek Polacek <polacek@redhat.com>
14911 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
14914 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
14916 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
14918 2017-09-05 Richard Biener <rguenther@suse.de>
14920 PR tree-optimization/82084
14921 * fold-const.c (can_native_encode_string_p): Handle wide characters.
14923 2017-09-05 Richard Biener <rguenther@suse.de>
14925 PR tree-optimization/82102
14926 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
14928 2017-09-05 Martin Liska <mliska@suse.cz>
14930 PR tree-optimization/82032
14931 * tree-cfg.c (generate_range_test): New function.
14932 * tree-cfg.h (generate_range_test): Declared here.
14933 * tree-cfgcleanup.c (convert_single_case_switch): New function.
14934 (cleanup_control_expr_graph): Use it.
14935 * tree-switch-conversion.c (try_switch_expansion): Remove
14937 (emit_case_nodes): Use generate_range_test.
14939 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
14942 * config/i386/i386.md (*<btsc><mode>_mask): Add
14943 TARGET_USE_BT to insn constraint.
14944 (*btr<mode>_mask): Ditto.
14946 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
14948 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
14949 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
14951 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
14954 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
14955 TARGET_NEON and TARGET_IWMMXT.
14956 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
14957 TARGET_NEON and TARGET_IWMMXT.
14958 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
14960 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
14962 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
14963 (ix86_rewrite_tls_address): Ditto.
14964 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
14965 (ix86_rewrite_tls_address_1): Ditto.
14966 (ix86_rewrite_tls_address): Ditto.
14967 * config/i386/predicates.md (tls_address_pattern): New predicate.
14968 * config/i386/i386.md (TLS address splitter): New splitter.
14970 2017-09-04 Richard Biener <rguenther@suse.de>
14972 PR tree-optimization/82084
14973 * fold-const.h (can_native_encode_string_p): Declare.
14974 * fold-const.c (can_native_encode_string_p): Factor out from ...
14975 (native_encode_string): ... here.
14976 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
14977 vectorizing stores from constants we later cannot handle.
14979 2017-09-04 Marek Polacek <polacek@redhat.com>
14982 * doc/invoke.texi: Update -Wtautological-compare documentation.
14984 2017-09-04 Jeff Law <law@redhat.com>
14986 PR tree-optimization/64910
14987 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
14988 swap the first and last operand if the last is a constant.
14990 2017-09-04 Marek Polacek <polacek@redhat.com>
14993 * convert.c (do_narrow): When sanitizing signed integer overflows,
14994 bail out for signed types.
14995 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
14997 2017-09-04 Richard Biener <rguenther@suse.de>
14999 PR tree-optimization/82060
15000 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15001 Move devirtualization after stmt folding and before EH/AB/noreturn
15002 cleanup to get the stmt refs canonicalized. Use a bool instead
15003 of gimple_modified_p since that doesn't work for NOPs. Schedule
15004 NOPs generated by folding for removal.
15006 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15007 Alan Hayward <alan.hayward@arm.com>
15008 David Sherwood <david.sherwood@arm.com>
15010 * coretypes.h (pad_direction): New enum.
15011 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
15012 (FUNCTION_ARG_PADDING): Likewise.
15013 * target.def (function_arg_padding): New hook.
15014 * targhooks.h (default_function_arg_padding): Declare.
15015 * targhooks.c (default_function_arg_padding): New function.
15016 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
15017 (TARGET_FUNCTION_ARG_PADDING): ...this.
15018 * doc/tm.texi: Regenerate.
15019 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
15020 instead of direction.
15021 (compute_argument_addresses): Likewise.
15022 (load_register_parameters): Likewise.
15023 (emit_library_call_value_1): Likewise.
15024 (store_one_arg): Use targetm.calls.function_arg_padding instead
15025 of FUNCTION_ARG_PADDING.
15026 (must_pass_in_stack_var_size_or_pad): Likewise.
15027 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
15028 (emit_group_store): Likewise.
15029 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
15030 instead of FUNCTION_ARG_PADDING.
15031 (emit_push_insn): Likewise, and propagate enum change throughout
15033 * function.h (direction): Delete.
15034 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
15036 * function.c (assign_parm_find_stack_rtl): Likewise.
15037 (assign_parm_setup_block_p): Likewise.
15038 (assign_parm_setup_block): Likewise.
15039 (gimplify_parameters): Likewise.
15040 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
15041 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
15043 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
15044 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15045 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
15046 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
15047 (aarch64_function_arg_padding): ...this new function.
15048 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
15049 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15050 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
15051 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15052 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
15053 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15054 (arm_pad_arg_upward): Replace with...
15055 (arm_function_arg_padding): ...this new function.
15056 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
15058 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
15059 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
15060 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15061 (ia64_hpux_function_arg_padding): Replace with...
15062 (ia64_function_arg_padding): ...this new function. Use pad_direction
15063 instead of direction. Check for TARGET_HPUX.
15064 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
15065 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15066 (iq2000_function_arg_padding): New function.
15067 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
15068 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
15069 (mips_function_arg_padding): ...this new function.
15070 (mips_pad_reg_upward): Update accordingly.
15071 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15072 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
15073 targetm.calls.function_arg_padding.
15074 (FUNCTION_ARG_PADDING): Delete.
15075 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15076 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
15077 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15078 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
15079 (nios2_block_reg_padding): Return pad_direction instead of direction.
15080 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
15081 instead of direction.
15082 (nios2_function_arg_padding): Likewise. Make static.
15083 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15084 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
15085 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
15086 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
15087 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15088 (pa_function_arg_padding): Make static. Return pad_direction instead
15090 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
15091 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15092 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
15093 instead of direction. Use targetm.calls.function_arg_padding.
15094 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
15095 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
15096 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
15097 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
15098 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
15100 (function_arg_padding): Rename to...
15101 (rs6000_function_arg_padding): ...this. Make static. Return
15102 pad_direction instead of direction.
15103 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
15104 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
15105 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15106 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
15107 instead of direction. Use targetm.calls.function_arg_padding.
15108 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
15109 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
15110 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
15111 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
15112 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15113 (function_arg_padding): Rename to...
15114 (rs6000_function_arg_padding): ...this. Make static. Return
15115 pad_direction instead of direction.
15116 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
15117 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
15118 * config/s390/s390.c (s390_function_arg_padding): New function.
15119 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15120 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
15121 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
15122 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15123 (function_arg_padding): Rename to...
15124 (sparc_function_arg_padding): ...this. Make static. Return
15125 pad_direction instead of direction.
15126 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
15127 * config/spu/spu.c (spu_function_arg_padding): New function.
15128 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15129 * system.h (FUNCTION_ARG_PADDING): Poison.
15131 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15132 Alan Hayward <alan.hayward@arm.com>
15133 David Sherwood <david.sherwood@arm.com>
15135 * target.def (modes_tieable_p): New hook.
15136 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
15137 (TARGET_MODES_TIEABLE_P): ...this.
15138 * doc/tm.texi.in: Regenerate.
15139 * hooks.h (hook_bool_mode_mode_true): Declare.
15140 * hooks.c (hook_bool_mode_mode_true): New function.
15141 * combine.c (subst): Use targetm.modes_tieable_p instead of
15143 * dse.c (find_shift_sequence): Likewise.
15144 * expmed.c (extract_low_bits): Likewise.
15145 * lower-subreg.c: Include target.h.
15146 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
15148 * rtlanal.c (rtx_cost): Likewise.
15149 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
15150 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
15151 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
15152 (TARGET_MODES_TIEABLE_P): Redefine.
15153 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
15154 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
15155 (TARGET_MODES_TIEABLE_P): Redefine.
15156 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
15157 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
15158 (arc_modes_tieable_p): New function.
15159 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
15160 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
15161 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
15162 (arm_modes_tieable_p): Make static.
15163 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
15164 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
15165 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
15166 (TARGET_MODES_TIEABLE_P): Redefine.
15167 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
15168 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
15169 (TARGET_MODES_TIEABLE_P): Redefine.
15170 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
15171 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
15172 (cr16_modes_tieable_p): New function.
15173 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
15174 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
15175 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
15176 (TRULY_NOOP_TRUNCATION): Update comment.
15177 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
15178 (TRULY_NOOP_TRUNCATION): Update comment.
15179 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
15180 (frv_modes_tieable_p): New function.
15181 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
15182 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
15183 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
15184 (TARGET_MODES_TIEABLE_P): Redefine.
15185 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
15186 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
15187 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
15188 (TARGET_MODES_TIEABLE_P): Redefine.
15189 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
15190 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
15191 (ia64_modes_tieable_p): New function.
15192 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
15193 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
15194 (iq2000_modes_tieable_p): New function.
15195 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
15196 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
15197 (lm32_modes_tieable_p): New function.
15198 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
15199 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
15200 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
15201 (TARGET_MODES_TIEABLE_P): Redefine.
15202 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
15203 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
15204 (m32r_modes_tieable_p): New function.
15205 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
15206 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
15207 (m68k_modes_tieable_p): New function.
15208 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
15209 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
15210 (mcore_modes_tieable_p): New function.
15211 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
15212 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
15214 (TARGET_MODES_TIEABLE_P): Redefine.
15215 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
15216 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
15217 * config/mips/mips.c (mips_modes_tieable_p): Make static.
15218 (TARGET_MODES_TIEABLE_P): Redefine.
15219 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
15220 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
15221 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
15222 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
15223 (mn10300_modes_tieable_p): ...this and make static.
15224 (TARGET_MODES_TIEABLE_P): Redefine.
15225 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
15226 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
15227 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
15228 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
15229 (msp430_modes_tieable_p): Make static.
15230 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
15231 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
15232 (TARGET_MODES_TIEABLE_P): Redefine.
15233 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
15234 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
15235 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
15236 (TARGET_MODES_TIEABLE_P): Redefine.
15237 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
15238 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
15239 * config/pa/pa.c (pa_modes_tieable_p): Make static.
15240 (TARGET_MODES_TIEABLE_P): Redefine.
15241 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
15242 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
15243 (pdp11_modes_tieable_p): New function.
15244 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
15245 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
15246 (rs6000_modes_tieable_p): New function.
15247 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
15248 * config/powerpcspe/powerpcspe.md: Update comment.
15249 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
15250 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
15251 (TARGET_MODES_TIEABLE_P): Redefine.
15252 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
15253 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
15254 (rl78_modes_tieable_p): New function.
15255 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
15256 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
15257 (rs6000_modes_tieable_p): New function.
15258 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
15259 * config/rs6000/rs6000.md: Update comment.
15260 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
15261 * config/rx/rx.c (rx_modes_tieable_p): New function.
15262 (TARGET_MODES_TIEABLE_P): Redefine.
15263 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
15264 * config/s390/s390.c (s390_modes_tieable_p): New function.
15265 (TARGET_MODES_TIEABLE_P): Redefine.
15266 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
15267 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
15268 (sh_modes_tieable_p): New function.
15269 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
15270 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
15271 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
15272 (sparc_modes_tieable_p): Make static.
15273 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
15274 * config/spu/spu.c (spu_modes_tieable_p): New function.
15275 (TARGET_MODES_TIEABLE_P): Redefine.
15276 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
15277 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
15278 (TARGET_MODES_TIEABLE_P): Redefine.
15279 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
15280 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
15281 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
15282 * config/v850/v850.c (v850_modes_tieable_p): New function.
15283 (TARGET_MODES_TIEABLE_P): Redefine.
15284 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
15285 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
15286 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
15287 (visium_modes_tieable_p): New function.
15288 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
15289 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
15290 (xtensa_modes_tieable_p): New function.
15291 * system.h (MODES_TIEABLE_P): Poison.
15293 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15294 Alan Hayward <alan.hayward@arm.com>
15295 David Sherwood <david.sherwood@arm.com>
15297 * target.def (hard_regno_mode_ok): New hook.
15298 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
15299 (TARGET_HARD_REGNO_MODE_OK): ...this.
15300 * doc/tm.texi.in: Regenerate.
15301 * hooks.h (hook_bool_uint_mode_true): Declare.
15302 * hooks.c (hook_bool_uint_mode_true): New function.
15303 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
15304 HARD_REGNO_MODE_OK.
15305 * genpreds.c (write_insn_preds_c): Add an include of target.h.
15306 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
15307 instead of HARD_REGNO_MODE_OK.
15308 * caller-save.c: Include target.h.
15309 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
15310 HARD_REGNO_MODE_OK.
15311 * combine.c (can_combine_p): Likewise.
15312 (combinable_i3pat): Likewise.
15313 (can_change_dest_mode): Likewise.
15314 * expr.c (init_expr_target): Likewise.
15315 (convert_move): Likewise.
15316 (convert_modes): Likewise.
15317 * ira.c (setup_prohibited_class_mode_regs): Likewise.
15318 (setup_prohibited_mode_move_regs): Likewise.
15319 * ira.h (target_ira): Likewise.
15320 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15321 * lra-constraints.c (process_alt_operands): Likewise.
15322 (split_reg): Likewise.
15323 * recog.c (peep2_find_free_register): Likewise.
15324 * ree.c (combine_reaching_defs): Likewise.
15325 * regcprop.c (maybe_mode_change): Likewise.
15326 * reginfo.c (init_reg_sets_1): Likewise.
15327 (choose_hard_reg_mode): Likewise.
15328 (simplifiable_subregs): Likewise.
15329 * regrename.c (check_new_reg_p): Likewise.
15330 * reload.c (find_valid_class): Likewise.
15331 (find_valid_class_1): Likewise.
15332 (reload_inner_reg_of_subreg): Likewise.
15333 (push_reload): Likewise.
15334 (combine_reloads): Likewise.
15335 (find_dummy_reload): Likewise.
15336 (find_reloads): Likewise.
15337 * reload1.c (find_reg): Likewise.
15338 (set_reload_reg): Likewise.
15339 (allocate_reload_reg): Likewise.
15340 (choose_reload_regs): Likewise.
15341 (reload_adjust_reg_for_temp): Likewise.
15342 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
15343 (simplify_subreg_regno): Likewise.
15344 * sel-sched.c (init_regs_for_mode): Likewise.
15345 * varasm.c (make_decl_rtl): Likewise.
15346 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
15347 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
15348 HARD_REGNO_MODE_OK.
15349 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
15350 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
15351 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15352 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
15353 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
15354 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15355 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
15356 (arc_mode_class): Delete.
15357 (HARD_REGNO_MODE_OK): Delete.
15358 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15359 (arc_hard_regno_mode_ok): Rename old array to...
15360 (arc_hard_regno_mode_ok_modes): ...this.
15361 (arc_conditional_register_usage): Update accordingly.
15362 (arc_mode_class): Make static.
15363 (arc_hard_regno_mode_ok): New function.
15364 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
15365 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
15366 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15367 (arm_hard_regno_mode_ok): Make static.
15368 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
15369 HARD_REGNO_MODE_OK.
15370 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
15371 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
15372 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
15374 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15375 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
15376 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
15377 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
15378 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
15379 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15380 * config/bfin/predicates.md (valid_reg_operand): Use
15381 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
15382 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
15383 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
15384 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15385 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
15386 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
15387 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15388 (cr16_hard_regno_mode_ok): Make static and return a bool.
15389 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
15390 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15391 (cris_hard_regno_mode_ok): New function.
15392 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
15393 (epiphany_mode_class): Delete.
15394 (HARD_REGNO_MODE_OK): Delete.
15395 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
15396 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15397 (hard_regno_mode_ok): Rename to...
15398 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
15399 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
15400 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
15401 HARD_REGNO_MODE_OK.
15402 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
15403 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
15404 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15405 (frv_hard_regno_mode_ok): Make static and return a bool.
15406 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
15407 HARD_REGNO_MODE_OK.
15408 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
15409 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
15410 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
15411 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
15413 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15414 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
15415 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
15416 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
15418 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15419 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
15420 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15421 (ia64_hard_regno_mode_ok): New function.
15422 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
15423 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15424 (iq2000_hard_regno_mode_ok): New function.
15425 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
15426 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15427 (lm32_hard_regno_mode_ok): New function.
15428 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
15429 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
15430 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
15431 instead of HARD_REGNO_MODE_OK.
15432 (m32c_hard_regno_ok): Rename to...
15433 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
15434 (m32c_cannot_change_mode_class): Update accordingly.
15435 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15436 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
15437 (m32r_mode_class): Delete.
15438 (HARD_REGNO_MODE_OK): Delete.
15439 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15440 (m32r_hard_regno_mode_ok): Rename to...
15441 (m32r_hard_regno_modes): ...this.
15442 (m32r_mode_class): Make static.
15443 (m32r_hard_regno_mode_ok): New function.
15444 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
15445 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
15446 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15447 (m68k_hard_regno_mode_ok): Make static.
15448 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
15449 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15450 (mcore_hard_regno_mode_ok): New function.
15451 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
15452 (HARD_REGNO_MODE_OK): Delete.
15453 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
15455 (microblaze_hard_regno_mode_ok_p): ...this and make static.
15456 (microblaze_hard_regno_mode_ok): New function.
15457 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15458 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
15459 (mips_hard_regno_mode_ok): Delete.
15460 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
15461 (mips_hard_regno_mode_ok_p): ...this and make static.
15462 (mips_hard_regno_mode_ok_p): Rename to...
15463 (mips_hard_regno_mode_ok_uncached): ...this.
15464 (mips_hard_regno_mode_ok): New function.
15465 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
15466 of HARD_REGNO_MODE_OK.
15467 (mips_option_override): Update after above name changes.
15468 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15469 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
15470 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
15471 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
15472 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
15473 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15474 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
15475 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
15476 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
15477 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15478 (msp430_hard_regno_mode_ok): Make static and return a bool.
15479 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
15480 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
15481 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
15483 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15484 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
15485 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
15486 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
15487 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
15488 (PA_HARD_REGNO_MODE_OK): ...this
15489 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
15490 (PA_HARD_REGNO_MODE_OK): ...this.
15491 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15492 (pa_hard_regno_mode_ok): New function.
15493 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
15494 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15495 (pdp11_hard_regno_mode_ok): New function.
15496 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
15497 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
15499 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
15501 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15502 (rs6000_hard_regno_mode_ok): Rename to...
15503 (rs6000_hard_regno_mode_ok_uncached): ...this.
15504 (rs6000_init_hard_regno_mode_ok): Update accordingly.
15505 (rs6000_hard_regno_mode_ok): New function.
15506 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
15507 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
15508 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
15509 (riscv_hard_regno_mode_ok): ...this and make static.
15510 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15511 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
15512 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
15513 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15514 (rl78_hard_regno_mode_ok): Make static and return bool.
15515 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
15516 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
15518 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
15519 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15520 (rs6000_hard_regno_mode_ok): Rename to...
15521 (rs6000_hard_regno_mode_ok_uncached): ...this.
15522 (rs6000_init_hard_regno_mode_ok): Update accordingly.
15523 (rs6000_hard_regno_mode_ok): New function.
15524 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
15525 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
15526 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15527 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
15528 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
15529 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
15530 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15531 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
15532 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
15533 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15534 (sh_hard_regno_mode_ok): Make static.
15535 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
15536 instead of HARD_REGNO_MODE_OK.
15537 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
15538 (sparc_mode_class): Delete.
15539 (HARD_REGNO_MODE_OK): Delete.
15540 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15541 (hard_regno_mode_classes): Make static.
15542 (sparc_mode_class): Likewise.
15543 (sparc_hard_regno_mode_ok): New function.
15544 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
15545 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
15546 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
15548 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15549 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
15550 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
15551 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
15552 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
15553 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15554 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
15555 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
15556 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15557 (visium_hard_regno_mode_ok): New function.
15558 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
15559 instead of HARD_REGNO_MODE_OK.
15560 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
15561 (HARD_REGNO_MODE_OK): Delete.
15562 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
15563 (xtensa_hard_regno_mode_ok_p): ...this and make static.
15564 (xtensa_option_override): Update accordingly.
15565 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15566 (xtensa_hard_regno_mode_ok): New function.
15567 * system.h (HARD_REGNO_MODE_OK): Poison.
15569 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15570 Alan Hayward <alan.hayward@arm.com>
15571 David Sherwood <david.sherwood@arm.com>
15573 * target.def (hard_regno_call_part_clobbered): New hook.
15574 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
15575 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
15576 * doc/tm.texi: Regenerate.
15577 * hooks.h (hook_bool_uint_mode_false): Declare.
15578 * hooks.c (hook_bool_uint_mode_false): New function.
15579 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15580 * cselib.c (cselib_process_insn): Use
15581 targetm.hard_regno_call_part_clobbered instead of
15582 HARD_REGNO_CALL_PART_CLOBBERED.
15583 * ira-conflicts.c (ira_build_conflicts): Likewise.
15584 * ira-costs.c (ira_tune_allocno_costs): Likewise.
15585 * lra-constraints.c (need_for_call_save_p): Likewise.
15586 * lra-lives.c: Include target.h.
15587 (check_pseudos_live_through_calls): Use
15588 targetm.hard_regno_call_part_clobbered instead of
15589 HARD_REGNO_CALL_PART_CLOBBERED.
15590 * regcprop.c: Include target.h.
15591 (copyprop_hardreg_forward_1): Use
15592 targetm.hard_regno_call_part_clobbered instead of
15593 HARD_REGNO_CALL_PART_CLOBBERED.
15594 * reginfo.c (choose_hard_reg_mode): Likewise.
15595 * regrename.c (check_new_reg_p): Likewise.
15596 * reload.c (find_equiv_reg): Likewise.
15597 * reload1.c (emit_reload_insns): Likewise.
15598 * sched-deps.c (deps_analyze_insn): Likewise.
15599 * sel-sched.c (init_regs_for_mode): Likewise.
15600 (mark_unavailable_hard_regs): Likewise.
15601 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
15602 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15603 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
15605 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15606 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15607 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
15609 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
15611 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15612 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15613 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
15615 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15616 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15617 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
15619 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15620 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
15622 * config/powerpcspe/powerpcspe.c
15623 (rs6000_hard_regno_call_part_clobbered): New function.
15624 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15625 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15626 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
15628 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15629 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15630 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
15632 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15633 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15634 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
15636 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15637 Alan Hayward <alan.hayward@arm.com>
15638 David Sherwood <david.sherwood@arm.com>
15640 * rtl.h (subreg_memory_offset): Declare.
15641 * emit-rtl.c (subreg_memory_offset): New function.
15642 * expmed.c (store_bit_field_1): Use it.
15643 * expr.c (undefined_operand_subword_p): Likewise.
15644 * simplify-rtx.c (simplify_subreg): Likewise.
15646 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
15648 PR rtl-optimization/57448
15651 * optabs.c (expand_atomic_load): Place compiler memory barriers if
15652 using atomic_load pattern.
15653 (expand_atomic_store): Likewise.
15655 2017-09-04 Jakub Jelinek <jakub@redhat.com>
15658 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
15659 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
15660 handling. Use replace_call_with_value with NULL instead of
15661 gsi_replace, unlink_stmt_vdef and release_defs.
15663 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
15666 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
15668 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15671 * rtl.h (emit_library_call_value_1): Declare.
15672 (emit_library_call): Replace declaration with a series of overloads.
15673 Remove the parameter count argument.
15674 (emit_library_call_value): Likewise.
15675 * calls.c (emit_library_call_value_1): Make global. Replace varargs
15676 with an "rtx_mode_t *".
15677 (emit_library_call_value): Delete.
15678 (emit_library_call): Likewise.
15679 * asan.c (asan_emit_stack_protection): Update calls accordingly.
15680 (asan_emit_allocas_unpoison): Likewise.
15681 * builtins.c (expand_builtin_powi): Likewise.
15682 (expand_asan_emit_allocas_unpoison): Likewise.
15683 * cfgexpand.c (expand_main_function): Likewise.
15684 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
15685 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
15686 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
15687 * config/arm/arm.c (arm_trampoline_init): Likewise.
15688 (arm_call_tls_get_addr): Likewise.
15689 (arm_expand_divmod_libfunc): Likewise.
15690 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
15691 (smulsi3_highpart): Likewise.
15692 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
15693 (c6x_expand_compare): Likewise.
15694 (c6x_expand_movmem): Likewise.
15695 * config/frv/frv.c (frv_trampoline_init): Likewise.
15696 * config/i386/i386.c (ix86_trampoline_init): Likewise.
15697 (ix86_expand_divmod_libfunc): Likewise.
15698 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
15699 (ia64_expand_compare): Likewise.
15700 (ia64_profile_hook): Likewise.
15701 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
15702 (nonlocal_goto): Likewise.
15703 (restore_stack_nonlocal): Likewise.
15704 * config/m32r/m32r.c (block_move_call): Likewise.
15705 (m32r_trampoline_init): Likewise.
15706 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
15707 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
15708 (m68k_call_m68k_read_tp): Likewise.
15709 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
15710 (microblaze_expand_divide): Likewise.
15711 * config/mips/mips.h (mips_args): Likewise.
15712 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
15713 (MIPS_ICACHE_SYNC): Likewise.
15714 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
15715 (nios2_trampoline_init): Likewise.
15716 * config/pa/pa.c (hppa_tls_call): Likewise.
15717 (pa_trampoline_init): Likewise.
15718 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
15719 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
15720 (expand_strn_compare): Likewise.
15721 (rs6000_generate_compare): Likewise.
15722 (rs6000_expand_float128_convert): Likewise.
15723 (output_profile_hook): Likewise.
15724 (rs6000_trampoline_init): Likewise.
15725 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
15726 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
15727 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
15728 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
15729 (rs6000_generate_compare): Likewise.
15730 (rs6000_expand_float128_convert): Likewise.
15731 (output_profile_hook): Likewise.
15732 (rs6000_trampoline_init): Likewise.
15733 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
15734 * config/sh/sh.c (sh_trampoline_init): Likewise.
15735 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
15736 (sparc_emit_float_lib_cmp): Likewise.
15737 (sparc32_initialize_trampoline): Likewise.
15738 (sparc64_initialize_trampoline): Likewise.
15739 (sparc_profile_hook): Likewise.
15740 * config/spu/spu.c (ea_load_store): Likewise.
15741 * config/spu/spu.md (floatunssidf2): Likewise.
15742 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
15743 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
15744 * config/visium/visium.c (expand_block_move_4): Likewise.
15745 (expand_block_move_2): Likewise.
15746 (expand_block_move_1): Likewise.
15747 (expand_block_set_4): Likewise.
15748 (expand_block_set_2): Likewise.
15749 (expand_block_set_1): Likewise.
15750 (visium_trampoline_init): Likewise.
15751 (visium_profile_hook): Likewise.
15752 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
15753 (xtensa_setup_frame_addresses): Likewise.
15754 (xtensa_trampoline_init): Likewise.
15755 * except.c (sjlj_emit_function_enter): Likewise.
15756 (sjlj_emit_function_exit): Likewise.
15757 * explow.c (allocate_dynamic_stack_space): Likewise.
15758 (probe_stack_range): Likewise.
15759 * expr.c (convert_mode_scalar): Likewise.
15760 * optabs.c (expand_binop): Likewise.
15761 (expand_twoval_binop_libfunc): Likewise.
15762 (expand_unop): Likewise.
15763 (prepare_cmp_insn): Likewise.
15764 (prepare_float_lib_cmp): Likewise.
15765 (expand_float): Likewise.
15766 (expand_fix): Likewise.
15767 (expand_fixed_convert): Likewise.
15768 (maybe_emit_sync_lock_test_and_set): Likewise.
15769 (expand_atomic_compare_and_swap): Likewise.
15770 (expand_mem_thread_fence): Likewise.
15771 (expand_atomic_fetch_op): Likewise.
15773 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
15775 * doc/generic.texi (OpenACC): Adjust URL.
15776 * doc/invoke.texi (C Dialect Options): Ditto.
15778 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
15780 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
15781 predicate for operand 1. Add (m,<S>) constraint.
15782 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
15783 Prevent memory operand 1 with register operand 2.
15785 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
15787 PR rtl-optimization/82024
15788 * combine.c (try_combine): If the combination result is a PARALLEL,
15789 and we only need to retain the SET in there that would be placed
15790 at I2, check that we can place that at I3 instead, before doing so.
15792 2017-09-01 Jakub Jelinek <jakub@redhat.com>
15795 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
15797 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
15798 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
15801 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
15802 Jeff Law <law@redhat.com>
15804 * varasm.c (bss_initializer_p): Do not put constants into .bss
15805 (categorize_decl_for_section): Handle bss_initializer_p returning
15806 false when DECL_INITIAL is NULL.
15808 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15811 * config/s390/s390.c (s390_can_inline_p): New function.
15813 2017-09-01 Jeff Law <law@redhat.com>
15815 PR tree-optimization/82052
15816 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
15817 Always initialize the returned slot after a hash table miss
15818 when INSERT is true.
15820 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
15822 * config/s390/s390.md (mem_signal_fence): Remove.
15823 * doc/md.texi (mem_signal_fence): Remove.
15824 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
15826 * target-insns.def (mem_signal_fence): Remove.
15828 2017-09-01 Jakub Jelinek <jakub@redhat.com>
15831 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
15834 * asan.c (create_odr_indicator): Strip name encoding from assembler
15835 name before appending it after __odr_asan_.
15837 2017-09-01 Martin Liska <mliska@suse.cz>
15839 PR tree-optimization/82059
15840 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
15841 frequency only when an edge is redirected.
15843 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15845 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
15846 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
15847 (arc_conditional_register_usage): Remove ARC600 lp_count
15849 (arc_file_start): Emit Tag_ARC_CPU_variation.
15850 (arc_can_use_doloop_p): New conditions to use ZOLs.
15851 (hwloop_fail): New function.
15852 (hwloop_optimize): Likewise.
15853 (hwloop_pattern_reg): Likewise.
15854 (arc_doloop_hooks): New struct, to be used with reorg_loops.
15855 (arc_reorg_loops): New function, calls reorg_loops.
15856 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
15857 (arc600_corereg_hazard): Remove ZOL checking, case handled by
15859 (arc_loop_hazard): Remove function, functionality moved into
15861 (arc_hazard): Remove arc_loop_hazard call.
15862 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
15863 into hwloop_optimize.
15864 (arc_label_align): Remove ZOL handling.
15865 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
15866 * config/arc/arc.md (doloop_begin): Remove pattern.
15867 (doloop_begin_i): Likewise.
15868 (doloop_end_i): Likewise.
15869 (doloop_fallback): Likewise.
15870 (doloop_fallback_m): Likewise.
15871 (doloop_end): Reimplement expand.
15872 (arc_lp): New pattern for LP instruction.
15873 (loop_end): New pattern.
15874 (loop_fail): Likewise.
15875 (decrement_and_branch_until_zero): Likewise.
15876 * config/arc/arc.opt (mlpc-width): New option.
15877 * doc/invoke.texi (mlpc-width): Document option.
15879 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15881 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
15882 (arc_ccfsm_advance): Fix checking for delay slots.
15883 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
15885 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15887 * config/arc/arc.md (movqi_insn): Add stores to save constant long
15889 (movhi_insn): Update store instruction constraint which are saving
15890 6-bit short immediates.
15891 (movsi_insn): Consider also short scaled load operations.
15892 (zero_extendhisi2_i): Use Usd constraint instead of T.
15893 (extendhisi2_i): Add q constraint.
15894 (arc_clzsi2): Add type and length attributes.
15895 (arc_ctzsi2): Likewise.
15896 * config/arc/constraints.md (Usc): Update constraint, the
15897 assembler can parse two relocations for a single instruction.
15899 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15901 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
15902 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
15904 2017-08-31 Olivier Hainque <hainque@adacore.com>
15906 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
15907 match as powerpc-wrs-vxworks*.
15909 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
15911 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
15912 register constraint for by-element operand.
15913 (aarch64_mls_elt_merge<mode>): Likewise.
15915 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15917 * config/arc/arc.c (arc_can_follow_jump): Check for short
15920 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15922 * config.gcc: Use g.opt for arc.
15923 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
15924 functionality moved to ...
15925 (legitimate_scaled_address_p): New function, ...here.
15926 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
15927 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
15928 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
15930 (arc_override_options): Handle G option.
15931 (arc_output_pic_addr_const): Correct function definition.
15932 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
15933 (arc_decl_anon_ns_mem_p): Delete.
15934 (arc_in_small_data_p): Overhaul this function to take into
15935 consideration the value given via G option.
15936 (arc_rewrite_small_data_1): Renamed and corrected old
15937 arc_rewrite_small_data function.
15938 (arc_rewrite_small_data): New function.
15939 (small_data_pattern): Don't use pic_offset_table_rtx.
15940 * config/arc/arc.h (CC1_SPEC): Recognize G option.
15941 * config/arc/simdext.md (movmisalignv2hi): Use
15942 prepare_move_operands function.
15944 (movmisalign*): Likewise.
15945 * doc/invoke.texi (ARC options): Document -G option.
15947 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15949 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
15951 * config/arc/arc.c (arc_print_operand): Output scalled address for
15952 sdata whenever is possible.
15953 (arc_in_small_data_p): Allow sdata for 64bit datum when double
15954 load/stores are available.
15955 (compact_sda_memory_operand): Check for the alignment required by
15956 code density instructions.
15957 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
15959 * config/arc/constraints.md (Usd): Update constraint.
15960 (Us0): New constraint.
15961 (Usc): Update constraint.
15963 2017-08-31 Richard Biener <rguenther@suse.de>
15965 PR middle-end/82054
15966 * dwarf2out.c (dwarf2out_early_global_decl): Process each
15967 function only once.
15969 2017-08-31 Tamar Christina <tamar.christina@arm.com>
15971 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
15972 Resize type_signature.
15974 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15975 Alan Hayward <alan.hayward@arm.com>
15976 David Sherwood <david.sherwood@arm.com>
15978 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
15979 subregs whose inner modes can be stored in GPRs.
15980 (aarch64_classify_index): Likewise.
15982 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15983 Alan Hayward <alan.hayward@arm.com>
15984 David Sherwood <david.sherwood@arm.com>
15986 * config/aarch64/iterators.md (V_cmp_result): Rename to...
15987 (V_INT_EQUIV): ...this.
15988 (v_cmp_result): Rename to...
15989 (v_int_equiv): ...this.
15990 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
15991 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
15992 (copysign<mode>3): Likewise.
15993 (aarch64_simd_bsl<mode>_internal): Likewise.
15994 (aarch64_simd_bsl<mode>): Likewise.
15995 (vec_cmp<mode><mode>): Likewise.
15996 (vcond<mode><mode>): Likewise.
15997 (vcond<v_cmp_mixed><mode>): Likewise.
15998 (vcondu<mode><v_cmp_mixed>): Likewise.
15999 (aarch64_cm<optab><mode>): Likewise.
16000 (aarch64_cmtst<mode>): Likewise.
16001 (aarch64_fac<optab><mode>): Likewise.
16002 (vec_perm_const<mode>): Likewise.
16003 (vcond_mask_<mode><v_cmp_result>): Rename to...
16004 (vcond_mask_<mode><v_int_equiv>): ...this.
16005 (vec_cmp<mode><v_cmp_result>): Rename to...
16006 (vec_cmp<mode><v_int_equiv>): ...this.
16008 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16009 Alan Hayward <alan.hayward@arm.com>
16010 David Sherwood <david.sherwood@arm.com>
16012 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
16014 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
16015 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
16016 (UNSPEC_LD4_DREG): New unspecs.
16017 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
16018 (aarch64_ld2<mode>_dreg_be): Replace with...
16019 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
16021 (aarch64_ld3<mode>_dreg_le)
16022 (aarch64_ld3<mode>_dreg_be): Replace with...
16023 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
16025 (aarch64_ld4<mode>_dreg_le)
16026 (aarch64_ld4<mode>_dreg_be): Replace with...
16027 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
16030 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16032 PR tree-optimization/81987
16033 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
16034 insert an initializer in a location not dominated by the stride
16037 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
16039 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
16040 on the entire header of the finally block in the fallthru case.
16042 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
16044 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
16046 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
16048 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
16049 rs6000_emit_move_from_cr and call renamed function.
16050 (rs6000_emit_prologue): Call renamed functions.
16051 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
16052 movesi_from_cr, remove volatile CRs.
16054 2017-08-30 Jon Beniston <jon@beniston.com>
16055 Richard Biener <rguenther@suse.de>
16057 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
16058 of VECTOR_MODE_P check.
16059 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
16060 element vector types.
16062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16064 * df.h (df_read_modify_subreg_p): Remove in favor of...
16065 * rtl.h (read_modify_subreg_p): ...this new function. Take a
16066 const_rtx instead of an rtx.
16067 * cprop.c (local_cprop_find_used_regs): Update accordingly.
16068 * df-problems.c (df_word_lr_mark_ref): Likewise.
16069 * ira-lives.c (mark_pseudo_reg_live): Likewise.
16070 (mark_pseudo_reg_dead): Likewise.
16071 (mark_ref_dead): Likewise.
16072 * reginfo.c (init_subregs_of_mode): Likewise.
16073 * sched-deps.c (sched_analyze_1): Likewise.
16074 * df-scan.c (df_def_record_1): Likewise.
16075 (df_uses_record): Likewise.
16076 (df_read_modify_subreg_p): Remove in favor of...
16077 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
16078 const_rtx instead of an rtx.
16080 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16081 Alan Hayward <alan.hayward@arm.com>
16082 David Sherwood <david.sherwood@arm.com>
16084 * rtl.h (partial_subreg_p): New function.
16085 * caller-save.c (save_call_clobbered_regs): Use it.
16086 * calls.c (expand_call): Likewise.
16087 * combine.c (combinable_i3pat): Likewise.
16088 (simplify_set): Likewise.
16089 (make_extraction): Likewise.
16090 (make_compound_operation_int): Likewise.
16091 (gen_lowpart_or_truncate): Likewise.
16092 (force_to_mode): Likewise.
16093 (make_field_assignment): Likewise.
16094 (reg_truncated_to_mode): Likewise.
16095 (record_truncated_value): Likewise.
16096 (move_deaths): Likewise.
16097 * cse.c (record_jump_cond): Likewise.
16098 (cse_insn): Likewise.
16099 * cselib.c (cselib_lookup_1): Likewise.
16100 * expmed.c (extract_bit_field_using_extv): Likewise.
16101 * function.c (assign_parm_setup_reg): Likewise.
16102 * ifcvt.c (noce_convert_multiple_sets): Likewise.
16103 * ira-build.c (create_insn_allocnos): Likewise.
16104 * lra-coalesce.c (merge_pseudos): Likewise.
16105 * lra-constraints.c (match_reload): Likewise.
16106 (simplify_operand_subreg): Likewise.
16107 (curr_insn_transform): Likewise.
16108 * lra-lives.c (process_bb_lives): Likewise.
16109 * lra.c (new_insn_reg): Likewise.
16110 (lra_substitute_pseudo): Likewise.
16111 * regcprop.c (mode_change_ok): Likewise.
16112 (maybe_mode_change): Likewise.
16113 (copyprop_hardreg_forward_1): Likewise.
16114 * reload.c (push_reload): Likewise.
16115 (find_reloads): Likewise.
16116 (find_reloads_subreg_address): Likewise.
16117 * reload1.c (alter_reg): Likewise.
16118 (eliminate_regs_1): Likewise.
16119 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16121 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
16123 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
16124 back to if statements, including unpack.
16126 2017-08-30 Martin Liska <mliska@suse.cz>
16128 PR inline-asm/82001
16129 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
16131 (func_checker::compare_asm_inputs_outputs): ... this function.
16132 (func_checker::compare_gimple_asm): Use the function to compare
16133 also ASM constrains.
16134 * ipa-icf-gimple.h: Rename the function.
16136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16137 Alan Hayward <alan.hayward@arm.com>
16138 David Sherwood <david.sherwood@arm.com>
16140 * coretypes.h (complex_mode): New type.
16141 * gdbhooks.py (build_pretty_printer): Handle it.
16142 * machmode.h (complex_mode): New class.
16143 (complex_mode::includes_p): New function.
16144 (is_complex_int_mode): Likewise.
16145 (is_complex_float_mode): Likewise.
16146 * genmodes.c (get_mode_class): Handle complex mode classes.
16147 * function.c (expand_function_end): Use is_complex_int_mode.
16149 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16150 Alan Hayward <alan.hayward@arm.com>
16151 David Sherwood <david.sherwood@arm.com>
16153 * coretypes.h (scalar_mode_pod): New typedef.
16154 * gdbhooks.py (build_pretty_printer): Handle it.
16155 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
16156 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
16157 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
16158 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
16159 as_a <scalar_mode>.
16161 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16162 Alan Hayward <alan.hayward@arm.com>
16163 David Sherwood <david.sherwood@arm.com>
16165 * machmode.h (mode_for_vector): Take a scalar_mode instead
16167 * stor-layout.c (mode_for_vector): Likewise.
16168 * explow.c (promote_mode): Use as_a <scalar_mode>.
16169 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
16171 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16172 Alan Hayward <alan.hayward@arm.com>
16173 David Sherwood <david.sherwood@arm.com>
16175 * target.def (preferred_simd_mode): Take a scalar_mode
16176 instead of a machine_mode.
16177 * targhooks.h (default_preferred_simd_mode): Likewise.
16178 * targhooks.c (default_preferred_simd_mode): Likewise.
16179 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
16180 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
16181 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
16182 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
16183 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
16184 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
16185 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
16186 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
16188 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
16189 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
16190 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
16191 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
16192 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
16193 * doc/tm.texi: Regenerate.
16194 * optabs-query.c (can_vec_mask_load_store_p): Return false for
16197 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16198 Alan Hayward <alan.hayward@arm.com>
16199 David Sherwood <david.sherwood@arm.com>
16201 * target.def (scalar_mode_supported_p): Take a scalar_mode
16202 instead of a machine_mode.
16203 * targhooks.h (default_scalar_mode_supported_p): Likewise.
16204 * targhooks.c (default_scalar_mode_supported_p): Likewise.
16205 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
16206 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16207 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
16208 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
16209 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
16210 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
16211 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
16212 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
16213 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
16214 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
16215 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
16216 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
16218 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
16219 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
16220 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16221 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16222 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
16224 * doc/tm.texi: Regenerate.
16226 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16227 Alan Hayward <alan.hayward@arm.com>
16228 David Sherwood <david.sherwood@arm.com>
16230 * coretypes.h (opt_scalar_mode): New typedef.
16231 * gdbhooks.py (build_pretty_printers): Handle it.
16232 * machmode.h (mode_iterator::get_2xwider): Add overload for
16234 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
16236 * expr.c (convert_mode_scalar): Likewise.
16237 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16238 * optabs.c (expand_float): Likewise.
16239 (expand_fix): Likewise.
16240 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16243 Alan Hayward <alan.hayward@arm.com>
16244 David Sherwood <david.sherwood@arm.com>
16246 * optabs.c (expand_float): Explicitly check for scalars before
16247 using a branching expansion.
16248 (expand_fix): Likewise.
16250 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16251 Alan Hayward <alan.hayward@arm.com>
16252 David Sherwood <david.sherwood@arm.com>
16254 * expr.c (convert_mode): Split scalar handling out into...
16255 (convert_mode_scalar): ...this new function. Treat the modes
16258 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16259 Alan Hayward <alan.hayward@arm.com>
16260 David Sherwood <david.sherwood@arm.com>
16262 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
16264 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
16266 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16267 Alan Hayward <alan.hayward@arm.com>
16268 David Sherwood <david.sherwood@arm.com>
16270 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
16271 rather than a machine_mode.
16272 (fixed_from_string): Likewise.
16273 (fixed_convert): Likewise.
16274 (fixed_convert_from_int): Likewise.
16275 (fixed_convert_from_real): Likewise.
16276 (real_convert_from_fixed): Likewise.
16277 * fixed-value.c (fixed_from_double_int): Likewise.
16278 (fixed_from_string): Likewise.
16279 (fixed_convert): Likewise.
16280 (fixed_convert_from_int): Likewise.
16281 (fixed_convert_from_real): Likewise.
16282 (real_convert_from_fixed): Likewise.
16283 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
16285 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16286 Alan Hayward <alan.hayward@arm.com>
16287 David Sherwood <david.sherwood@arm.com>
16289 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
16290 of separate mode class checks. Do not allow vector modes here.
16291 (immed_wide_int_const): Use as_a <scalar_mode>.
16292 * explow.c (trunc_int_for_mode): Likewise.
16293 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
16294 (wi::shwi): Likewise.
16295 (wi::min_value): Likewise.
16296 (wi::max_value): Likewise.
16297 * dwarf2out.c (loc_descriptor): Likewise.
16298 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
16299 for CONST_WIDE_INT.
16301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16302 Alan Hayward <alan.hayward@arm.com>
16303 David Sherwood <david.sherwood@arm.com>
16305 * tree.h (SCALAR_TYPE_MODE): New macro.
16306 * expr.c (expand_expr_addr_expr_1): Use it.
16307 (expand_expr_real_2): Likewise.
16308 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
16309 (fold_convert_const_fixed_from_int): Likewise.
16310 (fold_convert_const_fixed_from_real): Likewise.
16311 (native_encode_fixed): Likewise
16312 (native_encode_complex): Likewise
16313 (native_encode_vector): Likewise.
16314 (native_interpret_fixed): Likewise.
16315 (native_interpret_real): Likewise.
16316 (native_interpret_complex): Likewise.
16317 (native_interpret_vector): Likewise.
16318 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
16319 (simd_clone_adjust_argument_types): Likewise.
16320 (simd_clone_init_simd_arrays): Likewise.
16321 (simd_clone_adjust): Likewise.
16322 * stor-layout.c (layout_type): Likewise.
16323 * tree.c (build_minus_one_cst): Likewise.
16324 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16325 * tree-inline.c (estimate_move_cost): Likewise.
16326 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
16327 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
16328 (vectorizable_reduction): Likewise.
16329 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
16330 (vect_recog_mixed_size_cond_pattern): Likewise.
16331 (check_bool_pattern): Likewise.
16332 (adjust_bool_pattern): Likewise.
16333 (search_type_for_mask_1): Likewise.
16334 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
16335 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16336 (vectorizable_load): Likewise.
16337 (vectorizable_store): Likewise.
16338 * ubsan.c (ubsan_encode_value): Likewise.
16339 * varasm.c (output_constant): Likewise.
16341 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16342 Alan Hayward <alan.hayward@arm.com>
16343 David Sherwood <david.sherwood@arm.com>
16345 * coretypes.h (scalar_mode): New class.
16346 * machmode.h (scalar_mode): Likewise.
16347 (scalar_mode::includes_p): New function.
16348 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
16349 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
16350 * genmodes.c (get_mode_class): Handle remaining scalar modes.
16351 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
16352 * expmed.c (store_bit_field_1): Likewise.
16353 (extract_bit_field_1): Likewise.
16354 * expr.c (write_complex_part): Likewise.
16355 (read_complex_part): Likewise.
16356 (emit_move_complex_push): Likewise.
16357 (expand_expr_real_2): Likewise.
16358 * function.c (assign_parm_setup_reg): Likewise.
16359 (assign_parms_unsplit_complex): Likewise.
16360 * optabs.c (expand_binop): Likewise.
16361 * rtlanal.c (subreg_get_info): Likewise.
16362 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16363 * varasm.c (output_constant_pool_2): Likewise.
16365 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16366 Alan Hayward <alan.hayward@arm.com>
16367 David Sherwood <david.sherwood@arm.com>
16369 * expmed.c (extract_high_half): Use scalar_int_mode and remove
16371 (expmed_mult_highpart_optab): Likewise.
16372 (expmed_mult_highpart): Likewise.
16374 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16375 Alan Hayward <alan.hayward@arm.com>
16376 David Sherwood <david.sherwood@arm.com>
16378 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
16379 instead of a machine_mode.
16380 (builtin_memset_read_str): Likewise.
16381 * builtins.c (c_readstr): Likewise.
16382 (builtin_memcpy_read_str): Likewise.
16383 (builtin_strncpy_read_str): Likewise.
16384 (builtin_memset_read_str): Likewise.
16385 (builtin_memset_gen_str): Likewise.
16386 (expand_builtin_signbit): Use scalar_int_mode for local variables.
16387 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
16388 instead of a machine_mode.
16389 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
16391 (make_extraction): Likewise.
16392 (try_widen_shift_mode): Take and return scalar_int_modes instead
16394 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
16395 a scalar_int_mode instead of a machine_mode.
16396 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
16397 (avr_addr_space_pointer_mode): Likewise.
16398 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
16399 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
16400 (msp430_unwind_word_mode): Likewise.
16401 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
16402 (spu_addr_space_pointer_mode): Likewise.
16403 (spu_addr_space_address_mode): Likewise.
16404 (spu_libgcc_cmp_return_mode): Likewise.
16405 (spu_libgcc_shift_count_mode): Likewise.
16406 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
16407 (rl78_addr_space_pointer_mode): Likewise.
16408 (fl78_unwind_word_mode): Likewise.
16409 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
16411 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
16412 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
16413 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
16414 (mips_valid_pointer_mode): Likewise.
16415 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
16416 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
16417 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
16419 (ft32_addr_space_address_mode): Likewise.
16420 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
16421 scalar_int_mode instead of a machine_mode.
16422 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
16424 (m32c_addr_space_address_mode): Likewise.
16425 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
16426 (rs6000_eh_return_filter_mode): Likewise.
16427 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
16428 (rs6000_eh_return_filter_mode): Likewise.
16429 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
16430 (s390_libgcc_shift_count_mode): Likewise.
16431 (s390_unwind_word_mode): Likewise.
16432 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
16434 * target.def (mode_rep_extended): Likewise.
16435 (valid_pointer_mode): Likewise.
16436 (addr_space.valid_pointer_mode): Likewise.
16437 (eh_return_filter_mode): Return a scalar_int_mode rather than
16439 (libgcc_cmp_return_mode): Likewise.
16440 (libgcc_shift_count_mode): Likewise.
16441 (unwind_word_mode): Likewise.
16442 (addr_space.pointer_mode): Likewise.
16443 (addr_space.address_mode): Likewise.
16444 * doc/tm.texi: Regenerate.
16445 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
16447 (do_jump): Use scalar_int_mode for local variables.
16448 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
16449 rather than a machine_mode.
16450 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
16451 (scompare_loc_descriptor_wide): Likewise.
16452 (scompare_loc_descriptor_narrow): Likewise.
16453 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
16455 * except.c (sjlj_emit_dispatch_table): Likewise.
16456 (expand_builtin_eh_copy_values): Likewise.
16457 * explow.c (convert_memory_address_addr_space_1): Likewise.
16458 Take a scalar_int_mode rather than a machine_mode.
16459 (convert_memory_address_addr_space): Take a scalar_int_mode rather
16460 than a machine_mode.
16461 (memory_address_addr_space): Use scalar_int_mode for local variables.
16462 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
16463 rather than a machine_mode.
16464 * expmed.c (mask_rtx): Likewise.
16465 (init_expmed_one_conv): Likewise.
16466 (expand_mult_highpart_adjust): Likewise.
16467 (extract_high_half): Likewise.
16468 (expmed_mult_highpart_optab): Likewise.
16469 (expmed_mult_highpart): Likewise.
16470 (expand_smod_pow2): Likewise.
16471 (expand_sdiv_pow2): Likewise.
16472 (emit_store_flag_int): Likewise.
16473 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
16475 (extract_low_bits): Likewise.
16476 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
16478 * expr.c (pieces_addr::adjust): Likewise.
16479 (can_store_by_pieces): Likewise.
16480 (store_by_pieces): Likewise.
16481 (clear_by_pieces_1): Likewise.
16482 (expand_expr_addr_expr_1): Likewise.
16483 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
16484 (expand_expr_real_1): Likewise.
16485 (try_casesi): Likewise.
16486 * final.c (shorten_branches): Likewise.
16487 * fold-const.c (fold_convert_const_int_from_fixed): Change the
16488 type of "mode" to machine_mode.
16489 * internal-fn.c (expand_arith_overflow_result_store): Take a
16490 scalar_int_mode rather than a machine_mode.
16491 (expand_mul_overflow): Use scalar_int_mode for local variables.
16492 * loop-doloop.c (doloop_modify): Likewise.
16493 (doloop_optimize): Likewise.
16494 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
16495 than a machine_mode.
16496 (expand_doubleword_shift_condmove): Likewise.
16497 (expand_doubleword_shift): Likewise.
16498 (expand_doubleword_clz): Likewise.
16499 (expand_doubleword_popcount): Likewise.
16500 (expand_doubleword_parity): Likewise.
16501 (expand_absneg_bit): Use scalar_int_mode for local variables.
16502 (prepare_float_lib_cmp): Likewise.
16503 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
16504 rather than a machine_mode.
16505 (convert_memory_address_addr_space): Likewise.
16506 (get_mode_bounds): Likewise.
16507 (get_address_mode): Return a scalar_int_mode rather than a
16509 * rtlanal.c (get_address_mode): Likewise.
16510 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
16511 than a machine_mode.
16512 * targhooks.c (default_mode_rep_extended): Likewise.
16513 (default_valid_pointer_mode): Likewise.
16514 (default_addr_space_valid_pointer_mode): Likewise.
16515 (default_eh_return_filter_mode): Return a scalar_int_mode rather
16516 than a machine_mode.
16517 (default_libgcc_cmp_return_mode): Likewise.
16518 (default_libgcc_shift_count_mode): Likewise.
16519 (default_unwind_word_mode): Likewise.
16520 (default_addr_space_pointer_mode): Likewise.
16521 (default_addr_space_address_mode): Likewise.
16522 * targhooks.h (default_eh_return_filter_mode): Likewise.
16523 (default_libgcc_cmp_return_mode): Likewise.
16524 (default_libgcc_shift_count_mode): Likewise.
16525 (default_unwind_word_mode): Likewise.
16526 (default_addr_space_pointer_mode): Likewise.
16527 (default_addr_space_address_mode): Likewise.
16528 (default_mode_rep_extended): Take a scalar_int_mode rather than
16530 (default_valid_pointer_mode): Likewise.
16531 (default_addr_space_valid_pointer_mode): Likewise.
16532 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
16534 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
16535 rather than a machine_mode.
16536 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
16537 for local variables.
16538 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16539 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
16540 than a machine_mode.
16542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16543 Alan Hayward <alan.hayward@arm.com>
16544 David Sherwood <david.sherwood@arm.com>
16546 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
16547 the mode argument to scalar_int_mode.
16548 (do_jump_by_parts_zero_rtx): Likewise.
16549 (do_jump_by_parts_equality_rtx): Likewise.
16550 (do_jump_by_parts_greater): Take a mode argument.
16551 (do_jump_by_parts_equality): Likewise.
16552 (do_jump_1): Update calls accordingly.
16554 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16555 Alan Hayward <alan.hayward@arm.com>
16556 David Sherwood <david.sherwood@arm.com>
16558 * is-a.h (safe_dyn_cast): New function.
16559 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
16560 (jump_table_for_label): Likewise.
16561 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
16562 instead of an rtx_insn *.
16563 (shorten_branches): Use dyn_cast instead of LABEL_P and
16564 JUMP_TABLE_DATA_P. Use jump_table_for_label and
16565 rtx_jump_table_data::get_data_mode.
16566 (final_scan_insn): Likewise.
16568 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16569 Alan Hayward <alan.hayward@arm.com>
16570 David Sherwood <david.sherwood@arm.com>
16572 * combine.c (try_combine): Use is_a <scalar_int_mode> when
16573 trying to combine a full-register integer set with a subreg
16576 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16577 Alan Hayward <alan.hayward@arm.com>
16578 David Sherwood <david.sherwood@arm.com>
16580 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
16581 that is always either address_mode or pointer_mode.
16583 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16584 Alan Hayward <alan.hayward@arm.com>
16585 David Sherwood <david.sherwood@arm.com>
16587 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
16588 when the two are known to be equal.
16590 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16591 Alan Hayward <alan.hayward@arm.com>
16592 David Sherwood <david.sherwood@arm.com>
16594 * simplify-rtx.c (simplify_const_unary_operation): Use
16595 is_a <scalar_int_mode> instead of checking for a nonzero
16596 precision. Forcibly convert op_mode to a scalar_int_mode
16597 in that case. More clearly differentiate the operand and
16598 result modes and use the former when deciding what the value
16599 of a count-bits operation should be. Use is_int_mode instead
16600 of checking for a MODE_INT. Remove redundant check for whether
16601 this mode has a zero precision.
16603 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16604 Alan Hayward <alan.hayward@arm.com>
16605 David Sherwood <david.sherwood@arm.com>
16607 * optabs.c (widen_leading): Change the type of the mode argument
16608 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
16609 (widen_bswap): Likewise.
16610 (expand_parity): Likewise.
16611 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
16612 (expand_ffs): Likewise.
16613 (epand_unop): Check for scalar integer modes before calling the
16616 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16617 Alan Hayward <alan.hayward@arm.com>
16618 David Sherwood <david.sherwood@arm.com>
16620 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
16622 (expand_expr_real_1): Update call accordingly.
16624 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16625 Alan Hayward <alan.hayward@arm.com>
16626 David Sherwood <david.sherwood@arm.com>
16628 * expmed.c (store_bit_field_using_insv): Add op0_mode and
16629 value_mode arguments. Use scalar_int_mode internally.
16630 (store_bit_field_1): Rename the new integer mode from imode
16631 to op0_mode and use it instead of GET_MODE (op0). Update calls
16632 to store_split_bit_field, store_bit_field_using_insv and
16633 store_fixed_bit_field.
16634 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
16635 Use scalar_int_mode internally. Use a bit count rather than a mode
16636 when calculating the largest bit size for get_best_mode.
16637 Update calls to store_split_bit_field and store_fixed_bit_field_1.
16638 (store_fixed_bit_field_1): Add mode and value_mode arguments.
16639 Remove assertion that OP0 has a scalar integer mode.
16640 (store_split_bit_field): Add op0_mode and value_mode arguments.
16641 Update calls to extract_fixed_bit_field.
16642 (extract_bit_field_using_extv): Add an op0_mode argument.
16643 Use scalar_int_mode internally.
16644 (extract_bit_field_1): Rename the new integer mode from imode to
16645 op0_mode and use it instead of GET_MODE (op0). Update calls to
16646 extract_split_bit_field, extract_bit_field_using_extv and
16647 extract_fixed_bit_field.
16648 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
16649 to extract_split_bit_field and extract_fixed_bit_field_1.
16650 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
16651 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
16652 on the target mode.
16653 (extract_split_bit_field): Add an op0_mode argument. Update call
16654 to extract_fixed_bit_field.
16656 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16657 Alan Hayward <alan.hayward@arm.com>
16658 David Sherwood <david.sherwood@arm.com>
16660 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
16661 * explow.c (hard_function_value): Likewise.
16662 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
16663 convert_to_mode call outside the loop.
16664 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
16665 for the mode iterator. Require the mode specified by max_pieces
16667 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
16669 (copy_blkmode_to_reg): Likewise.
16670 (set_storage_via_setmem): Likewise.
16671 * optabs.c (prepare_cmp_insn): Likewise.
16672 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16673 * stor-layout.c (finish_bitfield_representative): Likewise.
16675 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16676 Alan Hayward <alan.hayward@arm.com>
16677 David Sherwood <david.sherwood@arm.com>
16679 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
16680 * expr.c (convert_move): Use them.
16681 (convert_modes): Likewise.
16682 (store_expr_with_bounds): Likewise.
16684 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16685 Alan Hayward <alan.hayward@arm.com>
16686 David Sherwood <david.sherwood@arm.com>
16688 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
16689 parameter for the mode of "x". Remove the "known_x", "known_mode"
16690 and "known_ret" arguments. Change the type of the mode argument
16691 to scalar_int_mode.
16692 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
16693 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
16694 (reg_num_sign_bit_copies_for_combine): Likewise.
16695 * rtlanal.c (nonzero_bits1): Likewise.
16696 (num_sign_bit_copies1): Likewise.
16697 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
16698 (reg_num_sign_bit_copies_general): Likewise.
16699 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
16700 (reg_nonzero_bits_general): Likewise.
16702 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16703 Alan Hayward <alan.hayward@arm.com>
16704 David Sherwood <david.sherwood@arm.com>
16706 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
16707 than in subroutines. Return 1 for non-integer modes.
16708 (cached_num_sign_bit_copies): Change the type of the mode parameter
16709 to scalar_int_mode.
16710 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
16711 classes. Handle CONST_INT_P first and then check whether X also
16712 has a scalar integer mode. Check the same thing for inner registers
16713 of a SUBREG and for values that are being extended or truncated.
16715 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16716 Alan Hayward <alan.hayward@arm.com>
16717 David Sherwood <david.sherwood@arm.com>
16719 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
16720 in subroutines. Return the mode mask for non-integer modes.
16721 (cached_nonzero_bits): Change the type of the mode parameter
16722 to scalar_int_mode.
16723 (nonzero_bits1): Likewise. Remove early exit for other mode
16724 classes. Handle CONST_INT_P first and then check whether X
16725 also has a scalar integer mode.
16727 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16728 Alan Hayward <alan.hayward@arm.com>
16729 David Sherwood <david.sherwood@arm.com>
16731 * expr.c (widest_int_mode_for_size): Make the comment match the code.
16732 Return a scalar_int_mode and assert that the size is greater than
16734 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
16735 (op_by_pieces_d::op_by_pieces_d): Likewise.
16736 (op_by_pieces_d::run): Likewise.
16737 (can_store_by_pieces): Likewise.
16739 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16740 Alan Hayward <alan.hayward@arm.com>
16741 David Sherwood <david.sherwood@arm.com>
16743 * combine.c (extract_left_shift): Add a mode argument and update
16745 (make_compound_operation_int): Change the type of the mode parameter
16746 to scalar_int_mode and update the call to extract_left_shift.
16748 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16749 Alan Hayward <alan.hayward@arm.com>
16750 David Sherwood <david.sherwood@arm.com>
16752 * combine.c (simplify_and_const_int): Change the type of the mode
16753 parameter to scalar_int_mode.
16754 (simplify_and_const_int_1): Likewise. Update recursive call.
16756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16757 Alan Hayward <alan.hayward@arm.com>
16758 David Sherwood <david.sherwood@arm.com>
16760 * combine.c (simplify_compare_const): Check that the mode is a
16761 scalar_int_mode (rather than VOIDmode) before testing its
16763 (simplify_comparison): Move COMPARISON_P handling out of the
16764 loop and restrict the latter part of the loop to scalar_int_modes.
16765 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
16766 and when considering SUBREG_REGs. Use is_int_mode instead of
16767 checking GET_MODE_CLASS against MODE_INT.
16769 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16770 Alan Hayward <alan.hayward@arm.com>
16771 David Sherwood <david.sherwood@arm.com>
16773 * combine.c (try_widen_shift_mode): Move check for equal modes to...
16774 (simplify_shift_const_1): ...here. Use scalar_int_mode for
16775 shift_unit_mode and for modes involved in scalar shifts.
16777 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16778 Alan Hayward <alan.hayward@arm.com>
16779 David Sherwood <david.sherwood@arm.com>
16781 * combine.c (force_int_to_mode): New function, split out from...
16782 (force_to_mode): ...here. Keep xmode up-to-date and use it
16783 instead of GET_MODE (x).
16785 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16786 Alan Hayward <alan.hayward@arm.com>
16787 David Sherwood <david.sherwood@arm.com>
16789 * optabs-query.h (extraction_insn::struct_mode): Change type to
16790 opt_scalar_int_mode and update comment.
16791 (extraction_insn::field_mode): Change type to scalar_int_mode.
16792 (extraction_insn::pos_mode): Likewise.
16793 * combine.c (make_extraction): Update accordingly.
16794 * optabs-query.c (get_traditional_extraction_insn): Likewise.
16795 (get_optab_extraction_insn): Likewise.
16796 * recog.c (simplify_while_replacing): Likewise.
16797 * expmed.c (narrow_bit_field_mem): Change the type of the mode
16798 parameter to opt_scalar_int_mode.
16800 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16801 Alan Hayward <alan.hayward@arm.com>
16802 David Sherwood <david.sherwood@arm.com>
16804 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
16805 to a scalar_int_mode instead of a machine_mode.
16806 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
16807 (get_best_mode): Return a boolean and use a pointer argument to store
16808 the selected mode. Replace the limit mode parameter with a bit limit.
16809 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
16810 for the values returned by bit_field_mode_iterator::next_mode.
16811 (store_bit_field): Update call to get_best_mode.
16812 (store_fixed_bit_field): Likewise.
16813 (extract_fixed_bit_field): Likewise.
16814 * expr.c (optimize_bitfield_assignment_op): Likewise.
16815 * fold-const.c (optimize_bit_field_compare): Likewise.
16816 (fold_truth_andor_1): Likewise.
16817 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
16818 Update for new type of m_mode.
16819 (get_best_mode): As above.
16821 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16822 Alan Hayward <alan.hayward@arm.com>
16823 David Sherwood <david.sherwood@arm.com>
16825 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
16826 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
16827 (store_bit_field): Check is_a <scalar_int_mode> before calling
16828 strict_volatile_bitfield_p.
16829 (extract_bit_field): Likewise.
16831 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16832 Alan Hayward <alan.hayward@arm.com>
16833 David Sherwood <david.sherwood@arm.com>
16835 * target.def (cstore_mode): Return a scalar_int_mode.
16836 * doc/tm.texi: Regenerate.
16837 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
16838 * targhooks.h (default_cstore_mode): Likewise.
16839 * targhooks.c (default_cstore_mode): Likewise, using a forced
16841 * expmed.c (emit_cstore): Expect the target of the cstore to be
16844 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16845 Alan Hayward <alan.hayward@arm.com>
16846 David Sherwood <david.sherwood@arm.com>
16848 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
16850 (niter_desc): Likewise mode.
16851 (iv_analyze): Add a mode parameter.
16853 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
16854 and change its type to scalar_int_mode.
16855 * loop-iv.c: Update commentary at head of file.
16856 (iv_constant): Pass the mode paraeter before the rtx it describes
16857 and change its type to scalar_int_mode. Remove VOIDmode handling.
16858 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
16859 (iv_extend): Likewise.
16860 (shorten_into_mode): Likewise.
16861 (iv_add): Use scalar_int_mode.
16862 (iv_mult): Likewise.
16863 (iv_shift): Likewise.
16864 (canonicalize_iv_subregs): Likewise.
16865 (get_biv_step_1): Pass the outer_mode parameter before the rtx
16866 it describes and change its mode to scalar_int_mode. Also change
16867 the type of the returned inner_mode to scalar_int_mode.
16868 (get_biv_step): Likewise, turning outer_mode from a pointer
16869 into a direct parameter. Update call to get_biv_step_1.
16870 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
16871 iv_constant and get_biv_step.
16872 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
16873 and change its type to scalar_int_mode. Don't initialise iv->mode
16874 to VOIDmode and remove later checks for its still being VOIDmode.
16875 Update calls to iv_analyze_op and iv_analyze_expr. Check
16876 is_a <scalar_int_mode> when changing the mode under consideration.
16877 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
16878 Update call to iv_analyze_expr.
16879 (iv_analyze_op): Add a mode parameter. Reject subregs whose
16880 inner register is not also a scalar_int_mode. Update call to
16882 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
16883 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
16884 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
16885 separate mode class checks. Update calls to iv_analyze. Remove
16886 fix-up of VOIDmodes after iv_analyze_biv.
16887 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
16888 don't have a scalar_int_mode. Update call to biv_p.
16890 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16891 Alan Hayward <alan.hayward@arm.com>
16892 David Sherwood <david.sherwood@arm.com>
16894 * cfgexpand.c (convert_debug_memory_address): Use
16895 as_a <scalar_int_mode>.
16896 * combine.c (expand_compound_operation): Likewise.
16897 (make_extraction): Likewise.
16898 (change_zero_ext): Likewise.
16899 (simplify_comparison): Likewise.
16900 * cse.c (cse_insn): Likewise.
16901 * dwarf2out.c (minmax_loc_descriptor): Likewise.
16902 (mem_loc_descriptor): Likewise.
16903 (loc_descriptor): Likewise.
16904 * expmed.c (init_expmed_one_mode): Likewise.
16905 (synth_mult): Likewise.
16906 (emit_store_flag_1): Likewise.
16907 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
16908 of a comparison with size.
16909 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
16910 (reduce_to_bit_field_precision): Likewise.
16911 * function.c (expand_function_end): Likewise.
16912 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
16913 * loop-doloop.c (doloop_modify): Likewise.
16914 * optabs.c (expand_binop): Likewise.
16915 (expand_unop): Likewise.
16916 (expand_copysign_absneg): Likewise.
16917 (prepare_cmp_insn): Likewise.
16918 (maybe_legitimize_operand): Likewise.
16919 * recog.c (const_scalar_int_operand): Likewise.
16920 * rtlanal.c (get_address_mode): Likewise.
16921 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16922 (simplify_cond_clz_ctz): Likewise.
16923 * tree-nested.c (get_nl_goto_field): Likewise.
16924 * tree.c (build_vector_type_for_mode): Likewise.
16925 * var-tracking.c (use_narrower_mode): Likewise.
16927 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16928 Alan Hayward <alan.hayward@arm.com>
16929 David Sherwood <david.sherwood@arm.com>
16931 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
16932 * builtins.c (expand_builtin_signbit): Use it.
16933 * cfgexpand.c (expand_debug_expr): Likewise.
16934 * dojump.c (do_jump): Likewise.
16935 (do_compare_and_jump): Likewise.
16936 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
16937 * expmed.c (make_tree): Likewise.
16938 * expr.c (expand_expr_real_2): Likewise.
16939 (expand_expr_real_1): Likewise.
16940 (try_casesi): Likewise.
16941 * fold-const-call.c (fold_const_call_ss): Likewise.
16942 * fold-const.c (unextend): Likewise.
16943 (extract_muldiv_1): Likewise.
16944 (fold_single_bit_test): Likewise.
16945 (native_encode_int): Likewise.
16946 (native_encode_string): Likewise.
16947 (native_interpret_int): Likewise.
16948 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
16949 * internal-fn.c (expand_addsub_overflow): Likewise.
16950 (expand_neg_overflow): Likewise.
16951 (expand_mul_overflow): Likewise.
16952 (expand_arith_overflow): Likewise.
16953 * match.pd: Likewise.
16954 * stor-layout.c (layout_type): Likewise.
16955 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16956 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
16957 * tree-ssanames.c (get_range_info): Likewise.
16958 * tree-switch-conversion.c (array_value_type) Likewise.
16959 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
16960 (vect_recog_divmod_pattern): Likewise.
16961 (vect_recog_mixed_size_cond_pattern): Likewise.
16962 * tree-vrp.c (extract_range_basic): Likewise.
16963 (simplify_float_conversion_using_ranges): Likewise.
16964 * tree.c (int_fits_type_p): Likewise.
16965 * ubsan.c (instrument_bool_enum_load): Likewise.
16966 * varasm.c (mergeable_string_section): Likewise.
16967 (narrowing_initializer_constant_valid_p): Likewise.
16968 (output_constant): Likewise.
16970 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16971 Alan Hayward <alan.hayward@arm.com>
16972 David Sherwood <david.sherwood@arm.com>
16974 * machmode.h (NARROWEST_INT_MODE): New macro.
16975 * expr.c (alignment_for_piecewise_move): Use it instead of
16976 GET_CLASS_NARROWEST_MODE (MODE_INT).
16977 (push_block): Likewise.
16978 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
16980 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16982 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16983 Alan Hayward <alan.hayward@arm.com>
16984 David Sherwood <david.sherwood@arm.com>
16986 * postreload.c (move2add_valid_value_p): Change the type of the
16987 mode parameter to scalar_int_mode.
16988 (move2add_use_add2_insn): Add a mode parameter and use it instead
16990 (move2add_use_add3_insn): Likewise.
16991 (reload_cse_move2add): Update accordingly.
16993 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16994 Alan Hayward <alan.hayward@arm.com>
16995 David Sherwood <david.sherwood@arm.com>
16997 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
16999 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
17000 * optabs.c (expand_unop): Likewise.
17002 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17003 Alan Hayward <alan.hayward@arm.com>
17004 David Sherwood <david.sherwood@arm.com>
17006 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
17007 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
17008 (popcount_loc_descriptor): Likewise.
17009 (bswap_loc_descriptor): Likewise.
17010 (rotate_loc_descriptor): Likewise.
17011 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
17012 calling the functions above.
17014 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17015 Alan Hayward <alan.hayward@arm.com>
17016 David Sherwood <david.sherwood@arm.com>
17018 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
17020 (try_combine): Likewise.
17021 (simplify_if_then_else): Likewise.
17022 * cse.c (cse_insn): Likewise.
17023 * dwarf2out.c (mem_loc_descriptor): Likewise.
17024 * emit-rtl.c (gen_lowpart_common): Likewise.
17025 * simplify-rtx.c (simplify_truncation): Likewise.
17026 (simplify_binary_operation_1): Likewise.
17027 (simplify_const_relational_operation): Likewise.
17028 (simplify_ternary_operation): Likewise.
17029 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
17031 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17032 Alan Hayward <alan.hayward@arm.com>
17033 David Sherwood <david.sherwood@arm.com>
17035 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
17036 * reload.c (push_reload): Likewise.
17037 (find_reloads): Likewise.
17039 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17040 Alan Hayward <alan.hayward@arm.com>
17041 David Sherwood <david.sherwood@arm.com>
17043 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
17044 (make_compound_operation_int): Likewise.
17045 (change_zero_ext): Likewise.
17046 * expr.c (convert_move): Likewise.
17047 (convert_modes): Likewise.
17048 * fwprop.c (forward_propagate_subreg): Likewise.
17049 * loop-iv.c (get_biv_step_1): Likewise.
17050 * optabs.c (widen_operand): Likewise.
17051 * postreload.c (move2add_valid_value_p): Likewise.
17052 * recog.c (simplify_while_replacing): Likewise.
17053 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17054 (simplify_binary_operation_1): Likewise. Remove redundant
17055 mode equality check.
17057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17058 Alan Hayward <alan.hayward@arm.com>
17059 David Sherwood <david.sherwood@arm.com>
17061 * combine.c (combine_simplify_rtx): Add checks for
17062 is_a <scalar_int_mode>.
17063 (simplify_if_then_else): Likewise.
17064 (make_field_assignment): Likewise.
17065 (simplify_comparison): Likewise.
17066 * ifcvt.c (noce_try_bitop): Likewise.
17067 * loop-invariant.c (canonicalize_address_mult): Likewise.
17068 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17070 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17071 Alan Hayward <alan.hayward@arm.com>
17072 David Sherwood <david.sherwood@arm.com>
17074 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
17075 is_a <scalar_int_mode> instead of != BLKmode.
17077 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17078 Alan Hayward <alan.hayward@arm.com>
17079 David Sherwood <david.sherwood@arm.com>
17081 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
17082 instead of != VOIDmode.
17083 * combine.c (if_then_else_cond): Likewise.
17084 (change_zero_ext): Likewise.
17085 * dwarf2out.c (mem_loc_descriptor): Likewise.
17086 (loc_descriptor): Likewise.
17087 * rtlanal.c (canonicalize_condition): Likewise.
17088 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
17090 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17091 Alan Hayward <alan.hayward@arm.com>
17092 David Sherwood <david.sherwood@arm.com>
17094 * simplify-rtx.c (simplify_binary_operation_1): Use
17095 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
17097 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17098 Alan Hayward <alan.hayward@arm.com>
17099 David Sherwood <david.sherwood@arm.com>
17101 * wide-int.h (int_traits<unsigned char>) New class.
17102 (int_traits<unsigned short>) Likewise.
17103 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
17104 Use GET_MODE_UNIT_PRECISION and remove redundant test for
17106 * combine.c (set_nonzero_bits_and_sign_copies): Use
17107 is_a <scalar_int_mode>.
17108 (find_split_point): Likewise.
17109 (combine_simplify_rtx): Likewise.
17110 (simplify_logical): Likewise.
17111 (expand_compound_operation): Likewise.
17112 (expand_field_assignment): Likewise.
17113 (make_compound_operation): Likewise.
17114 (extended_count): Likewise.
17115 (change_zero_ext): Likewise.
17116 (simplify_comparison): Likewise.
17117 * dwarf2out.c (scompare_loc_descriptor): Likewise.
17118 (ucompare_loc_descriptor): Likewise.
17119 (minmax_loc_descriptor): Likewise.
17120 (mem_loc_descriptor): Likewise.
17121 (loc_descriptor): Likewise.
17122 * expmed.c (init_expmed_one_mode): Likewise.
17123 * lra-constraints.c (lra_constraint_offset): Likewise.
17124 * optabs.c (prepare_libcall_arg): Likewise.
17125 * postreload.c (move2add_note_store): Likewise.
17126 * reload.c (operands_match_p): Likewise.
17127 * rtl.h (load_extend_op): Likewise.
17128 * rtlhooks.c (gen_lowpart_general): Likewise.
17129 * simplify-rtx.c (simplify_truncation): Likewise.
17130 (simplify_unary_operation_1): Likewise.
17131 (simplify_binary_operation_1): Likewise.
17132 (simplify_const_binary_operation): Likewise.
17133 (simplify_const_relational_operation): Likewise.
17134 (simplify_subreg): Likewise.
17135 * stor-layout.c (bitwise_mode_for_mode): Likewise.
17136 * var-tracking.c (adjust_mems): Likewise.
17137 (prepare_call_arguments): Likewise.
17139 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17140 Alan Hayward <alan.hayward@arm.com>
17141 David Sherwood <david.sherwood@arm.com>
17143 * machmode.h (is_int_mode): New fuction.
17144 * combine.c (find_split_point): Use it.
17145 (combine_simplify_rtx): Likewise.
17146 (simplify_if_then_else): Likewise.
17147 (simplify_set): Likewise.
17148 (simplify_shift_const_1): Likewise.
17149 (simplify_comparison): Likewise.
17150 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
17151 * cse.c (notreg_cost): Likewise.
17152 (cse_insn): Likewise.
17153 * cselib.c (cselib_lookup_1): Likewise.
17154 * dojump.c (do_jump_1): Likewise.
17155 (do_compare_rtx_and_jump): Likewise.
17156 * dse.c (get_call_args): Likewise.
17157 * dwarf2out.c (rtl_for_decl_init): Likewise.
17158 (native_encode_initializer): Likewise.
17159 * expmed.c (emit_store_flag_1): Likewise.
17160 (emit_store_flag): Likewise.
17161 * expr.c (convert_modes): Likewise.
17162 (store_field): Likewise.
17163 (expand_expr_real_1): Likewise.
17164 * fold-const.c (fold_read_from_constant_string): Likewise.
17165 * gimple-ssa-sprintf.c (get_format_string): Likewise.
17166 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
17167 * optabs.c (expand_binop): Likewise.
17168 (expand_unop): Likewise.
17169 (expand_abs_nojump): Likewise.
17170 (expand_one_cmpl_abs_nojump): Likewise.
17171 * simplify-rtx.c (mode_signbit_p): Likewise.
17172 (val_signbit_p): Likewise.
17173 (val_signbit_known_set_p): Likewise.
17174 (val_signbit_known_clear_p): Likewise.
17175 (simplify_relational_operation_1): Likewise.
17176 * tree.c (vector_type_mode): Likewise.
17178 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17179 Alan Hayward <alan.hayward@arm.com>
17180 David Sherwood <david.sherwood@arm.com>
17182 * machmode.h (smallest_mode_for_size): Fix formatting.
17183 (smallest_int_mode_for_size): New function.
17184 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
17185 instead of smallest_mode_for_size.
17186 * combine.c (make_extraction): Likewise.
17187 * config/arc/arc.c (arc_expand_movmem): Likewise.
17188 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
17189 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
17190 * config/s390/s390.c (s390_expand_insv): Likewise.
17191 * config/sparc/sparc.c (assign_int_registers): Likewise.
17192 * config/spu/spu.c (spu_function_value): Likewise.
17193 (spu_function_arg): Likewise.
17194 * coverage.c (get_gcov_type): Likewise.
17195 (get_gcov_unsigned_t): Likewise.
17196 * dse.c (find_shift_sequence): Likewise.
17197 * expmed.c (store_bit_field_1): Likewise.
17198 * expr.c (convert_move): Likewise.
17199 (store_field): Likewise.
17200 * internal-fn.c (expand_arith_overflow): Likewise.
17201 * optabs-query.c (get_best_extraction_insn): Likewise.
17202 * optabs.c (expand_twoval_binop_libfunc): Likewise.
17203 * stor-layout.c (layout_type): Likewise.
17204 (initialize_sizetypes): Likewise.
17205 * targhooks.c (default_get_mask_mode): Likewise.
17206 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
17208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17209 Alan Hayward <alan.hayward@arm.com>
17210 David Sherwood <david.sherwood@arm.com>
17212 * machmode.h (opt_mode::else_blk): New function.
17213 (int_mode_for_mode): Declare.
17214 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
17215 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
17217 * cfgexpand.c (expand_debug_expr): Likewise.
17218 * combine.c (gen_lowpart_or_truncate): Likewise.
17219 (gen_lowpart_for_combine): Likewise.
17220 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
17221 * config/avr/avr.c (avr_to_int_mode): Likewise.
17222 (avr_out_plus_1): Likewise.
17223 (avr_out_plus): Likewise.
17224 (avr_out_round): Likewise.
17225 * config/i386/i386.c (ix86_split_to_parts): Likewise.
17226 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
17227 (s390_expand_vcond): Likewise.
17228 * config/spu/spu.c (spu_split_immediate): Likewise.
17229 (spu_expand_mov): Likewise.
17230 * dse.c (get_stored_val): Likewise.
17231 * expmed.c (store_bit_field_1): Likewise.
17232 (convert_extracted_bit_field): Use int_mode_for_mode instead of
17234 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
17235 (extract_low_bits): Likewise.
17236 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
17237 handling rather than repeating the check.
17238 (emit_group_store): Likewise.
17239 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
17240 * optabs.c (expand_absneg_bit): Likewise.
17241 (expand_copysign_absneg): Likewise.
17242 (expand_copysign_bit): Likewise.
17243 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
17244 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
17245 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
17246 * var-tracking.c (prepare_call_arguments): Likewise.
17247 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
17248 int_mode_for_mode instead of mode_for_size.
17249 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
17251 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17252 Alan Hayward <alan.hayward@arm.com>
17253 David Sherwood <david.sherwood@arm.com>
17255 * machmode.h (int_mode_for_size): New function.
17256 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
17257 instead of mode_for_size.
17258 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
17260 * combine.c (expand_field_assignment): Use int_mode_for_size
17261 instead of mode_for_size.
17262 (make_extraction): Likewise.
17263 (simplify_shift_const_1): Likewise.
17264 (simplify_comparison): Likewise.
17265 * dojump.c (do_jump): Likewise.
17266 * dwarf2out.c (mem_loc_descriptor): Likewise.
17267 * emit-rtl.c (init_derived_machine_modes): Likewise.
17268 * expmed.c (flip_storage_order): Likewise.
17269 (convert_extracted_bit_field): Likewise.
17270 * expr.c (copy_blkmode_from_reg): Likewise.
17271 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
17272 * internal-fn.c (expand_mul_overflow): Likewise.
17273 * lower-subreg.c (simple_move): Likewise.
17274 * optabs-libfuncs.c (init_optabs): Likewise.
17275 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17276 * tree.c (vector_type_mode): Likewise.
17277 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
17278 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
17279 * tree-vect-generic.c (expand_vector_parallel): Likewise.
17280 * tree-vect-stmts.c (vectorizable_load): Likewise.
17281 (vectorizable_store): Likewise.
17283 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17284 Alan Hayward <alan.hayward@arm.com>
17285 David Sherwood <david.sherwood@arm.com>
17287 * coretypes.h (pod_mode): New type.
17288 (scalar_int_mode_pod): New typedef.
17289 * machmode.h (pod_mode): New class.
17290 (int_n_data_t::m): Change type to scalar_int_mode_pod.
17291 * genmodes.c (emit_mode_int_n): Update accordingly.
17292 * lower-subreg.h (target_lower_subreg): Change type to
17293 scalar_int_mode_pod.
17294 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
17295 scalar_int_mode_pod.
17297 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17298 Alan Hayward <alan.hayward@arm.com>
17299 David Sherwood <david.sherwood@arm.com>
17301 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
17302 machine_mode to scalar_int_mode.
17303 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
17304 (rs6000_option_override_internal): Remove cast to int.
17305 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
17306 machine_mode to scalar_int_mode.
17307 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
17308 (rs6000_option_override_internal): Remove cast to int.
17309 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
17310 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
17312 * config/s390/s390.c (s390_expand_builtin): Likewise.
17313 * coretypes.h (scalar_int_mode): New type.
17314 (opt_scalar_int_mode): New typedef.
17315 * machmode.h (scalar_int_mode): New class.
17316 (scalar_int_mode::includes_p): New function.
17317 (byte_mode): Change type to scalar_int_mode.
17318 (word_mode): Likewise.
17319 (ptr_mode): Likewise.
17320 * emit-rtl.c (byte_mode): Likewise.
17321 (word_mode): Likewise.
17322 (ptr_mode): Likewise.
17323 (init_derived_machine_modes): Update accordingly.
17324 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
17325 and MODE_PARTIAL_INT.
17326 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
17327 opt_scalar_int_mode.
17329 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17330 Alan Hayward <alan.hayward@arm.com>
17331 David Sherwood <david.sherwood@arm.com>
17333 * target.def (libgcc_floating_mode_supported_p): Take a
17335 * doc/tm.texi: Regenerate.
17336 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
17338 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
17339 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
17342 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17343 Alan Hayward <alan.hayward@arm.com>
17344 David Sherwood <david.sherwood@arm.com>
17346 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
17347 * doc/tm.texi: Regenerate.
17348 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
17349 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
17350 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
17351 * targhooks.h (default_floatn_mode): Likewise.
17352 * targhooks.c (default_floatn_mode): Likewise.
17353 * tree.c (build_common_tree_nodes): Update accordingly.
17355 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17356 Alan Hayward <alan.hayward@arm.com>
17357 David Sherwood <david.sherwood@arm.com>
17359 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
17360 (mode_iterator::iterate_p): Likewise.
17361 (mode_iterator::get_wider): Likewise.
17362 * expr.c (init_expr_target): Use opt_scalar_float_mode.
17364 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17365 Alan Hayward <alan.hayward@arm.com>
17366 David Sherwood <david.sherwood@arm.com>
17368 * coretypes.h (opt_scalar_float_mode): New typedef.
17369 * machmode.h (float_mode_for_size): New function.
17370 * emit-rtl.c (double_mode): Delete.
17371 (init_emit_once): Use float_mode_for_size.
17372 * stor-layout.c (layout_type): Likewise.
17373 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
17375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17376 Alan Hayward <alan.hayward@arm.com>
17377 David Sherwood <david.sherwood@arm.com>
17379 * output.h (assemble_real): Take a scalar_float_mode.
17380 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
17381 * config/arm/arm.md (consttable_4): Likewise.
17382 (consttable_8): Likewise.
17383 (consttable_16): Likewise.
17384 * config/mips/mips.md (consttable_float): Likewise.
17385 * config/s390/s390.c (s390_output_pool_entry): Likewise.
17386 * varasm.c (assemble_real): Take a scalar_float_mode.
17387 (output_constant_pool_2): Update accordingly.
17388 (output_constant): Likewise.
17390 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17391 Alan Hayward <alan.hayward@arm.com>
17392 David Sherwood <david.sherwood@arm.com>
17394 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
17395 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
17396 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
17397 (native_encode_real): Likewise.
17398 (native_interpret_real): Likewise.
17399 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
17400 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17402 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17403 Alan Hayward <alan.hayward@arm.com>
17404 David Sherwood <david.sherwood@arm.com>
17406 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
17407 <scalar_float_mode>. Simplify.
17408 (gen_extend_conv_libfunc): Likewise.
17410 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17411 Alan Hayward <alan.hayward@arm.com>
17412 David Sherwood <david.sherwood@arm.com>
17414 * coretypes.h (scalar_float_mode): New type.
17415 * machmode.h (mode_traits::from_int): Use machine_mode if
17416 USE_ENUM_MODES is defined.
17417 (is_a): New function.
17419 (dyn_cast): Likewise.
17420 (scalar_float_mode): New class.
17421 (scalar_float_mode::includes_p): New function.
17422 (is_float_mode): Likewise.
17423 * gdbhooks.py (MachineModePrinter): New class.
17424 (build_pretty_printer): Use it for scalar_float_mode.
17425 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
17426 (format_helper::format_helper): Turn into a template.
17427 * genmodes.c (get_mode_class): New function.
17428 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
17429 or machine_mode if none.
17430 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
17431 as_a <scalar_float_mode>.
17432 * dwarf2out.c (mem_loc_descriptor): Likewise.
17433 (insert_float): Likewise.
17434 (add_const_value_attribute): Likewise.
17435 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17436 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
17437 (expand_unop): Update accordingly.
17438 (expand_abs_nojump): Likewise.
17439 (expand_copysign_absneg): Take a scalar_float_mode.
17440 (expand_copysign_bit): Likewise.
17441 (expand_copysign): Update accordingly.
17443 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17444 Alan Hayward <alan.hayward@arm.com>
17445 David Sherwood <david.sherwood@arm.com>
17447 * coretypes.h (opt_mode): New class.
17448 * machmode.h (opt_mode): Likewise.
17449 (opt_mode::else_void): New function.
17450 (opt_mode::require): Likewise.
17451 (opt_mode::exists): Likewise.
17452 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
17453 (GET_MODE_2XWIDER_MODE): Likewise.
17454 (mode_iterator::get_wider): Update accordingly.
17455 (mode_iterator::get_2xwider): Likewise.
17456 (mode_iterator::get_known_wider): Likewise, turning into a template.
17457 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
17458 forcing a wider mode to exist.
17459 * config/cr16/cr16.h (LONG_REG_P): Likewise.
17460 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17461 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
17462 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
17463 * lower-subreg.c (init_lower_subreg): Likewise.
17464 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
17465 on the final iteration.
17466 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
17467 a wider mode exists before asking for a move pattern.
17468 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
17469 forcing a wider mode to exist.
17470 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
17471 returning false if no such mode exists.
17472 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
17473 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
17474 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
17475 Avoid checking for a MODE_INT if we already know the mode is not a
17477 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
17478 forcing a wider mode to exist.
17479 (expmed_mult_highpart_optab): Likewise.
17480 (expmed_mult_highpart): Likewise.
17481 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
17483 * lto-streamer-in.c (lto_input_mode_table): Likewise.
17484 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
17485 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
17486 * internal-fn.c (expand_mul_overflow): Update use of
17487 GET_MODE_2XWIDER_MODE.
17488 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17489 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
17490 GET_MODE_WIDER_MODE.
17491 (convert_plusminus_to_widen): Likewise.
17492 * tree-switch-conversion.c (array_value_type): Likewise.
17493 * var-tracking.c (emit_note_insn_var_location): Likewise.
17494 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17495 Return false inside rather than outside the loop if no wider mode
17497 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
17498 and GET_MODE_2XWIDER_MODE
17499 (can_compare_p): Use else_void.
17500 * gdbhooks.py (OptMachineModePrinter): New class.
17501 (build_pretty_printer): Use it for opt_mode.
17503 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17504 Alan Hayward <alan.hayward@arm.com>
17505 David Sherwood <david.sherwood@arm.com>
17507 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
17508 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
17510 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17511 Alan Hayward <alan.hayward@arm.com>
17512 David Sherwood <david.sherwood@arm.com>
17514 * machmode.h (mode_traits): New structure.
17515 (get_narrowest_mode): New function.
17516 (mode_iterator::start): Likewise.
17517 (mode_iterator::iterate_p): Likewise.
17518 (mode_iterator::get_wider): Likewise.
17519 (mode_iterator::get_known_wider): Likewise.
17520 (mode_iterator::get_2xwider): Likewise.
17521 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
17522 (FOR_EACH_MODE): Likewise.
17523 (FOR_EACH_MODE_FROM): Likewise.
17524 (FOR_EACH_MODE_UNTIL): Likewise.
17525 (FOR_EACH_WIDER_MODE): Likewise.
17526 (FOR_EACH_2XWIDER_MODE): Likewise.
17527 * builtins.c (expand_builtin_strlen): Use new mode iterators.
17528 * combine.c (simplify_comparison): Likewise
17529 * config/i386/i386.c (type_natural_mode): Likewise.
17530 * cse.c (cse_insn): Likewise.
17531 * dse.c (find_shift_sequence): Likewise.
17532 * emit-rtl.c (init_derived_machine_modes): Likewise.
17533 (init_emit_once): Likewise.
17534 * explow.c (hard_function_value): Likewise.
17535 * expmed.c (extract_fixed_bit_field_1): Likewise.
17536 (extract_bit_field_1): Likewise.
17537 (expand_divmod): Likewise.
17538 (emit_store_flag_1): Likewise.
17539 * expr.c (init_expr_target): Likewise.
17540 (convert_move): Likewise.
17541 (alignment_for_piecewise_move): Likewise.
17542 (widest_int_mode_for_size): Likewise.
17543 (emit_block_move_via_movmem): Likewise.
17544 (copy_blkmode_to_reg): Likewise.
17545 (set_storage_via_setmem): Likewise.
17546 (compress_float_constant): Likewise.
17547 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17548 * optabs-query.c (get_best_extraction_insn): Likewise.
17549 * optabs.c (expand_binop): Likewise.
17550 (expand_twoval_unop): Likewise.
17551 (expand_twoval_binop): Likewise.
17552 (widen_leading): Likewise.
17553 (widen_bswap): Likewise.
17554 (expand_parity): Likewise.
17555 (expand_unop): Likewise.
17556 (prepare_cmp_insn): Likewise.
17557 (prepare_float_lib_cmp): Likewise.
17558 (expand_float): Likewise.
17559 (expand_fix): Likewise.
17560 (expand_sfix_optab): Likewise.
17561 * postreload.c (move2add_use_add2_insn): Likewise.
17562 * reg-stack.c (reg_to_stack): Likewise.
17563 * reginfo.c (choose_hard_reg_mode): Likewise.
17564 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17565 * stor-layout.c (mode_for_size): Likewise.
17566 (smallest_mode_for_size): Likewise.
17567 (mode_for_vector): Likewise.
17568 (finish_bitfield_representative): Likewise.
17569 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
17570 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
17571 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17572 * var-tracking.c (prepare_call_arguments): Likewise.
17574 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17575 Alan Hayward <alan.hayward@arm.com>
17576 David Sherwood <david.sherwood@arm.com>
17578 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
17579 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
17580 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
17581 * machmode.h (mode_size): Move earlier in file.
17582 (mode_precision): Likewise.
17583 (mode_inner): Likewise.
17584 (mode_nunits): Likewise.
17585 (mode_unit_size): Likewise.
17586 (unit_unit_precision): Likewise.
17587 (mode_wider): Likewise.
17588 (mode_2xwider): Likewise.
17589 (machine_mode): New class.
17590 (mode_to_bytes): New function.
17591 (mode_to_bits): Likewise.
17592 (mode_to_precision): Likewise.
17593 (mode_to_inner): Likewise.
17594 (mode_to_unit_size): Likewise.
17595 (mode_to_unit_precision): Likewise.
17596 (mode_to_nunits): Likewise.
17597 (GET_MODE_SIZE): Use mode_to_bytes.
17598 (GET_MODE_BITSIZE): Use mode_to_bits.
17599 (GET_MODE_PRECISION): Use mode_to_precision.
17600 (GET_MODE_INNER): Use mode_to_inner.
17601 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
17602 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
17603 (GET_MODE_NUNITS): Use mode_to_nunits.
17604 * system.h (ALWAYS_INLINE): New macro.
17605 * config/powerpcspe/powerpcspe-c.c
17606 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
17607 int for arg1_mode and arg2_mode.
17609 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17610 Alan Hayward <alan.hayward@arm.com>
17611 David Sherwood <david.sherwood@arm.com>
17613 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17614 Prefix mode names with E_ in case statements.
17615 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17616 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
17617 (aarch64_split_simd_move): Likewise.
17618 (aarch64_gen_storewb_pair): Likewise.
17619 (aarch64_gen_loadwb_pair): Likewise.
17620 (aarch64_gen_store_pair): Likewise.
17621 (aarch64_gen_load_pair): Likewise.
17622 (aarch64_get_condition_code_1): Likewise.
17623 (aarch64_constant_pool_reload_icode): Likewise.
17624 (get_rsqrte_type): Likewise.
17625 (get_rsqrts_type): Likewise.
17626 (get_recpe_type): Likewise.
17627 (get_recps_type): Likewise.
17628 (aarch64_gimplify_va_arg_expr): Likewise.
17629 (aarch64_simd_container_mode): Likewise.
17630 (aarch64_emit_load_exclusive): Likewise.
17631 (aarch64_emit_store_exclusive): Likewise.
17632 (aarch64_expand_compare_and_swap): Likewise.
17633 (aarch64_gen_atomic_cas): Likewise.
17634 (aarch64_emit_bic): Likewise.
17635 (aarch64_emit_atomic_swap): Likewise.
17636 (aarch64_emit_atomic_load_op): Likewise.
17637 (aarch64_evpc_trn): Likewise.
17638 (aarch64_evpc_uzp): Likewise.
17639 (aarch64_evpc_zip): Likewise.
17640 (aarch64_evpc_ext): Likewise.
17641 (aarch64_evpc_rev): Likewise.
17642 (aarch64_evpc_dup): Likewise.
17643 (aarch64_gen_ccmp_first): Likewise.
17644 (aarch64_gen_ccmp_next): Likewise.
17645 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17646 (alpha_emit_xfloating_libcall): Likewise.
17647 (emit_insxl): Likewise.
17648 (alpha_arg_type): Likewise.
17649 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
17650 (arc_preferred_simd_mode): Likewise.
17651 (arc_secondary_reload): Likewise.
17652 (get_arc_condition_code): Likewise.
17653 (arc_print_operand): Likewise.
17654 (arc_legitimate_constant_p): Likewise.
17655 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17656 * config/arc/arc.md (casesi_load): Likewise.
17657 (casesi_compact_jump): Likewise.
17658 * config/arc/predicates.md (proper_comparison_operator): Likewise.
17659 (cc_use_register): Likewise.
17660 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17661 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17662 (arm_init_iwmmxt_builtins): Likewise.
17663 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
17664 (neon_expand_vector_init): Likewise.
17665 (arm_attr_length_move_neon): Likewise.
17666 (maybe_get_arm_condition_code): Likewise.
17667 (arm_emit_vector_const): Likewise.
17668 (arm_preferred_simd_mode): Likewise.
17669 (arm_output_iwmmxt_tinsr): Likewise.
17670 (thumb1_output_casesi): Likewise.
17671 (thumb2_output_casesi): Likewise.
17672 (arm_emit_load_exclusive): Likewise.
17673 (arm_emit_store_exclusive): Likewise.
17674 (arm_expand_compare_and_swap): Likewise.
17675 (arm_evpc_neon_vuzp): Likewise.
17676 (arm_evpc_neon_vzip): Likewise.
17677 (arm_evpc_neon_vrev): Likewise.
17678 (arm_evpc_neon_vtrn): Likewise.
17679 (arm_evpc_neon_vext): Likewise.
17680 (arm_validize_comparison): Likewise.
17681 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
17682 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
17683 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
17684 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
17685 (c6x_preferred_simd_mode): Likewise.
17686 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
17687 (epiphany_rtx_costs): Likewise.
17688 * config/epiphany/predicates.md (proper_comparison_operator):
17690 * config/frv/frv.c (condexec_memory_operand): Likewise.
17691 (frv_emit_move): Likewise.
17692 (output_move_single): Likewise.
17693 (output_condmove_single): Likewise.
17694 (frv_hard_regno_mode_ok): Likewise.
17695 (frv_matching_accg_mode): Likewise.
17696 * config/h8300/h8300.c (split_adds_subs): Likewise.
17697 (h8300_rtx_costs): Likewise.
17698 (h8300_print_operand): Likewise.
17699 (compute_mov_length): Likewise.
17700 (output_logical_op): Likewise.
17701 (compute_logical_op_length): Likewise.
17702 (compute_logical_op_cc): Likewise.
17703 (h8300_shift_needs_scratch_p): Likewise.
17704 (output_a_shift): Likewise.
17705 (compute_a_shift_length): Likewise.
17706 (compute_a_shift_cc): Likewise.
17707 (expand_a_rotate): Likewise.
17708 (output_a_rotate): Likewise.
17709 * config/i386/i386.c (classify_argument): Likewise.
17710 (function_arg_advance_32): Likewise.
17711 (function_arg_32): Likewise.
17712 (function_arg_64): Likewise.
17713 (function_value_64): Likewise.
17714 (ix86_gimplify_va_arg): Likewise.
17715 (ix86_legitimate_constant_p): Likewise.
17716 (put_condition_code): Likewise.
17717 (split_double_mode): Likewise.
17718 (ix86_avx256_split_vector_move_misalign): Likewise.
17719 (ix86_expand_vector_logical_operator): Likewise.
17720 (ix86_split_idivmod): Likewise.
17721 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
17722 (ix86_build_const_vector): Likewise.
17723 (ix86_build_signbit_mask): Likewise.
17724 (ix86_match_ccmode): Likewise.
17725 (ix86_cc_modes_compatible): Likewise.
17726 (ix86_expand_branch): Likewise.
17727 (ix86_expand_sse_cmp): Likewise.
17728 (ix86_expand_sse_movcc): Likewise.
17729 (ix86_expand_int_sse_cmp): Likewise.
17730 (ix86_expand_vec_perm_vpermi2): Likewise.
17731 (ix86_expand_vec_perm): Likewise.
17732 (ix86_expand_sse_unpack): Likewise.
17733 (ix86_expand_int_addcc): Likewise.
17734 (ix86_split_to_parts): Likewise.
17735 (ix86_vectorize_builtin_gather): Likewise.
17736 (ix86_vectorize_builtin_scatter): Likewise.
17737 (avx_vpermilp_parallel): Likewise.
17738 (inline_memory_move_cost): Likewise.
17739 (ix86_tieable_integer_mode_p): Likewise.
17740 (x86_maybe_negate_const_int): Likewise.
17741 (ix86_expand_vector_init_duplicate): Likewise.
17742 (ix86_expand_vector_init_one_nonzero): Likewise.
17743 (ix86_expand_vector_init_one_var): Likewise.
17744 (ix86_expand_vector_init_concat): Likewise.
17745 (ix86_expand_vector_init_interleave): Likewise.
17746 (ix86_expand_vector_init_general): Likewise.
17747 (ix86_expand_vector_set): Likewise.
17748 (ix86_expand_vector_extract): Likewise.
17749 (emit_reduc_half): Likewise.
17750 (ix86_emit_i387_round): Likewise.
17751 (ix86_mangle_type): Likewise.
17752 (ix86_expand_round_sse4): Likewise.
17753 (expand_vec_perm_blend): Likewise.
17754 (canonicalize_vector_int_perm): Likewise.
17755 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
17756 (expand_vec_perm_1): Likewise.
17757 (expand_vec_perm_interleave3): Likewise.
17758 (expand_vec_perm_even_odd_pack): Likewise.
17759 (expand_vec_perm_even_odd_1): Likewise.
17760 (expand_vec_perm_broadcast_1): Likewise.
17761 (ix86_vectorize_vec_perm_const_ok): Likewise.
17762 (ix86_expand_vecop_qihi): Likewise.
17763 (ix86_expand_mul_widen_hilo): Likewise.
17764 (ix86_expand_sse2_abs): Likewise.
17765 (ix86_expand_pextr): Likewise.
17766 (ix86_expand_pinsr): Likewise.
17767 (ix86_preferred_simd_mode): Likewise.
17768 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
17769 * config/i386/sse.md (*andnot<mode>3): Likewise.
17770 (<mask_codefor><code><mode>3<mask_name>): Likewise.
17771 (*<code><mode>3): Likewise.
17772 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
17773 (ia64_expand_atomic_op): Likewise.
17774 (ia64_arg_type): Likewise.
17775 (ia64_mode_to_int): Likewise.
17776 (ia64_scalar_mode_supported_p): Likewise.
17777 (ia64_vector_mode_supported_p): Likewise.
17778 (expand_vec_perm_broadcast): Likewise.
17779 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
17780 (iq2000_function_arg_advance): Likewise.
17781 (iq2000_function_arg): Likewise.
17782 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
17783 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
17784 (m68k_libcall_value): Likewise.
17785 (m68k_function_value): Likewise.
17786 (sched_attr_op_type): Likewise.
17787 * config/mcore/mcore.c (mcore_output_move): Likewise.
17788 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
17790 (microblaze_function_arg): Likewise.
17791 * config/mips/mips.c (mips16_build_call_stub): Likewise.
17792 (mips_print_operand): Likewise.
17793 (mips_mode_ok_for_mov_fmt_p): Likewise.
17794 (mips_vector_mode_supported_p): Likewise.
17795 (mips_preferred_simd_mode): Likewise.
17796 (mips_expand_vpc_loongson_even_odd): Likewise.
17797 (mips_expand_vec_unpack): Likewise.
17798 (mips_expand_vi_broadcast): Likewise.
17799 (mips_expand_vector_init): Likewise.
17800 (mips_expand_vec_reduc): Likewise.
17801 (mips_expand_msa_cmp): Likewise.
17802 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
17803 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
17804 (cc_flags_for_mode): Likewise.
17805 * config/msp430/msp430.c (msp430_print_operand): Likewise.
17806 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
17807 (nds32_output_casesi_pc_relative): Likewise.
17808 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17809 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
17810 (nvptx_gen_unpack): Likewise.
17811 (nvptx_gen_pack): Likewise.
17812 (nvptx_gen_shuffle): Likewise.
17813 (nvptx_gen_wcast): Likewise.
17814 (nvptx_preferred_simd_mode): Likewise.
17815 * config/pa/pa.c (pa_secondary_reload): Likewise.
17816 * config/pa/predicates.md (base14_operand): Likewise.
17817 * config/powerpcspe/powerpcspe-c.c
17818 (altivec_resolve_overloaded_builtin): Likewise.
17819 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
17821 (rs6000_preferred_simd_mode): Likewise.
17822 (output_vec_const_move): Likewise.
17823 (rs6000_expand_vector_extract): Likewise.
17824 (rs6000_split_vec_extract_var): Likewise.
17825 (reg_offset_addressing_ok_p): Likewise.
17826 (rs6000_legitimate_offset_address_p): Likewise.
17827 (rs6000_legitimize_address): Likewise.
17828 (rs6000_emit_set_const): Likewise.
17829 (rs6000_const_vec): Likewise.
17830 (rs6000_emit_move): Likewise.
17831 (spe_build_register_parallel): Likewise.
17832 (rs6000_darwin64_record_arg_recurse): Likewise.
17833 (swap_selector_for_mode): Likewise.
17834 (spe_init_builtins): Likewise.
17835 (paired_init_builtins): Likewise.
17836 (altivec_init_builtins): Likewise.
17837 (do_load_for_compare): Likewise.
17838 (rs6000_generate_compare): Likewise.
17839 (rs6000_expand_float128_convert): Likewise.
17840 (emit_load_locked): Likewise.
17841 (emit_store_conditional): Likewise.
17842 (rs6000_output_function_epilogue): Likewise.
17843 (rs6000_handle_altivec_attribute): Likewise.
17844 (rs6000_function_value): Likewise.
17845 (emit_fusion_gpr_load): Likewise.
17846 (emit_fusion_p9_load): Likewise.
17847 (emit_fusion_p9_store): Likewise.
17848 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
17849 (fusion_gpr_mem_load): Likewise.
17850 (fusion_addis_mem_combo_load): Likewise.
17851 (fusion_addis_mem_combo_store): Likewise.
17852 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
17853 (fusion_gpr_mem_load): Likewise.
17854 (fusion_addis_mem_combo_load): Likewise.
17855 (fusion_addis_mem_combo_store): Likewise.
17856 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17858 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
17859 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
17860 (rs6000_preferred_simd_mode): Likewise.
17861 (output_vec_const_move): Likewise.
17862 (rs6000_expand_vector_extract): Likewise.
17863 (rs6000_split_vec_extract_var): Likewise.
17864 (reg_offset_addressing_ok_p): Likewise.
17865 (rs6000_legitimate_offset_address_p): Likewise.
17866 (rs6000_legitimize_address): Likewise.
17867 (rs6000_emit_set_const): Likewise.
17868 (rs6000_const_vec): Likewise.
17869 (rs6000_emit_move): Likewise.
17870 (rs6000_darwin64_record_arg_recurse): Likewise.
17871 (swap_selector_for_mode): Likewise.
17872 (paired_init_builtins): Likewise.
17873 (altivec_init_builtins): Likewise.
17874 (rs6000_expand_float128_convert): Likewise.
17875 (emit_load_locked): Likewise.
17876 (emit_store_conditional): Likewise.
17877 (rs6000_output_function_epilogue): Likewise.
17878 (rs6000_handle_altivec_attribute): Likewise.
17879 (rs6000_function_value): Likewise.
17880 (emit_fusion_gpr_load): Likewise.
17881 (emit_fusion_p9_load): Likewise.
17882 (emit_fusion_p9_store): Likewise.
17883 * config/rx/rx.c (rx_gen_move_template): Likewise.
17884 (flags_from_mode): Likewise.
17885 * config/s390/predicates.md (s390_alc_comparison): Likewise.
17886 (s390_slb_comparison): Likewise.
17887 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
17888 (s390_vector_mode_supported_p): Likewise.
17889 (s390_cc_modes_compatible): Likewise.
17890 (s390_match_ccmode_set): Likewise.
17891 (s390_canonicalize_comparison): Likewise.
17892 (s390_emit_compare_and_swap): Likewise.
17893 (s390_branch_condition_mask): Likewise.
17894 (s390_rtx_costs): Likewise.
17895 (s390_secondary_reload): Likewise.
17896 (__SECONDARY_RELOAD_CASE): Likewise.
17897 (s390_expand_cs): Likewise.
17898 (s390_preferred_simd_mode): Likewise.
17899 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
17900 * config/sh/sh.c (sh_print_operand): Likewise.
17901 (dump_table): Likewise.
17902 (sh_secondary_reload): Likewise.
17903 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17904 * config/sh/sh.md (casesi_worker_1): Likewise.
17905 (casesi_worker_2): Likewise.
17906 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
17907 (fcc_comparison_operator): Likewise.
17908 * config/sparc/sparc.c (sparc_expand_move): Likewise.
17909 (emit_soft_tfmode_cvt): Likewise.
17910 (sparc_preferred_simd_mode): Likewise.
17911 (output_cbranch): Likewise.
17912 (sparc_print_operand): Likewise.
17913 (sparc_expand_vec_perm_bmask): Likewise.
17914 (vector_init_bshuffle): Likewise.
17915 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17916 (spu_vector_mode_supported_p): Likewise.
17917 (spu_expand_insv): Likewise.
17918 (spu_emit_branch_or_set): Likewise.
17919 (spu_handle_vector_attribute): Likewise.
17920 (spu_builtin_splats): Likewise.
17921 (spu_builtin_extract): Likewise.
17922 (spu_builtin_promote): Likewise.
17923 (spu_expand_sign_extend): Likewise.
17924 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17925 (tilegx_simd_int): Likewise.
17926 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
17927 (tilepro_simd_int): Likewise.
17928 * config/v850/v850.c (const_double_split): Likewise.
17929 (v850_print_operand): Likewise.
17930 (ep_memory_offset): Likewise.
17931 * config/vax/vax.c (vax_rtx_costs): Likewise.
17932 (vax_output_int_move): Likewise.
17933 (vax_output_int_add): Likewise.
17934 (vax_output_int_subtract): Likewise.
17935 * config/visium/predicates.md (visium_branch_operator): Likewise.
17936 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
17937 (visium_print_operand_address): Likewise.
17938 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17939 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
17940 (xtensa_expand_conditional_branch): Likewise.
17941 (xtensa_copy_incoming_a7): Likewise.
17942 (xtensa_output_literal): Likewise.
17943 * dfp.c (decimal_real_maxval): Likewise.
17944 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
17946 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17947 Alan Hayward <alan.hayward@arm.com>
17948 David Sherwood <david.sherwood@arm.com>
17950 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
17951 (mode_nunits_inline): Likewise.
17952 (mode_inner_inline): Likewise.
17953 (mode_unit_size_inline): Likewise.
17954 (mode_unit_precision_inline): Likewise.
17955 (emit_insn_modes_h): Likewise. Also emit a #define of the
17957 (emit_mode_wider): Add an E_ prefix to mode names.
17958 (emit_mode_complex): Likewise.
17959 (emit_mode_inner): Likewise.
17960 (emit_mode_adjustments): Likewise.
17961 (emit_mode_int_n): Likewise.
17962 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
17963 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
17964 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
17965 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
17966 (CRC32_BUILTIN, ENTRY): Likewise.
17967 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
17968 (aarch64_pop_regs): Likewise.
17969 (aarch64_process_components): Likewise.
17970 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
17971 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
17972 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
17973 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
17974 * config/arm/arm.c (arm_init_libfuncs): Likewise.
17975 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
17977 * config/i386/i386-builtin.def (pcmpestr): Likewise.
17978 (pcmpistr): Likewise.
17979 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
17980 * config/mmix/mmix.c (mmix_output_condition): Likewise.
17981 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17983 * config/rl78/rl78.c (mduc_regs): Likewise.
17984 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17985 (htm_expand_builtin): Likewise.
17986 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
17987 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
17988 * config/xtensa/xtensa.c (print_operand): Likewise.
17989 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
17990 (NUM_MODE_VECTOR_INT): Likewise.
17991 * genoutput.c (null_operand): Likewise.
17992 (output_operand_data): Likewise.
17993 * genrecog.c (print_parameter_value): Likewise.
17994 * lra.c (debug_operand_data): Likewise.
17996 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17998 * dwarf2out.c (scompare_loc_descriptor_wide)
17999 (scompare_loc_descriptor_narrow): New functions, split out from...
18000 (scompare_loc_descriptor): ...here.
18001 * expmed.c (emit_store_flag_int): New function, split out from...
18002 (emit_store_flag): ...here.
18004 2017-08-30 Richard Biener <rguenther@suse.de>
18006 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
18007 (dwarf2out_early_finish): Move setting of AT_pubnames from
18008 early debug output to early finish.
18010 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18012 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
18013 and -mdata-region to the assembler.
18015 2017-08-30 Richard Biener <rguenther@suse.de>
18017 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
18019 (gen_subprogram_die): Add DW_AT_object_pointer only early.
18020 (dwarf2out_early_global_decl): Only generate a DIE for the
18021 abstract origin if it doesn't already exist or is a declaration DIE.
18022 (resolve_addr): Do not add the linkage name twice when
18023 generating a stub DIE for the DW_TAG_GNU_call_site target.
18025 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18027 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18028 Use machine_mode rather than int for arg1_mode.
18030 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
18033 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
18034 that the second argument of the built-in functions to unpack
18035 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
18036 switch statement instead a lot of if statements.
18037 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
18038 Allow 64-bit values to be in Altivec registers as well as
18039 traditional floating point registers.
18040 (pack<mode>, FMOVE128_VSX iterator): Likewise.
18042 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
18044 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
18045 MAX_REGS_PER_ADDRESS == 1.
18047 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
18049 * config/i386/i386.opt (flag_fentry): Do not init to -1.
18050 * config/i386/i386.c (ix86_option_override_internal): Simplify
18051 setting of opts->x_flag_entry.
18053 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18054 Jakub Jelinek <jakub@redhat.com>
18055 Richard Biener <rguenther@suse.de>
18057 PR tree-optimization/81503
18058 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
18059 folded constant fits in the target type; reorder tests for clarity.
18061 2017-08-29 Martin Liska <mliska@suse.cz>
18063 * passes.def: Include pass_lower_switch.
18064 * stmt.c (dump_case_nodes): Remove and move to
18065 tree-switch-conversion.
18066 (case_values_threshold): Likewise.
18067 (expand_switch_as_decision_tree_p): Likewise.
18068 (emit_case_decision_tree): Likewise.
18069 (expand_case): Likewise.
18070 (balance_case_nodes): Likewise.
18071 (node_has_low_bound): Likewise.
18072 (node_has_high_bound): Likewise.
18073 (node_is_bounded): Likewise.
18074 (emit_case_nodes): Likewise.
18075 (struct simple_case_node): New struct.
18076 (add_case_node): Remove.
18077 (emit_case_dispatch_table): Use vector instead of case_list.
18078 (reset_out_edges_aux): Remove.
18079 (compute_cases_per_edge): Likewise.
18080 (expand_case): Build list of simple_case_node.
18081 (expand_sjlj_dispatch_table): Use it.
18082 * tree-switch-conversion.c (struct case_node): Moved from
18083 stmt.c and adjusted.
18084 (emit_case_nodes): Likewise.
18085 (node_has_low_bound): Likewise.
18086 (node_has_high_bound): Likewise.
18087 (node_is_bounded): Likewise.
18088 (case_values_threshold): Likewise.
18089 (reset_out_edges_aux): Likewise.
18090 (compute_cases_per_edge): Likewise.
18091 (add_case_node): Likewise.
18092 (dump_case_nodes): Likewise.
18093 (balance_case_nodes): Likewise.
18094 (expand_switch_as_decision_tree_p): Likewise.
18095 (emit_jump): Likewise.
18096 (emit_case_decision_tree): Likewise.
18097 (try_switch_expansion): Likewise.
18098 (do_jump_if_equal): Likewise.
18099 (emit_cmp_and_jump_insns): Likewise.
18100 (fix_phi_operands_for_edge): New function.
18101 (record_phi_operand_mapping): Likewise.
18102 (class pass_lower_switch): New pass.
18103 (pass_lower_switch::execute): New function.
18104 (make_pass_lower_switch): Likewise.
18105 (conditional_probability):
18106 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
18107 * tree-pass.h: Add make_pass_lower_switch.
18109 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18112 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
18115 2017-08-29 Richard Biener <rguenther@suse.de>
18117 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
18118 we do not add a DW_AT_inline attribute twice.
18119 (gen_subprogram_die): Remove code setting DW_AT_inline on
18120 DECL_ABSTRACT_P nodes.
18122 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
18124 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
18125 calls to internal functions.
18126 (gimplify_modify_expr): Likewise.
18127 * tree-call-cdce.c (use_internal_fn): Likewise.
18128 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
18129 (convert_to_divmod): Set the nothrow flag.
18130 * tree-if-conv.c (predicate_mem_writes): Likewise.
18131 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
18132 (vectorizable_call): Likewise.
18133 (vectorizable_store): Likewise.
18134 (vectorizable_load): Likewise.
18135 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
18136 (vect_recog_mask_conversion_pattern): Likewise.
18138 2017-08-29 Martin Liska <mliska@suse.cz>
18141 * gcc.c (driver_handle_option): Add new argument.
18142 * opts-common.c (handle_option): Pass
18143 target_option_override_hook.
18144 * opts-global.c (lang_handle_option): Add new option.
18145 (set_default_handlers): Add new argument.
18146 (decode_options): Likewise.
18147 * opts.c (target_handle_option): Likewise.
18148 (common_handle_option): Call target_option_override_hook.
18149 * opts.h (struct cl_option_handler_func): Add hook for
18150 target option override.
18151 (struct cl_option_handlers): Likewise.
18152 (set_default_handlers): Add new argument.
18153 (decode_options): Likewise.
18154 (common_handle_option): Likewise.
18155 (target_handle_option): Likewise.
18156 * toplev.c (toplev::main): Pass targetm.target_option.override
18159 2017-08-29 Richard Biener <rguenther@suse.de>
18160 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
18162 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
18163 life to the active subtree.
18165 2017-08-28 Jeff Law <law@redhat.com>
18167 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
18168 derive_equivalences.
18169 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
18170 Code moved into....
18171 (edge_info::derive_equivalences): New private member function
18173 * tree-ssa-dom.c (class edge_info): Changed from a struct
18174 to a class. Add ctor/dtor, methods and data members.
18175 (edge_info::edge_info): Renamed from allocate_edge_info.
18176 Initialize additional members.
18177 (edge_info::~edge_info): New.
18178 (free_dom_edge_info): Delete the edge info.
18179 (record_edge_info): Use new class & associated member functions.
18180 Tighten forms for testing for edge equivalences.
18181 (record_temporary_equivalences): Iterate over the simple
18182 equivalences rather than assuming there's only one per edge.
18183 (cprop_into_successor_phis): Iterate over the simple
18184 equivalences rather than assuming there's only one per edge.
18185 (optimize_stmt): Use operand_equal_p rather than pointer
18186 equality for mini-DSE code.
18188 2017-08-28 Nathan Sidwell <nathan@acm.org>
18190 * gcc.c (execute): Fold SIGPIPE handling into switch
18191 statement. Adjust internal error message.
18193 2017-08-28 Richard Biener <rguenther@suse.de>
18196 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
18197 Do nothing for removed DIEs.
18199 2017-08-28 Richard Biener <rguenther@suse.de>
18201 PR tree-optimization/81977
18202 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
18205 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
18208 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
18209 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
18210 using targetm.gen_mem_thread_fence.
18212 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
18215 * config/i386/i386.md (*<btsc><mode>): Change operand 2
18216 predicate to register_operand. Reorder operands.
18217 (*btr<mode>): Ditto.
18218 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
18219 (*btr<mode>_mask): Ditto.
18221 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
18223 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
18224 * config/rs6000/xmmintrin.h: New file.
18225 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
18227 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18230 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
18231 parameter and_insn and return it.
18232 (recombine_lvx_pattern): Insert a copy to ensure availability of
18233 the base register of the copied masking operation at the point of
18234 the instruction replacement.
18235 (recombine_stvx_pattern): Likewise.
18237 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
18239 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
18240 undocumented switches.
18241 (-mpower9-dform-vector): Likewise.
18242 (-mpower9-dform): Likewise.
18243 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
18244 comments to delete references to -mpower9-dform* switches.
18245 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
18246 Delete reference to -mpower9-dform* switches, test for
18247 -mpower9-vector instead.
18248 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
18249 (OTHER_P9_VECTOR_MASKS): Likewise.
18250 (POWERPC_MASKS): Likewise.
18251 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
18252 tests against -mpower9-dform* to -mpower9-vector. Delete code
18253 that checked for -mpower9-dform* consistancy with other options.
18254 Add test for -mpower9-misc to enable other power9 switches.
18255 (rs6000_init_hard_regno_mode_ok): Likewise.
18256 (rs6000_option_override_internal): Likewise.
18257 (rs6000_emit_prologue): Likewise.
18258 (rs6000_emit_epilogue): Likewise.
18259 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
18260 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
18261 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
18263 (emit_fusion_p9_store): Likewise.
18264 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
18265 resetting these macros if the assembler does not support ISA 3.0
18267 (TARGET_P9_DFORM_VECTOR): Likewise.
18268 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
18269 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
18271 2017-08-25 Alan Modra <amodra@gmail.com>
18273 PR rtl-optimization/81747
18274 * cse.c (cse_extended_basic_block): Don't attempt to record
18275 equivalences for degenerate conditional branches that branch
18276 to their fall-through.
18278 2017-08-24 Martin Sebor <msebor@redhat.com>
18280 PR middle-end/81908
18281 * gimple-fold.c (size_must_be_zero_p): New function.
18282 (gimple_fold_builtin_memory_op): Call it.
18284 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
18286 * config/rs6000/mm_malloc.h: New file.
18288 2017-08-24 Bin Cheng <bin.cheng@arm.com>
18290 PR tree-optimization/81913
18291 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
18292 analysis when either IVs in condition can wrap.
18294 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
18296 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
18297 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
18299 2017-08-24 Richard Biener <rguenther@suse.de>
18302 * targhooks.c (default_target_can_inline_p): Properly
18303 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
18304 is present and always compare.
18305 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
18306 infer -mfpmath=sse from TARGET_SSE_P.
18307 (ix86_can_inline_p): Properly use target_option_default_node when
18308 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
18310 2017-08-24 Richard Biener <rguenther@suse.de>
18313 * dwarf2out.c (output_die): Handle flag_generate_offload like
18315 (output_comp_unit): Likewise.
18316 (gen_array_type_die): Likewise.
18317 (dwarf2out_early_finish): Likewise.
18318 (note_variable_value_in_expr): Likewise.
18319 (dwarf2out_finish): Likewise. Adjust assert.
18320 * cgraphunit.c (symbol_table::compile): Move setting of
18321 flag_generate_offload earlier ...
18322 (symbol_table::finalize_compilation_unit): ... here, before
18323 early debug finalization.
18325 2017-08-24 Richard Biener <rguenther@suse.de>
18327 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
18328 and ipa-fnsummary.h.
18329 (ix86_can_inline_p): When ix86_fpmath flags do not match
18330 check whether the callee uses FP math at all.
18332 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
18334 PR middle-end/81931
18335 * tree-ssanames.c (get_nonzero_bits): Use element_precision
18336 instead of TYPE_PRECISION.
18338 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
18339 Alan Hayward <alan.hayward@arm.com>
18340 David Sherwood <david.sherwood@arm.com>
18342 * combine.c (make_extraction): Use subreg_offset_from_lsb.
18344 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
18346 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
18348 (ix86_frame::stack_realign_allocate): New field.
18349 (struct machine_frame_state): Modify comments.
18350 (machine_frame_state::sp_realigned_fp_end): New field.
18351 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
18352 layout calculation.
18353 (sp_valid_at): Add assertion to assure no attempt to access invalid
18354 offset of a realigned stack.
18355 (fp_valid_at): Likewise.
18356 (choose_baseaddr): Modify comments.
18357 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
18358 ix86_expand_prologue.
18359 (ix86_expand_prologue): Modify stack realignment and allocation.
18360 (ix86_expand_epilogue): Modify comments.
18361 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
18362 avx2_runtime, avx512f, and avx512f_runtime.
18364 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
18366 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
18367 (mstackrealign): Do not init to -1.
18368 * config/i386/i386.c (ix86_option_override_internal):
18369 Check opts_set, not opts when setting default value of
18370 opts->x_ix86_force_align_arg_pointer.
18372 2017-08-23 Richard Biener <rguenther@suse.de>
18374 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
18375 lang_hooks.decl_printable_name.
18376 * print-rtl-function.c (print_rtx_function): Likewise.
18377 * tree-pretty-print.c (dump_function_header): Likewise.
18379 2017-08-23 Richard Biener <rguenther@suse.de>
18382 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
18383 -g0 at compile-time.
18385 2017-08-23 Tamar Christina <tamar.christina@arm.com>
18387 PR middle-end/19706
18388 * doc/sourcebuild.texi (Other hardware attributes):
18391 2017-08-23 Tamar Christina <tamar.christina@arm.com>
18393 PR middle-end/19706
18394 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
18395 Fix single-use check.
18397 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
18399 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
18401 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
18403 * doc/install.texi: Modify to add more details on running selected
18406 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
18408 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
18409 is combined with -mabi=ms.
18410 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
18413 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18415 PR tree-optimization/81488
18416 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
18417 and cached_basis fields.
18418 (MAX_SPREAD): New constant.
18419 (alloc_cand_and_find_basis): Initialize new fields.
18420 (clear_visited): New function.
18421 (create_phi_basis_1): Rename from create_phi_basis, set visited
18422 and cached_basis fields.
18423 (create_phi_basis): New wrapper function.
18424 (phi_add_costs_1): Rename from phi_add_costs, add spread
18425 parameter, set visited field, short-circuit when limits reached.
18426 (phi_add_costs): New wrapper function.
18427 (record_phi_increments_1): Rename from record_phi_increments, set
18429 (record_phi_increments): New wrapper function.
18430 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
18431 (phi_incr_cost): New wrapper function.
18432 (all_phi_incrs_profitable_1): Rename from
18433 all_phi_incrs_profitable, set visited field.
18434 (all_phi_incrs_profitable): New wrapper function.
18436 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
18437 Alan Hayward <alan.hayward@arm.com>
18438 David Sherwood <david.sherwood@arm.com>
18440 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
18441 that takes the outer and inner modes.
18442 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
18443 comparison as the canonical test for a paradoxical subreg.
18444 * combine.c (simplify_set): Use paradoxical_subreg_p.
18445 (make_extraction): Likewise.
18446 (force_to_mode): Likewise.
18447 (rtx_equal_for_field_assignment_p): Likewise.
18448 (gen_lowpart_for_combine): Likewise.
18449 (simplify_comparison): Likewise.
18450 * cse.c (equiv_constant): Likewise.
18451 * expmed.c (store_bit_field_1): Likewise.
18452 * final.c (alter_subreg): Likewise.
18453 * fwprop.c (propagate_rtx): Likewise.
18454 (forward_propagate_subreg): Likewise.
18455 * ira-conflicts.c (ira_build_conflicts): Likewise.
18456 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
18457 * lra-constraints.c (curr_insn_transform): Likewise.
18458 (split_reg): Likewise.
18459 * lra-eliminations.c (move_plus_up): Likewise.
18460 (lra_eliminate_regs_1): Likewise.
18461 * recog.c (general_operand): Likewise.
18462 * ree.c (combine_reaching_defs): Likewise.
18463 * reload.c (push_reload): Likewise.
18464 (find_reloads): Likewise.
18465 * reload1.c (elimination_effects): Likewise.
18466 (compute_reload_subreg_offset): Likewise.
18467 (choose_reload_regs): Likewise.
18468 * rtlanal.c (subreg_lsb_1): Likewise.
18469 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18470 (simplify_subreg): Likewise.
18471 * var-tracking.c (track_loc_p): Likewise.
18472 * emit-rtl.c (byte_lowpart_offset): Likewise.
18473 (paradoxical_subreg_p): Delete out-of-line definition.
18475 2017-08-22 Jeff Law <law@redhat.com>
18477 PR tree-optimization/81741
18478 PR tree-optimization/71947
18479 * tree-ssa-dom.c: Include tree-inline.h.
18480 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
18481 equivalences if one is more expensive to compute than the other.
18482 * tree-ssa-scopedtables.h (class const_or_copies): Make
18483 record_const_or_copy_raw method private.
18484 (class avail_exprs_stack): New method simplify_binary_operation.
18485 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
18486 avail_exprs_stack::simplify_binary_operation as needed.
18487 (avail_exprs_stack::simplify_binary_operation): New function.
18489 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
18491 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
18492 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
18493 (DOT_SYMBOLS): Likewise.
18494 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
18495 (RELOCATABLE_NEEDS_FIXUP): Likewise.
18496 (RS6000_ABI_NAME): Likewise.
18497 (TARGET_CMODEL): Likewise.
18498 (TOC_SECTION_ASM_OP): Likewise.
18499 (SET_CMODEL): New macro.
18500 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
18502 2017-08-22 Richard Biener <rguenther@suse.de>
18504 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
18505 to allow for free-lang-data replacements similar to verify_type_variant.
18507 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
18508 Alan Hayward <alan.hayward@arm.com>
18509 David Sherwood <david.sherwood@arm.com>
18511 * config/aarch64/aarch64.md (casesi): Use DImode rather than
18512 VOIDmode for the LABEL_REF.
18514 2017-08-22 Richard Biener <rguenther@suse.de>
18516 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
18518 2017-08-22 Richard Biener <rguenther@suse.de>
18520 * common.opt (feliminate-dwarf2-dups): Ignore.
18521 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
18522 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
18523 same_die_p_wrap, compute_section_prefix,
18524 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
18525 (comdat_symbol_id, comdat_symbol_number): Likewise.
18526 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
18528 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
18529 (output_die): Mark unreachable path unreachable.
18530 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
18531 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
18532 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
18533 (dwarf2out_early_finish): Likewise.
18535 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
18537 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
18539 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
18542 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
18543 not VAR_P. Filter attribute warnings with OPT_Wattributes.
18544 (avr_attribute_table) <io, io_low, address>: Initialize
18545 .decl_required with true.
18547 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
18549 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
18550 undocumented debugging options.
18551 (-mvsx-scalar-double): Likewise.
18552 (-mallow-df-permute): Likewise.
18553 (-mvectorize-builtins): Likewise.
18554 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18555 (rs6000_builtin_vectorized_function): Likewise.
18556 (rs6000_builtin_md_vectorized_function): Likewise.
18557 (rs6000_opt_vars): Likewise.
18559 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
18562 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
18563 (*btrq_imm): Rename from *btrq.
18564 (*btcq_imm): Rename from *btcq.
18565 (btsc): New code attribute.
18566 (*<btsc><mode>): New insn pattern.
18567 (*btr<mode>): Ditto.
18568 (*<btsc><mode>_mask): New insn_and_split pattern.
18569 (*btr<mode>_mask): Ditto.
18571 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18572 Alan Hayward <alan.hayward@arm.com>
18573 David Sherwood <david.sherwood@arm.com>
18575 * function.c (pad_below): Simplify padding calculation.
18577 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18578 Alan Hayward <alan.hayward@arm.com>
18579 David Sherwood <david.sherwood@arm.com>
18581 * target.def (function_prologue): Remove frame size argument.
18582 (function_epilogue): Likewise.
18583 * doc/tm.texi: Regenerate.
18584 * final.c (final_start_function): Update call to function_prologue.
18585 (final_end_function): Update call to function_epilogue.
18586 (default_function_pro_epilogue): Remove frame size argument.
18587 * output.h (default_function_pro_epilogue): Likewise.
18588 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
18589 (arm_output_function_prologue): Likewise.
18590 * config/frv/frv.c (frv_function_prologue): Likewise.
18591 (frv_function_epilogue): Likewise.
18592 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
18593 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
18594 (ia64_output_function_epilogue): Likewise.
18595 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
18596 (m32r_output_function_epilogue): Likewise.
18597 * config/microblaze/microblaze.c (microblaze_function_prologue)
18598 (microblaze_function_epilogue): Likewise.
18599 * config/mips/mips.c (mips_output_function_prologue): Likewise.
18600 (mips_output_function_epilogue): Likewise.
18601 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
18602 (mmix_target_asm_function_epilogue): Likewise.
18603 * config/msp430/msp430.c (msp430_start_function): Likewise.
18604 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
18605 (nds32_asm_function_epilogue): Likewise.
18606 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
18607 * config/pa/pa.c (pa_output_function_prologue): Likewise.
18608 (pa_output_function_epilogue): Likewise.
18609 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
18610 (rs6000_output_function_epilogue): Likewise.
18611 * config/rl78/rl78.c (rl78_start_function): Likewise.
18612 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
18613 (rs6000_output_function_epilogue): Likewise.
18614 * config/rx/rx.c (rx_output_function_prologue): Likewise.
18615 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
18616 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
18617 (sparc_asm_function_epilogue): Likewise.
18619 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18621 * tree.h (type_has_mode_precision_p): New function.
18622 * convert.c (convert_to_integer_1): Use it.
18623 * expr.c (expand_expr_real_2): Likewise.
18624 (expand_expr_real_1): Likewise.
18625 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
18626 * match.pd: Likewise.
18627 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
18628 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
18629 * tree-tailcall.c (process_assignment): Likewise.
18630 * tree-vect-loop.c (vectorizable_reduction): Likewise.
18631 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
18632 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
18633 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18634 (vectorizable_assignment): Likewise.
18635 (vectorizable_shift): Likewise.
18636 (vectorizable_operation): Likewise.
18637 * tree-vrp.c (register_edge_assert_for_2): Likewise.
18639 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
18641 * match.pd: Add pow (C, x) simplification.
18643 2017-08-21 Richard Biener <rguenther@suse.de>
18645 PR tree-optimization/81900
18646 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
18647 for blocks with abnormal predecessors.
18648 (compute_antic): Do not set visited flag prematurely.
18650 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
18653 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
18655 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18657 * stor-layout.h (vector_type_mode): Move to...
18658 * tree.h (vector_type_mode): ...here.
18659 * stor-layout.c (vector_type_mode): Move to...
18660 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
18662 2017-08-21 Richard Biener <rguenther@suse.de>
18664 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
18665 register_external_die hooks.
18666 (debug_false_tree_charstarstar_uhwistar): Declare.
18667 (debug_nothing_tree_charstar_uhwi): Likewise.
18668 * debug.c (do_nothing_debug_hooks): Adjust.
18669 (debug_false_tree_charstarstar_uhwistar): New do nothing.
18670 (debug_nothing_tree_charstar_uhwi): Likewise.
18671 * dbxout.c (dbx_debug_hooks): Adjust.
18672 (xcoff_debug_hooks): Likewise.
18673 * sdbout.c (sdb_debug_hooks): Likewise.
18674 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
18675 * dwarf2out.c (macinfo_label_base): New global.
18676 (dwarf2out_register_external_die): New function for the
18677 register_external_die hook.
18678 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
18679 (dwarf2_debug_hooks): Use them.
18680 (dwarf2_lineno_debug_hooks): Adjust.
18681 (struct die_struct): Add with_offset flag.
18682 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
18683 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
18684 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
18685 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
18686 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
18687 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
18688 defining section names for the early LTO debug variants.
18689 (reset_indirect_string): New helper.
18690 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
18691 (print_dw_val): Add support for offsetted symbol references.
18692 (get_ultimate_context): Split out from is_cxx.
18693 (is_cxx): Use get_ultimate_context.
18694 (is_fortran): Add decl overload.
18695 (compute_comp_unit_symbol): Split out worker from
18696 compute_section_prefix.
18697 (compute_section_prefix): Call compute_comp_unit_symbol and
18698 set comdat_type_p here.
18699 (output_die): Skip DIE symbol output for the LTO added one.
18700 Handle DIE symbol references with offset.
18701 (output_comp_unit): Guard section name mangling properly.
18702 For LTO debug sections emit a symbol at the section beginning
18703 which we use to refer to its DIEs.
18704 (add_abstract_origin_attribute): For DIEs registered via
18705 dwarf2out_register_external_die directly refer to the early
18706 DIE rather than indirectly through the shadow one we created.
18707 Remove obsolete call to dwarf2out_abstract_function for
18708 non-function/block origins.
18709 (gen_array_type_die): When generating early LTO debug do
18710 not emit DW_AT_string_length.
18711 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
18712 late when in LTO. As suggested place a gcc_unreachable for
18713 the DECL_ABSTRACT_P case.
18714 (gen_subprogram_die): Avoid another specification DIE
18715 for early built declarations/definitions for the late LTO case.
18716 (gen_variable_die): Add type references for late duplicated VLA dies
18718 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
18719 we have the abstract instance already.
18720 (process_scope_var): Adjust decl DIE contexts in LTO which
18721 first puts them in limbo.
18722 (gen_decl_die): Do not generate type DIEs late apart from
18723 types for VLAs or for decls we do not yet have a DIE. Do not
18724 call dwarf2out_abstract_function late.
18725 (dwarf2out_early_global_decl): Make sure to create DIEs
18726 for abstract instances of a decl first.
18727 (dwarf2out_late_global_decl): Adjust comment.
18728 (output_macinfo_op): With multiple macro sections use
18729 macinfo_label_base to distinguish labels.
18730 (output_macinfo): Likewise. Update macinfo_label_base.
18731 Pass in the line info label.
18732 (note_variable_value_in_expr): When generating LTO resolve
18733 all variable values here by generating DIEs as needed.
18734 (init_sections_and_labels): Add early LTO debug flag parameter
18735 and generate different sections and names if set. Add generation
18736 counter for the labels so we can have multiple of them.
18737 (reset_dies): Helper to allow DIEs to be output multiple times.
18738 (dwarf2out_finish): When outputting DIEs to the fat part of an
18739 LTO object first reset DIEs.
18740 (dwarf2out_early_finish): Output early DIEs when generating LTO.
18741 (modified_type_die): Check for decl_ultimate_origin being self
18743 (gen_type_die_with_usage): Likewise.
18744 (gen_typedef_die): Allow decl_ultimate_origin being self.
18745 (set_decl_abstract_flags): Remove.
18746 (set_block_abstract_flags): Likewise.
18747 (dwarf2out_abstract_function): Treat the early generated DIEs
18748 as the abstract copy and only add DW_AT_inline and
18749 DW_AT_artificial here and call set_decl_origin_self.
18750 If the DIE has an abstract origin don't do anything.
18751 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
18752 if we have none yet (Go fails to build one, PR78628).
18753 (variably_modified_type_p): Prevent endless recursion for Ada
18754 cyclic pointer types.
18755 * lto-streamer-in.c: Include debug.h.
18756 (dref_queue): New global.
18757 (lto_read_tree_1): Stream in DIE references.
18758 (lto_input_tree): Register DIE references.
18759 (input_function): Stream DECL_DEBUG_ARGS.
18760 * lto-streamer-out.c: Include debug.h.
18761 (lto_write_tree_1): Output DIE references.
18762 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
18763 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
18764 (output_function): Stream DECL_DEBUG_ARGS.
18765 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
18766 Stream DECL_ABSTRACT_ORIGIN.
18767 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
18768 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
18769 DECL_CONTEXT for file-scope decls.
18770 * lto-streamer.h (struct dref_entry): Declare.
18771 (dref_queue): Likewise.
18772 * cfgexpand.c (pass_expand::execute): Do not call the
18773 outlining_inline_function hook here.
18774 * lto-wrapper.c (debug_obj): New global.
18775 (tool_cleanup): Unlink it if required.
18776 (debug_objcopy): New function.
18777 (run_gcc): Handle early debug sections in the IL files by
18778 extracting them to separate files, partially linkin them and
18779 feeding the result back as result to the linker.
18780 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
18781 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
18782 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
18783 sections into a separate segment.
18784 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
18786 (darwin_asm_dwarf_section): Likewise.
18787 (darwin_asm_output_dwarf_offset): Likewise.
18788 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
18790 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18791 Alan Hayward <alan.hayward@arm.com>
18792 David Sherwood <david.sherwood@arm.com>
18794 * read-md.h (md_reader::record_potential_iterator_use): Replace
18795 pointer argument with an rtx and an index.
18796 * read-rtl.c (iterator_group::apply_iterator): Likewise.
18797 (apply_mode_iterator): Likewise.
18798 (apply_code_iterator): Likewise.
18799 (apply_int_iterator): Likewise.
18800 (apply_subst_iterator): Likewise.
18801 (record_iterator_use): Likewise.
18802 (record_attribute_use): Likewise.
18803 (md_reader::record_potential_iterator_use): Likewise. Update calls
18804 to record_iterator_use and apply_iterator.
18805 (iterator_use): Replace ptr with x and index.
18806 (attribute_use): Likewise.
18807 (apply_attribute_uses): Update calls to apply_iterator.
18808 (apply_iterators): Likewise. Update initialization of iterator_use.
18809 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
18810 and record_potential_iterator_use.
18811 (rtx_reader::read_rtx_operand): Likewise.
18813 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18814 Alan Hayward <alan.hayward@arm.com>
18815 David Sherwood <david.sherwood@arm.com>
18817 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
18820 2017-08-21 Richard Biener <rguenther@suse.de>
18822 PR middle-end/81884
18823 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
18824 at struct end conservatively when comparing common bases.
18826 2017-08-21 Richard Biener <rguenther@suse.de>
18828 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
18829 (mem_ref_in_stmt): Remove.
18830 (determine_max_movement): Use ref index to get at the reference.
18831 (invariantness_dom_walker::before_dom_children): Deal with
18832 lim data already initialized.
18833 (gather_mem_refs_stmt): Initialize lim data and record ref index.
18835 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
18837 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
18838 (TARGET_ISA_ROUND): Ditto.
18839 (TARGET_ROUND): Ditto.
18840 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
18841 * config/i386/i386.md: Ditto.
18842 * config/i386/sse.md: Ditto.
18843 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
18844 with OPTION_MASK_ISA_SSE4_1.
18846 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
18849 * doc/extend.texi (x86 Built-in Functions): Correct the name of
18850 __builtin_ia32_lzcnt_u16.
18852 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
18855 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
18856 (rs6000_set_current_function): Rewrite function to use it.
18858 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
18861 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
18862 and TYPE_WARN_IF_NOT_ALIGN.
18863 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
18864 (handle_warn_if_not_align): New.
18865 (place_union_field): Call handle_warn_if_not_align.
18866 (place_field): Call handle_warn_if_not_align.
18867 Copy TYPE_WARN_IF_NOT_ALIGN.
18868 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
18869 (layout_type): Likewise.
18870 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
18872 (tree_decl_common): Add warn_if_not_align.
18873 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
18874 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
18875 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
18876 (DECL_WARN_IF_NOT_ALIGN): Likewise.
18877 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
18878 * doc/extend.texi: Document warn_if_not_aligned attribute.
18879 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
18881 2017-08-17 Martin Liska <mliska@suse.cz>
18884 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
18885 (get_data_dependence): Use it as pointer type.
18886 (distribute_loop): Likewise.
18888 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18890 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
18891 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
18892 (p8_vmrgow_<mode>_direct): New define_insn.
18893 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
18894 handle endianness for vmrgew and vmrgow permute patterns.
18896 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
18898 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
18899 * config/rs6000/rs6000-cpus.def: Remove comment.
18900 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
18901 (POWERPC_MASKS): Likewise.
18902 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
18903 use of TARGET_VSX_TIMODE.
18904 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
18905 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
18906 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
18907 (rs6000_option_override_internal): Remove dead code.
18908 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
18909 (rs6000_legitimize_reload_address): Likewise.
18910 (rs6000_legitimate_address_p): Likewise.
18911 (rs6000_opt_masks): Delete "vsx-timode".
18912 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
18913 from function comment.
18914 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
18915 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
18916 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
18918 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
18919 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
18920 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
18921 (VSX_TI): Likewise.
18923 (define_peephole2): Likewise.
18925 2017-08-17 Martin Sebor <msebor@redhat.com>
18928 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
18929 past the end of an array.
18930 (test_pp_format): Add test cases.
18932 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
18934 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
18935 missing ECF_NOTHROW flags.
18937 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
18940 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
18941 operands residing in integer registers.
18942 (*vsx_le_perm_load_<mode>): Likewise.
18943 (*vsx_le_perm_store_<mode>): Likewise.
18944 (define_peephole2): Add peepholes to optimize the above.
18946 2017-08-17 Marek Polacek <polacek@redhat.com>
18948 PR middle-end/81814
18949 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
18950 to mimic what shorten_compare did. Change the return type to bool.
18951 (fold_cond_expr_with_comparison): Update call to
18952 operand_equal_for_comparison_p.
18953 (fold_ternary_loc): Likewise.
18955 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
18957 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
18959 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
18961 2017-08-17 Richard Biener <rguenther@suse.de>
18963 * tree-ssa-structalias.c (solve_graph): When propagating
18964 to successors update the graphs succ edges and avoid duplicate work.
18966 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
18969 * config/i386/i386.c (ix86_option_override_internal): Save target
18970 specific options after ix86_stack_protector_guard_reg was changed.
18972 2017-08-17 Richard Biener <rguenther@suse.de>
18974 PR tree-optimization/81827
18975 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
18976 (new_var_info): Initialize it conservatively.
18977 (get_call_vi): Mark register vars.
18978 (new_scalar_tmp_constraint_exp): Likewise.
18979 (handle_rhs_call): Likewise.
18980 (handle_const_call): Likewise.
18981 (create_function_info_for): Likewise.
18982 (solve_constraints): Sort varinfos to separate register from
18983 non-register vars to pack points-to solution bitmaps during
18986 2017-08-17 Marek Polacek <polacek@redhat.com>
18988 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
18990 2017-08-17 Richard Biener <rguenther@suse.de>
18992 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
18993 to true when overflow is undefined and we saturated the result.
18995 2017-08-17 Alan Modra <amodra@gmail.com>
18998 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
18999 Don't use store multiple if only one reg needs saving.
19000 (interesting_frame_related_regno): New function.
19001 (rs6000_frame_related): Don't emit frame info for regs that
19003 (rs6000_emit_epilogue): Likewise.
19005 2017-08-16 Nathan Sidwell <nathan@acm.org>
19007 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
19008 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
19009 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
19010 * tree.c (free_lang_data_in_type): Use else-if chain. Always
19011 clear TYPE_LANG_1. Remove obsolete member-function stripping.
19012 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
19013 (verify_type): Adjust for TYPE_BINFO move.
19014 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
19015 process TYPE_BINFO directly.
19016 (hash_tree): Likewise.
19017 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
19019 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
19022 2017-08-16 David Malcolm <dmalcolm@redhat.com>
19024 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
19026 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
19029 * config/i386/i386.md (*anddi_1_btr): Change predicates of
19030 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
19031 Add ix86_binary_operator_ok to insn constraint.
19032 (*iordi_1_bts): Ditto.
19033 (*xordi_1_btc): Ditto.
19034 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
19035 Update corresponding peephole2 pattern.
19039 2017-08-16 Bin Cheng <bin.cheng@arm.com>
19041 PR tree-optimization/81832
19042 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
19043 copy loop header which has IFN_LOOP_DIST_ALIAS call.
19045 2017-08-16 Marek Polacek <polacek@redhat.com>
19048 * fold-const.c (fold_indirect_ref_1): Restore original behavior
19049 regarding size_zero_node.
19051 2017-08-16 Martin Liska <mliska@suse.cz>
19054 * config.gcc: Respect previously set extra_objs in case
19057 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
19059 PR tree-optimization/81835
19060 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
19061 the phi SSA_NAME. Check that the condition in a COND_EXPR does
19062 not depend on the phi.
19064 2017-08-16 Alan Modra <amodra@gmail.com>
19066 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
19069 2017-08-16 Alan Modra <amodra@gmail.com>
19071 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
19072 (save_reg_p): ..into this. Update all callers.
19073 (first_reg_to_save): Simplify.
19075 2017-08-16 Alan Modra <amodra@gmail.com>
19077 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
19080 2017-08-15 Joseph Myers <joseph@codesourcery.com>
19084 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
19085 constant count if that count is less than 32.
19087 2017-08-15 Nathan Sidwell <nathan@acm.org>
19089 * gcc.c (execute): Emit friendlier message if inferior is killed
19090 by an external cause.
19092 2017-08-15 Richard Biener <rguenther@suse.de>
19094 PR tree-optimization/81790
19095 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
19096 CONSTRUCTORs from simplifying and VN.
19098 2017-08-14 Martin Sebor <msebor@redhat.com>
19100 * builtin-attrs.def: Add comments.
19102 2017-08-14 Martin Sebor <msebor@redhat.com>
19105 * doc/extend.texi (attribute nonstring): Document new attribute.
19107 2017-08-14 Martin Sebor <msebor@redhat.com>
19110 * tree-diagnostic.c (default_tree_printer): Handle %G.
19111 * gimple-pretty-print.h (percent_G_format): Declare new function.
19112 * gimple-pretty-print.c (percent_G_format): Define.
19113 * tree-pretty-print.c (percent_K_format): Add argument.
19115 2017-08-14 Martin Sebor <msebor@redhat.com>
19117 PR translation/79998
19118 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
19119 Remove a stray space.
19121 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
19124 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
19125 (*iordi_1_bts): Ditto.
19126 (*xordi_1_btc): Ditto.
19128 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19131 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
19132 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19134 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
19135 quoted strings, and make more translator-friendly.
19136 (darwin_rs6000_override_options): Likewise.
19137 (rs6000_option_override_internal): Likewise.
19138 (rs6000_return_in_memory): Fix overlong line.
19139 (init_cmulative_args): Use quoted strings, and make more
19140 translator-friendly.
19141 (rs6000_pass_by_reference): Fix overlong line.
19142 (def_builtin): Use quoted strings.
19143 (altivec_expand_predicate_builtin): Use quoted strings, and make
19144 more translator-friendly.
19145 (htm_expand_builtin): Use quoted strings.
19146 (cpu_expand_builtin): Use quoted strings, and make more
19147 translator-friendly.
19148 (altivec_expand_builtin): Likewise.
19149 (paired_expand_predicate_builtin): Likewise.
19150 (rs6000_invalid_builtin): Likewise.
19151 (builtin_function_type): Use quoted strings.
19152 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
19153 more translator-friendly.
19154 (rs6000_trampoline_init): Likewise.
19155 (rs6000_handle_altivec_attribute): Likewise.
19156 (rs6000_inner_target_options): Use quoted strings.
19157 (rs6000_disable_incompatible_switches): Likewise.
19158 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
19159 strings, and make more translator-friendly.
19160 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
19162 2017-08-14 Bin Cheng <bin.cheng@arm.com>
19164 PR tree-optimization/81799
19165 * tree-loop-distribution.c (version_loop_by_alias_check): Force
19166 cond_expr to simple gimple operand.
19168 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
19170 PR middle-end/46932
19171 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
19173 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
19177 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
19178 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
19179 TARGET_GASISR_PROLOGUES.
19180 * config/avr/avr.c (avr_option_override): Same.
19181 (avr_pass_pre_proep::execute): Same.
19183 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
19186 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
19187 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
19189 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
19191 * config/i386/i386.md (*load_tp_<mode>): Redefine as
19192 define_insn_and_split. Split to a memory load from 0 in
19193 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
19194 using PTR mode iterator.
19195 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
19196 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
19197 (*add_tp_<mode>): Redefine as define_insn_and_split.
19198 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
19199 address space. Merge with *add_tp_x32 using PTR mode iterator.
19200 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
19201 Split to an add with a memory load from 0 in
19202 DEFAULT_TLS_SEG_REG address space.
19204 2017-08-12 Andrew Pinski <apinski@cavium.com>
19206 * config/aarch64/aarch64-option-extensions.def (rdma):
19207 Fix feature string to what Linux prints out in /proc/cpuinfo.
19209 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
19212 * dwarf2out.c (modified_type_die): For C typedef types that have
19213 an ultimate origin, process the ultimate origin instead of the
19215 (gen_typedef_die): Assert that input DECLs have no ultimate
19217 (gen_type_die_with_usage): For typedef variants that have an
19218 ultimate origin, just call gen_decl_die on the original DECL.
19219 (process_scope_var): Avoid creating DIEs for local typedefs and
19220 concrete static variables.
19222 2017-08-12 Alan Modra <amodra@gmail.com>
19226 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
19227 match gnu-user.h startfile.
19228 (ENDFILE_LINUX_SPEC): Similarly.
19230 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
19233 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
19235 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
19237 2017-08-11 Tamar Christina <tamar.christina@arm.com>
19238 * config/aarch64/aarch64.md (mov<mode>): Change.
19239 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19240 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
19241 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
19243 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
19245 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
19246 for storage order barriers.
19248 2017-08-11 Martin Liska <mliska@suse.cz>
19251 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
19252 variables of void type.
19254 2017-08-11 Martin Liska <mliska@suse.cz>
19256 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
19257 TARGET_SUPPORTS_ALIASES.
19258 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
19259 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
19260 (optimize_weakref): Likewise.
19261 * symtab.c (symtab_node::noninterposable_alias): Likewise.
19262 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
19263 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
19265 2017-08-11 Martin Liska <mliska@suse.cz>
19268 * config/i386/i386.c (make_resolver_func): Do complete
19269 refactoring of the function.
19271 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
19274 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
19275 * config/i386/i386.c (ix86_stack_protect_guard): Use
19276 ix86_stack_protect_guard_symbol_str to generate varible declaration.
19277 * doc/invoke.texi (x86 Options): Document
19278 -mstack-protector-guard-symbol= option.
19280 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
19282 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
19283 * config/i386/i386.c (ix86_split_stack_guard): New function.
19284 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
19285 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
19286 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
19287 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
19288 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
19289 (split_stack_space_check): Call ix86_split_stack_guard.
19291 2017-08-10 Martin Sebor <msebor@redhat.com>
19293 * print-tree.c (print_node): Print location using the established
19295 Replace spaces with colons.
19296 (debug_raw, debug): Ditto.
19298 2017-08-10 Martin Sebor <msebor@redhat.com>
19301 * pretty-print.c (pp_format): Correct the handling of %s precision.
19303 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
19306 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
19308 (ix86_finalize_stack_frame_flags): This. Also clear
19309 frame_pointer_needed if -fno-omit-frame-pointer is used without
19311 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
19312 with ix86_finalize_stack_frame_flags.
19313 (ix86_expand_epilogue): Likewise.
19314 (ix86_expand_split_stack_prologue): Likewise.
19315 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
19317 2017-08-10 Martin Liska <mliska@suse.cz>
19320 * c-attribs.c (handle_target_attribute):
19321 Report warning for an empty string argument of target attribute.
19323 2017-08-09 Jakub Jelinek <jakub@redhat.com>
19326 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
19328 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
19329 or DECL_NONLOCAL labels.
19330 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
19331 or DECL_NONLOCAL labels here.
19333 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
19335 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
19336 to indicate when early gimple folding has been disabled.
19337 (rs6000_gimple_fold_builtin): Add debug content.
19338 (rs6000_invalid_builtin): Fix whitespace.
19339 (rs6000_expand_builtin): Fix whitespace.
19340 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
19342 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
19345 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
19346 SAVE_MULTIPLE if not all the registers that saves, should be saved.
19348 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
19350 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
19351 (qdf24xx): Likewise.
19352 * config/aarch64/aarch64.md: Include falkor.md.
19353 * config/aarch64/falkor.md: New.
19355 2017-08-09 Marek Polacek <polacek@redhat.com>
19358 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
19359 * diagnostic.c (emit_diagnostic): Add a comment.
19360 (emit_diagnostic_valist): New function.
19362 2017-08-09 Marek Polacek <polacek@redhat.com>
19365 * input.c (make_location): New overload.
19366 * input.h (make_location): Declare.
19368 2017-08-08 Alan Modra <amodra@gmail.com>
19369 H.J. Lu <hongjiu.lu@intel.com>
19372 * gcc.c (NO_PIE_SPEC): Delete.
19373 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
19375 (LINK_PIE_SPEC): ..to here.
19376 (LINK_COMMAND_SPEC): Support -no-pie.
19377 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
19378 chain of crtbegin*.o selection, update for PIE_SPEC changes and
19380 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
19381 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
19382 (ENDFILE_CRTEND_SPEC): Similarly.
19384 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
19387 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
19388 (mstack-protector-guard-offset=): Ditto.
19389 * config/i386/i386.c (ix86_option_override): Handle
19390 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
19392 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
19393 ix86_stack_protect_guard_offset variables.
19394 (TARGET_STACK_PROTECT_GUARD): Always define.
19395 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
19396 and -mstack-protector-guard-offset= options.
19398 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19400 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
19401 boundary case for the last candidate.
19403 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19405 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
19406 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
19408 2017-08-08 Tamar Christina <tamar.christina@arm.com>
19410 PR middle-end/19706
19411 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
19412 * config/aarch64/aarch64-builtins.c
19413 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
19414 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
19415 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
19417 2017-08-08 Tamar Christina <tamar.christina@arm.com>
19418 Andrew Pinski <pinskia@gmail.com>
19420 PR middle-end/19706
19421 * internal-fn.def (XORSIGN): New.
19422 * optabs.def (xorsign_optab): New.
19423 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
19424 (convert_expand_mult_copysign): New.
19425 (pass_optimize_widening_mul::execute): Call
19426 convert_expand_mult_copysign.
19428 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19430 PR tree-optimization/81354
19431 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
19432 Insert on edges rather than explicitly creating landing pads.
19433 (analyze_candidates_and_replace): Commit edge inserts.
19435 2017-08-08 Richard Biener <rguenther@suse.de>
19437 PR middle-end/81719
19438 * tree-ssa-loop-niter.c: Include tree-dfa.h.
19439 (expand_simple_operations): Also look through ADDR_EXPRs with
19440 MEM_REF bases treating them as POINTER_PLUS_EXPR.
19442 2017-08-08 Richard Biener <rguenther@suse.de>
19444 PR tree-optimization/81723
19445 * tree-vect-slp.c (struct bst_traits): New hash traits.
19446 (bst_fail): New global.
19447 (vect_build_slp_tree_2): New worker, split out from ...
19448 (vect_build_slp_tree): ... this now wrapping it with using
19449 bst_fail set to cache SLP tree build fails. Properly handle
19451 (vect_analyze_slp_instance): Allocate and free bst_fail.
19453 2017-08-08 Martin Liska <mliska@suse.cz>
19456 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
19457 LABEL_DECLs that can be from a different function.
19459 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19461 PR tree-optimization/81744
19462 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
19463 loop's number of iterations.
19465 2017-08-08 Martin Liska <mliska@suse.cz>
19467 * asan.c: Include header files.
19468 * attribs.c (build_decl_attribute_variant): New function moved
19470 (build_type_attribute_qual_variant): Likewise.
19471 (cmp_attrib_identifiers): Likewise.
19472 (simple_cst_list_equal): Likewise.
19473 (omp_declare_simd_clauses_equal): Likewise.
19474 (attribute_value_equal): Likewise.
19475 (comp_type_attributes): Likewise.
19476 (build_type_attribute_variant): Likewise.
19477 (lookup_ident_attribute): Likewise.
19478 (remove_attribute): Likewise.
19479 (merge_attributes): Likewise.
19480 (merge_type_attributes): Likewise.
19481 (merge_decl_attributes): Likewise.
19482 (merge_dllimport_decl_attributes): Likewise.
19483 (handle_dll_attribute): Likewise.
19484 (attribute_list_equal): Likewise.
19485 (attribute_list_contained): Likewise.
19486 * attribs.h (lookup_attribute): New function moved from tree.[ch].
19487 (lookup_attribute_by_prefix): Likewise.
19488 * bb-reorder.c: Include header files.
19489 * builtins.c: Likewise.
19490 * calls.c: Likewise.
19491 * cfgexpand.c: Likewise.
19492 * cgraph.c: Likewise.
19493 * cgraphunit.c: Likewise.
19494 * convert.c: Likewise.
19495 * dwarf2out.c: Likewise.
19496 * final.c: Likewise.
19497 * fold-const.c: Likewise.
19498 * function.c: Likewise.
19499 * gimple-expr.c: Likewise.
19500 * gimple-fold.c: Likewise.
19501 * gimple-pretty-print.c: Likewise.
19502 * gimple.c: Likewise.
19503 * gimplify.c: Likewise.
19504 * hsa-common.c: Likewise.
19505 * hsa-gen.c: Likewise.
19506 * internal-fn.c: Likewise.
19507 * ipa-chkp.c: Likewise.
19508 * ipa-cp.c: Likewise.
19509 * ipa-devirt.c: Likewise.
19510 * ipa-fnsummary.c: Likewise.
19511 * ipa-inline.c: Likewise.
19512 * ipa-visibility.c: Likewise.
19514 * lto-cgraph.c: Likewise.
19515 * omp-expand.c: Likewise.
19516 * omp-general.c: Likewise.
19517 * omp-low.c: Likewise.
19518 * omp-offload.c: Likewise.
19519 * omp-simd-clone.c: Likewise.
19520 * opts-global.c: Likewise.
19521 * passes.c: Likewise.
19522 * predict.c: Likewise.
19523 * sancov.c: Likewise.
19524 * sanopt.c: Likewise.
19525 * symtab.c: Likewise.
19526 * toplev.c: Likewise.
19527 * trans-mem.c: Likewise.
19528 * tree-chkp.c: Likewise.
19529 * tree-eh.c: Likewise.
19530 * tree-into-ssa.c: Likewise.
19531 * tree-object-size.c: Likewise.
19532 * tree-parloops.c: Likewise.
19533 * tree-profile.c: Likewise.
19534 * tree-ssa-ccp.c: Likewise.
19535 * tree-ssa-live.c: Likewise.
19536 * tree-ssa-loop.c: Likewise.
19537 * tree-ssa-sccvn.c: Likewise.
19538 * tree-ssa-structalias.c: Likewise.
19539 * tree-ssa.c: Likewise.
19540 * tree-streamer-in.c: Likewise.
19541 * tree-vectorizer.c: Likewise.
19542 * tree-vrp.c: Likewise.
19543 * tsan.c: Likewise.
19544 * ubsan.c: Likewise.
19545 * varasm.c: Likewise.
19546 * varpool.c: Likewise.
19547 * tree.c: Remove functions moved to attribs.[ch].
19548 * tree.h: Likewise.
19549 * config/aarch64/aarch64.c: Add attrs.h header file.
19550 * config/alpha/alpha.c: Likewise.
19551 * config/arc/arc.c: Likewise.
19552 * config/arm/arm.c: Likewise.
19553 * config/avr/avr.c: Likewise.
19554 * config/bfin/bfin.c: Likewise.
19555 * config/c6x/c6x.c: Likewise.
19556 * config/cr16/cr16.c: Likewise.
19557 * config/cris/cris.c: Likewise.
19558 * config/darwin.c: Likewise.
19559 * config/epiphany/epiphany.c: Likewise.
19560 * config/fr30/fr30.c: Likewise.
19561 * config/frv/frv.c: Likewise.
19562 * config/ft32/ft32.c: Likewise.
19563 * config/h8300/h8300.c: Likewise.
19564 * config/i386/winnt.c: Likewise.
19565 * config/ia64/ia64.c: Likewise.
19566 * config/iq2000/iq2000.c: Likewise.
19567 * config/lm32/lm32.c: Likewise.
19568 * config/m32c/m32c.c: Likewise.
19569 * config/m32r/m32r.c: Likewise.
19570 * config/m68k/m68k.c: Likewise.
19571 * config/mcore/mcore.c: Likewise.
19572 * config/microblaze/microblaze.c: Likewise.
19573 * config/mips/mips.c: Likewise.
19574 * config/mmix/mmix.c: Likewise.
19575 * config/mn10300/mn10300.c: Likewise.
19576 * config/moxie/moxie.c: Likewise.
19577 * config/msp430/msp430.c: Likewise.
19578 * config/nds32/nds32-isr.c: Likewise.
19579 * config/nds32/nds32.c: Likewise.
19580 * config/nios2/nios2.c: Likewise.
19581 * config/nvptx/nvptx.c: Likewise.
19582 * config/pa/pa.c: Likewise.
19583 * config/pdp11/pdp11.c: Likewise.
19584 * config/powerpcspe/powerpcspe.c: Likewise.
19585 * config/riscv/riscv.c: Likewise.
19586 * config/rl78/rl78.c: Likewise.
19587 * config/rx/rx.c: Likewise.
19588 * config/s390/s390.c: Likewise.
19589 * config/sh/sh.c: Likewise.
19590 * config/sol2.c: Likewise.
19591 * config/sparc/sparc.c: Likewise.
19592 * config/spu/spu.c: Likewise.
19593 * config/stormy16/stormy16.c: Likewise.
19594 * config/tilegx/tilegx.c: Likewise.
19595 * config/tilepro/tilepro.c: Likewise.
19596 * config/v850/v850.c: Likewise.
19597 * config/vax/vax.c: Likewise.
19598 * config/visium/visium.c: Likewise.
19599 * config/xtensa/xtensa.c: Likewise.
19601 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
19604 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
19605 constraints since the -mupper-regs-* switches have been
19607 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
19608 into a vector from a double word element that was extracted from
19609 another vector, and eliminate extra XXPERMDI instructions.
19610 (vsx_concat_<mode>_2): Likewise.
19611 (vsx_concat_<mode>_3): Likewise.
19612 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
19613 concat to allow optimizing inserts from previous extracts.
19615 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
19617 * config/i386/i386.c (ix86_stack_protect_guard): Generate
19618 memory reference to a SSP offset in TLS address space.
19619 (ix86_print_operand) <case '@'>: Remove.
19620 (ix86_print_operand_punct_valid_p): Remove '@' code.
19621 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
19622 UNSPEC_SP_TLS_TEST.
19623 (stack_tls_protect_set_<mode>): Remove.
19624 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
19625 (stack_tls_protect_test_<mode>): Remove.
19626 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
19628 2017-08-07 Olivier Hainque <hainque@adacore.com>
19631 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
19633 2017-08-07 Douglas Rupp <rupp@adacore.com>
19635 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
19636 variable was referenced as multidir in command.
19638 2017-08-07 Jakub Jelinek <jakub@redhat.com>
19641 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
19644 2017-08-07 Martin Liska <mliska@suse.cz>
19646 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
19647 * config/rl78/rl78.c: Add include of attribs.h.
19648 * config/sh/sh.c: Likewise.
19649 * config/v850/v850.c: Likewise.
19651 2017-08-07 Tom de Vries <tom@codesourcery.com>
19653 PR middle-end/78266
19654 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
19656 2017-08-07 Martin Liska <mliska@suse.cz>
19658 * config/mips/mips.c: Include attribs.h.
19660 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
19663 * doc/invoke.texi: Document change in behvaior for -Ofast for
19666 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
19668 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
19670 (aarch64_pop_regs): Likewise.
19671 (aarch64_gen_load_pair): Likewise.
19672 (aarch64_save_callee_saves): Likewise.
19673 (aarch64_restore_callee_saves): Likewise.
19675 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19677 * config/i386/i386.c: Revert the last change.
19679 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19682 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
19684 (ix86_finalize_stack_frame_flags): This. Also clear
19685 frame_pointer_needed if -fno-omit-frame-pointer is used without
19687 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
19688 with ix86_finalize_stack_frame_flags.
19689 (ix86_expand_epilogue): Likewise.
19690 (ix86_expand_split_stack_prologue): Likewise.
19692 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19695 * config/i386/i386.c (get_builtin_code_for_version): Set priority
19696 to P_AES for Westmere.
19698 2017-08-07 Jonathan Yong <10walls@gmail.com>
19700 * config/i386/mingw.opt (fset-stack-executable): Removed.
19701 * config/i386/cygming.opt (fset-stack-executable): Moved
19703 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
19705 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
19707 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
19709 2017-08-07 Marek Polacek <polacek@redhat.com>
19711 PR middle-end/81737
19712 * fold-const.c (fold_indirect_ref_1): Check type_domain.
19714 2017-08-07 Martin Liska <mliska@suse.cz>
19716 * attribs.h (canonicalize_attr_name): New function.
19717 (cmp_attribs): Move from c-format.c and adjusted.
19718 (is_attribute_p): Moved from tree.h.
19719 * tree-inline.c: Add new includes.
19720 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
19721 (private_is_attribute_p): Remove.
19722 (private_lookup_attribute): Likewise.
19723 (private_lookup_attribute_by_prefix): Simplify.
19724 (remove_attribute): Use is_attribute_p.
19725 * tree.h: Remove removed declarations.
19727 2017-08-07 Jakub Jelinek <jakub@redhat.com>
19729 PR middle-end/81698
19730 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
19731 instead of computing it in the function. Formatting fix.
19732 (expand_case): Don't rely on default_edge being the first edge,
19733 clear it if removing it, pass default_edge to
19734 emit_case_dispatch_table.
19735 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
19738 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
19740 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
19741 insn in the function, emit NOP after the insn.
19743 2017-08-06 Tom de Vries <tom@codesourcery.com>
19745 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
19748 2017-08-06 Tom de Vries <tom@codesourcery.com>
19750 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
19753 2017-08-04 Yury Gribov <tetra2005@gmail.com>
19755 PR tree-optimization/57371
19756 * match.pd: New pattern.
19758 2017-08-04 Marek Polacek <polacek@redhat.com>
19760 PR middle-end/81695
19761 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
19762 perform the computation in offset_int.
19764 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19766 PR tree-optimization/81136
19767 * tree-vectorizer.h: Include tree-hash-traits.h.
19768 (vec_base_alignments): New typedef.
19769 (vec_info): Add a base_alignments field.
19770 (vect_record_base_alignments): Declare.
19771 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
19773 (DR_IS_CONDITIONAL_IN_STMT): New macro.
19774 (create_data_ref): Add an is_conditional_in_stmt argument.
19775 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
19776 the is_conditional_in_stmt field.
19777 (data_ref_loc): Add an is_conditional_in_stmt field.
19778 (get_references_in_stmt): Set the is_conditional_in_stmt field.
19779 (find_data_references_in_stmt): Update call to create_data_ref.
19780 (graphite_find_data_references_in_stmt): Likewise.
19781 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
19782 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
19783 (vect_record_base_alignment): New function.
19784 (vect_record_base_alignments): Likewise.
19785 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
19786 for nested statements even if we fail to compute a misalignment.
19787 Use pooled base alignments for unconditional references.
19788 (vect_find_same_alignment_drs): Compare base addresses instead
19790 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
19791 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
19793 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19795 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
19796 Add an explicit name for the enum. Use auto_vec for slp_instances
19797 and grouped_stores.
19798 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
19800 (_bb_vec_info): Add a constructor and destructor.
19801 (vinfo_for_stmt): Return NULL for uids of -1 as well.
19802 (destroy_loop_vec_info): Delete.
19803 (vect_destroy_datarefs): Likewise.
19804 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
19805 (vec_info::vec_info): New function.
19806 (vec_info::~vec_info): Likewise.
19807 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
19808 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
19809 destroy_loop_vec_info.
19810 * tree-vect-loop.c (new_loop_vec_info): Replace with...
19811 (_loop_vec_info::_loop_vec_info): ...this.
19812 (destroy_loop_vec_info): Replace with...
19813 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
19814 the stmt_vec_infos. Leave handling of vec_info information to its
19815 destructor. Remove explicit vector releases.
19816 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
19817 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
19818 * tree-vect-slp.c (new_bb_vec_info): Replace with...
19819 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
19820 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
19821 (destroy_bb_vec_info): Replace with...
19822 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
19823 information to its destructor.
19824 (vect_slp_analyze_bb_1): Use new and delete instead of
19825 new_bb_vec_info and destroy_bb_vec_info.
19826 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
19829 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19831 * tree-data-ref.h (subscript): Add access_fn field.
19832 (data_dependence_relation): Add could_be_independent_p.
19833 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
19834 (same_access_functions): Move to tree-data-ref.c.
19835 * tree-data-ref.c (ref_contains_union_access_p): New function.
19836 (access_fn_component_p): Likewise.
19837 (access_fn_components_comparable_p): Likewise.
19838 (dr_analyze_indices): Add a reference to access_fn_component_p.
19839 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
19841 (constant_access_functions): Likewise.
19842 (add_other_self_distances): Likewise.
19843 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
19844 (initialize_data_dependence_relation): Use XCNEW and remove
19845 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
19846 of access functions that have the same type. Allow the
19847 subsequence to end with different bases in some circumstances.
19848 Record the chosen access functions in SUB_ACCESS_FN.
19849 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
19850 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
19851 (subscript_dependence_tester_1): Likewise dra and drb.
19852 (build_classic_dist_vector): Update calls accordingly.
19853 (subscript_dependence_tester): Likewise.
19854 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
19855 DDR_COULD_BE_INDEPENDENT_P.
19856 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
19857 comp_alias_ddrs instead of may_alias_ddrs.
19858 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
19860 (vect_analyze_data_ref_dependence): Use it if
19861 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
19862 distance vectors if that fails.
19863 (dependence_distance_ge_vf): New function.
19864 (vect_prune_runtime_alias_test_list): Use it. Don't clear
19865 LOOP_VINFO_MAY_ALIAS_DDRS.
19867 2017-08-04 Richard Biener <rguenther@suse.de>
19869 PR middle-end/81705
19870 * fold-const.c (fold_binary_loc): Properly restrict
19871 minus_var0 && minus_var1 case when associating undefined overflow
19874 2017-08-04 Tom de Vries <tom@codesourcery.com>
19876 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
19878 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19880 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19881 Don't start diagnostic messages with a capital letter.
19882 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19884 (rs6000_invalid_builtin): Likewise.
19885 (rs6000_trampoline_init): Likewise.
19887 2017-08-03 Jakub Jelinek <jakub@redhat.com>
19890 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
19891 after setting changeable df flags.
19893 2017-08-03 Richard Biener <rguenther@suse.de>
19895 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
19896 up if the use is in USE - X.
19898 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
19900 * toplev.c (dumpfile.h): New include.
19901 (internal_error_reentered): New static function. Use it...
19902 (internal_error_function): ...here to handle reentered internal_error.
19904 2017-08-03 Richard Biener <rguenther@suse.de>
19906 PR middle-end/81148
19907 * fold-const.c (split_tree): Add minus_var and minus_con
19908 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
19909 here but always use minus_*.
19910 (associate_trees): Assert we never associate with MINUS_EXPR
19911 and NULL first operand. Do not recurse for PLUS_EXPR operands
19912 when associating as MINUS_EXPR either.
19913 (fold_binary_loc): Track minus_var and minus_con.
19915 2017-08-03 Tom de Vries <tom@codesourcery.com>
19918 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
19919 ACCEL_COMPILER, apply finish_options on
19920 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
19922 2017-08-03 Tom de Vries <tom@codesourcery.com>
19925 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
19926 function_entry_patch_area_size > 0.
19928 2017-08-03 Jakub Jelinek <jakub@redhat.com>
19931 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
19932 instead of 10??LU, perform unit multiplication in wide_int,
19933 don't change alloc_object_size_limit if the limit is larger
19936 PR tree-optimization/81655
19937 PR tree-optimization/81588
19938 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
19939 the case when ranges[i].low and high are 1 for unsigned type with
19942 PR middle-end/81052
19943 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
19944 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
19946 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19948 * tree-vrp.h: Add include guard.
19950 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
19953 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
19954 (ud2): New insn pattern.
19955 * config/i386/i386.c (ix86_expand_epilogue):
19956 For naked functions, generate ud2 instead of trap insn.
19958 2017-08-02 Marek Polacek <polacek@redhat.com>
19961 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
19963 2017-08-02 Tom de Vries <tom@codesourcery.com>
19964 Cesar Philippidis <cesar@codesourcery.com>
19966 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
19967 Add missing edge probabilities.
19969 2017-08-02 Tamar Christina <tamar.christina@arm.com>
19971 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
19972 Correct endianness.
19974 2017-08-02 Jakub Jelinek <jakub@redhat.com>
19976 PR middle-end/79499
19977 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
19978 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
19979 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
19980 prologue_seq sequences - if any.
19982 2017-08-02 Richard Biener <rguenther@suse.de>
19984 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
19985 via vectors if supported, integer extracts via punning if supported
19986 or otherwise vector extracts.
19988 2017-08-02 Richard Biener <rguenther@suse.de>
19990 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
19992 (bitmap_insert_into_set): ... this.
19994 2017-08-02 Richard Biener <rguenther@suse.de>
19996 PR tree-optimization/81633
19998 2015-08-17 Alan Hayward <alan.hayward@arm.com>
20000 PR tree-optimization/71752
20001 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
20003 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
20005 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
20006 (machine_function::call_ms2sysv_pad_out): Remove field.
20007 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
20008 (ix86_compute_frame_layout): Likewise.
20010 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
20013 * config/i386/i386.c (ix86_set_func_type): Disallow naked
20014 attribute with interrupt attribute.
20016 2017-08-01 Andrew Pinski <apinski@cavium.com>
20018 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
20019 BIT_INSERT_EXPR's operand 1
20020 to see if the types precision matches.
20022 2017-08-01 Martin Liska <mliska@suse.cz>
20024 PR middle-end/70140
20025 * builtins.c (expand_builtin_memcpy_args): Remove.
20026 (expand_builtin_memcpy): Call newly added function
20027 expand_builtin_memory_copy_args.
20028 (expand_builtin_memcpy_with_bounds): Likewise.
20029 (expand_builtin_mempcpy): Remove last argument.
20030 (expand_builtin_mempcpy_with_bounds): Likewise.
20031 (expand_builtin_memory_copy_args): New function created from
20032 expand_builtin_mempcpy_args with small modifications.
20033 (expand_builtin_mempcpy_args): Remove.
20034 (expand_builtin_stpcpy): Remove unused argument.
20035 (expand_builtin): Likewise.
20036 (expand_builtin_with_bounds): Likewise.
20038 2017-08-01 Martin Liska <mliska@suse.cz>
20041 Make mempcpy more optimal (PR middle-end/70140).
20043 2017-08-01 Jakub Jelinek <jakub@redhat.com>
20046 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
20047 __builtin_vec_cmpne verify both arguments are compatible vectors
20048 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
20049 verify arg1_type is a pointer or array type. For __builtin_vec_st,
20050 move computation of aligned to after checking the argument types.
20054 * config/rs6000/vsx.md (vextract_fp_from_shorth,
20055 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
20058 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
20059 Doug Rupp <rupp@adacore.com>
20060 Olivier Hainque <hainque@adacore.com>
20062 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
20063 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
20065 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
20066 for TARGET_OS_CPP_BUILTIN.
20067 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
20068 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
20070 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
20071 passing required abi options to the assembler for EABI configurations.
20072 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
20073 of .text.hot and .text.unlikely sections for kernel modules when
20074 using ARM style exceptions.
20075 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
20076 options. Add EXTRA_CC1_SPEC.
20077 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
20079 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
20081 (ARM_TARGET2_DWARF_FORMAT): Define.
20082 * config/arm/t-vxworks: Adjust multilib control to removal of the
20083 Diab command line options.
20085 2017-08-01 Martin Liska <mliska@suse.cz>
20087 PR gcov-profile/81561
20088 * gcov.c (unblock): Make unblocking safe as we need to preserve
20089 index correspondence of blocks and block_lists.
20091 2017-08-01 Richard Biener <rguenther@suse.de>
20093 PR tree-optimization/81181
20094 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
20095 (compute_antic): ... end of iteration here.
20097 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
20099 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
20100 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
20101 (ftree-slp-vectorize): Likewise.
20102 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
20103 can no longer be set independent of flag_tree_loop_vectorize.
20104 * omp-general.c (emp_max_vf): Likewise.
20105 * opts.c (enable_fdo_optimizations): Remove references to
20106 flag_tree_vectorize, these are now implicit.
20107 (common_handle_option): Remove handling for OPT_ftree_vectorize,
20108 and leave it for the options machinery.
20110 2017-08-01 Martin Liska <mliska@suse.cz>
20112 PR middle-end/70140
20113 * builtins.c (expand_builtin_memcpy_args): Remove.
20114 (expand_builtin_memcpy): Call newly added function
20115 expand_builtin_memory_copy_args.
20116 (expand_builtin_memcpy_with_bounds): Likewise.
20117 (expand_builtin_mempcpy): Remove last argument.
20118 (expand_builtin_mempcpy_with_bounds): Likewise.
20119 (expand_builtin_memory_copy_args): New function created from
20120 expand_builtin_mempcpy_args with small modifications.
20121 (expand_builtin_mempcpy_args): Remove.
20122 (expand_builtin_stpcpy): Remove unused argument.
20123 (expand_builtin): Likewise.
20124 (expand_builtin_with_bounds): Likewise.
20126 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
20129 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
20130 print "ds:" only for immediates in generic address space.
20132 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
20135 * config/i386/i386.c (ix86_funciton_naked): New prototype.
20136 (ix86_function_ok_for_sibcall): Return false for naked functions.
20138 2017-08-01 Richard Biener <rguenther@suse.de>
20140 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
20141 (compute_antic): Seed worklist with exit block predecessors.
20142 * cfganal.c (dfs_find_deadend): For a cycle return the source
20143 of the edge closing it.
20145 2017-08-01 Tamar Christina <tamar.christina@arm.com>
20147 * config/aarch64/aarch64.c
20148 (aarch64_can_const_movi_rtx_p): Move 0 check.
20150 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20152 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
20153 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
20155 * match.pd: Ditto in address comparison pattern.
20157 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20159 PR tree-optimization/81627
20160 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
20161 closed ssa form for store-store chain.
20163 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20165 PR tree-optimization/81620
20166 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
20167 for store-store chain.
20169 2017-08-01 Jakub Jelinek <jakub@redhat.com>
20171 PR tree-optimization/81588
20172 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
20173 ranges[i].in_p, invert comparison code ccode. For >/>=,
20174 swap rhs1 and rhs2 and comparison code unconditionally,
20175 for </<= don't do that. Don't swap rhs1/rhs2 again if
20176 ranges[i].in_p, instead invert comparison code ccode if
20177 opcode or oe->rank is BIT_IOR_EXPR.
20180 * optabs.def (vec_extract_optab, vec_init_optab): Change from
20181 a direct optab to conversion optab.
20182 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
20183 with GET_MODE_INNER as last argument instead of optab_handler.
20184 * expmed.c (extract_bit_field_1): Likewise. Use vector from
20185 vector extraction if possible and optab is available.
20186 * expr.c (store_constructor): Use convert_optab_handler instead
20187 of optab_handler. Use vector initialization from smaller
20188 vectors if possible and optab is available.
20189 * tree-vect-stmts.c (vectorizable_load): Likewise.
20190 * doc/md.texi (vec_extract, vec_init): Document that the optabs
20191 now have two modes.
20192 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
20193 of vec_init from half-sized vectors with the same element mode.
20194 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
20195 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
20196 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
20197 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
20198 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
20199 after mode in gen_vec_extract* calls.
20200 (vec_extract<mode>): Renamed to ...
20201 (vec_extract<mode><ssescalarmodelower>): ... this.
20202 (vec_extract<mode><ssehalfvecmodelower>): New expander.
20203 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
20204 element mode after mode in gen_vec_init* calls.
20205 (VEC_INIT_HALF_MODE): New mode iterator.
20206 (vec_init<mode>): Renamed to ...
20207 (vec_init<mode><ssescalarmodelower>): ... this.
20208 (vec_init<mode><ssehalfvecmodelower>): New expander.
20209 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
20210 (vec_extractv2sfsf): ... this.
20211 (vec_initv2sf): Renamed to ...
20212 (vec_initv2sfsf): ... this.
20213 (vec_extractv2si): Renamed to ...
20214 (vec_extractv2sisi): ... this.
20215 (vec_initv2si): Renamed to ...
20216 (vec_initv2sisi): ... this.
20217 (vec_extractv4hi): Renamed to ...
20218 (vec_extractv4hihi): ... this.
20219 (vec_initv4hi): Renamed to ...
20220 (vec_initv4hihi): ... this.
20221 (vec_extractv8qi): Renamed to ...
20222 (vec_extractv8qiqi): ... this.
20223 (vec_initv8qi): Renamed to ...
20224 (vec_initv8qiqi): ... this.
20225 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
20226 (vec_init<mode>): Renamed to ...
20227 (vec_init<mode><VEC_base_l>): ... this.
20228 (vec_extract<mode>): Renamed to ...
20229 (vec_extract<mode><VEC_base_l>): ... this.
20230 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
20231 (vec_initv2sfsf): ... this.
20232 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
20233 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
20234 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
20235 element mode after mode in gen_vec_init* calls.
20236 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
20237 (vec_init<mode><Vel>): ... this.
20238 (vec_extract<mode>): Renamed to ...
20239 (vec_extract<mode><Vel>): ... this.
20240 * config/aarch64/iterators.md (Vel): New mode attribute.
20241 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
20242 Add element mode after mode in gen_vec_extract* calls.
20243 * config/s390/vector.md (non_vec_l): New mode attribute.
20244 (vec_extract<mode>): Renamed to ...
20245 (vec_extract<mode><non_vec_l>): ... this.
20246 (vec_init<mode>): Renamed to ...
20247 (vec_init<mode><non_vec_l>): ... this.
20248 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
20249 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
20251 * config/arm/iterators.md (V_elem_l): New mode attribute.
20252 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
20253 (vec_extract<mode><V_elem_l>): ... this.
20254 (vec_extractv2di): Renamed to ...
20255 (vec_extractv2didi): ... this.
20256 (vec_init<mode>): Renamed to ...
20257 (vec_init<mode><V_elem_l>): ... this.
20258 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
20259 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
20260 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
20261 Add element mode after gen_vec_extract* calls.
20262 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
20263 (vec_init<mode><unitmode>): ... this.
20264 (vec_extract<mode>): Renamed to ...
20265 (vec_extract<mode><unitmode>): ... this.
20266 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
20267 (vec_init<mode><unitmode>): ... this.
20268 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
20269 (vec_initv2sfsf): ... this.
20270 (vec_extractv2sf): Renamed to ...
20271 (vec_extractv2sfsf): ... this.
20272 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
20273 Add element mode after gen_vec_extract* calls.
20274 * config/mips/mips.md (unitmode): New mode iterator.
20275 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
20276 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
20277 * config/spu/spu.md (inner_l): New mode attribute.
20278 (vec_init<mode>): Renamed to ...
20279 (vec_init<mode><inner_l>): ... this.
20280 (vec_extract<mode>): Renamed to ...
20281 (vec_extract<mode><inner_l>): ... this.
20282 * config/sparc/sparc.md (veltmode): New mode iterator.
20283 (vec_init<VMALL:mode>): Renamed to ...
20284 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
20285 * config/ia64/vect.md (vec_initv2si): Renamed to ...
20286 (vec_initv2sisi): ... this.
20287 (vec_initv2sf): Renamed to ...
20288 (vec_initv2sfsf): ... this.
20289 (vec_extractv2sf): Renamed to ...
20290 (vec_extractv2sfsf): ... this.
20291 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
20292 (vec_init<mode>): Renamed to ...
20293 (vec_init<mode><VEC_base_l>): ... this.
20294 (vec_extract<mode>): Renamed to ...
20295 (vec_extract<mode><VEC_base_l>): ... this.
20296 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
20297 (vec_initv2sfsf): ... this.
20298 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
20299 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
20300 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
20301 gen_vec_init* calls.
20303 2017-08-01 Richard Biener <rguenther@suse.de>
20305 PR tree-optimization/81297
20306 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
20307 TREE_OVERFLOW from INTEGER_CSTs.
20309 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
20311 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
20313 2017-07-31 Carl Love <cel@us.ibm.com>
20315 * config/rs6000/rs6000-c: Add support for built-in functions
20316 vector signed char vec_xl_be (signed long long, signed char *);
20317 vector unsigned char vec_xl_be (signed long long, unsigned char *);
20318 vector signed int vec_xl_be (signed long long, signed int *);
20319 vector unsigned int vec_xl_be (signed long long, unsigned int *);
20320 vector signed long long vec_xl_be (signed long long, signed long long *);
20321 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
20322 vector signed short vec_xl_be (signed long long, signed short *);
20323 vector unsigned short vec_xl_be (signed long long, unsigned short *);
20324 vector double vec_xl_be (signed long long, double *);
20325 vector float vec_xl_be (signed long long, float *);
20326 * config/rs6000/altivec.h (vec_xl_be): Add #define.
20327 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
20328 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
20330 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
20331 (altivec_expand_builtin): Add switch statement to call
20332 altivec_expand_xl_be for each builtin.
20333 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
20334 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
20335 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
20336 __builtin_vsx_le_be_v16qi.
20337 * doc/extend.texi: Update the built-in documentation file for the
20338 new built-in functions.
20340 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
20343 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
20345 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
20347 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20349 * config.gcc: Add z14.
20350 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
20351 CPU model numbers for z13s and z14.
20352 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
20354 * config/s390/s390-opts.h (enum processor_type): Rename
20355 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
20356 * config/s390/s390.c (processor_table): Add field for CPU name to
20357 be passed to Binutils.
20358 (s390_asm_output_machine_for_arch): Use the new field in
20359 processor_table for Binutils.
20360 (s390_expand_builtin): Replace arch12 with z14.
20361 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
20362 (s390_get_sched_attrmask): Likewise.
20363 (s390_get_unit_mask): Likewise.
20364 * config/s390/s390.opt: Add z14 to processor_type enum.
20366 2017-07-31 Martin Jambor <mjambor@suse.cz>
20369 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
20370 regardless of optimization level.
20372 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
20373 Martin Liska <mliska@suse.cz>
20375 * predict.def: Remove old comment and adjust probability.
20376 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
20377 PREDICT statements.
20379 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
20382 * config/i386/i386.c (ix86_function_naked): New function.
20383 (ix86_can_use_return_insn_p): Return false for naked functions.
20384 (ix86_expand_prologue): Skip prologue for naked functions.
20385 (ix86_expand_epilogue): Skip epilogue for naked functions
20386 and emit trap instruction.
20387 (ix86_warn_func_return): New function.
20388 (ix86_attribute_table): Add "naked" attribute specification.
20389 (TARGET_WARN_FUNC_RETURN): Define.
20390 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
20392 2017-07-31 Martin Liska <mliska@suse.cz>
20394 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
20395 (dump_gimple_bb_header): Always dump BB info.
20396 (pp_cfg_jump): Do not append info about BB when dumping a jump.
20398 2017-07-31 Martin Liska <mliska@suse.cz>
20401 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
20402 also with current_function_decl non-null equality.
20404 2017-07-31 Jakub Jelinek <jakub@redhat.com>
20407 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
20408 change type to the element type, instead add eltype variable and
20409 use it where we are interested in the element type.
20411 PR tree-optimization/81603
20412 * ipa-polymorphic-call.c
20413 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
20414 offset arithmetic in offset_int, bail out if the resulting bit offset
20415 doesn't fit into shwi.
20417 2017-07-31 Martin Liska <mliska@suse.cz>
20419 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
20420 (gimplify_save_expr): Fix comment.
20422 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
20425 * config/i386/i386.c (ix86_function_arg): Update arguments for
20427 (ix86_compute_frame_layout): Set the initial stack offset to
20428 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
20429 INCOMING_FRAME_SP_OFFSET.
20430 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
20431 stack before exception handler returns.
20432 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
20433 the 'ERROR_CODE' for exception handler.
20435 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
20437 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
20438 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
20439 (ASM_OUTPUT_REG_POP): Ditto.
20440 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
20441 instead of asm_fprintf to output pure string.
20443 2017-07-29 Jakub Jelinek <jakub@redhat.com>
20445 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
20446 to imported_module_or_decl hook.
20447 (debug_nothing_tree_tree_tree_bool): Remove.
20448 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
20449 * debug.c (do_nothing_debug_hooks): Use
20450 debug_nothing_tree_tree_tree_bool_bool instead of
20451 debug_nothing_tree_tree_tree_bool.
20452 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20453 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
20454 * sdbout.c (sdb_debug_hooks): Likewise.
20455 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
20456 (gen_namespace_die): Add DW_AT_export_symbols attribute if
20458 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
20459 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
20460 attribute, don't add anything.
20462 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20464 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
20465 (fold_build2_stat_loc): Likewise.
20466 (fold_build3_stat_loc): Likewise.
20467 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
20468 (fold_build1_loc): Remove macro.
20469 (fold_build2_loc): Likewise.
20470 (fold_build3_loc): Likewise.
20472 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20474 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
20475 (gimple_build_debug_bind_source_stat): Likewise.
20476 * gimple.h (gimple_build_debug_bind): Remove macro.
20477 (gimple_build_debug_bind_source): Likewise.
20479 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20481 * bitmap.c (bitmap_alloc): Adjust.
20482 (bitmap_gc_alloc): Likewise.
20483 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
20485 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20487 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
20488 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
20489 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
20490 (bitmap_gc_alloc_stat): Likewise.
20491 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
20493 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20495 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
20496 * rtl.h (shallow_copy_rtx): Remove macro.
20498 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20500 * emit-rtl.c (gen_raw_REG): Adjust.
20501 * gengenrtl.c (gendef): Likewise.
20502 * rtl.c (rtx_alloc_stat): Remove _stat from name.
20503 * rtl.h (rtx_alloc): Remove macro.
20505 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20507 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
20508 (build_tree_list_stat): Likewise.
20509 * tree.h (build_tree_list): Remove macro.
20510 (build_tree_list_vec): Likewise.
20512 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20514 * tree.c (make_vector_stat): Remove _stat from name.
20515 (build_vector_stat): Likewise.
20516 * tree.h (make_vector_stat): Remove macro.
20517 (build_vector_stat): Likewise.
20519 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20521 * tree.h (build_var_debug_value): Remove prototype.
20523 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20525 * tree.c (tree_cons_stat): Remove _stat from name.
20526 * tree.h (tree_cons): Remove macro.
20528 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20530 * tree.c (build_vl_exp_stat): Remove _stat from name.
20531 * tree.h (build_vl_exp): Remove macro.
20533 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20535 * tree.c (build_decl_stat): Remove _stat from name.
20536 * tree.h (build_decl): Remove macro.
20538 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20540 * gimple.c (gimple_build_with_ops_stat): Adjust.
20541 (gimple_alloc_stat): Remove _stat from name.
20542 * gimple.h (gimple_alloc): Remove macro.
20544 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20546 * tree.c (make_tree_vec_stat): Remove _stat from name.
20547 (grow_tree_vec_stat): Likewise.
20548 * tree.h (make_tree_vec_stat): Adjust prototype.
20549 (grow_tree_vec_stat): Likewise.
20550 (make_tree_vec): Remove macro.
20551 (grow_tree_vec): Likewise.
20553 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20555 * fold-const.c (fold_build1_stat_loc): Adjust.
20556 (fold_build2_stat_loc): Likewise.
20557 (fold_build3_stat_loc): Likewise.
20558 * tree.c (build0_stat): Remove _stat from name.
20559 (build1_stat): Likewise.
20560 (build2_stat): Likewise.
20561 (build3_stat): Likewise.
20562 (build4_stat): Likewise.
20563 (build5_stat): Likewise.
20564 * tree.h (build1_loc): Remove macro, and rename _stat function
20566 (build2_loc): Likewise.
20567 (build3_loc): Likewise.
20568 (build4_loc): Likewise.
20569 (build5_loc): Likewise.
20571 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20573 * tree.c (make_int_cst_stat): Remove _stat from name.
20574 * tree.h (make_int_cst_stat): Adjust prototype.
20575 (make_int_cst): Remove macro.
20577 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20579 * tree.c (make_tre_binfo_stat): Remove _stat from name.
20580 * tree.h (make_tree_binfo_stat): Adjust prototype.
20581 (make_tree_binfo): Remove.
20583 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20585 * tree.c (copy_node_stat): Rename to copy_node.
20586 (build_distinct_type_copy): Adjust.
20587 * tree.h (copy_node_stat): Adjust prototype.
20588 (copy_node): Remove macro.
20590 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20592 * tree.c (make_node_stat): rename to make_node.
20593 (build_tree_list_stat): Adjust.
20594 (build0_stat): Likewise.
20595 (build2_stat): Likewise.
20596 (build3_stat): Likewise.
20597 (build4_stat): Likewise.
20598 (build5_stat): Likewise.
20599 (build_decl_stat): Likewise.
20600 * tree.h (make_node_stat): Adjust prototype.
20601 (make_node): remove macro.
20603 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
20605 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
20606 (PPC_FEATURE2_SCV): Likewise.
20607 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
20609 2017-07-28 Tamar Christina <tamar.christina@arm.com>
20611 * config/aarch64/aarch64.c
20612 (aarch64_internal_mov_immediate): Add new special pattern.
20613 * config/aarch64/aarch64.md (*movdi_aarch64):
20614 Add reg/32bit const mov case.
20616 2017-07-28 Tamar Christina <tamar.christina@arm.com>
20617 Richard Sandiford <richard.sandiford@linaro.org>
20619 * config/aarch64/aarch64.md (mov<mode>): Generalize.
20620 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20621 Add integer and movi cases.
20622 (movi-split-hf-df-sf split, fp16): New.
20623 (enabled): Added TARGET_FP_F16INST.
20624 * config/aarch64/iterators.md (GPF_HF): New.
20625 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
20627 2017-07-28 Tamar Christina <tamar.christina@arm.com>
20629 * config/aarch64/aarch64.c
20630 (aarch64_simd_container_mode): Add prototype.
20631 (aarch64_expand_mov_immediate): Add HI support.
20632 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
20633 (aarch64_can_const_movi_rtx_p): New.
20634 (aarch64_preferred_reload_class):
20635 Remove restrictions of using FP registers for certain SIMD operations.
20636 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
20637 (aarch64_valid_floating_const): Add integer move validation.
20638 (aarch64_simd_imm_scalar_p): Remove.
20639 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
20640 (aarch64_legitimate_constant_p): Expand list of supported cases.
20641 * config/aarch64/aarch64-protos.h
20642 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
20643 (aarch64_reinterpret_float_as_int): New.
20644 (aarch64_simd_imm_scalar_p): Remove.
20645 * config/aarch64/constraints.md (Uvi): New.
20646 (Dd): Split into Ds and new Dd.
20647 * config/aarch64/aarch64.md (*movsi_aarch64):
20649 (*movdi_aarch64): Add SIMD mov case.
20651 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20653 * tree-predcom.c: (struct chain): Handle store-store chain in which
20654 stores for elimination only store loop invariant values.
20655 (execute_pred_commoning_chain): Ditto.
20656 (prepare_initializers_chain_store_elim): Ditto.
20657 (prepare_finalizers): Ditto.
20658 (is_inv_store_elimination_chain): New function.
20659 (initialize_root_vars_store_elim_1): New function.
20661 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20663 * tree-predcom.c: Revise general description of the pass.
20664 (enum chain_type): New enum type for store elimination.
20665 (struct chain): New field supporting store elimination.
20666 (struct component): Ditto.
20667 (dump_chain): Dump store-stores chain.
20668 (release_chain): Release resources.
20669 (split_data_refs_to_components): Compute and create component
20670 contains only stores for elimination.
20671 (get_chain_last_ref_at): New function.
20672 (make_invariant_chain): Initialization.
20673 (make_rooted_chain): Specify chain type in parameter and record it.
20674 (add_looparound_copies): Skip for store-stores chain.
20675 (determine_roots_comp): Compute type of chain and pass it to
20677 (initialize_root_vars_store_elim_2): New function.
20678 (finalize_eliminated_stores): New function.
20679 (remove_stmt): Handle store for elimination.
20680 (execute_pred_commoning_chain): Execute predictive commoning on
20681 store-store chains.
20682 (determine_unroll_factor): Skip unroll for store-stores chain.
20683 (prepare_initializers_chain_store_elim): New function.
20684 (prepare_initializers_chain): Hanlde store-store chain.
20685 (prepare_finalizers_chain, prepare_finalizers): New function.
20686 (tree_predictive_commoning_loop): Return integer value indicating
20687 if loop is unrolled or lcssa form is corrupted.
20688 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
20690 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20692 * tree-predcom.c (initialize_root): Delete.
20693 (execute_pred_commoning_chain): Initialize root vars and replace
20694 reference of non-combined chain directly, rather than call above
20697 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20699 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
20700 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
20702 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20704 * tree-predcom.c (struct chain): New field init_seq.
20705 (release_chain): Release init_seq.
20706 (prepare_initializers_chain): Record intialization stmts in above
20708 (insert_init_seqs): New function.
20709 (tree_predictive_commoning_loop): Call insert_init_seqs.
20711 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20713 * tree-predcom.c (determine_roots_comp): Skip trivial components.
20715 2017-07-28 Richard Biener <rguenther@suse.de>
20717 * match.pd: Remove superfluous :c.
20718 * genmatch.c (simplify::id): Add member.
20719 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
20721 (current_id): New global.
20722 (dt_node::parent): Move from ...
20723 (dt_operand::parent): ... here. Add for_id member.
20724 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
20725 (decision_tree::find_node): Relax order requirement when
20726 merging DT_TRUE nodes to ones inbetween the current simplify
20727 and the one we try to merge with. Add diagnostic whenever
20728 we need to enforce pattern order by not merging.
20729 (decision_tree::insert): Set current_id.
20730 (decision_tree::print_node): Dump parent node and for_id.
20731 (parser::last_id): Add member.
20732 (parser::push_simplify): Assign unique id.
20733 (parser::parser): Initialize last_id.
20735 2017-07-28 Martin Liska <mliska@suse.cz>
20738 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
20739 gimple_build_debug_bind.
20741 2017-07-28 Richard Biener <rguenther@suse.de>
20743 PR tree-optimization/81502
20744 * match.pd: Add pattern combining BIT_INSERT_EXPR with
20746 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
20748 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
20749 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
20750 for BIT_FIELD_REF args.
20751 * fold-const.c (make_bit_field_ref): Likewise.
20752 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
20754 2017-07-28 Jakub Jelinek <jakub@redhat.com>
20757 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
20758 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
20759 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
20760 Or it into SANITIZER_UNDEFINED.
20761 * ubsan.c: Include gimple-fold.h and varasm.h.
20762 (ubsan_expand_ptr_ifn): New function.
20763 (instrument_pointer_overflow): New function.
20764 (maybe_instrument_pointer_overflow): New function.
20765 (instrument_object_size): Formatting fix.
20766 (pass_ubsan::execute): Call instrument_pointer_overflow
20767 and maybe_instrument_pointer_overflow.
20768 * internal-fn.c (expand_UBSAN_PTR): New function.
20769 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
20770 * sanitizer.def (__ubsan_handle_pointer_overflow,
20771 __ubsan_handle_pointer_overflow_abort): New builtins.
20772 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
20773 * internal-fn.def (UBSAN_PTR): New internal function.
20774 * opts.c (sanitizer_opts): Add pointer-overflow.
20775 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
20776 * fold-const.c (build_range_check): Compute pointer range check in
20777 integral type if pointer arithmetics would be needed. Formatting
20780 2017-07-28 Martin Liska <mliska@suse.cz>
20783 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
20784 parameters that are of a variable-length.
20786 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
20788 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
20790 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
20791 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
20792 (CRT_CALL_STATIC_FUNCTION): Likewise.
20793 (ASM_DEFAULT_SPEC): New define.
20794 (ASM_SPEC32): Likewise.
20795 (ASM_SPEC64): Likewise.
20796 (ASM_SPEC_COMMON): Likewise.
20797 (ASM_SPEC): Likewise.
20798 (INVALID_64BIT): Likewise.
20799 (LINK_OS_DEFAULT_SPEC): Likewise.
20800 (LINK_OS_SPEC32): Likewise.
20801 (LINK_OS_SPEC64): Likewise.
20802 (POWERPC_LINUX): Likewise.
20803 (PTRDIFF_TYPE): Likewise.
20804 (RESTORE_FP_PREFIX): Likewise.
20805 (RESTORE_FP_SUFFIX): Likewise.
20806 (SAVE_FP_PREFIX): Likewise.
20807 (SAVE_FP_SUFFIX): Likewise.
20808 (SIZE_TYPE): Likewise.
20809 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
20810 (TARGET_64BIT): Likewise.
20811 (TARGET_64BIT): Likewise.
20812 (TARGET_AIX): Likewise.
20813 (WCHAR_TYPE_SIZE): Likewise.
20814 (WCHAR_TYPE): Undefine.
20815 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
20816 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
20817 (CPP_OS_RTEMS_SPEC): Delete.
20818 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
20819 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
20821 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
20823 2017-07-28 Jakub Jelinek <jakub@redhat.com>
20825 PR tree-optimization/81578
20826 * tree-parloops.c (build_new_reduction): Bail out if
20827 reduction_code isn't one of the standard OpenMP reductions.
20828 Move the details printing after that decision.
20830 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20832 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
20833 related to reload_in_progress.
20834 (splat_input_operand): Likewise.
20835 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
20837 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
20839 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
20840 (TARGET_INSTANTIATE_DECLS): Likewise.
20841 (legitimate_indexed_address_p): Delete reload_in_progress code.
20842 (rs6000_debug_legitimate_address_p): Likewise.
20843 (rs6000_eliminate_indexed_memrefs): Likewise.
20844 (rs6000_emit_le_vsx_store): Likewise.
20845 (rs6000_emit_move_si_sf_subreg): Likewise.
20846 (rs6000_emit_move): Likewise.
20847 (register_to_reg_type): Likewise.
20848 (rs6000_pre_atomic_barrier): Likewise.
20849 (rs6000_machopic_legitimize_pic_address): Likewise.
20850 (rs6000_allocate_stack_temp): Likewise.
20851 (rs6000_address_for_fpconvert): Likewise.
20852 (rs6000_address_for_altivec): Likewise.
20853 (rs6000_secondary_memory_needed_rtx): Delete function.
20854 (rs6000_check_sdmode): Likewise.
20855 (rs6000_alloc_sdmode_stack_slot): Likewise.
20856 (rs6000_instantiate_decls): Likewise.
20857 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
20858 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
20859 Delete reload_in_progress.
20860 (*vec_reload_and_plus_<mptrsize>): Likewise.
20861 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
20862 (vsx_div_v2di): Likewise.
20863 (vsx_udiv_v2di): Likewise.
20865 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20867 * config/rs6000/rs6000.opt (mlra): Replace with stub.
20868 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
20869 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
20870 (rs6000_debug_reg_global): Delete print of LRA status.
20871 (rs6000_option_override_internal): Delete dead LRA related code.
20872 (rs6000_lra_p): Delete function.
20873 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
20875 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
20877 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
20878 * config/riscv/rtems.h: New file.
20880 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20881 Sudakshina Das <sudi.das@arm.com>
20883 * config/aarch64/aarch64.md
20884 (define_split for and<mode>3nr_compare): Move
20885 non aarch64_logical_operand to a register.
20886 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
20887 register immediate operand to a register.
20888 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
20890 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20892 PR middle-end/81564
20893 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
20895 2017-07-27 Richard Biener <rguenther@suse.de>
20897 PR tree-optimization/81573
20898 PR tree-optimization/81494
20899 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
20900 multi defuse cycle case.
20902 2017-07-27 Richard Biener <rguenther@suse.de>
20904 PR tree-optimization/81571
20905 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
20908 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
20910 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
20911 earlier and only if MASK_FPU is set. Adjust formatting.
20913 2017-07-27 Martin Liska <mliska@suse.cz>
20915 * opt-functions.awk: Add validation of value of Init.
20916 * optc-gen.awk: Pass new argument.
20918 2017-07-27 Martin Liska <mliska@suse.cz>
20920 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
20921 Fix wrong condition.
20923 2017-07-27 Martin Liska <mliska@suse.cz>
20925 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
20926 BBs and edges seen by autoFDO.
20928 2017-07-27 Richard Biener <rguenther@suse.de>
20930 PR tree-optimization/81502
20931 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
20932 with incompatible but same sized type.
20933 (execute_update_addresses_taken): Likewise.
20935 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
20937 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
20938 flag_tree_loop_vectorize rather than flag_tree_vectorize.
20940 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20943 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
20944 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
20945 Change s_operand to memory_operand.
20947 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
20949 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
20950 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
20951 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
20952 Emit instructions rather than returning an expression. Handle TFmode
20953 and KFmode by casting to TImode.
20954 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
20955 (rs6000_emit_le_vsx_store): Likewise.
20956 * config/rs6000/vsx.md (VSX_TI): New iterator.
20957 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
20958 (*vsx_le_undo_permute_<mode>): Likewise.
20959 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
20960 emit the split sequence.
20961 (*vsx_le_perm_store_<mode>): Likewise.
20963 2017-07-27 Jakub Jelinek <jakub@redhat.com>
20965 PR tree-optimization/81555
20966 PR tree-optimization/81556
20967 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
20968 if true, force CHANGED for the recursive invocation.
20969 (reassociate_bb): Remember original length of ops array, pass
20970 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
20972 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
20973 attributes for noipa attribute. For naked attribute use
20974 lookup_attribute first before lookup_attribute_spec.
20975 * final.c (rest_of_handle_final): Disable IPA RA for functions with
20977 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
20978 for functions with noipa attribute.
20979 (cgraph_externally_visible_p): Return true for functions with noipa
20981 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
20982 for functions with noipa attribute.
20983 * doc/extend.texi: Document noipa function attribute.
20984 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
20985 also for functions with noipa attribute.
20986 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
20988 2017-07-26 Andrew Pinski <apinski@cavium.com>
20990 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
20991 vec_unalign_load_cost and vec_unalign_store_cost.
20993 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
20995 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
20996 -mvsx-small-integer option.
20997 (ISA_3_0_MASKS_IEEE): Likewise.
20998 (OTHER_VSX_VECTOR_MASKS): Likewise.
20999 (POWERPC_MASKS): Likewise.
21000 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
21001 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
21002 code, only testing for DImode being allowed in non-VSX floating
21004 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
21005 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
21007 (rs6000_option_override_internal): Delete -mvsx-small-integer.
21008 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
21009 TARGET_P8_VECTOR test.
21010 (rs6000_secondary_reload_simple_move): Likewise.
21011 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
21012 since TARGET_P9_VECTOR was already tested.
21013 (rs6000_opt_masks): Remove -mvsx-small-integer.
21014 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
21015 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21017 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
21018 test for TARGET_VEXTRACTUB was used, and that uses
21020 (p9 extract splitter): Likewise.
21021 (vsx_extract_<mode>_di_p9): Likewise.
21022 (vsx_extract_<mode>_store_p9): Likewise.
21023 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
21024 for TARGET_P9_VECTOR was used. Delete code that is now dead with
21025 the elimination of TARGET_VSX_SMALL_INTEGER.
21026 (vsx_extract_<mode>_p8): Likewise.
21027 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
21028 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
21029 (vsx_set_<mode>_p9): Likewise.
21030 (vsx_set_v4sf_p9): Likewise.
21031 (vsx_set_v4sf_p9_zero): Likewise.
21032 (vsx_insert_extract_v4sf_p9): Likewise.
21033 (vsx_insert_extract_v4sf_p9_2): Likewise.
21034 * config/rs6000/rs6000.md (sign extend splitter): Change
21035 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
21036 (floatsi<mode>2_lfiwax_mem): Likewise.
21037 (floatunssi<mode>2_lfiwzx_mem): Likewise.
21038 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
21039 since a test for TARGET_P9_VECTOR was used.
21040 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21041 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
21042 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21043 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
21044 TARGET_P8_VECTOR test.
21045 (fix_trunc<mode>si2_stfiwx): Likewise.
21046 (fix_trunc<mode>si2_internal): Likewise.
21047 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
21048 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21050 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
21051 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
21052 TARGET_P8_VECTOR test.
21053 (fixuns_trunc<mode>si2_stfiwx): Likewise.
21054 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
21055 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21057 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
21058 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
21059 since a test for TARGET_P9_VECTOR was used.
21060 (splitter for loading small constants): Likewise.
21062 2017-07-26 Andrew Pinski <apinski@cavium.com>
21064 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
21067 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
21070 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
21071 (fp_valid_at): Likewise.
21073 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
21075 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
21076 (qdf24xx_addrcost_table): Likewise.
21077 (cortexa57_tunings): Update to use generic_branch_cost.
21078 (cortexa72_tunings): Likewise.
21079 (cortexa73_tunings): Likewise.
21080 (qdf24xx_tunings): Likewise.
21082 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
21084 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
21085 (thunderx2t99_branch_cost): Likewise.
21086 (cortexa35_tunings): Update to use generic_branch_cost.
21087 (cortexa53_tunings): Likewise.
21088 (cortexa57_tunings): Likewise.
21089 (cortexa72_tunings): Likewise.
21090 (cortexa73_tunings): Likewise.
21091 (thunderx2t99_tunings): Likewise.
21093 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
21095 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
21096 (sparc_option_override): Honour MASK_FSMULD.
21097 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
21098 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
21099 * config/sparc/sparc.opt (mfsmuld): New option.
21100 * doc/invoke.texi (mfsmuld): Document option.
21102 2017-07-26 Marek Polacek <polacek@redhat.com>
21104 PR middle-end/70992
21105 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
21107 2017-07-26 Richard Biener <rguenther@suse.de>
21109 * gimple-match-head.c (do_valueize): Return OP if valueize
21111 (get_def): New helper to get at the def stmt of a SSA name
21112 if valueize allows.
21113 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
21114 do_valueize to get at the def stmt.
21115 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
21117 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
21119 PR middle-end/46932
21120 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
21122 2017-07-26 Martin Liska <mliska@suse.cz>
21125 * function.c (expand_function_start): Make expansion of
21126 nonlocal_goto_save_area after parm_birth_insn.
21128 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
21130 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
21131 from all CPU target flags enable members.
21133 2017-07-26 Richard Biener <rguenther@suse.de>
21135 * genmatch.c (dt_simplify::gen): Make iterator vars const.
21136 (decision_tree::gen): Make 'type' const.
21137 (write_predicate): Likewise.
21139 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
21141 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
21142 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
21143 (rs6000_option_override_internal): Likewise.
21144 (rs6000_expand_vector_set): Likewise.
21145 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
21146 (TARGET_UPPER_REGS_SF): Likewise.
21147 (TARGET_UPPER_REGS_DI): Likewise.
21148 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
21149 (TARGET_DIRECT_MOVE_64BIT): Likewise.
21150 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
21151 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21152 (Splitters for DI constants in Altivec registers): Likewise.
21153 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
21154 (vsx_set_v4sf_p9): Likewise.
21155 (vsx_set_v4sf_p9_zero): Likewise.
21156 (vsx_insert_extract_v4sf_p9): Likewise.
21157 (vsx_insert_extract_v4sf_p9_2): Likewise.
21159 2017-07-25 Carl Love <cel@us.ibm.com>
21161 * doc/extend.texi: Update the built-in documentation file for the
21162 existing built-in functions
21163 vector signed char vec_cnttz (vector signed char);
21164 vector unsigned char vec_cnttz (vector unsigned char);
21165 vector signed short vec_cnttz (vector signed short);
21166 vector unsigned short vec_cnttz (vector unsigned short);
21167 vector signed int vec_cnttz (vector signed int);
21168 vector unsigned int vec_cnttz (vector unsigned int);
21169 vector signed long long vec_cnttz (vector signed long long);
21170 vector unsigned long long vec_cnttz (vector unsigned long long);
21172 2017-07-25 Andrew Pinski <apinski@cavium.com>
21174 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
21175 accesses where the use is for the first operand of a BIT_INSERT.
21177 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
21180 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
21181 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
21183 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
21185 * config/i386/gstabs.h: Delete.
21186 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
21188 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
21190 * config/i386/i386.c (ix86_decompose_address): Do not check for
21191 register RTX when looking at index_reg or base_reg.
21192 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
21194 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
21196 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
21197 to update EH info here.
21199 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21201 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
21203 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21205 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
21207 2017-07-25 Torsten Duwe <duwe@suse.de>
21209 * common.opt: Introduce -fpatchable-function-entry
21210 command line option, and its variables function_entry_patch_area_size
21211 and function_entry_patch_area_start.
21212 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
21213 including a two-value parser.
21214 * target.def (print_patchable_function_entry): New target hook.
21215 * targhooks.h (default_print_patchable_function_entry): New function.
21216 * targhooks.c (default_print_patchable_function_entry): Likewise.
21217 * toplev.c (process_options): Switch off IPA-RA if
21218 patchable function entries are being generated.
21219 * varasm.c (assemble_start_function): Look at the
21220 patchable-function-entry command line switch and current
21221 function attributes and maybe generate NOP instructions by
21222 calling the print_patchable_function_entry hook.
21223 * doc/extend.texi: Document patchable_function_entry attribute.
21224 * doc/invoke.texi: Document -fpatchable_function_entry
21225 command line option.
21226 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
21228 * doc/tm.texi: Re-generate.
21230 2017-07-25 Jakub Jelinek <jakub@redhat.com>
21233 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
21234 TARGET_AVX512DQ rather than TARGET_AVX512BW.
21236 2017-07-25 Tamar Christina <tamar.christina@arm.com>
21238 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
21240 2017-07-25 Richard Biener <rguenther@suse.de>
21242 PR tree-optimization/81455
21243 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
21244 not walk in cycles when looking for guards.
21246 2017-07-25 Richard Biener <rguenther@suse.de>
21248 PR tree-optimization/81529
21249 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
21250 when optimizing backedge uses.
21252 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
21254 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
21256 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
21257 to dl_section_ref. On AIX, append an expression to subtract
21258 the size of the section length to dl_section_ref.
21260 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
21262 * configure.ac: If any of the config.* scripts fail, exit 1.
21263 * configure: Regenerate.
21265 2017-07-25 Richard Biener <rguenther@suse.de>
21267 PR middle-end/81546
21268 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
21269 of immediate uses, be more verbose on errors.
21271 2017-07-25 Richard Biener <rguenther@suse.de>
21273 PR tree-optimization/81510
21274 * tree-vect-loop.c (vect_is_simple_reduction): When the
21275 reduction stmt is not inside the loop bail out.
21277 2017-07-25 Richard Biener <rguenther@suse.de>
21279 PR tree-optimization/81303
21280 * tree-vect-loop-manip.c (vect_loop_versioning): Build
21281 profitability check against LOOP_VINFO_NITERSM1.
21283 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21285 * domwalk.c (cmp_bb_postorder): Simplify.
21286 (sort_bbs_postorder): New function. Use it...
21287 (dom_walker::walk): ...here to optimize common cases.
21289 2017-07-25 Martin Liska <mliska@suse.cz>
21292 * ipa-visibility.c (function_and_variable_visibility): Make the
21293 redirection just on target that supports aliasing.
21294 Fix GNU coding style.
21296 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
21299 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
21301 * config/arm/rtems.h: Likewise.
21302 * config/bfin/rtems.h: Likewise.
21303 * config/i386/rtemself.h: Likewise.
21304 * config/lm32/rtems.h: Likewise.
21305 * config/m32c/rtems.h: Likewise.
21306 * config/m68k/rtemself.h: Likewise.
21307 * config/microblaze/rtems.h: Likewise.
21308 * config/mips/rtems.h: Likewise.
21309 * config/moxie/rtems.h: Likewise.
21310 * config/nios2/rtems.h: Likewise.
21311 * config/powerpcspe/rtems.h: Likewise.
21312 * config/rs6000/rtems.h: Likewise.
21313 * config/rtems.h: Likewise.
21314 * config/sh/rtems.h: Likewise.
21315 * config/sh/rtemself.h: Likewise.
21316 * config/sparc/rtemself.h: Likewise.
21318 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
21321 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
21322 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
21323 * tree-ssa-structalias.c (alias_get_name): Same.
21325 2017-07-25 Bin Cheng <bin.cheng@arm.com>
21328 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
21329 instructions if no du chain is found.
21331 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
21333 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
21335 2017-07-25 Richard Biener <rguenther@suse.de>
21337 PR middle-end/81505
21338 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
21341 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
21343 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
21344 upper-regs options.
21345 (ISA_2_7_MASKS_SERVER): Likewise.
21346 (ISA_3_0_MASKS_IEEE): Likewise.
21347 (OTHER_P8_VECTOR_MASKS): Likewise.
21348 (OTHER_VSX_VECTOR_MASKS): Likewise.
21349 (POWERPC_MASKS): Likewise.
21350 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
21351 duplicate list of options.
21352 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
21353 explicit -mupper-regs options.
21354 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
21355 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
21356 alias for -mupper-regs-df.
21357 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
21358 (rs6000_init_hard_regno_mode_ok): Likewise.
21359 (rs6000_option_override_internal): Likewise.
21360 (rs6000_opt_masks): Likewise.
21361 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
21362 options in terms of whether -mvsx or -mpower8-vector was used.
21363 (TARGET_UPPER_REGS_DI): Likewise.
21364 (TARGET_UPPER_REGS_SF): Likewise.
21365 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
21366 -mupper-regs-* options.
21368 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
21370 * passes.c (emergency_dump_function): Print some empty lines and a
21371 header before the RTL dump.
21373 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
21375 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
21377 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
21380 * config/aarch64/aarch64.c (aarch64_classify_symbol):
21381 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
21383 2017-07-24 Carl Love <cel@us.ibm.com>
21385 * config/rs6000/rs6000-c.c: Add support for built-in functions
21386 vector float vec_extract_fp32_from_shorth (vector unsigned short);
21387 vector float vec_extract_fp32_from_shortl (vector unsigned short);
21388 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
21389 vec_extract_fp_from_shortl): Add defines for the two builtins.
21390 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
21391 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
21393 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
21394 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
21395 * doc/extend.texi: Update the built-in documentation file for the
21396 new built-in function.
21398 2017-07-24 Jakub Jelinek <jakub@redhat.com>
21401 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
21403 * doc/generic.texi: Likewise.
21404 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
21405 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
21407 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
21409 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
21410 (aarch64_mls_elt_merge<mode>): Likewise.
21412 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
21414 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21415 having __cxa_atexit.
21417 2017-07-23 Michael Collison <michael.collison@arm.com>
21419 * config/arm/arm.c (arm_option_override): Deprecate
21420 use of -mstructure-size-boundary.
21421 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
21422 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
21424 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21427 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
21428 Reduce cost estimate for direct moves.
21430 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
21433 * config/i386/i386.c (ix86_option_override_internal): Disable
21434 BMI, BMI2 and TBM instructions for -m16.
21436 2017-07-21 Carl Love <cel@us.ibm.com>
21438 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21439 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21440 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21441 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21442 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21443 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21444 VMULOSW): New enum "unspec" values.
21445 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21446 altivec_vmulosw): New patterns.
21447 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21448 VMULOSW): Add definitions.
21450 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
21452 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
21453 (qdf24xx): Likewise.
21454 * config/aarch64/aarch64-options-extensions.def (rdma); New.
21455 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
21456 (AARCH64_FL_V8_1): Renumber.
21457 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
21458 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
21459 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
21460 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
21461 rdma to feature modifiers list.
21463 2017-07-21 Yury Gribov <tetra2005@gmail.com>
21465 PR middle-end/56727
21466 * ipa-visibility (function_and_variable_visibility): Convert
21467 recursive PLT call to direct call if appropriate.
21469 2017-07-21 Andrew Pinski <apinski@cavium.com>
21471 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
21472 operand 1 to see if the types precision matches.
21473 * fold-const.c (operand_equal_p): Likewise.
21475 2017-07-21 Richard Biener <rguenther@suse.de>
21477 PR tree-optimization/81303
21478 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
21479 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
21480 (vect_peeling_hash_get_lowest_cost): Adjust.
21481 (vect_enhance_data_refs_alignment): Likewise. Use
21482 vect_get_peeling_costs_all_drs to compute the penalty for no
21483 peeling to match up costs.
21485 2017-07-21 Richard Biener <rguenther@suse.de>
21487 PR tree-optimization/81500
21488 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
21489 we didn't identify a reduction path.
21491 2017-07-21 Tom de Vries <tom@codesourcery.com>
21492 Cesar Philippidis <cesar@codesourcery.com>
21494 PR gcov-profile/81442
21495 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
21498 2017-07-21 Tom de Vries <tom@codesourcery.com>
21501 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
21503 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
21504 nvptx_override_options_after_change.
21506 2017-07-21 Ulrich Drepper <drepper@redhat.com>
21508 * dwarf2out.c (output_file_names): Avoid double testing for
21509 dwarf_version >= 5.
21511 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
21513 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
21515 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
21517 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
21519 (try_crossjump_to_edge): Do not punt on partitioned functions.
21521 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
21523 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
21524 Put all BBs reachable only via paths crossing cold region to cold
21526 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
21528 2016-07-21 Richard Biener <rguenther@suse.de>
21530 PR tree-optimization/81303
21531 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
21532 into account prologue and epilogue iterations when raising
21533 min_profitable_iters to sth at least covering one vector iteration.
21535 2017-07-21 Tamar Christina <tamar.christina@arm.com>
21537 * config/arm/arm.c (arm_test_cpu_arch_dat):
21540 2017-07-20 Nathan Sidwell <nathan@acm.org>
21542 Remove TYPE_METHODS.
21543 * tree.h (TYPE_METHODS): Delete.
21544 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
21545 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
21546 (dbxout_type_methods): Scan TYPE_FIELDS.
21547 (dbxout_type): Don't check TYPE_METHODS here.
21548 * function.c (use_register_for_decl): Always ignore register for
21549 class types when not optimizing.
21550 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
21551 * tree.c (free_lang_data_in_type): Stitch out member functions and
21552 templates from TYPE_FIELDS.
21553 (build_distinct_type_copy, verify_type_variant,
21554 verify_type): Member fns are on TYPE_FIELDS.
21555 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
21556 * tree-pretty-print.c (dump_generic_node): Likewise.
21558 2017-07-20 Jakub Jelinek <jakub@redhat.com>
21561 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
21562 V2TImode and V4TImode.
21563 (ix86_expand_vector_extract): Likewise.
21564 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
21565 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
21566 (ssescalarmode): Handle V4TImode and V2TImode.
21567 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
21568 (*vec_extractv2ti, *vec_extractv4ti): New insns.
21569 (VEXTRACTI128_MODE): New mode iterator.
21570 (splitter for *vec_extractv?ti first element): New.
21571 (VEC_INIT_MODE): New mode iterator.
21572 (vec_init<mode>): Consolidate 3 expanders into one using
21573 VEC_INIT_MODE mode iterator.
21575 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
21577 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
21578 non_spilled_static_chain_regno_p.
21580 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
21582 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
21584 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
21586 * bb-reorder.c (connect_traces): Allow copying of blocks within
21589 2017-07-20 Richard Biener <rguenther@suse.de>
21591 * gimple.h (gimple_phi_result): Add gphi * overload.
21592 (gimple_phi_result_ptr): Likewise.
21593 (gimple_phi_arg): Likewise. Adjust index assert to only
21594 allow actual argument accesses rather than all slots available
21596 (gimple_phi_arg_def): Add gphi * overload.
21597 * tree-phinodes.c (make_phi_node): Initialize only actual
21599 (resize_phi_node): Clear memory not covered by old node,
21600 do not initialize excess argument slots.
21601 (reserve_phi_args_for_new_edge): Initialize new argument slot
21604 2017-07-20 Bin Cheng <bin.cheng@arm.com>
21606 PR tree-optimization/81388
21608 2016-07-21 Bin Cheng <bin.cheng@arm.com>
21610 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
21611 by removing computation of may_be_zero.
21613 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21614 Tom de Vries <tom@codesourcery.com>
21616 PR middle-end/81030
21617 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
21618 when gimple level profile disagrees with what RTL expander did.
21620 2017-07-20 Richard Biener <rguenther@suse.de>
21622 PR tree-optimization/61171
21623 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
21624 (vect_analyze_stmt): Add slp instance parameter.
21625 (vectorizable_reduction): Likewise.
21626 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
21627 (vect_is_simple_reduction): Deal with chains not detected
21628 as SLP reduction chain, specifically not properly associated
21629 chains containing a mix of plus/minus.
21630 (get_reduction_op): Remove.
21631 (get_initial_defs_for_reduction): Simplify, pass in whether
21632 this is a reduction chain, pass in the SLP node for the PHIs.
21633 (vect_create_epilog_for_reduction): Get the SLP instance as
21635 (vectorizable_reduction): Get the SLP instance as arg.
21636 During analysis remember the SLP node with the PHIs in the
21637 instance. Simplify getting at the vectorized reduction PHIs.
21638 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
21639 through SLP instance.
21640 (vect_slp_analyze_operations): Likewise.
21641 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
21642 (vect_transform_stmt): Likewise.
21644 2017-07-20 Tom de Vries <tom@codesourcery.com>
21646 PR tree-optimization/81489
21647 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
21648 read of phi arg location to before loop that modifies phi.
21650 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21652 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
21655 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
21657 PR middle-end/81331
21658 * except.c (execute): Fix ordering issue.
21660 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
21662 PR rtl-optimization/81423
21663 * combine.c (make_compound_operation_int): Don't try to optimize
21664 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
21666 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
21668 PR rtl-optimization/81423
21669 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
21670 with a constant that is -1 in the truncated to mode.
21672 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
21674 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
21675 (determine_unlikely_bbs): ... here.
21676 * predict.h (propagate_unlikely_bbs_forward): Declare.
21677 * cfgexpand.c (pass_expand::execute): Use it.
21678 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
21680 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
21681 propagate_unlikely_bbs_forward.
21683 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
21685 PR middle-end/81331
21686 * except.c (maybe_add_nop_after_section_switch): New function.
21689 2017-07-19 Tom de Vries <tom@codesourcery.com>
21691 * gimple.h (gimple_phi_set_arg): Make assert more strict.
21693 2017-07-19 Tom de Vries <tom@codesourcery.com>
21695 * gimple.h (gimple_phi_arg): Make assert more strict.
21697 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
21699 * config.gcc (powerpc*-*-*): Add mmintrin.h.
21700 * config/rs6000/mmintrin.h: New file.
21701 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
21703 2017-07-19 Jakub Jelinek <jakub@redhat.com>
21705 PR tree-optimization/81346
21706 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
21708 2017-07-19 Tom de Vries <tom@codesourcery.com>
21710 * config/nvptx/nvptx.md (VECIM): Add V2DI.
21712 2017-07-19 Tom de Vries <tom@codesourcery.com>
21714 * config/nvptx/nvptx-modes.def: Add V2DImode.
21715 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
21716 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
21717 (nvptx_output_mov_insn): Handle lack of mov.b128.
21718 (nvptx_print_operand): Handle 'H' and 'L' codes.
21719 (nvptx_vector_mode_supported): Allow V2DImode.
21720 (nvptx_preferred_simd_mode): New function.
21721 (nvptx_data_alignment): New function.
21722 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
21723 nvptx_preferred_simd_mode.
21724 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
21726 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
21728 2017-07-19 Tom de Vries <tom@codesourcery.com>
21730 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
21731 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
21732 (nvptx_vector_mode_supported): New function. Allow V2SImode.
21733 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
21734 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
21735 (mov<VECIM>_insn): New define_insn.
21736 (define_expand "mov<VECIM>): New define_expand.
21738 2017-07-19 Tom de Vries <tom@codesourcery.com>
21740 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
21742 2017-07-19 Jakub Jelinek <jakub@redhat.com>
21744 PR tree-optimization/81346
21745 * fold-const.h (fold_div_compare, range_check_type): Declare.
21746 * fold-const.c (range_check_type): New function.
21747 (build_range_check): Use range_check_type.
21748 (fold_div_compare): No longer static, rewritten into
21749 a match.pd helper function.
21750 (fold_comparison): Don't call fold_div_compare here.
21751 * match.pd (X / C1 op C2): New optimization using fold_div_compare
21752 as helper function.
21754 2017-07-19 Nathan Sidwell <nathan@acm.org>
21756 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
21757 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
21758 * tree.c (find_decls_types_r, verify_type): Use
21759 TYPE_{MIN,MAX}_VALUE_RAW.
21760 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
21761 (hash_tree): Likewise.
21762 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21764 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21767 2017-07-18 Tom de Vries <tom@codesourcery.com>
21769 PR middle-end/81464
21770 * omp-expand.c (expand_omp_for_static_chunk): Handle
21771 equal-argument loop exit phi.
21773 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
21776 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
21777 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
21778 operand 2 predicate.
21779 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
21780 operand 2 predicate.
21781 (ror,rol -> rorx splitters): Use const_int_operand as
21782 operand 2 predicate.
21784 2017-06-18 Richard Biener <rguenther@suse.de>
21786 PR tree-optimization/81410
21787 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
21788 the gap in the ! slp_perm SLP case after each group.
21790 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21792 PR middle-end/81463
21793 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
21796 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21798 PR middle-end/81462
21799 * predict.c (set_even_probabilities): Cleanup; do not affect
21800 probabilities that are already known.
21801 (combine_predictions_for_bb): Call even when count is set.
21803 2017-07-18 Nathan Sidwell <nathan@acm.org>
21805 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
21808 2017-07-18 Bin Cheng <bin.cheng@arm.com>
21811 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
21812 optimization for loop niter analysis.
21814 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
21817 * config/avr/avr.c (avr_optimize_casesi): Don't use
21818 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
21820 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
21822 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
21823 body_cost_vec from _vect_peel_extended_info.
21824 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
21825 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
21828 2017-07-18 Bin Cheng <bin.cheng@arm.com>
21830 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
21832 2017-07-18 Richard Biener <rguenther@suse.de>
21834 PR tree-optimization/80620
21835 PR tree-optimization/81403
21836 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
21837 info when re-using a VN table entry.
21839 2017-07-18 Richard Biener <rguenther@suse.de>
21841 PR tree-optimization/81418
21842 * tree-vect-loop.c (vectorizable_reduction): Properly compute
21843 vectype_in. Verify that with lane-reducing reduction operations
21844 we have a single def-use cycle.
21846 2017-07-17 Carl Love <cel@us.ibm.com>
21848 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
21850 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21851 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21852 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21853 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21854 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21855 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21856 VMULOSW): New enum "unspec" values.
21857 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
21858 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
21859 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21860 altivec_vmulosw): New patterns.
21861 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21862 VMULOSW): Add definitions.
21864 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
21866 * config/alpha/alpha.c: Include predict.h.
21868 2017-07-17 Yury Gribov <tetra2005@gmail.com>
21870 * tree-vrp.c (compare_assert_loc): Fix comparison function
21871 to return predictable results.
21873 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21875 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
21877 (subdi3): Likewise.
21878 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
21879 * doc/invoke.texi (mexpand-adddi): Update text.
21881 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21883 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
21884 that also clobbers the CC register. The old expand code is moved
21886 (*arc_clzsi2): ... here.
21887 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
21888 the CC register. The old expand code is moved to ...
21889 (arc_ctzsi2): ... here.
21891 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21893 * config/arc/arc.opt (mindexed-loads): Use initial value
21894 TARGET_INDEXED_LOADS_DEFAULT.
21895 (mauto-modify-reg): Use initial value
21896 TARGET_AUTO_MODIFY_REG_DEFAULT.
21897 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
21898 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
21899 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
21900 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
21902 2017-07-17 Martin Liska <mliska@suse.cz>
21905 * opts.c (finish_options): Do not allow -fgnu-tm
21906 w/ -fsanitize={kernel-,}address. Say sorry.
21908 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21911 * tree-loop-distribution.c (classify_partition): Only assert on
21912 numer of iterations.
21913 (merge_dep_scc_partitions): Delete prameter. Update function call.
21914 (distribute_loop): Remove code handling loop with unknown niters.
21915 (pass_loop_distribution::execute): Skip loop with unknown niters.
21917 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21920 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
21921 function sort_partitions_by_post_order.
21923 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21925 PR tree-optimization/81374
21926 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
21927 the max index of basic blocks, rather than number of basic blocks.
21929 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21931 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
21933 (arc_legitimate_pic_operand_p): Likewise.
21934 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
21936 (arc_needs_pcl_p): Likewise.
21937 (arc_legitimate_pc_offset_p): Likewise.
21938 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
21939 function is also used in constrains.md.
21940 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
21941 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
21942 PLUS. Only return true/false in known cases, otherwise assert.
21943 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
21944 is already called in arc_legitimate_constant_p.
21945 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
21947 (LEGITIMATE_PIC_OPERAND_P): Use
21948 arc_raw_symbolic_reference_mentioned_p function.
21949 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
21954 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21955 Andrew Burgess <andrew.burgess@embecosm.com>
21957 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
21958 (arc_return_address_register): New function.
21959 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
21960 (arc_handle_fndecl_attribute): Add naked attribute.
21961 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21962 (TARGET_WARN_FUNC_RETURN): Likewise.
21963 (arc_allocate_stack_slots_for_args): New function.
21964 (arc_warn_func_return): Likewise.
21965 (machine_function): Change type fn_type.
21966 (arc_compute_function_type): Consider new naked function type,
21967 change function return type.
21968 (arc_must_save_register): Adapt to handle new
21969 arc_compute_function_type's return type.
21970 (arc_expand_prologue): Likewise.
21971 (arc_expand_epilogue): Likewise.
21972 (arc_return_address_regs): Delete.
21973 (arc_return_address_register): New function.
21974 (arc_epilogue_uses): Use above function.
21975 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
21976 (arc_function_type): Change encoding, add naked type.
21977 (ARC_INTERRUPT_P): Change to handle the new encoding.
21978 (ARC_FAST_INTERRUPT_P): Likewise.
21979 (ARC_NORMAL_P): Define.
21980 (ARC_NAKED_P): Likewise.
21981 (arc_compute_function_type): Delete prototype.
21982 * config/arc/arc.md (in_ret_delay_slot): Use
21983 arc_return_address_register function.
21984 (simple_return): Likewise.
21985 (p_return_i): Likewise.
21987 2017-07-17 Jakub Jelinek <jakub@redhat.com>
21989 PR tree-optimization/81428
21990 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
21991 can't be built for those types.
21993 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
21995 Remove stuff dead since r239246.
21997 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
21998 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
21999 (avr_inform_devices): Remove dead stuff.
22001 2017-07-17 Tamar Christina <tamar.christina@arm.com>
22003 * config/arm/arm_neon.h: Fix softp typo.
22005 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22007 PR tree-optimization/81365
22008 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
22009 aggregate moves onto bb predecessor edges, make sure there are no
22010 loads that could alias the lhs in between the start of bb and the
22013 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22016 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
22017 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
22018 [LSHIFTRT, outer_code = TRUNCATE]: Same.
22020 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22022 PR tree-optimization/81396
22023 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
22024 (init_symbolic_number): Initialize it to 1.
22025 (perform_symbolic_merge): Add n_ops from both operands into the new
22027 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
22028 without base_addr as useless if they need more than one operation.
22029 (bswap_replace): Handle !bswap case for NULL base_addr.
22031 2017-07-17 Tom de Vries <tom@codesourcery.com>
22034 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
22037 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
22039 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
22040 conditional builtin define __FIX_LEON3FT_B2BST.
22042 2017-07-17 Daniel Cederman <cederman@gaisler.com>
22044 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
22045 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
22048 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
22050 PR rtl-optimization/81424
22051 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
22052 to remove potential trapping from operands if -fnon-call-exceptions.
22054 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22056 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
22057 profile_proability for scalling.
22058 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
22060 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22062 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
22064 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22066 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
22067 fixpoint arithmetics.
22069 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22071 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
22072 fixpoint arithmetics.
22074 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22076 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
22077 fixpoint arithmetics.
22079 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22081 * profile-count.h (profile_probability::from_reg_br_prob_note,
22082 profile_probability::to_reg_br_prob_note): New functions.
22083 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
22084 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
22085 * predict.c (probability_reliable_p): Update.
22086 (edge_probability_reliable_p): Update.
22087 (br_prob_note_reliable_p): Update.
22088 (invert_br_probabilities): Update.
22089 (add_reg_br_prob_note): New function.
22090 (combine_predictions_for_insn): Update.
22091 * asan.c (asan_clear_shadow): Update.
22092 * cfgbuild.c (compute_outgoing_frequencies): Update.
22093 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22094 (update_br_prob_note): Update.
22095 (rtl_verify_edges): Update.
22096 (purge_dead_edges): Update.
22097 (fixup_reorder_chain): Update.
22098 * emit-rtl.c (try_split): Update.
22099 * ifcvt.c (cond_exec_process_insns): Update.
22100 (cond_exec_process_if_block): Update.
22101 (dead_or_predicable): Update.
22102 * internal-fn.c (expand_addsub_overflow): Update.
22103 (expand_neg_overflow): Update.
22104 (expand_mul_overflow): Update.
22105 * loop-doloop.c (doloop_modify): Update.
22106 * loop-unroll.c (compare_and_jump_seq): Update.
22107 * optabs.c (emit_cmp_and_jump_insn_1): Update.
22108 * predict.h: Update.
22109 * reorg.c (mostly_true_jump): Update.
22111 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
22112 * config/alpha/alpha.c (emit_unlikely_jump): Update.
22113 * config/arc/arc.c: (emit_unlikely_jump): Update.
22114 * config/arm/arm.c: (emit_unlikely_jump): Update.
22115 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
22116 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
22117 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
22118 (ix86_print_operand): Update.
22119 (ix86_split_fp_branch): Update.
22120 (predict_jump): Update.
22121 * config/ia64/ia64.c (ia64_print_operand): Update.
22122 * config/mmix/mmix.c (mmix_print_operand): Update.
22123 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
22124 (rs6000_expand_split_stack_prologue): Update.
22125 * config/rs6000/rs6000.c: Update.
22126 * config/s390/s390.c (s390_expand_vec_strlen): Update.
22127 (s390_expand_vec_movstr): Update.
22128 (s390_expand_cs_tdsi): Update.
22129 (s390_expand_split_stack_prologue): Update.
22130 * config/sh/sh.c (sh_print_operand): Update.
22131 (expand_cbranchsi4): Update.
22132 (expand_cbranchdi4): Update.
22133 * config/sparc/sparc.c (output_v9branch): Update.
22134 * config/spu/spu.c (get_branch_target): Update.
22135 (ea_load_store_inline): Update.
22136 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
22137 * config/tilepro/tilepro.c: Update.
22139 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
22141 * gimplify.c (mostly_copy_tree_r): Revert latest change.
22142 (gimplify_save_expr): Likewise.
22144 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22146 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
22148 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22150 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
22152 * timevar.def (TV_IPA_FNSUMMARY): Define.
22154 2017-07-16 Daniel Cederman <cederman@gaisler.com>
22156 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
22157 to back store errata sensitive sequence from being generated.
22158 (sqrtdf2_fix): Likewise.
22160 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22162 * tree-ssa-threadupdate.c (compute_path_counts,
22163 update_joiner_offpath_counts): Use profile_probability.
22165 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
22168 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22170 * config/arm/arm-c.c (arm_cpu_builtins): Define
22171 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
22173 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
22175 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22176 array entries to represent __ieee128 versions of the
22177 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
22178 scalar_extract_sig, and scalar_insert_exp built-in functions.
22179 (altivec_resolve_overloaded_builtin): Add special case handling
22180 for the __builtin_scalar_insert_exp function, as represented by
22181 the P9V_BUILTIN_VEC_VSIEDP constant.
22182 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
22183 exponent support for __ieee128 argument.
22184 (VSESQP): Add scalar extract signature support for __ieee128
22186 (VSTDCNQP): Add scalar test negative support for __ieee128
22188 (VSIEQP): Add scalar insert exponent support for __int128 argument
22189 with __ieee128 result.
22190 (VSIEQPF): Add scalar insert exponent support for __ieee128
22191 argument with __ieee128 result.
22192 (VSTDCQP): Add scalar test data class support for __ieee128
22194 (VSTDCNQP): Add overload support for scalar test negative with
22195 __ieee128 argument.
22196 (VSTDCQP): Add overload support for scalar test data class
22197 __ieee128 argument.
22198 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
22199 UNSPEC_VSX_SXSIGDP.
22200 (UNSPEC_VSX_SIEXPQP): New constant.
22201 (xsxexpqp): New insn for VSX scalar extract exponent quad
22203 (xsxsigqp): New insn for VSX scalar extract significand quad
22205 (xsiexpqpf): New insn for VSX scalar insert exponent quad
22206 precision with floating point argument.
22207 (xststdcqp): New expand for VSX scalar test data class quad
22209 (xststdcnegqp): New expand for VSX scalar test negative quad
22211 (xststdcqp): New insn to match expansions for VSX scalar test data
22212 class quad precision and VSX scalar test negative quad precision.
22213 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
22214 special case operand checking to enforce that second operand of
22215 VSX scalar test data class with quad precision argument is a 7-bit
22217 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22218 prototypes and descriptions of __ieee128 versions of
22219 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
22220 scalar_test_data_class, and scalar_test_neg built-in functions.
22222 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22224 PR tree-optimization/81162
22225 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
22226 replace a negate with an add.
22228 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
22230 * doc/invoke.texi (arm/-mcpu): Document +crypto.
22232 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22234 * config/arm/arm-c.c (arm_cpu_builtins): Define
22235 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
22237 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22239 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
22240 (armv8-r): Set ARM Cortex-R52 as default CPU.
22241 * config/arm/arm-tables.opt: Regenerate.
22242 * config/arm/arm-tune.md: Regenerate.
22243 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22245 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
22246 extension for -mcpu=cortex-r52.
22248 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22250 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
22251 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
22252 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
22253 (fp-armv8): Define it as FP_ARMv8 only.
22254 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
22255 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
22257 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
22258 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
22259 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
22260 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
22261 than TARGET_FPU_ARMV8.
22262 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
22263 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
22264 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
22266 * config/arm/neon.md (neon_vrint): Likewise.
22267 (neon_vcvt): Likewise.
22268 (neon_<fmaxmin_op><mode>): Likewise.
22269 (<fmaxmin><mode>3): Likewise.
22270 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
22271 * config/arm/predicates.md (arm_cond_move_operator): Check against
22272 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
22274 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
22276 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
22277 to top of function.
22279 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22281 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
22282 loop in comment with memset.
22284 2017-07-14 Martin Liska <mliska@suse.cz>
22286 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
22287 * dwarf2out.c (is_java): Remove the function.
22288 (output_pubname): Remove usage of the function.
22289 (lower_bound_default): Remove usage of DW_LANG_Java.
22290 (gen_compile_unit_die): Likewise.
22291 * gcc.c: Remove compiler defaults for .java and .zip files.
22292 * gimple-expr.c (remove_suffix): Change as there's no longer
22293 extension than 4-letter one.
22294 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
22295 (gimplify_save_expr): Likewise.
22296 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
22297 as it's possible even for other languages than Java.
22298 * langhooks.h (struct lang_hooks): Remove Java from a comment.
22299 * lto-opts.c (lto_write_options): Remove reference to Java.
22300 * opts.c (strip_off_ending): Update file extension handling.
22301 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
22302 * tree-eh.c (lower_resx): Likewise.
22303 * tree.c (free_lang_data_in_type): Remove dead code.
22304 (find_decls_types_r): Likewise.
22305 (build_common_builtin_nodes): Remove Java from a comment.
22306 (verify_type): Remove dead code.
22307 * varasm.c (assemble_external): Remove Java from a comment.
22309 2017-07-14 Martin Liska <mliska@suse.cz>
22311 * opts.c (finish_options): Add quotes.
22312 (common_handle_option): Likewise.
22314 2017-07-14 Martin Liska <mliska@suse.cz>
22316 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
22317 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
22318 Remove N_SO_PASCAL.
22319 * dwarf2out.c (lower_bound_default): Do not handle
22321 (gen_compile_unit_die): Likewise.
22322 * gcc.c: Remove default extension binding for GNU Pascal.
22323 * stmt.c: Remove Pascal language from a comment.
22324 * xcoffout.c: Likewise.
22326 2017-07-13 David Malcolm <dmalcolm@redhat.com>
22329 * diagnostic-show-locus.c (fixit_cmp): New function.
22330 (layout::layout): Sort m_fixit_hints.
22331 (column_range::column_range): Assert that the values are valid.
22332 (struct char_span): New struct.
22333 (correction::overwrite): New method.
22334 (struct source_line): New struct.
22335 (line_corrections::add_hint): Add assertions. Reimplement memcpy
22336 calls in terms of classes source_line and char_span, and
22337 correction::overwrite.
22338 (selftest::test_overlapped_fixit_printing_2): New function.
22339 (selftest::diagnostic_show_locus_c_tests): Call it.
22341 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
22343 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
22344 early if there is no lhs.
22346 2017-07-13 Martin Liska <mliska@suse.cz>
22348 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
22349 (gen_reference_type_die): Likewise.
22350 * stor-layout.c: Remove Pascal-related comment.
22352 2017-07-13 Martin Liska <mliska@suse.cz>
22354 * opts.c (finish_options): Add quotes to error messages.
22355 (parse_sanitizer_options): Likewise.
22357 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
22359 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
22361 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
22363 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
22365 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
22367 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
22369 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
22371 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
22374 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
22375 provides the hardware capability bits, define the macro
22376 __BUILTIN_CPU_SUPPORTS__.
22377 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
22378 if GLIBC does not provide the hardware capability bits. Add a
22379 gcc_unreachable call if the built-in cpu function is neither
22380 __builtin_cpu_is nor __builtin_cpu_supports.
22381 (rs6000_get_function_versions_dispatcher): Change the warning
22382 that an old GLIBC is used which does not export the capability
22383 bits to be an error.
22384 * doc/extend.texi (target_clones attribute): Document the
22385 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
22386 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
22387 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
22388 the macros defined by GCC if the newer GLIBC is available.
22390 2017-07-12 Jeff Law <law@redhat.com>
22392 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
22393 remaining includes slightly.
22394 * config/riscv/riscv-builtins.c: Include profile-count.h.
22396 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
22399 * config/avr/avr.c (avr_set_current_function): In diagnostic
22400 messages: Quote keywords and (parts of) identifiers.
22401 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
22404 2017-07-12 Carl Love <cel@us.ibm.com>
22406 * config/rs6000/rs6000-c.c: Add support for built-in functions
22407 vector bool char vec_revb (vector bool char);
22408 vector bool short vec_revb (vector short char);
22409 vector bool int vec_revb (vector bool int);
22410 vector bool long long vec_revb (vector bool long long);
22411 * doc/extend.texi: Update the built-in documentation file for the
22412 new built-in functions.
22414 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22416 * config/s390/s390.md: Remove movcc splitter.
22418 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22420 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
22421 load/store on condition.
22423 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
22426 * config/avr/avr.c (avr_encode_section_info)
22427 [progmem && !TREE_READONLY]: Error if progmem object needs
22430 2017-07-11 Michael Collison <michael.collison@arm.com>
22432 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
22435 2017-07-11 Carl Love <cel@us.ibm.com>
22437 * config/rs6000/rs6000-c.c: Add support for builtins
22438 vector unsigned int vec_parity_lsbb (vector signed int);
22439 vector unsigned int vec_parity_lsbb (vector unsigned int);
22440 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
22441 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
22442 vector unsigned long long vec_parity_lsbb (vector signed long long);
22443 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
22444 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
22445 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
22446 * doc/extend.texi: Update the built-in documentation file for the
22447 new built-in functions.
22449 2017-07-11 David Malcolm <dmalcolm@redhat.com>
22451 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
22452 (layout::m_primary_loc): New field.
22453 (layout::layout): Initialize new field. Move location filtering
22454 logic from here to...
22455 (layout::maybe_add_location_range): ...this new method. Add
22456 support for filtering to just the lines already specified by other
22458 (layout::will_show_line_p): New method.
22459 (gcc_rich_location::add_location_if_nearby): New method.
22460 (selftest::test_add_location_if_nearby): New test function.
22461 (selftest::diagnostic_show_locus_c_tests): Call it.
22462 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
22465 2017-07-11 Tom de Vries <tom@codesourcery.com>
22467 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
22468 (bb_first_real_insn): New function.
22469 (nvptx_single): Add extra initialization of broadcasted condition
22472 2017-07-11 Nathan Sidwell <nathan@acm.org>
22474 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
22476 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
22478 * doc/extend.texi (AVR Function Attributes): Remove weblink to
22479 Binutils doc as TEXI will mess them up.
22480 * doc/invoke.texi (AVR Options): Same here.
22482 2017-07-11 Daniel Cederman <cederman@gaisler.com>
22484 * config/sparc/sparc.opt (mfix-ut700): New option.
22485 (mfix-gr712rc): Likewise.
22486 (sparc_fix_b2bst): New variable.
22487 * doc/invoke.texi (SPARC options): Document them.
22488 (ARM options): Fix warnings.
22489 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
22490 instructions to prevent sequences that can trigger the store-store
22491 errata for certain LEON3FT processors.
22492 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
22493 (sparc_option_override): Set sparc_fix_b2bst appropriately.
22494 * config/sparc/sparc.md (fix_b2bst): New attribute.
22495 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
22497 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
22500 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
22502 (*rsqrtsf2_sse): Ditto.
22504 (div<mode>3): Macroize insn from divdf3 and divsf3
22505 using MODEF mode iterator.
22507 2017-07-10 Martin Sebor <msebor@redhat.com>
22509 PR tree-optimization/80397
22510 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
22511 instead of testing for equality to INTEGER_TYPE.
22513 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
22515 * config.gcc: Remove uclibc from arc target spec.
22517 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
22519 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
22521 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22524 * lto-wrapper.c (remove_option): New function.
22525 (merge_and_complain): Merge PIC/PIE options more realistically.
22527 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
22529 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
22533 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
22534 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
22535 * config.in: Regenerate.
22536 * configure: Regenerate.
22537 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
22538 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
22539 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
22540 (TARGET_GASISR_PROLOGUES): ...target mask.
22541 * common/config/avr/avr-common.c
22542 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
22543 Set -mgas-isr-prologues.
22544 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
22545 INSERT_PASS_BEFORE for it.
22546 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
22547 * config/avr/avr.c (avr_option_override)
22548 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
22549 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
22550 (avr_attribute_table) <no_gccisr>: Add new function attribute.
22551 (avr_set_current_function) <is_no_gccisr>: Init machine field.
22552 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
22554 (make_avr_pass_pre_proep): New function.
22555 (emit_push_sfr) <treg>: Add argument to function and use it
22556 instead of TMP_REG.
22557 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
22558 and set machine->gasisr.yes.
22559 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
22560 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
22561 __gcc_isr.n_pushed to .L__stack_usage.
22562 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
22563 (avr_asm_final_postscan_insn): ...this new static function.
22564 * config/avr/avr.h (machine_function)
22565 <is_no_gccisr, use_L__stack_usage>: New fields.
22566 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
22567 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
22568 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
22569 (gasisr, *gasisr): New expander and insn.
22570 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
22571 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
22572 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
22574 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
22576 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
22579 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
22581 Move jump-tables out of .text again.
22584 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
22585 (ASM_OUTPUT_ADDR_VEC): New function.
22586 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
22587 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
22588 INSN_ADDRESSes as asm comment.
22589 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
22590 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
22591 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
22592 * config/avr/avr.md (*tablejump): Adjust comment.
22593 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
22594 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
22596 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
22597 (avr_output_addr_vec): New proto.
22598 (avr_log_t) <insn_addresses>: New field.
22600 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
22603 * config/i386/i386.c (ix86_function_arg_advance): Set
22604 outgoing_args_on_stack to true if there are outgoing arguments
22606 (ix86_function_arg): Likewise.
22607 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
22608 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
22609 * config/i386/i386.h (machine_function): Add
22610 outgoing_args_on_stack.
22612 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
22614 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22615 supporting pthreds.
22616 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
22618 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
22620 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
22621 (REAL_H): Remove $(MACHMODE_H).
22622 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
22624 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
22625 $(MACHMODE_H) and double-int.h.
22626 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
22628 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
22631 2017-07-07 Andrew Pinski <apinski@cavium.com>
22633 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
22634 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
22636 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
22638 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
22639 Add warning if GCC was not configured to link against a GLIBC that
22640 exports the hardware capability bits.
22641 (make_resolver_func): Make resolver function private and not a
22642 COMDAT function. Create the name with clone_function_name instead
22643 of make_unique_name.
22646 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
22647 correct operand in doing the split.
22649 2017-07-07 Carl Love <cel@us.ibm.com>
22651 * config/rs6000/rs6000-c: Add support for built-in function
22652 vector unsigned short vec_pack_to_short_fp32 (vector float,
22654 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
22655 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
22656 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
22657 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
22658 (convert_4f32_8i16): Add define_expand.
22659 * doc/extend.texi: Update the built-in documentation file for the
22660 new built-in function.
22662 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22664 * config/sparc/m8.md: New file.
22665 * config/sparc/sparc.md: Include m8.md.
22667 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22669 * config/sparc/sparc.opt: New option -mvis4b.
22670 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
22671 (sparc_option_override): Handle VIS4B.
22672 (enum sparc_builtins): Define
22673 SPARC_BUILTIN_DICTUNPACK{8,16,32},
22674 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
22675 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
22676 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
22677 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
22678 (check_constant_argument): New function.
22679 (sparc_vis_init_builtins): Define builtins
22680 __builtin_vis_dictunpack{8,16,32},
22681 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
22682 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
22683 __builtin_vis_fpcmpde{8,16,32}shl and
22684 __builtin_vis_fpcmpur{8,16,32}shl.
22685 (sparc_expand_builtin): Check that the constant operands to
22686 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
22687 constant and in range.
22688 * config/sparc/sparc-c.c (sparc_target_macros): Handle
22690 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
22691 (SPARC_IMM5_P): Likewise.
22692 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
22693 (enabled): Handle vis4b.
22694 (UNSPEC_DICTUNPACK): New unspec.
22695 (UNSPEC_FPCMPSHL): Likewise.
22696 (UNSPEC_FPUCMPSHL): Likewise.
22697 (UNSPEC_FPCMPDESHL): Likewise.
22698 (UNSPEC_FPCMPURSHL): Likewise.
22699 (cpu_feature): New CPU feature `vis4b'.
22700 (dictunpack{8,16,32}): New insns.
22701 (FPCSMODE): New mode iterator.
22702 (fpcscond): New code iterator.
22703 (fpcsucond): Likewise.
22704 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
22705 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
22706 (fpcmpde{8,16,32}{si,di}shl): Likewise.
22707 (fpcmpur{8,16,32}{si,di}shl): Likewise.
22708 * config/sparc/constraints.md: Define constraints `q' for unsigned
22709 2-bit integer constants and `t' for unsigned 5-bit integer
22711 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
22713 (imm5_operand_dictunpack16): Likewise.
22714 (imm5_operand_dictunpack32): Likewise.
22715 (imm2_operand): Likewise.
22716 * doc/invoke.texi (SPARC Options): Document -mvis4b.
22717 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
22718 ditunpack* and fpcmp*shl builtins.
22720 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22722 * config.gcc: Handle m8 in --with-{cpu,tune} options.
22723 * config.in: Add HAVE_AS_SPARC6 define.
22724 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
22726 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
22728 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
22729 (CPP_CPU_SPEC): Handle m8.
22730 (ASM_CPU_SPEC): Likewise.
22731 * config/sparc/sparc-opts.h (enum processor_type): Add
22733 * config/sparc/sparc.c (m8_costs): New struct.
22734 (sparc_option_override): Handle TARGET_CPU_m8.
22735 (sparc32_initialize_trampoline): Likewise.
22736 (sparc64_initialize_trampoline): Likewise.
22737 (sparc_issue_rate): Likewise.
22738 (sparc_register_move_cost): Likewise.
22739 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
22740 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
22741 (ASM_CPU64_DEFAULT_SPEC): Likewise.
22742 (CPP_CPU_SPEC): Handle M8.
22743 (ASM_CPU_SPEC): Likewise.
22744 (AS_M8_FLAG): Define.
22745 * config/sparc/sparc.md: Add m8 to the cpu attribute.
22746 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
22747 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
22749 * configure: Regenerate.
22750 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
22753 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22755 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
22757 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
22758 ("*movdi_insn_sp32"): Do not set v3pipe.
22759 ("*movsi_insn"): Likewise.
22760 ("*movdi_insn_sp64"): Likewise.
22761 ("*movsf_insn"): Likewise.
22762 ("*movdf_insn_sp32"): Likewise.
22763 ("*movdf_insn_sp64"): Likewise.
22764 ("*zero_extendsidi2_insn_sp64"): Likewise.
22765 ("*sign_extendsidi2_insn"): Likewise.
22766 ("*mov<VM32:mode>_insn"): Likewise.
22767 ("*mov<VM64:mode>_insn_sp64"): Likewise.
22768 ("*mov<VM64:mode>_insn_sp32"): Likewise.
22769 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
22770 ("<vlop:code><VL:mode>3"): Likewise.
22771 ("*not_<vlop:code><VL:mode>3"): Likewise.
22772 ("*nand<VL:mode>_vis"): Likewise.
22773 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
22774 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
22775 ("one_cmpl<VL:mode>2"): Likewise.
22776 ("faligndata<VM64:mode>_vis"): Likewise.
22777 ("alignaddrsi_vis"): Likewise.
22778 ("alignaddrdi_vis"): Likweise.
22779 ("alignaddrlsi_vis"): Likewise.
22780 ("alignaddrldi_vis"): Likewise.
22781 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
22782 ("bmaskdi_vis"): Likewise.
22783 ("bmasksi_vis"): Likewise.
22784 ("bshuffle<VM64:mode>_vis"): Likewise.
22785 ("cmask8<P:mode>_vis"): Likewise.
22786 ("cmask16<P:mode>_vis"): Likewise.
22787 ("cmask32<P:mode>_vis"): Likewise.
22788 ("pdistn<P:mode>_vis"): Likewise.
22789 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
22791 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22793 * config/sparc/sparc.md ("subtype"): New insn attribute.
22794 ("*wrgsr_sp64"): Set insn subtype.
22795 ("*rdgsr_sp64"): Likewise.
22796 ("alignaddrsi_vis"): Likewise.
22797 ("alignaddrdi_vis"): Likewise.
22798 ("alignaddrlsi_vis"): Likewise.
22799 ("alignaddrldi_vis"): Likewise.
22800 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
22801 ("fexpand_vis"): Likewise.
22802 ("fpmerge_vis"): Likewise.
22803 ("faligndata<VM64:mode>_vis"): Likewise.
22804 ("bshuffle<VM64:mode>_vis"): Likewise.
22805 ("cmask8<P:mode>_vis"): Likewise.
22806 ("cmask16<P:mode>_vis"): Likewise.
22807 ("cmask32<P:mode>_vis"): Likewise.
22808 ("fchksm16_vis"): Likewise.
22809 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
22810 ("fmean16_vis"): Likewise.
22811 ("fp<plusminus_insn>64_vis"): Likewise.
22812 ("<plusminus_insn>v8qi3"): Likewise.
22813 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
22814 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
22815 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
22816 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
22817 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
22818 ("*movqi_insn"): Likewise.
22819 ("*movhi_insn"): Likewise.
22820 ("*movsi_insn"): Likewise.
22821 ("movsi_pic_gotdata_op"): Likewise.
22822 ("*movdi_insn_sp32"): Likewise.
22823 ("*movdi_insn_sp64"): Likewise.
22824 ("movdi_pic_gotdata_op"): Likewise.
22825 ("*movsf_insn"): Likewise.
22826 ("*movdf_insn_sp32"): Likewise.
22827 ("*movdf_insn_sp64"): Likewise.
22828 ("*zero_extendhisi2_insn"): Likewise.
22829 ("*zero_extendqihi2_insn"): Likewise.
22830 ("*zero_extendqisi2_insn"): Likewise.
22831 ("*zero_extendqidi2_insn"): Likewise.
22832 ("*zero_extendhidi2_insn"): Likewise.
22833 ("*zero_extendsidi2_insn_sp64"): Likewise.
22834 ("ldfsr"): Likewise.
22835 ("prefetch_64"): Likewise.
22836 ("prefetch_32"): Likewise.
22837 ("tie_ld32"): Likewise.
22838 ("tie_ld64"): Likewise.
22839 ("*tldo_ldub_sp32"): Likewise.
22840 ("*tldo_ldub1_sp32"): Likewise.
22841 ("*tldo_ldub2_sp32"): Likewise.
22842 ("*tldo_ldub_sp64"): Likewise.
22843 ("*tldo_ldub1_sp64"): Likewise.
22844 ("*tldo_ldub2_sp64"): Likewise.
22845 ("*tldo_ldub3_sp64"): Likewise.
22846 ("*tldo_lduh_sp32"): Likewise.
22847 ("*tldo_lduh1_sp32"): Likewise.
22848 ("*tldo_lduh_sp64"): Likewise.
22849 ("*tldo_lduh1_sp64"): Likewise.
22850 ("*tldo_lduh2_sp64"): Likewise.
22851 ("*tldo_lduw_sp32"): Likewise.
22852 ("*tldo_lduw_sp64"): Likewise.
22853 ("*tldo_lduw1_sp64"): Likewise.
22854 ("*tldo_ldx_sp64"): Likewise.
22855 ("*mov<VM32:mode>_insn"): Likewise.
22856 ("*mov<VM64:mode>_insn_sp64"): Likewise.
22857 ("*mov<VM64:mode>_insn_sp32"): Likewise.
22859 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22861 * config/sparc/sparc.md ("type"): New insn type viscmp.
22862 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
22864 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
22865 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
22866 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
22867 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
22869 ("n7_vis_logical_11cycle"): Likewise.
22870 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
22871 * config/sparc/niagara2.md ("niag3_vis": Likewise.
22872 * config/sparc/niagara.md ("niag_vis"): Likewise.
22873 * config/sparc/ultra3.md ("us3_fga"): Likewise.
22874 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
22876 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22878 * config/sparc/sparc.md: New instruction type `bmask'.
22879 (bmaskdi_vis): Use the `bmask' type.
22880 (bmasksi_vis): Likewise.
22881 * config/sparc/ultra3.md (us3_array): Likewise.
22882 * config/sparc/niagara7.md (n7_array): Likewise.
22883 * config/sparc/niagara4.md (n4_array): Likewise.
22884 * config/sparc/niagara2.md (niag2_vis): Likewise.
22885 (niag3_vis): Likewise.
22886 * config/sparc/niagara.md (niag_vis): Likewise.
22888 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22890 * ipa-comdats.c: Remove optimize check from gate.
22891 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
22892 for functions not optimized.
22893 (ipa_fn_summary_read): Skip optimize check.
22894 (ipa_fn_summary_write): Likewise.
22895 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
22897 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
22899 (can_inline_edge_p): Check flag_pcc_struct_return for match.
22900 (check_callers): Give up on caller which is not optimized.
22901 (inline_small_functions): Likewise.
22902 (ipa_inline): Do not give up when not optimizing.
22903 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
22904 away unoptimizes cdtors.
22905 (whole_program_function_and_variable_visibility): Do
22906 ipa_discover_readonly_nonaddressable_vars in LTO mode.
22907 * ipa.c (process_references): Do not check optimize.
22908 (symbol_table::remove_unreachable_nodes): Update optimize check.
22909 (set_writeonly_bit): Update optimize check.
22910 (pass_ipa_cdtor_merge::gate): Do not check optimize.
22911 (pass_ipa_single_use::gate): Remove.
22913 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22915 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
22916 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
22917 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
22918 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
22919 permute_load, permute_store, adjust_extract, adjust_splat,
22920 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
22921 replace_swap_with_copy, dump_swap_insn_table,
22922 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
22923 recombine_lvx_pattern, recombine_stvx_pattern,
22924 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
22925 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
22926 to file rs6000-p8swap.c.
22927 * config/rs6000/rs6000-p8swap.c: New file.
22928 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
22929 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
22930 and rs6000*-*-* targets.
22932 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22934 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
22936 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22938 * lto-wrapper.c (merge_and_complain): Do not merge
22939 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
22940 fsigned_zeros, ftrapping_math, fwrapv.
22941 (append_compiler_options): Do not track these options.
22942 (append_linker_options): Likewie
22944 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22946 * cgraphunit.c (cgraph_node::finalize_function): When
22947 !flag_toplevel_reorde set no_reorder flag.
22948 (varpool_node::finalize_decl): Likewise.
22949 (symbol_table::compile): Drop no toplevel reorder path.
22951 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22953 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
22954 edges; zero probability is not better than uninitialized.
22956 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
22958 * asan.h (asan_sanitize_allocas_p): Declare.
22959 * asan.c (asan_sanitize_allocas_p): New function.
22960 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
22961 (handle_builtin_alloca): Likewise.
22962 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
22963 if !asan_sanitize_allocas_p.
22964 * params.def (asan-instrument-allocas): Add new option.
22965 * params.h (ASAN_PROTECT_ALLOCAS): Define.
22966 * opts.c (common_handle_option): Disable allocas sanitization for
22969 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
22971 * asan.c: Include gimple-fold.h.
22972 (get_last_alloca_addr): New function.
22973 (handle_builtin_stackrestore): Likewise.
22974 (handle_builtin_alloca): Likewise.
22975 (asan_emit_allocas_unpoison): Likewise.
22976 (get_mem_refs_of_builtin_call): Add new parameter, remove const
22977 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
22978 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
22979 (instrument_builtin_call): Pass gimple iterator to
22980 get_mem_refs_of_builtin_call.
22981 (last_alloca_addr): New global.
22982 * asan.h (asan_emit_allocas_unpoison): Declare.
22983 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
22984 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
22985 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
22986 if function calls alloca.
22987 * gimple-fold.c (replace_call_with_value): Remove static keyword.
22988 * gimple-fold.h (replace_call_with_value): Declare.
22989 * internal-fn.c: Include asan.h.
22990 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
22991 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
22993 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22995 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
22996 (C_SELFTEST_FLAGS): New.
22997 (CPP_SELFTEST_FLAGS): New.
22998 (SELFTEST_DEPS): New, from deps of s-selftest.
22999 (C_SELFTEST_DEPS): New, from deps of s-selftest.
23000 (CPP_SELFTEST_DEPS): New.
23001 (selftest): Add dependency on s-selftest-c++.
23002 (s-selftest): Rename to...
23003 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
23004 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
23005 than SELFTEST_FLAGS.
23006 (selftest-gdb): Rename to...
23007 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
23009 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
23010 (selftest-valgrind): Rename to...
23011 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
23013 (selftest-valgrind): Reintroduce as an alias for
23014 selftest-c-valgrind.
23015 (s-selftest-c++): New.
23016 (selftest-c++-gdb): New.
23017 (selftest-c++-valgrind): New.
23019 2017-07-06 Olivier Hainque <hainque@adacore.com>
23021 * gcc.c (process_command): When deciding if undefined variables
23022 should be ignored when processing specs, accept "gcc -v" as well.
23024 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23026 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
23027 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
23029 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
23031 * config/arm/arm-cpus.in (armv8-r): Add new entry.
23032 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
23033 * config/arm/arm-tables.opt: Regenerate.
23034 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
23036 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
23038 2017-07-06 Carl Love <cel@us.ibm.com>
23040 * ChangeLog: Clean up from mid air collision
23042 2017-07-06 Carl Love <cel@us.ibm.com>
23044 * config/rs6000/rs6000-c.c: Add support for built-in functions
23045 vector signed int vec_subc (vector signed int, vector signed int);
23046 vector signed __int128 vec_subc (vector signed __int128,
23047 vector signed __int128);
23048 vector unsigned __int128 vec_subc (vector unsigned __int128,
23049 vector unsigned __int128);
23050 vector signed int vec_sube (vector signed int, vector signed int,
23051 vector signed int);
23052 vector unsigned int vec_sube (vector unsigned int,
23053 vector unsigned int,
23054 vector unsigned int);
23055 vector signed __int128 vec_sube (vector signed __int128,
23056 vector signed __int128,
23057 vector signed__int128);
23058 vector unsigned __int128 vec_sube (vector unsigned __int128,
23059 vector unsigned __int128,
23060 vector unsigned __int128);
23061 vector signed int vec_subec (vector signed int, vector signed int,
23062 vector signed int);
23063 vector unsigned int vec_subec (vector unsigned int,
23064 vector unsigned int,
23065 vector unsigned int);
23066 vector signed __int128 vec_subec (vector signed __int128,
23067 vector signed __int128,
23068 vector signed__int128);
23069 vector unsigned __int128 vec_subec (vector unsigned __int128,
23070 vector unsigned __int128,
23071 vector unsigned __int128);
23072 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
23073 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
23074 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
23075 BU_ALTIVEC_OVERLOAD_X definitions.
23076 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
23077 * doc/extend.texi: Update the built-in documentation file for the new
23078 built-in functions.
23080 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23083 * diagnostic-show-locus.c (layout::layout): Use start and finish
23084 spelling location for the start and finish of each range.
23085 * genmatch.c (linemap_client_expand_location_to_spelling_point):
23086 Add unused aspect param.
23087 * input.c (expand_location_1): Add "aspect" param, and use it
23088 to access the correct part of the location.
23089 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
23091 (expand_location_to_spelling_point): Likewise.
23092 (linemap_client_expand_location_to_spelling_point): Add "aspect"
23093 param, and pass it to expand_location_1.
23095 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
23097 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
23098 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
23099 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
23100 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
23101 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
23102 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
23103 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
23104 _mm_maskz_getmant_ss): New intrinsics.
23105 (__builtin_ia32_getexpss128_mask): Changed to ...
23106 __builtin_ia32_getexpss128_round ... this.
23107 (__builtin_ia32_getexpsd128_mask): Changed to ...
23108 __builtin_ia32_getexpsd128_round ... this.
23109 * config/i386/i386-builtin-types.def
23110 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
23111 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
23112 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
23113 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
23114 __builtin_ia32_getmantss_mask_round): New builtins.
23115 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
23116 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
23117 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
23118 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
23119 * config/i386/sse.md
23120 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
23121 avx512f_sgetexp<mode><mask_scalar_name>
23122 <round_saeonly_scalar_name> ... this.
23123 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
23124 %0, %1, %2<round_saeonly_op3>}): Changed to ...
23125 vgetexp<ssescalarmodesuffix>
23126 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23127 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
23128 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
23129 avx512f_vgetmant<mode><mask_scalar_name>
23130 <round_saeonly_scalar_name> ... this.
23131 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
23132 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
23133 vgetmant<ssescalarmodesuffix>
23134 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
23135 %0<mask_scalar_operand4>, %1, %2
23136 <round_saeonly_scalar_mask_op4>, %3} ... this.
23137 * config/i386/subst.md (mask_scalar_operand4,
23138 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
23139 round_saeonly_scalar_nimm_predicate): New subst attributes.
23141 2017-07-06 Julia Koval <julia.koval@intel.com>
23143 * config/i386/i386.c (ix86_erase_embedded_rounding):
23144 Remove code for old rounding pattern.
23146 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
23148 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
23150 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
23152 * doc/sourcebuild.texi (Test Directives, Variants of
23153 dg-require-support): Add documentation for dg-require-stack-check.
23155 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
23157 * config/i386/subst.md (mask_scalar, round_scalar,
23158 round_saeonly_scalar): New meta-templates.
23159 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
23160 round_scalar_mask_operand3, round_scalar_mask_op3,
23161 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
23162 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
23163 round_saeonly_scalar_constraint,
23164 round_saeonly_scalar_prefix): New subst attribute.
23165 * config/i386/sse.md
23166 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
23167 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
23168 <round_scalar_name> ... this.
23169 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
23170 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
23171 <round_scalar_name> ... this.
23172 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
23173 <sse>_vm<code><mode>3<mask_scalar_name>
23174 <round_saeonly_scalar_name> ... this.
23175 (v<plusminus_mnemonic><ssescalarmodesuffix>
23176 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
23177 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
23178 v<plusminus_mnemonic><ssescalarmodesuffix>
23179 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23180 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
23181 (v<multdiv_mnemonic><ssescalarmodesuffix>
23182 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
23183 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
23184 v<multdiv_mnemonic><ssescalarmodesuffix>
23185 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23186 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
23187 (v<maxmin_float><ssescalarmodesuffix>
23188 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
23189 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
23190 v<maxmin_float><ssescalarmodesuffix>
23191 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23192 %0<mask_scalar_operand3>, %1, %<iptr>2
23193 <round_saeonly_scalar_mask_op3>} ... this.
23195 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
23197 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
23198 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
23200 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
23201 Alan Hayward <alan.hayward@arm.com>
23202 David Sherwood <david.sherwood@arm.com>
23204 * combine.c (simplify_if_then_else): Remove "enum" before
23206 * compare-elim.c (can_eliminate_compare): Likewise.
23207 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
23209 (aarch64_lookup_simd_builtin_type): Likewise.
23210 (aarch64_simd_builtin_type): Likewise.
23211 (aarch64_init_simd_builtin_types): Likewise.
23212 (aarch64_simd_expand_args): Likewise.
23213 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
23215 (aarch64_reverse_mask): Likewise.
23216 (aarch64_simd_emit_reg_reg_move): Likewise.
23217 (aarch64_gen_adjusted_ldpstp): Likewise.
23218 (aarch64_ccmp_mode_to_code): Likewise.
23219 (aarch64_operands_ok_for_ldpstp): Likewise.
23220 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
23221 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
23223 (aarch64_min_divisions_for_recip_mul): Likewise.
23224 (aarch64_reassociation_width): Likewise.
23225 (aarch64_get_condition_code_1): Likewise.
23226 (aarch64_simd_emit_reg_reg_move): Likewise.
23227 (aarch64_simd_attr_length_rglist): Likewise.
23228 (aarch64_reverse_mask): Likewise.
23229 (aarch64_operands_ok_for_ldpstp): Likewise.
23230 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
23231 (aarch64_gen_adjusted_ldpstp): Likewise.
23232 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
23234 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
23235 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
23236 (arm_lookup_simd_builtin_type): Likewise.
23237 (arm_simd_builtin_type): Likewise.
23238 (arm_init_simd_builtin_types): Likewise.
23239 (arm_expand_builtin_args): Likewise.
23240 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
23241 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
23242 (ft32_setup_incoming_varargs): Likewise.
23243 (ft32_function_arg): Likewise.
23244 (ft32_function_arg_advance): Likewise.
23245 (ft32_pass_by_reference): Likewise.
23246 (ft32_arg_partial_bytes): Likewise.
23247 (ft32_valid_pointer_mode): Likewise.
23248 (ft32_addr_space_pointer_mode): Likewise.
23249 (ft32_addr_space_legitimate_address_p): Likewise.
23250 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
23252 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
23253 (ix86_emit_outlined_ms2sysv_restore): Likewise.
23254 (iamcu_alignment): Likewise.
23255 (canonicalize_vector_int_perm): Likewise.
23256 (ix86_noce_conversion_profitable_p): Likewise.
23257 (ix86_mpx_bound_mode): Likewise.
23258 (ix86_operands_ok_for_move_multiple): Likewise.
23259 * config/microblaze/microblaze-protos.h
23260 (microblaze_expand_conditional_branch_reg): Likewise.
23261 * config/microblaze/microblaze.c
23262 (microblaze_expand_conditional_branch_reg): Likewise.
23263 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
23265 (rs6000_reassociation_width): Likewise.
23266 (rs6000_invalid_binary_op): Likewise.
23267 (fusion_p9_p): Likewise.
23268 (emit_fusion_p9_load): Likewise.
23269 (emit_fusion_p9_store): Likewise.
23270 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
23272 (riscv_hard_regno_mode_ok_p): Likewise.
23273 (riscv_address_insns): Likewise.
23274 (riscv_split_symbol): Likewise.
23275 (riscv_legitimize_move): Likewise.
23276 (riscv_function_value): Likewise.
23277 (riscv_hard_regno_nregs): Likewise.
23278 (riscv_expand_builtin): Likewise.
23279 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
23280 (riscv_build_integer): Likewise.
23281 (riscv_split_integer): Likewise.
23282 (riscv_legitimate_constant_p): Likewise.
23283 (riscv_cannot_force_const_mem): Likewise.
23284 (riscv_regno_mode_ok_for_base_p): Likewise.
23285 (riscv_valid_base_register_p): Likewise.
23286 (riscv_valid_offset_p): Likewise.
23287 (riscv_valid_lo_sum_p): Likewise.
23288 (riscv_classify_address): Likewise.
23289 (riscv_legitimate_address_p): Likewise.
23290 (riscv_address_insns): Likewise.
23291 (riscv_load_store_insns): Likewise.
23292 (riscv_force_binary): Likewise.
23293 (riscv_split_symbol): Likewise.
23294 (riscv_force_address): Likewise.
23295 (riscv_legitimize_address): Likewise.
23296 (riscv_move_integer): Likewise.
23297 (riscv_legitimize_const_move): Likewise.
23298 (riscv_legitimize_move): Likewise.
23299 (riscv_address_cost): Likewise.
23300 (riscv_subword): Likewise.
23301 (riscv_output_move): Likewise.
23302 (riscv_canonicalize_int_order_test): Likewise.
23303 (riscv_emit_int_order_test): Likewise.
23304 (riscv_function_arg_boundary): Likewise.
23305 (riscv_pass_mode_in_fpr_p): Likewise.
23306 (riscv_pass_fpr_single): Likewise.
23307 (riscv_pass_fpr_pair): Likewise.
23308 (riscv_get_arg_info): Likewise.
23309 (riscv_function_arg): Likewise.
23310 (riscv_function_arg_advance): Likewise.
23311 (riscv_arg_partial_bytes): Likewise.
23312 (riscv_function_value): Likewise.
23313 (riscv_pass_by_reference): Likewise.
23314 (riscv_setup_incoming_varargs): Likewise.
23315 (riscv_print_operand): Likewise.
23316 (riscv_elf_select_rtx_section): Likewise.
23317 (riscv_save_restore_reg): Likewise.
23318 (riscv_for_each_saved_reg): Likewise.
23319 (riscv_register_move_cost): Likewise.
23320 (riscv_hard_regno_mode_ok_p): Likewise.
23321 (riscv_hard_regno_nregs): Likewise.
23322 (riscv_class_max_nregs): Likewise.
23323 (riscv_memory_move_cost): Likewise.
23324 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
23325 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
23326 (rl78_addr_space_address_mode): Likewise.
23327 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
23329 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
23330 (rs6000_reassociation_width): Likewise.
23331 (rs6000_invalid_binary_op): Likewise.
23332 (fusion_p9_p): Likewise.
23333 (emit_fusion_p9_load): Likewise.
23334 (emit_fusion_p9_store): Likewise.
23335 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
23336 (ok_for_simple_move_operands): Likewise.
23337 (ok_for_simple_move_strict_operands): Likewise.
23338 (ok_for_simple_arith_logic_operands): Likewise.
23339 (visium_legitimize_reload_address): Likewise.
23340 (visium_select_cc_mode): Likewise.
23341 (output_cbranch): Likewise.
23342 (visium_split_double_move): Likewise.
23343 (visium_expand_copysign): Likewise.
23344 (visium_expand_int_cstore): Likewise.
23345 (visium_expand_fp_cstore): Likewise.
23346 * config/visium/visium.c (visium_pass_by_reference): Likewise.
23347 (visium_function_arg): Likewise.
23348 (visium_function_arg_advance): Likewise.
23349 (visium_libcall_value): Likewise.
23350 (visium_setup_incoming_varargs): Likewise.
23351 (visium_legitimate_constant_p): Likewise.
23352 (visium_legitimate_address_p): Likewise.
23353 (visium_legitimize_address): Likewise.
23354 (visium_secondary_reload): Likewise.
23355 (visium_register_move_cost): Likewise.
23356 (visium_memory_move_cost): Likewise.
23357 (prepare_move_operands): Likewise.
23358 (ok_for_simple_move_operands): Likewise.
23359 (ok_for_simple_move_strict_operands): Likewise.
23360 (ok_for_simple_arith_logic_operands): Likewise.
23361 (visium_function_value_1): Likewise.
23362 (rtx_ok_for_offset_p): Likewise.
23363 (visium_legitimize_reload_address): Likewise.
23364 (visium_split_double_move): Likewise.
23365 (visium_expand_copysign): Likewise.
23366 (visium_expand_int_cstore): Likewise.
23367 (visium_expand_fp_cstore): Likewise.
23368 (visium_split_cstore): Likewise.
23369 (visium_select_cc_mode): Likewise.
23370 (visium_split_cbranch): Likewise.
23371 (output_cbranch): Likewise.
23372 (visium_print_operand_address): Likewise.
23373 * expmed.c (flip_storage_order): Likewise.
23374 * expmed.h (emit_cstore): Likewise.
23375 (flip_storage_order): Likewise.
23376 * genrecog.c (validate_pattern): Likewise.
23377 * hsa-gen.c (gen_hsa_addr): Likewise.
23378 * internal-fn.c (expand_arith_overflow): Likewise.
23379 * ira-color.c (allocno_copy_cost_saving): Likewise.
23380 * lra-assigns.c (find_hard_regno_for_1): Likewise.
23381 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
23382 (process_invariant_for_inheritance): Likewise.
23383 * lra-eliminations.c (move_plus_up): Likewise.
23384 * omp-low.c (lower_oacc_reductions): Likewise.
23385 * simplify-rtx.c (simplify_subreg): Likewise.
23386 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
23387 (TARGET_CHKP_BOUND_MODE): Likewise..
23388 * targhooks.c (default_chkp_bound_mode): Likewise.
23389 (default_setup_incoming_vararg_bounds): Likewise.
23390 * targhooks.h (default_chkp_bound_mode): Likewise.
23391 (default_setup_incoming_vararg_bounds): Likewise.
23392 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
23393 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
23394 (have_whole_vector_shift): Likewise.
23395 * tree-vect-stmts.c (vectorizable_load): Likewise.
23396 * doc/tm.texi: Regenerate.
23398 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
23400 Graceful degrade if Binutils PR21472 is not available.
23403 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
23404 .rodata in flash test fails.
23405 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
23406 * confgure: Regenerate.
23407 * config.in: Regenerate.
23408 * config/avr/avr.c (avr_asm_named_section)
23409 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
23410 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
23411 (avr_asm_init_sections): Same.
23413 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23415 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
23416 (fma<VH:mode>4_intrinsic): Likewise.
23417 (*fmsub<VCVTF:mode>4): Likewise.
23418 (*fmsub<VH:mode>4_intrinsic): Likewise.
23420 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
23423 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
23424 Don't depend on "optimize > 0".
23425 (out_movhi_r_mr, out_movqi_mr_r): Same.
23426 (out_movhi_mr_r, out_movqi_r_mr): Same.
23427 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
23428 io_address_operand on "optimize > 0".
23430 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23432 * tree-loop-distribution.c: Add general explanantion on the pass.
23433 (generate_loops_for_partition): Mark distributed loop.
23434 (pg_add_dependence_edges): New parameter. Handle alias data
23435 dependence specially and record it in the parameter if asked.
23436 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
23437 (init_partition_graph_vertices, add_partition_graph_edge): New.
23438 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
23439 (free_partition_graph_vdata, build_partition_graph): New.
23440 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
23441 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
23442 (data_ref_segment_size, latch_dominated_by_data_ref): New.
23443 (compute_alias_check_pairs, version_loop_by_alias_check): New.
23444 (version_for_distribution_p, finalize_partitions): New.
23445 (distribute_loop): Handle alias data dependence specially. Factor
23446 out loop fusion code as functions and call these functions.
23448 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23450 * tree-loop-distribution.c (classify_partition): New parameter and
23451 better handle reduction statement.
23452 (rdg_build_partitions): Revise comment.
23453 (distribute_loop): Compute statements in all partitions and pass it
23454 to classify_partition.
23456 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23458 * tree-loop-distribution.c (enum partition_type): New.
23459 (struct partition): New field type.
23460 (partition_merge_into): Add parameter. Update partition type.
23461 (data_dep_in_cycle_p, update_type_for_merge): New functions.
23462 (build_rdg_partition_for_vertex): Compute partition type.
23463 (rdg_build_partitions): Dump partition type.
23464 (distribute_loop): Update calls to partition_merge_into.
23466 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23468 * tree-loop-distribution.c (struct ddr_hasher): New.
23469 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
23471 (classify_partition): Call get_data_dependence.
23472 (pg_add_dependence_edges): Ditto.
23473 (distribute_loop): Release data dependence hash table.
23475 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23477 * tree-loop-distribution.c (ref_base_address): Delete.
23478 (similar_memory_accesses): Rename ...
23479 (share_memory_accesses): ... to this. Check if partitions access
23480 the same memory reference.
23481 (distribute_loop): Call share_memory_accesses.
23483 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23485 * tree-loop-distribution.c (struct partition): New field recording
23486 its data reference.
23487 (partition_alloc, partition_free): Init and release data refs.
23488 (partition_merge_into): Merge data refs.
23489 (build_rdg_partition_for_vertex): Collect data refs for partition.
23490 (pg_add_dependence_edges): Change parameters from vector to bitmap.
23492 (distribute_loop): Remve data refs from vertice data of partition
23495 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23497 * tree-loop-distribution.c (params.h): Include header file.
23498 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
23499 (datarefs_vec): New global var.
23500 (create_rdg_vertices): Use datarefs_vec directly.
23501 (free_rdg): Don't free data references.
23502 (build_rdg): Update use. Don't free data references.
23503 (distribute_loop): Compute global variable for data references.
23504 Bail out if there are too many data references.
23506 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23508 * tree-loop-distribution.c (loop_nest): New global var.
23509 (build_rdg): Use loop directly, rather than loop nest.
23510 (pg_add_dependence_edges): Remove loop nest parameter. Use global
23512 (distribute_loop): Compute global variable loop nest. Update use.
23514 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23516 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
23517 (partition_merge_into): New parameter. Dump reason for fusion.
23518 (distribute_loop): Update use of partition_merge_into.
23520 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23522 * tree-loop-distribution.c (bb_top_order_index): New.
23523 (bb_top_order_index_size, bb_top_order_cmp): New.
23524 (stmts_from_loop): Use topological order.
23525 (pass_loop_distribution::execute): Compute and release topological
23526 order for basic blocks.
23528 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23530 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
23533 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23535 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
23536 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
23537 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
23538 * internal-fn.def (LOOP_DIST_ALIAS): New.
23539 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
23540 (fold_loop_internal_call): ... this.
23541 (vect_loop_dist_alias_call): New function.
23542 (set_uid_loop_bbs): Call fold_loop_internal_call.
23543 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
23546 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
23549 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
23550 Require dead FLAGS_REG at the beginning of a peephole.
23552 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
23555 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
23556 arguments in the call to __builtin_ia32_sbb_u32.
23557 (_subborrow_u64): Swap _X and _Y arguments in the call to
23558 __builtin_ia32_sbb_u64.
23560 2017-07-04 Jakub Jelinek <jakub@redhat.com>
23563 * tree-vrp.c (compare_assert_loc): Turn into a function template
23564 with stable template parameter. Only test if a->e is NULL,
23565 !a->e == !b->e has been verified already. Use e == NULL or
23566 e != NULL instead of e or ! e tests. If stable is true, don't use
23567 iterative_hash_expr, on the other side allow a or b or both NULL
23568 and sort the NULLs last.
23569 (process_assert_insertions): Sort using compare_assert_loc<false>
23570 instead of compare_assert_loc, later sort using
23571 compare_assert_loc<true> before calling process_assert_insertions_for
23572 in a loop. Use break instead of continue once seen NULL pointer.
23574 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23576 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23577 Cortex-R7 and Cortex-R8 processors.
23579 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
23581 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
23582 uninitialized while src is not.
23584 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
23586 * common/config/arm/arm-common.c: Adjust include path for
23588 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
23589 (arm-cpu.h): Create in build directory. Adjust dependency rules.
23590 (arm-cpu-data.h): Likewise.
23591 (arm-cpu-cdata.h): Likewise.
23592 * config/arm/arm-cpu.h: Delete.
23593 * config/arm/arm-cpu-cdata.h: Delete.
23594 * config/arm/arm-cpu-data.h: Delete.
23596 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
23598 * config/arm/arm-cpus.in (cortex-a55): New.
23599 (cortex-a75): Likewise.
23600 (cortex-a75.cortex-a55): Likewise.
23601 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
23603 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
23604 * config/arm/arm-cpu-cdata.h: Regenerate.
23605 * config/arm/arm-cpu-data.h: Regenerate.
23606 * config/arm/arm-cpu.h: Regenerate.
23607 * config/arm/arm-tables.opt: Regenerate.
23608 * config/arm/arm-tune.md: Regenerate.
23610 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
23612 * haifa-sched.c (sched_create_recovery_edges): Update profile.
23614 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
23616 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
23619 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
23621 PR tree-optimization/81292
23622 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
23623 full_string_p, also call adjust_related_strinfos if the adjustment
23624 is simple, otherwise invalidate related strinfos.
23626 2017-07-04 Martin Liska <mliska@suse.cz>
23629 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
23630 newly created variable as DECL_IGNORED_P.
23632 2017-07-04 Martin Liska <mliska@suse.cz>
23635 * ipa-inline.c (inline_small_functions):
23636 Use xstrdup_for_dump.
23638 2017-07-04 Tom de Vries <tom@codesourcery.com>
23640 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
23642 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
23645 * config/darwin.c (darwin_function_switched_text_sections):
23648 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
23650 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
23652 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
23654 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
23656 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23658 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
23659 min_profitable_iters, and th as inclusive lower bounds.
23660 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
23661 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
23662 for min_profitable_iters and min_profitable_estimate.
23663 (vect_transform_loop): Treat th as an inclusive lower bound.
23664 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
23666 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
23669 * config/darwin.c (darwin_function_switched_text_sections):
23670 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
23671 in two pieces, and suppress the use of buf.
23673 2017-07-03 Nathan Sidwell <nathan@acm.org>
23675 * hash-table.h (hash_table_mod1): Fix indentation.
23677 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23679 PR middle-end/81290
23680 * predict.c (force_edge_cold): Be more careful about propagation
23682 * profile-count.h (profile_probability::guessed,
23683 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
23685 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
23687 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
23689 * doc/invoke.texi (rcpc architecture extension): Document it.
23691 2017-07-03 Richard Biener <rguenther@suse.de>
23693 PR tree-optimization/60510
23694 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
23695 the scalar reduction PHI and use it.
23696 (vectorizable_reduction): Properly guard the single_defuse_cycle
23697 path for non-SLP reduction chains where we cannot use it.
23698 Rework reduc_def/index and vector type deduction. Rework
23699 vector operand gathering during reduction op code-gen.
23700 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
23701 chains dissolve the chain and leave it to non-SLP reduction
23704 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23706 * tree-data-ref.h (dr_alignment): Declare.
23707 * tree-data-ref.c (dr_alignment): New function.
23708 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
23709 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
23711 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
23713 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23715 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
23716 and base_misalignment fields.
23717 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
23718 * tree-data-ref.c: Include builtins.h.
23719 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
23720 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
23721 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
23722 * tree-vect-data-refs.c: Include tree-cfg.h.
23723 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
23724 fields instead of calculating an alignment here.
23725 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
23726 innermost_loop_behavior fields.
23728 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23730 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
23732 (DR_STEP_ALIGNMENT): New macro.
23733 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
23734 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
23735 (create_data_ref): Print it.
23736 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
23737 to tell whether the step preserves vector (mis)alignment.
23738 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
23739 Move the check for an integer step and generalise to all INTEGER_CST.
23740 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
23741 Print the outer step alignment.
23743 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23745 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
23746 with offset_alignment.
23747 (DR_ALIGNED_TO): Delete.
23748 (DR_OFFSET_ALIGNMENT): New macro.
23749 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
23750 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
23751 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
23752 (create_data_ref): Likewise.
23753 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
23754 (vect_analyze_data_refs): Likewise.
23755 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
23756 creating dummy innermost behavior.
23758 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23760 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
23761 with a "innermost_loop_behavior *" and refeence tree.
23762 * tree-data-ref.c (dr_analyze_innermost): Likewise.
23763 (create_data_ref): Update call accordingly.
23764 * tree-predcom.c (find_looparound_phi): Likewise.
23766 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23768 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
23769 fields with dr_wrt_vec_loop.
23770 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
23771 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
23772 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
23773 (vect_dr_behavior): New function.
23774 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
23775 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
23776 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
23777 track whether the step preserves the misalignment.
23778 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
23779 Use vect_dr_behavior.
23780 (vect_setup_realignment): Update call accordingly.
23781 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
23782 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
23783 call to vect_create_addr_base_for_vector_ref.
23784 (vect_create_cond_for_align_checks): Likewise.
23785 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
23786 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
23787 (vect_recog_mask_conversion_pattern): Likewise.
23788 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
23789 (new_stmt_vec_info): Remove redundant zeroing.
23791 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
23793 * common/config/arm/arm-common.c (arm_be8_option): New function.
23794 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
23795 (ISA_ARMv6): Add isa_bit_be8.
23796 * config/arm/arm.h (arm_be8_option): Add prototype.
23797 (BE8_SPEC_FUNCTION): New define.
23798 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
23799 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
23800 (mlittle-endian): Similarly.
23801 (mbe8, mbe32): New options.
23802 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
23803 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
23805 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23807 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
23809 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23811 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
23812 (cleanup_tree_cfg_bb): Use it.
23813 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
23815 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
23817 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23820 * loop-doloop.c (add_test): Update profile.
23822 2017-07-03 Martin Liska <mliska@suse.cz>
23825 * sanopt.c (rewrite_usage_of_param): New function.
23826 (sanitize_rewrite_addressable_params): Likewise.
23827 (pass_sanopt::execute): Call rewrite_usage_of_param.
23829 2017-07-03 Richard Biener <rguenther@suse.de>
23831 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
23832 back to using VIEW_CONVERT_EXPR.
23834 2017-07-03 Martin Liska <mliska@suse.cz>
23837 * doc/extend.texi: Document when a resolver function is
23838 generated for target_clones.
23840 2017-07-03 Martin Liska <mliska@suse.cz>
23842 * asan.c (asan_emit_stack_protection): Unpoison just red zones
23843 and shadow memory of auto variables which are subject of
23844 use-after-scope sanitization.
23845 (asan_expand_mark_ifn): Add do set only when is_poison.
23847 2016-07-03 Richard Biener <rguenther@suse.de>
23849 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
23851 (vect_force_simple_reduction): Record reduction def -> phi mapping.
23852 (vectorizable_reduction): Perform reduction PHI creation when
23853 visiting a reduction PHI and adjust and simplify code generation
23854 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
23855 (vect_transform_loop): Visit reduction PHIs.
23856 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
23857 defs into the SLP tree.
23858 (vect_build_slp_tree): Reduction defs terminate the recursion.
23859 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
23861 (vect_get_vec_defs_for_stmt_copy): Export.
23862 (vect_get_vec_defs): Likewise.
23863 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
23865 (vect_get_vec_defs_for_stmt_copy): Declare.
23866 (vect_get_vec_defs): Likewise.
23868 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23870 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
23871 parameter with a "loop" parameter and use it instead of the
23872 loop containing DR_STMT. Don't check simple_iv when doing
23873 BB analysis. Describe the two analysis modes in the comment.
23875 2017-07-03 Tom de Vries <tom@codesourcery.com>
23877 PR tree-optimization/69468
23878 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
23879 (find_same_succ_bb): Handle ignore_edge_flags.
23881 2017-07-03 Tom de Vries <tom@codesourcery.com>
23883 PR tree-optimization/81192
23884 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
23886 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
23888 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
23890 2017-07-03 Tom de Vries <tom@codesourcery.com>
23892 PR tree-optimization/81192
23893 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
23894 BB_SAME_SUCC (bb) == NULL.
23896 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23898 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
23901 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23903 * dumpfile.c: Include profile-count.h
23904 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
23906 (insert_cond_bb): Update profile.
23907 * tree-cfg.h (insert_cond_bb): Update prototype.
23908 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
23909 * tree-dump.c: Do not include tree-cfg.
23911 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23913 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
23915 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23917 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
23920 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23922 * tree-complex.c (expand_complex_div_wide): update profile.
23924 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23925 Alan Hayward <alan.hayward@arm.com>
23926 David Sherwood <david.sherwood@arm.com>
23928 * Makefile.in (MACHMODE_H): Remove insn-modes.h
23929 (CORETYPES_H): New define.
23930 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
23931 (insn-modes-inline.h, s-modes-inline-h): New rules.
23932 (generated_files): Add insn-modes-inline.h.
23933 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
23934 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
23935 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
23936 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
23937 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
23938 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
23939 (build/gencodes.o, build/genconditions.o): Likewise.
23940 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
23941 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
23942 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
23943 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
23944 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
23945 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
23946 * coretypes.h: Include everything up to real.h for generators.
23947 Include insn-modes.h first. Include wide-int-print.h after
23948 wide-int.h. Include insn-modes-inline.h and then machmode.h.
23949 * machmode.h: Don't include insn-modes.h here.
23950 * function-tests.c: Remove includes of signop.h, machmode.h,
23951 double-int.h and wide-int.h.
23953 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
23955 * optc-save-gen.awk: Likewise.
23956 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
23957 * godump.c: Remove include of wide-int-print.h.
23958 * pretty-print.h: Likewise.
23959 * wide-int-print.cc: Likewise.
23960 * wide-int.cc: Likewise.
23961 * hash-map-tests.c: Remove include of signop.h.
23962 * hash-set-tests.c: Likewise.
23963 * rtl-tests.c: Likewise.
23964 * mkconfig.sh: Remove include of machmode.h.
23965 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
23967 (emit_insn_modes_inline_h): ...this new function. Emit the code
23968 into an insn-modes-inline.h header file, adding appropriate
23969 include guards and end comments.
23970 (emit_insn_modes_c_header): Remove include of machmode.h.
23971 (emit_min_insn_modes_c_header): Include coretypes.h rather than
23973 (main): Handle -i flag and call emit_insn_modes_inline_h when
23976 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23978 * tree-ssa-strlen.c (strinfo): Rename the length field to
23979 nonzero_chars. Add a full_string_p field.
23980 (compare_nonzero_chars, zero_length_string_p): New functions.
23981 (get_addr_stridx): Add an offset_out parameter.
23982 Use compare_nonzero_chars.
23983 (get_stridx): Update accordingly. Use compare_nonzero_chars.
23984 (new_strinfo): Update after above changes to strinfo.
23985 (set_endptr_and_length): Set full_string_p.
23986 (get_string_length): Update after above changes to strinfo.
23987 (unshare_strinfo): Update call to new_strinfo.
23988 (maybe_invalidate): Likewise.
23989 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
23990 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
23991 as a uhwi instead of an shwi. Update after above changes to
23992 strinfo and new_strinfo.
23993 (zero_length_string): Assert that chainsi contains full strings.
23994 Use zero_length_string_p. Update call to new_strinfo.
23995 (adjust_related_strinfos): Update after above changes to strinfo.
23996 Copy full_string_p from origsi.
23997 (adjust_last_stmt): Use zero_length_string_p.
23998 (handle_builtin_strlen): Update after above changes to strinfo and
23999 new_strinfo. Install the lhs as the string length if the previous
24000 entry didn't describe a full string.
24001 (handle_builtin_strchr): Update after above changes to strinfo
24003 (handle_builtin_strcpy): Likewise.
24004 (handle_builtin_strcat): Likewise.
24005 (handle_builtin_malloc): Likewise.
24006 (handle_pointer_plus): Likewise.
24007 (handle_builtin_memcpy): Likewise. Track nonzero characters
24008 that aren't necessarily followed by a nul terminator.
24009 (handle_char_store): Likewise.
24011 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24013 PR tree-optimization/80769
24014 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
24015 for malloc and calloc. Document the new invariant that all related
24016 strinfos have delayed lengths or none do.
24017 (verify_related_strinfos): Move earlier in file.
24018 (set_endptr_and_length): New function, split out from...
24019 (get_string_length): ...here. Also set the lengths of related
24021 (zero_length_string): Assert that chainsi has known (rather than
24023 (adjust_related_strinfos): Likewise.
24025 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24027 PR tree-optimization/81136
24028 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
24029 assert that two references with the same misalignment have the same
24030 compile-time misalignment if those compile-time misalignments
24033 2017-07-01 Andi Kleen <ak@linux.intel.com>
24035 * print-tree.c (print_node): Print all attributes.
24037 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24039 * cfg.c (scale_bbs_frequencies): New function.
24040 * cfg.h (scale_bbs_frequencies): Declare it.
24041 * cfgloopanal.c (single_likely_exit): Cleanup.
24042 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
24044 (scale_loop_profile): Likewise.
24045 (loop_version): Likewise.
24046 (create_empty_loop_on_edge): Update.
24047 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
24048 scale_loop_frequencies, scale_loop_profile, loopify,
24049 loop_version): Update prototypes.
24050 * modulo-sched.c (sms_schedule): Update.
24051 * predict.c (unlikely_executed_edge_p): Also check probability.
24052 (probably_never_executed_edge_p): Fix typo.
24053 * tree-if-conv.c (version_loop_for_if_conversion): Update.
24054 * tree-parloops.c (gen_parallel_loop): Update.
24055 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
24056 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24057 * tree-ssa-loop-split.c (split_loop): Update.
24058 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
24059 * tree-vect-loop-manip.c (vect_do_peeling): Update.
24060 (vect_loop_versioning): Update.
24061 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24063 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24065 * trans-mem.c (split_bb_make_tm_edge): Update profile.
24067 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24069 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
24070 to keep profile consistent.
24072 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24074 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
24075 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
24076 * profile-count.h (max_safe_multiplier): Make unsigned.
24077 (profile_count::guessed_zero): New.
24079 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24081 * bb-reorder.c (fix_up_crossing_landing_pad,
24082 fix_crossing_conditional_branches): Use make_single_succ_edge
24083 to keep profile consistent.
24085 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24087 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
24090 2017-07-01 Jakub Jelinek <jakub@redhat.com>
24093 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
24094 the right scopes, make sure cond_jump isn't preserved between multiple
24095 iterations. Search for fallthru edge whenever there are 3+ edges and
24096 use find_fallthru_edge for it.
24098 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24100 Patch by Alexander Monakov <amonakov@ispras.ru>
24101 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
24102 probabilities consistently.
24104 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24106 * pa.c (pa_expand_compare_and_swap_loop): Update call of
24107 emit_cmp_and_jump_insns.
24109 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24112 * tree-inline.c (expand_call_inline): Combine profile statuses.
24114 2017-06-30 Andrew Pinski <apinski@cavium.com>
24116 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
24117 fold_stmt returned true.
24119 2017-06-30 Nathan Sidwell <nathan@acm.org>
24121 * ggc.h (empty_string): Delete.
24122 * cfgexpand.c (expand_asm_stmt): Use plain "".
24123 * optabs.c (expand_asm_memory_barrier): Likewise.
24124 * stringpool.c (empty_string): Delete.
24125 (digit_vector, digit_string): Delete.
24126 (ggc_alloc_string): Use plain "", don't optimize single digit
24127 strings. Use ggc_alloc_atomic.
24129 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
24131 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
24132 comparison set and one other set, use the cost of the non-comparison
24135 2017-06-30 Nathan Sidwell <nathan@acm.org>
24137 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
24140 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
24142 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
24143 loops. Remove now unneeded calls to gimple_switch_set_label() that
24144 just set removed labels to NULL_TREE.
24146 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
24148 * tree-ssanames.c (set_range_info_raw): Abstract from ...
24149 (set_range_info): ...here. Only call set_range_info_raw if domain
24151 (set_nonzero_bits): Call set_range_info_raw.
24152 * tree-ssanames.h (set_range_info_raw): New.
24154 2017-06-30 Jakub Jelinek <jakub@redhat.com>
24157 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
24158 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
24159 of nonimmediate_operand and <store_mask_constraint> instead of m
24160 for the input operand. For V8FI iterator, always split if input
24161 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
24162 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
24163 <store_mask_predicate> instead of register_operand and
24164 <store_mask_constraint> instead of v for the input operand. Make
24165 sure both operands aren't MEMs for if not <mask_applied>.
24167 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
24169 * lto-wrapper.c (copy_file) Close both file descriptors before
24172 2017-06-30 Martin Liska <mliska@suse.cz>
24175 * multiple_target.c (create_dispatcher_calls): Make ifunc
24176 also for function that don't have calls or are not referenced.
24178 2017-06-30 Richard Biener <rguenther@suse.de>
24180 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
24181 analyze the first scalar stmt. Move vector type computation
24182 for the BB case here from ...
24183 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
24184 live operation processing in the SLP case properly.
24186 2017-06-30 Richard Biener <rguenther@suse.de>
24188 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
24190 2017-06-30 Martin Liska <mliska@suse.cz>
24193 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
24194 before BUILT_IN_UNWIND_RESUME when ASAN is used.
24196 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
24198 * doc/invoke.texi (AArch64): Add missing options and remove redundant
24201 2017-06-30 Richard Biener <rguenther@suse.de>
24203 PR tree-optimization/81249
24204 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
24205 condition reduction result to original scalar type.
24207 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24209 * profile-count.h (enum profile_quality): Fix typos and whitespace
24212 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24214 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
24215 type for branch probabilities.
24217 2017-06-29 Julian Brown <julian@codesourcery.com>
24218 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24220 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
24221 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
24222 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
24223 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
24225 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24227 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
24228 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
24229 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
24230 CC usage from generic code to here.
24231 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
24232 CC usage into the target macros.
24234 2017-06-29 Maya Rashish <coypu@sdf.org>
24236 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
24239 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24241 * arm/arm-builtins.c: Include profile-count.h
24242 * except.c (sjlj_emit_function_enter): Use
24243 profile_probability::unlikely.
24245 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24247 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
24248 and tocrel_offset be pointer args rather than implicitly using
24250 (legitimate_constant_pool_address_p, rs6000_emit_move,
24251 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
24252 tocrel_offset and use in toc_relative_expr_p call.
24253 (print_operand, print_operand_address): Use static tocrel_base_oac
24254 and tocrel_offset_oac.
24255 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
24258 2017-06-29 Maya Rashish <coypu@sdf.org>
24260 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
24262 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
24264 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
24265 objects, take into account only the alignment of 'op0' and 'mode1' if
24268 2017-06-29 Steve Ellcey <sellcey@cavium.com>
24270 * ccmp.c (ccmp_tree_comparison_p): New function.
24271 (ccmp_candidate_p): Update to use above function.
24272 (get_compare_parts): New function.
24273 (expand_ccmp_next): Update to use new functions.
24274 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
24276 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
24277 take mode as argument.
24278 * ccmp.h (expand_ccmp_expr): Add mode as argument.
24279 * expr.c (expand_expr_real_1): Pass mode as argument.
24281 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
24283 * combine.c (combine_instructions): Print insns to dump_file, together
24286 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24288 * asan.c (asan_emit_stack_protection): Update.
24289 (create_cond_insert_point): Update.
24290 * auto-profile.c (afdo_propagate_circuit): Update.
24291 * basic-block.h (struct edge_def): Turn probability to
24292 profile_probability.
24293 (EDGE_FREQUENCY): Update.
24294 * bb-reorder.c (find_traces_1_round): Update.
24295 (better_edge_p): Update.
24296 (sanitize_hot_paths): Update.
24297 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
24298 (make_single_succ_edge): Update.
24299 (check_bb_profile): Update.
24300 (dump_edge_info): Update.
24301 (update_bb_profile_for_threading): Update.
24302 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
24303 probabilitycount to 0.
24304 * cfgbuild.c (compute_outgoing_frequencies): Update.
24305 * cfgcleanup.c (try_forward_edges): Update.
24306 (outgoing_edges_match): Update.
24307 (try_crossjump_to_edge): Update.
24308 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
24309 (expand_gimple_tailcall): Update.
24310 (construct_init_block): Use make_single_succ_edge.
24311 (construct_exit_block): Use make_single_succ_edge.
24312 * cfghooks.c (verify_flow_info): Update.
24313 (redirect_edge_succ_nodup): Update.
24314 (split_edge): Update.
24315 (account_profile_record): Update.
24316 * cfgloopanal.c (single_likely_exit): Update.
24317 * cfgloopmanip.c (scale_loop_profile): Update.
24318 (set_zero_probability): Remove.
24319 (duplicate_loop_to_header_edge): Update.
24320 * cfgloopmanip.h (loop_version): Update prototype.
24321 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
24322 (force_nonfallthru_and_redirect): Update.
24323 (update_br_prob_note): Update.
24324 (rtl_verify_edges): Update.
24325 (purge_dead_edges): Update.
24326 (rtl_lv_add_condition_to_bb): Update.
24327 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
24328 * cgraphunit.c (init_lowered_empty_function): Update.
24329 (cgraph_node::expand_thunk): Update.
24330 * cilk-common.c: Include profile-count.h
24331 * dojump.c (inv): Remove.
24332 (jumpifnot): Update.
24333 (jumpifnot_1): Update.
24334 (do_jump_1): Update.
24336 (do_jump_by_parts_greater_rtx): Update.
24337 (do_compare_rtx_and_jump): Update.
24338 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
24339 do_jump_1. do_compare_rtx_and_jump): Update prototype.
24340 * dwarf2cfi.c: Include profile-count.h
24341 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
24342 (sjlj_emit_dispatch_table): Likewise.
24343 * explow.c: Include profile-count.h
24344 * expmed.c (emit_store_flag_force): Update.
24345 (do_cmp_and_jump): Update.
24346 * expr.c (compare_by_pieces_d::generate): Update.
24347 (compare_by_pieces_d::finish_mode): Update.
24348 (emit_block_move_via_loop): Update.
24349 (store_expr_with_bounds): Update.
24350 (store_constructor): Update.
24351 (expand_expr_real_2): Update.
24352 (expand_expr_real_1): Update.
24353 * expr.h (try_casesi, try_tablejump): Update prototypes.
24354 * gimple-pretty-print.c (dump_probability): Update.
24355 (dump_profile): New.
24356 (dump_gimple_label): Update.
24357 (dump_gimple_bb_header): Update.
24358 * graph.c (draw_cfg_node_succ_edges): Update.
24359 * hsa-gen.c (convert_switch_statements): Update.
24360 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
24361 (find_if_case_1): Update.
24362 (find_if_case_2): Update.
24363 * internal-fn.c (expand_arith_overflow_result_store): Update.
24364 (expand_addsub_overflow): Update.
24365 (expand_neg_overflow): Update.
24366 (expand_mul_overflow): Update.
24367 (expand_vector_ubsan_overflow): Update.
24368 * ipa-cp.c (good_cloning_opportunity_p): Update.
24369 * ipa-split.c (split_function): Use make_single_succ_edge.
24370 * ipa-utils.c (ipa_merge_profiles): Update.
24371 * loop-doloop.c (add_test): Update.
24372 (doloop_modify): Update.
24373 * loop-unroll.c (compare_and_jump_seq): Update.
24374 (unroll_loop_runtime_iterations): Update.
24375 * lra-constraints.c (lra_inheritance): Update.
24376 * lto-streamer-in.c (input_cfg): Update.
24377 * lto-streamer-out.c (output_cfg): Update.
24378 * mcf.c (adjust_cfg_counts): Update.
24379 * modulo-sched.c (sms_schedule): Update.
24380 * omp-expand.c (expand_omp_for_init_counts): Update.
24381 (extract_omp_for_update_vars): Update.
24382 (expand_omp_ordered_sink): Update.
24383 (expand_omp_for_ordered_loops): Update.
24384 (expand_omp_for_generic): Update.
24385 (expand_omp_for_static_nochunk): Update.
24386 (expand_omp_for_static_chunk): Update.
24387 (expand_cilk_for): Update.
24388 (expand_omp_simd): Update.
24389 (expand_omp_taskloop_for_outer): Update.
24390 (expand_omp_taskloop_for_inner): Update.
24391 * omp-simd-clone.c (simd_clone_adjust): Update.
24392 * optabs.c (expand_doubleword_shift): Update.
24393 (expand_abs): Update.
24394 (emit_cmp_and_jump_insn_1): Update.
24395 (expand_compare_and_swap_loop): Update.
24396 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
24397 * predict.c (predictable_edge_p): Update.
24398 (edge_probability_reliable_p): Update.
24399 (set_even_probabilities): Update.
24400 (combine_predictions_for_insn): Update.
24401 (combine_predictions_for_bb): Update.
24402 (propagate_freq): Update.
24403 (estimate_bb_frequencies): Update.
24404 (force_edge_cold): Update.
24405 * profile-count.c (profile_count::dump): Add missing space into dump.
24406 (profile_count::debug): Add newline.
24407 (profile_count::differs_from_p): Explicitly convert to unsigned.
24408 (profile_count::stream_in): Update.
24409 (profile_probability::dump): New member function.
24410 (profile_probability::debug): New member function.
24411 (profile_probability::differs_from_p): New member function.
24412 (profile_probability::differs_lot_from_p): New member function.
24413 (profile_probability::stream_in): New member function.
24414 (profile_probability::stream_out): New member function.
24415 * profile-count.h (profile_count_quality): Rename to ...
24416 (profile_quality): ... this one.
24417 (profile_probability): New.
24418 (profile_count): Update.
24419 * profile.c (compute_branch_probabilities): Update.
24420 * recog.c (peep2_attempt): Update.
24421 * sched-ebb.c (schedule_ebbs): Update.
24422 * sched-rgn.c (find_single_block_region): Update.
24423 (compute_dom_prob_ps): Update.
24424 (schedule_region): Update.
24425 * sel-sched-ir.c (compute_succs_info): Update.
24426 * stmt.c (struct case_node): Update.
24427 (do_jump_if_equal): Update.
24428 (get_outgoing_edge_probs): Update.
24429 (conditional_probability): Update.
24430 (emit_case_dispatch_table): Update.
24431 (expand_case): Update.
24432 (expand_sjlj_dispatch_table): Update.
24433 (emit_case_nodes): Update.
24434 * targhooks.c: Update.
24435 * tracer.c (better_p): Update.
24436 (find_best_successor): Update.
24437 * trans-mem.c (expand_transaction): Update.
24438 * tree-call-cdce.c: Update.
24439 * tree-cfg.c (gimple_split_edge): Upate.
24440 (move_sese_region_to_fn): Upate.
24441 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
24442 * tree-eh.c (lower_resx): Upate.
24443 (cleanup_empty_eh_move_lp): Upate.
24444 * tree-if-conv.c (version_loop_for_if_conversion): Update.
24445 * tree-inline.c (copy_edges_for_bb): Update.
24446 (copy_cfg_body): Update.
24447 * tree-parloops.c (gen_parallel_loop): Update.
24448 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
24449 (gimple_gen_time_profiler): Update.
24450 * tree-ssa-dce.c (remove_dead_stmt): Update.
24451 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
24452 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
24453 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
24454 (unloop_loops): Update.
24455 (try_peel_loop): Update.
24456 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24457 * tree-ssa-loop-split.c (connect_loops): Update.
24458 (split_loop): Update.
24459 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
24460 (hoist_guard): Update.
24461 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
24462 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
24463 (value_replacement): Update.
24464 * tree-ssa-reassoc.c (branch_fixup): Update.
24465 * tree-ssa-tail-merge.c (replace_block_by): Update.
24466 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
24467 (create_edge_and_update_destination_phis): Update.
24468 (compute_path_counts): Update.
24469 (recompute_probabilities): Update.
24470 (update_joiner_offpath_counts): Update.
24471 (freqs_to_counts_path): Update.
24472 (duplicate_thread_path): Update.
24473 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
24474 (struct switch_conv_info): Update.
24475 (gen_inbound_check): Update.
24476 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
24477 (vect_do_peeling): Update.
24478 (vect_loop_versioning): Update.
24479 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24480 (optimize_mask_stores): Update.
24481 * ubsan.c (ubsan_expand_null_ifn): Update.
24482 * value-prof.c (gimple_divmod_fixed_value): Update.
24483 (gimple_divmod_fixed_value_transform): Update.
24484 (gimple_mod_pow2): Update.
24485 (gimple_mod_pow2_value_transform): Update.
24486 (gimple_mod_subtract): Update.
24487 (gimple_mod_subtract_transform): Update.
24488 (gimple_ic): Update.
24489 (gimple_stringop_fixed_value): Update.
24490 (gimple_stringops_transform): Update.
24491 * value-prof.h: Update.
24493 2017-06-29 Carl Love <cel@us.ibm.com>
24495 * config/rs6000/rs6000-c.c: Add support for built-in functions
24496 vector signed int vec_signed (vector float);
24497 vector signed long long vec_signed (vector double);
24498 vector signed int vec_signed2 (vector double, vector double);
24499 vector signed int vec_signede (vector double);
24500 vector signed int vec_signedo (vector double);
24501 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
24502 instruction generator.
24503 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
24504 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
24505 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
24507 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
24508 vunsignede_v2df): Add define_expands.
24509 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
24510 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
24511 VEC_UNSIGNEDO): Add definitions.
24512 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
24513 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
24514 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
24515 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
24516 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
24517 * config/rs6000/altivec.h (vec_signed, vec_signed2,
24518 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
24519 vec_unsignede, vec_unsignedo): Add builtin defines.
24520 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
24522 * doc/extend.texi: Update the built-in documentation file for the
24523 new built-in functions.
24525 2017-06-29 Richard Biener <rguenther@suse.de>
24527 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
24528 reduction chains to LOOP_VINFO_REDUCTIONS.
24529 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
24530 SLP reductions after processing reduction chains.
24532 2017-06-29 Nathan Sidwell <nathan@acm.org>
24534 * builtins.c (fold_builtin_FUNCTION): Use
24535 lang_hooks.decl_printable_name.
24537 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
24539 PR middle-end/81194
24540 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
24541 with only one label.
24542 * stmt.c (expand_case): Assert NCASES is greater than one.
24544 2017-06-29 Richard Biener <rguenther@suse.de>
24546 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
24548 (group_case_labels): Likewise.
24549 (find_taken_edge): Push sanity checking on val to workers...
24550 (find_taken_edge_cond_expr): ... here
24551 (find_taken_edge_switch_expr): ... and here, handle cases
24552 with just a default label.
24553 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
24554 (group_case_labels): Likewise.
24555 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
24556 group_case_labels does anything cleanup the CFG again.
24558 2017-06-29 Bin Cheng <bin.cheng@arm.com>
24560 PR tree-optimization/81196
24561 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
24562 exit condition comparing two IVs.
24564 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
24566 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
24567 profile to the dummy entry at the end of the list of architectures.
24568 * config/arm/arm-cpu-cdata.h: Regenerated.
24570 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24571 Michael Collison <michael.collison@arm.com>
24574 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
24576 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
24577 (*aarch64_reg_<mode>3_minus_mask): New pattern.
24578 (*aarch64_<optab>_reg_di3_mask2): New pattern.
24579 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
24580 of shift when the shift amount is masked with constant equal to
24581 the size of the mode.
24582 * config/aarch64/predicates.md (subreg_lowpart_operator): New
24585 2017-06-29 Martin Liska <mliska@suse.cz>
24587 * config/i386/i386.opt: Change range from [1,5] to [0,5].
24589 2017-06-29 Yury Gribov <tetra2005@gmail.com>
24592 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
24595 (edge_growth_cache_entry::edge_growth_cache_entry): New
24597 (reset_edge_growth_cache): Update to use constructor.
24599 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24601 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
24602 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
24603 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
24605 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
24607 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
24608 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
24610 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
24612 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
24613 (*-linux-uclibc*): Add t-uclibc tmake_file.
24614 * config/t-musl: New.
24615 * config/t-uclibc: New.
24617 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
24619 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
24621 (gen_comm_data): Emit architectural setting of arch_prof.
24622 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
24624 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
24625 (armv8-m.base, armv8-m.main): Likewise.
24626 * arm-protos.h (arm_build_target): Add profile field.
24627 (arch_option): Likewise.
24628 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
24630 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
24631 arm_active_target.profile.
24633 2017-06-28 Richard Biener <rguenther@suse.de>
24635 PR middle-end/81227
24636 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
24637 TYPE_OVERFLOW_WRAPS.
24638 * match.pd (negate_expr_p): Likewise.
24639 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
24640 fold_build2, not fold_binary.
24642 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
24644 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
24645 Convert memory address to Pmode.
24646 (aarch64_print_operand): Assert MEM operands are always Pmode.
24648 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
24651 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
24652 Remove redundant if.
24653 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
24654 * config/arm/aarch-common-protos.h
24655 (aarch_forward_to_shift_is_not_shifted_re): Remove.
24656 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
24658 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
24661 * multiple_target.c (create_dispatcher_calls): Set the default
24662 clone to be static, not public.
24664 2017-06-28 Richard Biener <rguenther@suse.de>
24666 * tree-vect-loop.c (vectorizable_reduction): Move special
24667 cond reduction IV var creation ...
24668 (vect_create_epilog_for_reduction): ... here. Remove induction_index
24669 parameter. Use STMT_VINFO_VECTYPE.
24670 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
24673 2017-06-28 Martin Liska <mliska@suse.cz>
24676 * ipa-visibility.c (non_local_p): Handle visibility.
24678 2017-06-28 Martin Liska <mliska@suse.cz>
24681 * common.opt: Add IntegerRange to various options.
24682 * opt-functions.awk (integer_range_info): New function.
24683 * optc-gen.awk: Add integer_range_info to cl_options struct.
24684 * opts-common.c (decode_cmdline_option): Handle
24685 CL_ERR_INT_RANGE_ARG.
24686 (cmdline_handle_error): Likewise.
24687 * opts.c (print_filtered_help): Show valid interval in
24688 when --help is provided.
24689 * opts.h (struct cl_option): Add range_min and range_max fields.
24690 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
24692 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
24694 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
24695 (x * C EQ/NE y * C): New transformation.
24697 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
24699 * genmultilib (combination_space): Accept '+' in option names.
24701 2017-06-28 Martin Liska <mliska@suse.cz>
24704 * asan.c (instrument_derefs): Bail out inner references
24705 that are hard register variables.
24707 2017-06-28 Jakub Jelinek <jakub@redhat.com>
24710 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
24711 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
24713 2017-06-28 Richard Biener <rguenther@suse.de>
24715 * tree-vectorizer.h (vect_get_vec_defs): Remove.
24716 (vect_get_slp_defs): Adjust.
24717 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
24719 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
24721 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
24722 get_initial_defs_for_reduction instead of vect_get_vec_defs.
24723 (vectorizable_reduction): Adjust.
24724 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
24726 (vect_get_slp_defs): Likewise.
24727 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
24728 (vectorizable_bswap): Adjust.
24729 (vectorizable_call): Likewise.
24730 (vectorizable_conversion): Likewise.
24731 (vectorizable_assignment): Likewise.
24732 (vectorizable_shift): Likewise.
24733 (vectorizable_operation): Likewise.
24734 (vectorizable_store): Likewise.
24735 (vectorizable_condition): Likewise.
24736 (vectorizable_comparison): Likewise.
24738 2017-06-28 Michael Collison <michael.collison@arm.com>
24741 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
24743 (arm_gen_movmemqi): Removed unused variable 'i'.
24744 Convert 'for' loop into 'while' loop.
24745 (arm_expand_prologue): Remove last unnecessary set of insn.
24746 (thumb_pop): Remove unused variable 'pushed_words'.
24747 (thumb_exit): Remove last unnecessary set of regs_to_pop.
24749 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24751 * config/s390/predicates.md: Use s390_rel_address_ok_p.
24752 * config/s390/s390-protos.h: Add prototype of
24753 s390_rel_address_ok_p.
24754 * config/s390/s390.c (s390_got_symbol): New function.
24755 (s390_rel_address_ok_p): New function.
24756 (legitimize_pic_address): Use s390_rel_address_ok_p.
24757 (s390_load_got): Use s390_got_symbol.
24758 (s390_option_override): Issue error if
24759 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
24760 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
24762 * config/s390/s390.opt: New option mpic-data-is-text-relative.
24764 2017-06-27 Andrew Pinski <apinski@cavium.com>
24766 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
24767 (X * copysign (1.0, X)): New pattern.
24768 (X * copysign (1.0, -X)): New pattern.
24769 (copysign (-1.0, CST)): New pattern.
24771 2017-06-27 Joseph Myers <joseph@codesourcery.com>
24773 * genmultilib (combination_space): Remove variable.
24774 Validate reuse rules against regular expression for any sequence
24775 of multilib options in any order.
24777 2017-06-27 Michael Collison <michael.collison@arm.com>
24779 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
24780 call aarch64_split_simd_combine.
24781 * (aarch64_combine_internal<mode>): Delete pattern.
24782 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
24783 Allow register and subreg operands.
24785 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24787 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
24788 specific need, just fallback on defaults.
24789 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
24791 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24792 Olivier Hainque <hainque@adacore.com>
24794 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
24796 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
24797 targets. Pick a default if no particular attempt applied.
24798 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
24801 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24803 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
24804 (x86_64-wrs-vxworks7): Likewise.
24806 2017-06-27 Marek Polacek <polacek@redhat.com>
24809 * ubsan.c (instrument_null): Check get_base_address's result for null.
24811 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
24813 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
24815 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
24817 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
24818 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
24819 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
24820 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
24821 New function types.
24822 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
24823 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
24824 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
24825 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
24826 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
24827 BUILT_IN_FEUPDATEENV): New builtins.
24828 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
24829 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
24830 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
24831 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
24833 (builtin_structptr_types): Adjust size.
24834 * tree.c (builtin_structptr_types): Add four entries.
24836 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24837 Olivier Hainque <hainque@adacore.com>
24839 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
24840 (TLS_SYM): New local macro, forcing reference to __tls__ on
24841 link command lines for VxWorks 7 RTPs, triggering initialization
24843 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
24844 OS features TLS support, true for RTPs on VxWorks 7.
24845 * config/vxworks.c (vxworks_override_options): Setup emutls
24848 2017-06-27 Jakub Jelinek <jakub@redhat.com>
24850 * predict.c (test_prediction_value_range): Use -1U instead of -1
24851 to avoid narrowing conversion warning.
24852 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
24853 to avoid narrowing conversion warning.
24854 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
24856 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
24858 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24860 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
24861 64bit configurations.
24862 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
24863 (SIZE_TYPE): Likewise.
24864 * config/vxworks.c (vxworks_emutls_var_fields): Use
24865 long_unsigned_type_node instead of unsigned_type_node as the offset
24866 field type, which is "pointer" mode in emutls.c.
24868 2017-06-27 Jakub Jelinek <jakub@redhat.com>
24871 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
24873 PR middle-end/81207
24874 * gimple-fold.c (replace_call_with_call_and_fold): Handle
24875 gimple_vuse copying separately from gimple_vdef copying.
24877 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24879 * value-prof.c (free_hist): Remove call to memset and the enclosing if
24882 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
24883 Olivier Hainque <hainque@adacore.com>
24885 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
24886 for all vxworks7 targets.
24887 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
24888 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
24889 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
24890 variations for VX6/VX7 and 32/64bits later on in ...
24891 (VXWORKS_LIB_SPEC): Leverage new macros.
24892 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
24893 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
24895 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
24897 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
24898 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
24900 2017-06-26 Carl Love <cel@us.ibm.com>
24902 * config/rs6000/rs6000-c.c: Add support for built-in functions
24903 vector bool char vec_reve (vector bool char);
24904 vector signed char vec_reve (vector signed char);
24905 vector unsigned char vec_reve (vector unsigned char);
24906 vector bool int vec_reve (vector bool int);
24907 vector signed int vec_reve (vector signed int);
24908 vector unsigned int vec_reve (vector unsigned int);
24909 vector bool long long vec_reve (vector bool long long);
24910 vector signed long long vec_reve (vector signed long long);
24911 vector unsigned long long vec_reve (vector unsigned long long);
24912 vector bool short vec_reve (vector bool short);
24913 vector signed short vec_reve (vector signed short);
24914 vector double vec_reve (vector double);
24915 vector float vec_reve (vector float);
24916 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
24917 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
24918 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
24919 (altivec_vreve): New pattern.
24920 * config/rs6000/altivec.h (vec_reve): New define.
24921 * doc/extend.texi (vec_rev): Update the built-in documentation file
24922 for the new built-in functions.
24924 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24926 PR tree-optimization/71815
24927 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
24929 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
24931 (slsr_process_phi): Likewise.
24932 (replace_uncond_cands_and_profitable_phis): Don't replace a
24933 multiply candidate with a stride of 1 (copy or cast).
24934 (phi_incr_cost): Call uses_consumed_by_stmt rather than
24936 (lowest_cost_path): Likewise.
24937 (total_savings): Likewise.
24939 2017-06-26 Richard Biener <rguenther@suse.de>
24942 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
24943 Use def_builtin_pure for all gather builtins.
24945 2017-06-26 Richard Biener <rguenther@suse.de>
24947 PR tree-optimization/81203
24948 * tree-tailcall.c (find_tail_calls): Do not move stmts into
24949 non-dominating BBs.
24951 2017-06-26 Marek Polacek <polacek@redhat.com>
24954 * doc/invoke.texi: Document -Wmultistatement-macros.
24956 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
24958 * doc/sourcebuild.texi (ARM-specific attributes): Document new
24959 arm_neon_ok_no_float_abi effective target.
24961 2017-06-26 Richard Biener <rguenther@suse.de>
24963 PR tree-optimization/80928
24964 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
24965 (copy_bbs): Set BB_DUPLICATED flag early.
24966 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
24968 (execute_on_shrinking_pred): Likewise.
24969 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
24970 BB_DUPLICATED blocks.
24971 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
24972 iterate over all PHIs considering removal of *gsi.
24974 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
24976 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
24979 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24981 * config/rs6000/rs6000-string.c: (expand_block_clear,
24982 do_load_for_compare, select_block_compare_mode,
24983 compute_current_alignment, expand_block_compare,
24984 expand_strncmp_align_check, expand_strn_compare,
24985 expand_block_move, rs6000_output_load_multiple)
24986 Move functions related to string/block move/compare
24987 to a separate file.
24988 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
24989 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
24990 for this function which is now used in two files.
24991 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
24992 * config.gcc: Add rs6000-string.o to extra_objs for
24993 targets powerpc*-*-* and rs6000*-*-*.
24995 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
24998 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
24999 32-bit, since indexed is not valid for DImode.
25000 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
25001 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
25002 (define_peephole2 for Altivec d-form load): Add 32-bit support.
25003 (define_peephole2 for Altivec d-form store): Likewise.
25006 * multiple_target.c (create_dispatcher_calls): Only create the
25007 dispatcher call if the function is the default clone of a
25008 versioned function.
25010 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
25012 PR middle-end/80902
25013 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
25014 a call, force the call to not be a tail call.
25016 2017-06-23 Jeff Law <law@redhat.com>
25018 * doc/contrib.texi: Add entry for Steven Pemberton's work on
25021 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
25023 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
25024 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
25025 handling for early expansion of vector shifts (sl,sr,sra,rl).
25026 (builtin_function_type): Add vector shift right instructions
25027 to the unsigned argument list.
25029 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
25031 rtl-optimizatoin/79286
25032 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
25033 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
25034 trap. PIC register plus a const unspec without offset can never trap.
25036 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
25038 * tree.h (builtin_structptr_type): New type.
25039 (builtin_structptr_types): Declare new array.
25040 * tree.c (builtin_structptr_types): New array.
25041 (free_lang_data, build_common_tree_nodes): Use it.
25043 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
25046 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
25049 2017-06-22 Matt Turner <mattst88@gmail.com>
25051 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
25052 Lake models to skylake case. Assume skylake for unknown
25053 models with clflushopt.
25055 2017-06-22 Jeff Law <law@redhat.com>
25057 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
25058 frame sizes that do not satisfy aarch64_uimm12_shift.
25060 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
25062 * profile-count.h (apply_probability,
25063 apply_scale, probability_in): Fix checks for zero.
25065 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25067 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
25068 * doc/cppdiropts.texi (-I @var{dir}): Document it.
25070 2016-06-22 Richard Biener <rguenther@suse.de>
25072 * tree-vect-loop.c (vect_model_reduction_cost): Handle
25073 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
25074 REDUC_MAX_EXPR support.
25075 (vectorizable_reduction): Likewise.
25076 (vect_create_epilog_for_reduction): Likewise.
25078 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
25080 * match.pd (A / (1 << B) -> A >> B): New.
25081 * generic-match-head.c: Include optabs-tree.h.
25082 * gimple-match-head.c: Likewise.
25083 * optabs-tree.h (target_supports_op_p): New.
25084 * optabs-tree.c (target_supports_op_p): New.
25086 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25088 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
25089 $gcc_cv_ld --help output.
25090 (gcc_cv_ld_demangle): Likewise.
25091 (gcc_cv_ld_eh_frame_hdr): Likewise.
25092 (gcc_cv_ld_pie): Likewise.
25093 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
25094 (gcc_cv_ld_buildid): Likewise.
25095 (gcc_cv_ld_sysroot): Likewise.
25096 (ld_bndplt_support): Likewise.
25097 (ld_pushpopstate_support): Likewise.
25098 * configure: Regenerate.
25099 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
25101 2017-06-21 Jakub Jelinek <jakub@redhat.com>
25104 * config/i386/sse.md (round<mode>2): Renumber match_dup and
25105 operands indexes to avoid gap between operands and match_dups.
25107 2017-06-21 Andrew Pinski <apinski@cavium.com>
25109 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
25110 Increment Arith_shift and Arith_shift_reg by 1.
25111 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
25113 * config/aarch64/aarch64.c (thunderx_tunings): Enable
25114 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
25115 (aarch64_strip_extend): Add new argument and test for it.
25116 (aarch64_cheap_mult_shift_p): New function.
25117 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
25118 add a cost if it is true.
25119 Update calls to aarch64_strip_extend.
25120 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
25122 2017-06-21 Andrew Pinski <apinski@cavium.com>
25124 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
25126 (thunderxt88): Likewise.
25127 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
25128 (thunderx_prefetch_tune): New variable.
25129 (thunderx2t99_prefetch_tune): Update for the correct values.
25130 (thunderxt88_tunings): New variable.
25131 (thunderx_tunings): Use thunderx_prefetch_tune instead of
25132 generic_prefetch_tune.
25133 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
25135 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25137 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
25138 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
25139 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
25140 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
25141 (aarch64_atomic_cas<mode>, GPI): Likewise.
25143 2017-06-21 Martin Liska <mliska@suse.cz>
25145 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
25146 statements on cold and hot labels.
25147 * predict.c (tree_estimate_probability_bb): Remove the
25148 prediction from this place.
25150 2017-06-21 Martin Liska <mliska@suse.cz>
25152 PR tree-optimization/79489
25153 * gimplify.c (maybe_add_early_return_predict_stmt): New
25155 (gimplify_return_expr): Call the function.
25156 * predict.c (tree_estimate_probability_bb): Remove handling
25158 * predict.def: Update comment about early return predictor.
25159 * gimple-predict.h (is_gimple_predict): New function.
25160 * predict.def: Change default value of early return to 66.
25161 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
25163 * passes.def: Put pass_strip_predict_hints to the beginning of
25166 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
25168 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
25169 FUNCTION_DECL declarations.
25170 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
25172 (dwaf2out_decl): Likewise.
25173 * godump.c (go_early_global_decl): Skip call to the real debug hook
25174 for FUNCTION_DECL declarations.
25175 * passes.c (rest_of_decl_compilation): Skip call to the
25176 early_global_decl debug hook for FUNCTION_DECL declarations, unless
25177 -fdump-go-spec is passed.
25179 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
25181 * config/i386/i386.c (struct builtin_isa): New field pure_p.
25182 Reorder for compactness.
25183 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
25184 (def_builtin_pure, def_builtin_pure2): New functions.
25185 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
25187 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
25189 * match.pd (nop_convert): New predicate.
25190 ((A +- CST1) +- CST2): Allow some NOP conversions.
25192 2017-06-21 Jakub Jelinek <jakub@redhat.com>
25195 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
25196 with ctors/dtors if GOVD_SHARED is set.
25198 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25200 * config/aarch64/aarch64.md (movti_aarch64):
25201 Emit mov rather than orr.
25202 (movtf_aarch64): Likewise.
25203 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
25204 Emit mov rather than orr.
25206 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25208 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
25209 Swap alternatives, make integer dup more expensive.
25211 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25213 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
25214 Return true for non-tls symbols.
25216 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
25218 * config/aarch64/aarch64-cores.def (cortex-a55): New.
25219 (cortex-a75): Likewise.
25220 (cortex-a75.cortex-a55): Likewise.
25221 * config/aarch64/aarch64-tune.md: Regenerate.
25222 * doc/invoke.texi (-mtune): Document new values for -mtune.
25224 2017-06-21 Tom de Vries <tom@codesourcery.com>
25226 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
25227 stack_size feature.
25228 (Effective-Target Keywords, Other attributes): Suggest using
25229 dg-add-options stack_size feature to get stack limit in stack_size
25230 effective target documentation.
25232 2017-06-21 Julian Brown <julian@codesourcery.com>
25233 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25235 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
25236 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
25237 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
25239 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
25240 attribute type list for neon_multiply.
25241 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
25242 attribute type list for neon_multiply.
25243 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
25244 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
25245 attribute type list for neon_multiply.
25246 * config/arm/types.md (crypto_pmull): Add.
25247 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
25248 attribute type list.
25250 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
25252 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
25255 2017-06-20 Jakub Jelinek <jakub@redhat.com>
25257 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
25258 to make sure not to dereference a NULL cost_classes_ptr pointer.
25260 2017-06-20 Carl Love <cel@us.ibm.com>
25262 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25263 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
25264 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
25265 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
25266 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
25267 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
25268 VMULOSW): New enum "unspec" values.
25269 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
25270 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
25271 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
25272 altivec_vmulosw): New patterns.
25273 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
25274 VMULOSW): Add definitions.
25276 2017-06-20 Julia Koval <julia.koval@intel.com>
25278 * config/i386/i386.c: Fix rounding expand for new pattern.
25279 * config/i386/subst.md: Fix pattern (parallel -> unspec).
25281 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25283 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
25284 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
25286 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25288 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
25291 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25293 * config/aarch64/aarch64-cores.def: Rearrange to sort by
25294 architecture, then by implementer ID.
25295 * config/aarch64/aarch64-tune.md: Regenerate.
25297 2017-06-20 Richard Biener <rguenther@suse.de>
25299 PR middle-end/81097
25300 * fold-const.c (split_tree): Fold to type before negating.
25302 2017-06-20 David Malcolm <dmalcolm@redhat.com>
25304 * diagnostic-show-locus.c
25305 (selftest::test_fixit_deletion_affecting_newline): New function.
25306 (selftest::diagnostic_show_locus_c_tests): Call it.
25308 2017-06-20 Andreas Schwab <schwab@suse.de>
25311 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
25314 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25316 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
25317 __ARM_FEATURE_COPROC according to support.
25319 2017-06-20 Jakub Jelinek <jakub@redhat.com>
25321 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25322 Rewritten to avoid overflow for > 32-bit pointers.
25325 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
25326 by removing enum keyword.
25327 (ubsan_type_descriptor): Likewise. Formatting fix.
25330 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
25331 splitter): Require TARGET_SSE2 in the condition.
25333 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
25336 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
25337 for doing vector set of SFmode on ISA 3.0.
25338 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
25339 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
25341 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
25342 SFmode value into a V4SF variable that was extracted from another
25343 V4SF variable without converting the element to double precision
25344 and back to single precision vector format.
25345 (vsx_insert_extract_v4sf_p9_2): Likewise.
25347 2017-06-19 Jakub Jelinek <jakub@redhat.com>
25349 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
25350 in UWHI to avoid undefined overflow.
25353 * ubsan.h (enum ubsan_encode_value_phase): New.
25354 (ubsan_encode_value): Change second argument to
25355 enum ubsan_encode_value_phase with default value of
25356 UBSAN_ENCODE_VALUE_GENERIC.
25357 * ubsan.c (ubsan_encode_value): Change second argument to
25358 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
25359 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
25360 create_tmp_var_raw instead of create_tmp_var and use a
25362 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
25363 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
25364 ubsan_encode_value callers.
25367 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
25368 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
25369 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
25371 2017-06-19 Richard Biener <rguenther@suse.de>
25373 PR middle-end/81118
25374 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
25375 estimates if we changed anything.
25377 2017-06-19 Richard Biener <rguenther@suse.de>
25379 PR tree-optimization/80887
25380 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
25381 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
25382 simplified lookups, then reset mprts_hook.
25383 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
25385 (try_to_simplify): Likewise.
25387 2017-06-19 Martin Liska <mliska@suse.cz>
25390 * gimplify.c (gimplify_switch_expr):
25391 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
25393 2017-06-19 Martin Liska <mliska@suse.cz>
25395 * doc/install.texi: Document that PGO runs in 4 stages.
25397 2017-06-19 Martin Liska <mliska@suse.cz>
25400 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
25401 to dispatcher function name.
25402 * multiple_target.c (replace_function_decl): New function.
25403 (create_dispatcher_calls): Redirect both edges and references.
25405 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
25407 * profile-count.c (profile_count::dump): Dump quality.
25408 (profile_count::differs_from_p): Update for unsigned val.
25409 * profile-count.h (profile_count_quality): New enum.
25410 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
25412 2017-06-19 Richard Biener <rguenther@suse.de>
25414 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
25415 struct function as arg.
25416 (estimate_numbers_of_iterations): Export overload with loop arg.
25417 (free_numbers_of_iterations_estimates_loop): Use an overload of
25418 free_numbers_of_iterations_estimates instead.
25419 * tree-cfg.c (remove_bb): Adjust.
25420 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
25421 * tree-parloops.c (gen_parallel_loop): Likewise.
25422 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
25424 (tree_unroll_loops_completely): Likewise.
25425 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
25426 Use an overload instead and export.
25427 (estimated_loop_iterations): Adjust.
25428 (max_loop_iterations): Likewise.
25429 (likely_max_loop_iterations): Likewise.
25430 (estimate_numbers_of_iterations): Take struct function as arg
25432 (loop_exits_before_overflow): Adjust.
25433 (free_numbers_of_iterations_estimates_loop): Use an overload.
25434 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
25435 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
25437 2017-06-19 Richard Biener <rguenther@suse.de>
25440 * ipa-prop.c (find_constructor_constant_at_offset): Handle
25441 RANGE_EXPR conservatively.
25443 2017-06-16 Carl Love <cel@us.ibm.com>
25445 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25446 definitions for vec_float, vec_float2, vec_floato,
25447 vec_floate built-ins.
25448 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
25449 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
25451 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
25452 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
25453 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
25454 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
25455 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
25456 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
25457 vec_floato): Add builtin defines.
25458 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
25459 Update the built-in documentation file for the new built-in
25462 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25464 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
25465 (mthumb): Mark as the negative of -marm.
25467 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25469 * doc/invoke.texi (ARM Options, -mcpu): Document supported
25471 (ARM Options, -mtune): Document that this accepts the same
25472 extension options as -mcpu.
25473 (ARM Options, -mfpu): Document addition of -mfpu=auto.
25475 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25477 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
25478 permitted extensions.
25480 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25482 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
25483 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
25484 (armv8-m.main): Add option +nodsp.
25485 * config/arm/arm-cpu-cdata.h: Regenerated.
25487 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25489 * config/arm/t-fuchsia: New file.
25490 * config.gcc (arm*-*-fuchsia*): Use it.
25492 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25494 * config/arm/t-symbian: Rewrite for new option infrastructure.
25496 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25498 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
25499 (MULTILIB_REQUIRED): Likewise.
25501 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25503 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
25504 (MULTILIB_RESUE): Likewise.
25505 (MULTILIB_MATCHES): Likewise.
25506 (MULTLIB_REQUIRED): Likewise.
25508 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25510 * config/arm/t-rtems: Rewrite for new option framework.
25512 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25514 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
25515 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
25516 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
25517 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
25518 * config/arm/t-multilib: ... here.
25519 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
25520 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
25521 armv7-a and armv8*-a when A-profile libraries have not been built.
25522 * config/arm/t-rmprofile: Rewrite.
25524 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25526 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
25527 with a backslash. Remove the backslash after substituting unescaped
25529 * doc/fragments.texi (MULTILIB_REUSE): Document it.
25531 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25533 * config.gcc: (arm*-*-*): When building a-profile libraries, force
25534 the driver to pass through the default setting of -mfloat-abi.
25535 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
25537 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
25538 (all_feat_combs): New rule.
25539 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
25541 * config/arm/t-aprofile: Rewrite.
25543 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25545 * config/arm/arm.h (FPUTYPE_AUTO): Define.
25546 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
25547 fpu is not specified by the user/command-line.
25548 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
25549 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
25550 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
25551 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
25552 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
25553 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
25555 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25557 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
25558 * config/arm/t-arm-elf: Rewritten.
25560 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25562 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
25563 have some floating-point instructions.
25564 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
25565 (TARGET_MAYBE_HARD_FLOAT): New macro.
25566 * config/arm/arm-builtins.c (arm_init_builtins): Use
25567 TARGET_MAYBE_HARD_FLOAT.
25568 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
25570 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25572 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
25573 (configargs.h): Include it.
25574 (arm_print_hint_for_fpu_option): New function.
25575 (arm_parse_fpu_option): New function.
25576 (candidate_extension): New class.
25577 (arm_canon_for_multilib): New function.
25578 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
25579 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
25580 (ARCH_CANONICAL_SPECS): New macro.
25581 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
25583 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25585 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
25586 are set after handling multilib fragments. Set target_cpu_default2
25589 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25591 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
25593 (arm*-*-*): Set target_cpu_default2 to a quoted string.
25594 * config/arm/parsecpu.awk (check_cpu): Validate any extension
25596 (check_arch): Likewise.
25597 * config/arm/arm.c (arm_configure_build_target): Handle
25598 TARGET_CPU_DEFAULT being a string constant. Scan any feature
25599 options in the default.
25601 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25603 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
25604 when an option is an alias of another.
25605 * config/arm/parsecpu.awk (optalias): New parser token.
25606 (gen_comm_data): Mark non-alias options as such. Emit entries
25607 for extension aliases.
25608 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
25609 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
25610 (armv6kz, armv6zk, armv6t2): Likewise.
25611 (armv7): Make vfpv3-d16 an alias.
25612 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
25614 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
25615 Sort in canonical order.
25616 (armv8-a): Sort in canonical order.
25617 (armv8.1-a, armv8.2-a): Likewise.
25618 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
25620 (cortex-a9): Sort in canonical order.
25621 * config/arm/arm.c (selftests.h): Include it.
25622 (arm_test_cpu_arch_data): New function.
25623 (arm_run_self_tests): New function.
25624 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
25625 (targetm): Move declaration to the end of the file.
25626 * arm-cpu-cdata.h: Regenerated.
25628 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25630 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
25631 call to target_mode_check describing the type of option passed.
25632 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
25633 (arm_target_thumb_only): Use arm_parse_arch_option_name or
25634 arm_parse_cpu_option_name to match parameters against list of
25636 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
25637 arm_arch_core_flags data structure.
25638 * config/arm/arm-cpu_cdata.h: Regenerated.
25640 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25642 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
25644 (arm_print_hint_for_cpu_option): Likewise.
25645 (arm_print_hint_for_arch_option): Likewise.
25646 (arm_parse_cpu_option_name): Likewise.
25647 (arm_parse_arch_option_name): Likewise.
25648 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
25649 of entries in the all_fpus list.
25650 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
25651 (arm_parse_cpu_option_name): Declare.
25652 (arm_parse_arch_option_name): Declare.
25653 (arm_parse_option_features): Declare.
25654 (arm_intialize_isa): Declare.
25655 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
25657 (gen_comm_data): ... here. Make definitions non-static.
25658 * config/arm/arm-cpu-data.h: Regenerated.
25659 * config/arm/arm-cpu-cdata.h: Regenerated.
25661 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25663 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
25664 (cpu_arch_extension): New structure.
25665 (cpu_arch_option, arch_option, cpu_option): New structures.
25666 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
25667 architecture types.
25668 (gen_data): Generate new format data tables.
25669 * config/arm/arm.c (cpu_tune): New structure.
25670 (cpu_option, processors): Delete.
25671 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
25672 (arm_print_hint_for_cpu_option): ... this and ...
25673 (arm_print_hint_for_arch_option): ... this.
25674 (arm_parse_arch_cpu_name): Delete. Replace with ...
25675 (arm_parse_cpu_option_name): ... this and ...
25676 (arm_parse_arch_option_name): ... this.
25677 (arm_unrecognized_feature): Change type of target parameter to
25679 (arm_parse_arch_cpu_features): Delete. Replace with ...
25680 (arm_parse_option_features): ... this.
25681 (arm_configure_build_target): Rework to use new configuration data
25683 (arm_print_tune_info): Rework for new configuration data tables.
25684 * config/arm/arm-cpu-data.h: Regenerated.
25685 * config/arm/arm-cpu.h: Regenerated.
25687 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25689 * Makefile.in (OBJS): Move sbitmap.o from here ...
25690 (OBJS-libcommon): ... to here.
25692 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25694 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
25695 (ISA_ALL_CRYPTO): New macro.
25696 (ISA_ALL_SIMD): New macro
25697 (ISA_ALL_FP): New macro.
25698 * config/arm/arm.c (fpu_bitlist): Update initializer.
25699 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
25701 (arm9e): Add fpu. Add option for nofp
25702 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
25703 (arm926ej-s, arm1026ej-s): Likewise.
25704 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
25705 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
25706 neon-fp16, neon-vfpv4, nofp and nosimd.
25707 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
25708 (cortex-a8): Add fpu. Add option for nofp.
25709 (cortex-a9): Add fpu. Add options for nosimd and nofp.
25710 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
25711 (cortex-r4f): Add fpu.
25712 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
25713 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
25715 (cortex-r8): Likewise.
25716 (cortex-m4): Add fpu. Add option for nofp.
25717 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
25718 (cortex-a17.cortex-a7): Likewise.
25719 (cortex-a32): Add fpu. Add options for crypto and nofp.
25720 (cortex-a35, cortex-a53): Likewise.
25721 (cortex-a57): Add fpu. Add option for crypto.
25722 (cortex-a72, cortex-a73): Likewise.
25723 (exynos-m1): Likewise.
25724 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
25725 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
25726 (cortex-m33): Add fpu. Add option for nofp.
25727 * config/arm/arm-cpu-cdata.h: Regenerated
25728 * config/arm/arm-cpu-data.h: Regenerated.
25730 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25732 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
25733 (armv5te, armv5tej): Likewise.
25734 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
25735 (armv7): Add options fp and vfpv3-d16.
25736 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
25737 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
25739 (armv7ve): Likewise.
25740 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
25741 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
25742 (armv8-a): Add nocrypto option.
25743 (armv8.1-a, armv8.2-a): Likewise.
25744 (armv8-m.main): add options fp, fp.dp and nofp.
25746 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25748 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
25750 (armv8-a+crc): Delete.
25751 (armv8.1-a): Add options simd, crypto and nofp.
25752 (armv8.2-a): Add options fp16, simd, crypto and nofp.
25753 (armv8.2-a+fp16): Delete.
25754 (armv8-m.main): Add option dsp.
25755 (armv8-m.main+dsp): Delete.
25756 (cortex-a8): Add fpu. Add nofp option.
25757 (cortex-a9): Add fpu. Add nofp and nosimd options.
25758 * config/arm/parsecpu.awk (gen_data): Generate option tables and
25759 link to main cpu and architecture data structures.
25760 (gen_comm_data): Only put isa attributes from the main architecture
25762 (option): New statement for architecture and CPU entries.
25763 * arm.c (struct cpu_option): New structure.
25764 (struct processors): Add entry for options.
25765 (arm_unrecognized_feature): New function.
25766 (arm_parse_arch_cpu_name): Ignore any characters after the first
25768 (arm_parse_arch_cpu_feature): New function.
25769 (arm_configure_build_target): Separate out any CPU and architecture
25770 features and parse separately. Don't error out if -mfpu=auto is
25771 used with only an architecture string.
25772 (arm_print_asm_arch_directives): New function.
25773 (arm_file_start): Call it.
25774 * config/arm/arm-cpu-cdata.h: Regenerated.
25775 * config/arm/arm-cpu-data.h: Likewise.
25776 * config/arm/arm-tables.opt: Likewise.
25778 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25780 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
25781 assembler when it is not -mfpu=auto.
25783 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25785 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
25786 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
25787 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
25788 (ASM_CPU_SPEC): Rewrite.
25789 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
25790 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
25791 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
25792 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
25793 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
25794 copied string is NUL-terminated. Also strip any characters prefixed
25796 (arm_rewrite_selected_arch): New function.
25797 (arm_rewrite_march): New function.
25799 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25801 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
25802 (x_arm_cpu_string, x_arm_tune_string): Likewise.
25803 (march, mcpu, mtune): Convert to string-based options.
25804 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
25805 (arm_parse_arch_cpu_name): New function.
25806 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
25807 identify selected architecture or CPU.
25808 (arm_option_save): New function.
25809 (TARGET_OPTION_SAVE): Redefine.
25810 (arm_option_restore): Restore string options.
25811 (arm_option_print): Print string options.
25813 2017-06-16 Martin Sebor <msebor@redhat.com>
25815 PR tree-optimization/80933
25816 PR tree-optimization/80934
25817 * builtins.c (fold_builtin_3): Do not handle bcmp here.
25818 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
25819 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
25820 (gimple_fold_builtin): Call them.
25822 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25824 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
25825 as unlikely; update profile.
25827 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25829 * predict.c (force_edge_cold): Handle declaring edges impossible
25832 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25834 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
25836 (try_unroll_loop_completely): Fix reporting.
25838 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25840 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
25842 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
25845 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
25846 if given a non-constant argument for an intrinsic which requires a
25849 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25851 * profile.c (compare_freqs): New function.
25852 (branch_prob): Sort edge list.
25853 (find_spanning_tree): Assume that the list is priority sorted.
25855 2017-06-16 Richard Biener <rguenther@suse.de>
25857 PR tree-optimization/81090
25858 * passes.def (pass_record_bounds): Remove.
25859 * tree-pass.h (make_pass_record_bounds): Likewise.
25860 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
25861 make_pass_record_bounds): Likewise.
25862 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
25863 not free niter estimates at the beginning but at the end.
25864 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
25866 2017-06-16 Richard Biener <rguenther@suse.de>
25868 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
25869 initializer to workaround ICE in host GCC 4.8.
25871 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25873 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
25875 (clone_inlined_nodes): Update.
25877 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25879 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
25880 prefetch settings, and enable prefetching by default at -O3.
25882 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25884 * config/aarch64/aarch64.c (aarch64_override_options_internal):
25885 Set flag_prefetch_loop_arrays according to tuning data.
25887 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25889 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
25890 New tune structure.
25891 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
25892 [Unrelated to main purpose of the patch] Place the pointer field last
25893 to enable type checking errors when tune structure are wrongly merged.
25894 * config/aarch64/aarch64.c (generic_prefetch_tune,)
25895 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
25896 (thunderx2t99_prefetch_tune): New tune constants.
25897 (tune_params *_tunings): Update all tunings (no functional change).
25898 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
25899 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
25900 from tunings structures.
25902 2017-06-16 Jakub Jelinek <jakub@redhat.com>
25905 * ubsan.c (instrument_null): Add T argument, use it instead
25906 of computing it based on IS_LHS.
25907 (instrument_object_size): Likewise.
25908 (pass_ubsan::execute): Adjust instrument_null and
25909 instrument_object_size callers to pass gimple_get_lhs or
25910 gimple_assign_rhs1 result to it. Use instrument_null instead of
25911 calling get_base_address and instrument_mem_ref. Handle
25912 aggregate call arguments for object-size sanitization.
25914 2017-06-16 Yury Gribov <tetra2005@gmail.com>
25916 PR tree-optimization/81089
25917 * tree-vrp.c (is_masked_range_test): Validate operands of
25920 2017-06-15 Martin Sebor <msebor@redhat.com>
25923 * dumpfile.c (dump_register): Avoid calling memset to initialize
25924 a class with a default ctor.
25925 * gcc.c (struct compiler): Remove const qualification.
25926 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
25927 * hash-table.h: Ditto.
25928 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
25930 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
25931 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
25933 * params.h (struct param_info): Make struct members non-const.
25934 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
25935 with default initialization.
25936 * vec.h (vec_copy_construct, vec_default_construct): New helper
25938 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
25939 with vec_copy_construct.
25940 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
25941 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
25942 * doc/invoke.texi (-Wclass-memaccess): Document.
25944 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25946 * emit-rtl.h (is_leaf): Update comment about local
25947 register allocator.
25949 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
25952 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
25953 for a variable to have a section before checking if the section has a
25955 Set section to.persistent if persistent attribute is set.
25956 Warn if .persistent attribute is used on an automatic variable.
25958 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
25960 PR rtl-optimization/80474
25961 * reorg.c (update_block): Do not ignore instructions in a delay slot.
25963 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
25965 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
25968 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
25970 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
25971 (casesi): Emit bounds checking as RTL.
25972 (casesi_internal_mips16_<mode>): Remove bounds checking.
25974 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
25976 * config/xtensa/xtensa.c (xtensa_option_override): Append
25977 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
25978 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
25979 xtensa_doloop_hooks): Define unconditionally.
25980 (xtensa_reorg_loops): Only call reorg_loops in the presence of
25982 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
25983 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
25984 for it in xtensa_option_override.
25985 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
25986 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
25988 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
25990 * doc/cppopts.texi: Document '-' special value to -MF.
25992 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
25994 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
25995 (cortex_a53_fconst): Likewise.
25996 (cortex_a53_fpmul): Likewise.
25997 (cortex_a53_f_load_64): Likewise.
25998 (cortex_a53_f_load_many): Likewise.
25999 (cortex_a53_advsimd_alu): Likewise.
26000 (cortex_a53_advsimd_alu_q): Likewise.
26001 (cortex_a53_advsimd_mul): Likewise.
26002 (cortex_a53_advsimd_mul_q): Likewise.
26003 (fpmac bypass): Add new bypass for fpmac-fpmac case.
26004 Add missing fmul, r2f_cvt and fconst cases.
26006 2017-06-14 Richard Biener <rguenther@suse.de>
26008 PR middle-end/81088
26009 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
26011 (fold_binary_loc): When associating do not treat pre-existing
26012 TREE_OVERFLOW on literal constants as a reason to allow
26013 TREE_OVERFLOW on associated literal constants.
26015 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26017 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
26018 (MASK_FEATURES): New macro.
26019 * config/sparc/sparc.c (sparc_option_override): Remove the special
26020 handling of -mfpu and generalize it to all MASK_FEATURES switches.
26022 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26024 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
26025 a division of 0 if non-call exceptions are enabled.
26027 2017-06-14 Andrew Pinski <apinski@cavium.com>
26028 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26031 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
26032 Improve vector initialization code gen for only variable case.
26034 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26036 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
26038 2017-06-14 Richard Biener <rguenther@suse.de>
26040 PR tree-optimization/81083
26041 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
26044 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26046 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
26047 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
26048 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
26049 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
26050 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
26051 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
26053 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26055 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
26056 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
26058 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26060 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
26062 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26064 * config/rs6000/t-rtems: Don't handle SPE.
26066 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26068 * config/rs6000/t-linux: Don't handle SPE.
26070 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26072 * config/rs6000/eabispe.h: Delete file.
26074 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26076 * config/rs6000/t-spe: Delete file.
26078 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26080 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
26081 (rs6000_legitimate_offset_address_p): Return false for anything in
26082 V2SImode or V2SFmode.
26084 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26086 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
26087 except V2SF and V2SI. Rearrange the vector modes, and add comments.
26088 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
26090 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
26091 (rs6000_legitimate_offset_address_p): Ditto.
26092 (rs6000_emit_move): Ditto.
26093 (rs6000_init_builtins): Remove V4HI_type_node.
26095 2017-06-13 Martin Liska <mliska@suse.cz>
26098 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
26099 (gate_asan): Likewise.
26100 * asan.h (asan_no_sanitize_address_p): Remove the function.
26101 (sanitize_flags_p): New function.
26102 * builtins.def: Fix coding style.
26103 * common.opt: Use renamed enum value.
26104 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
26105 * doc/extend.texi: Document no_sanitize attribute.
26106 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
26107 to SANITIZE_UNDEFINED_NONDEFAULT.
26108 * gcc.c (sanitize_spec_function): Use the renamed enum value.
26109 * gimple-fold.c (optimize_atomic_compare_exchange_p):
26110 Use sanitize_flags_p.
26111 * gimplify.c (gimplify_function_tree): Likewise.
26112 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
26113 * opts.c (parse_no_sanitize_attribute): New function.
26114 (common_handle_option): Use renamed enum value.
26115 * opts.h (parse_no_sanitize_attribute): Declare.
26116 * tree.c (sanitize_flags_p): New function.
26117 * tree.h: Declared here.
26118 * tsan.c: Use sanitize_flags_p.
26119 * ubsan.c (ubsan_expand_null_ifn): Likewise.
26120 (instrument_mem_ref): Likewise.
26121 (instrument_bool_enum_load): Likewise.
26122 (do_ubsan_in_current_function): Remove the function.
26123 (pass_ubsan::execute): Use sanitize_flags_p.
26124 * ubsan.h: Remove do_ubsan_in_current_function
26125 * tree-cfg.c (print_no_sanitize_attr_value): New function.
26126 (dump_function_to_file): Use it here.
26128 2017-06-13 Martin Jambor <mjambor@suse.cz>
26130 PR tree-optimization/80803
26131 PR tree-optimization/81063
26132 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
26133 (propagate_subaccesses_across_link): Enqueue subtree whenever
26134 necessary instead of relying on the caller.
26136 2017-06-13 Martin Jambor <mjambor@suse.cz>
26138 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
26139 that have a first_link.
26140 (sort_and_splice_var_accesses): Do not check first_link before
26142 (subtree_mark_written_and_enqueue): Likewise.
26143 (propagate_all_subaccesses): Likewise and do not stop at first
26144 parent with a first_link.
26146 2017-06-13 Martin Jambor <mjambor@suse.cz>
26148 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
26151 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26153 * match.pd: New pattern.
26155 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26157 * tree-vrp.c (is_masked_range_test): New function.
26158 (register_edge_assert_for): Determine ranges for
26161 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26163 PR tree-optimization/67328
26164 * fold-const.c (maskable_range_p): New function.
26165 (build_range_check): Generate bittests if possible.
26167 2017-06-13 Martin Liska <mliska@suse.cz>
26169 * gimple-pretty-print.c (dump_probability): Add new argument.
26170 (dump_edge_probability): Dump both probability and count.
26171 (dump_gimple_label): Likewise.
26172 (dump_gimple_bb_header): Likewise.
26174 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
26177 * config/avr/avr-devices.c: Fix indentation.
26178 * config/avr/gen-avr-mmcu-specs.c: Dito.
26180 2017-06-13 Richard Biener <rguenther@suse.de>
26182 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
26183 instead get vector type from stmt_info.
26184 (vectorizable_reduction): Adjust. Remove dead code.
26186 2017-06-13 Richard Biener <rguenther@suse.de>
26188 PR middle-end/81065
26189 * fold-const.c (extract_muldiv_1): Remove bogus distribution
26190 case of C * (x * C2 + C3).
26191 (fold_addr_of_array_ref_difference): Properly fold index difference.
26193 2017-06-12 David S. Miller <davem@davemloft.net>
26196 * config/sparc/sparc.md (return expander): Emit frame blockage if
26197 function uses alloca.
26199 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
26201 * combine.c (make_field_assignment): Check len rather than the mode
26202 precision when calling force_to_mode.
26204 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
26206 Support multilibs and devices that see flash in RAM address range.
26209 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
26210 (avr_mcu_t) <flash_pm_offset>: New field.
26211 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
26212 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
26213 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
26214 (AVR_TINY_PM_OFFSET): Remove macro.
26215 * config/avr/avr.opt (-mshort-calls): New option.
26216 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
26217 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
26218 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
26219 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
26220 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
26221 instead of avr_arch->have_jmp_call.
26222 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
26223 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
26224 avr_arch->flash_pm_offset to define.
26225 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
26226 new field flash_pm_offset. Add entry for avrxmega3.
26227 (avr_texinfo): Add entry for avrxmega3.
26228 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
26229 attiny212, attiny214,
26230 attiny412, attiny414, attiny416, attiny417,
26231 attiny814, attiny816, attiny817,
26232 attiny1614, attiny1616, attiny1617,
26233 attiny3214, attiny3216, attiny3217.
26234 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
26235 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
26236 (avr_print_operand_address) [AVR_TINY]: Same.
26237 (avr_asm_init_sections) <readonly_data_section>: Only patch
26238 callback if avr_arch->flash_pm_offset = 0.
26239 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
26240 for rodata if avr_arch->flash_pm_offset != 0.
26241 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
26242 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
26243 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
26244 (m_options): Append opt_rcall.
26245 (m_dirnames): Append dir_rcall.
26246 * config/avr/t-multilib: Regenerate.
26248 * configure.ac [target=avr]: Check whether avrxmega3 default
26249 linker description file works as needed.
26250 * configure: Regenerate.
26251 * doc/avr-mmcu.texi: Regenerate.
26252 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
26253 <__AVR_ARCH__>: Document avrxmega3 and 103.
26254 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
26255 <__AVR_SHORT_CALLS__>: Document it.
26256 <__AVR_PM_BASE_ADDRESS__>: Document it.
26257 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
26258 (AVR Variable Attributes) <progmem>: Document this is
26259 not needed for avrxmega3.
26260 (AVR Named Address Spaces) <__flash>: Dito.
26262 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
26264 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
26266 2017-06-12 Doug Rupp <rupp@adacore.com>
26268 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
26269 Append vxworks-stdint.h to the tm_file list.
26270 * config/vxworks-stdint.h: New file.
26272 2017-06-12 Martin Liska <mliska@suse.cz>
26274 PR tree-optimization/81041
26275 * tree-profile.c (gimple_gen_ic_func_profiler):
26276 Create an extra BB in profile-generate
26277 (gimple_gen_time_profiler): Likewise.
26279 2017-06-12 Jakub Jelinek <jakub@redhat.com>
26281 PR tree-optimization/81003
26282 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
26283 (update_range_test): Use it instead of force_gimple_operand_gsi.
26285 2017-06-12 Richard Biener <rguenther@suse.de>
26287 PR tree-optimization/81053
26288 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
26289 with backedge value not defined in loop. Simplify def stmt
26292 2017-06-11 Tom de Vries <tom@codesourcery.com>
26295 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
26297 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
26298 nvptx_cannot_force_const_mem.
26300 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26302 * opts.c (finish_options): Move test for flag_split_stack after
26303 it has been initialized.
26305 2017-06-11 Jason Merrill <jason@redhat.com>
26307 * tree.h (id_equal): New.
26308 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
26309 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
26310 instead of strcmp of IDENTIFIER_POINTER.
26312 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26314 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
26315 (mark_all_inlined_calls_cdtor): Fix formating.
26316 (inline_transform): Rescale profile before inlining.
26318 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26320 * cgraph.h (cgraph_edge::clone): Update prototype.
26321 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
26322 (cgraph_node::create_clone): Update.
26323 (cgraph_node::create_version_clone): Update.
26324 * tree-inline.c (copy_bb): Update.
26325 (expand_call_inline): Update.
26327 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
26329 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
26330 factored out from ...
26331 (rs6000_emit_prologue): ... here.
26333 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
26335 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
26336 factored out from ...
26337 (rs6000_emit_prologue): ... here.
26339 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26341 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
26343 (handle_missing_profiles): Fix computation of tp_first_run.
26344 (counts_to_freqs): Do not touch freqs when count is 0.
26346 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26348 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
26351 2017-06-10 Tom de Vries <tom@codesourcery.com>
26353 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
26354 attributes): Document signal effective target.
26356 2017-06-10 Tom de Vries <tom@codesourcery.com>
26358 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
26359 Document effective target stack_size.
26361 2017-06-09 David Malcolm <dmalcolm@redhat.com>
26363 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
26364 to the edit_context if they can be auto-applied.
26366 2017-06-9 Ian Lance Taylor <iant@golang.org>
26368 * opts.c (finish_options): If -fsplit-stack, disable implicit
26369 -forder-blocks-and-partition.
26370 * doc/invoke.texi (Optimize Options): Document that when using
26371 -fsplit-stack -forder-blocks-and-partition is not implicitly
26374 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26376 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
26377 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
26378 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
26379 * builtins.def (abort, trap, unreachable): Declare cold.
26380 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
26381 * tree-core.h (ECF_COLD): New.
26382 * tree.c (set_call_expr_flags): Handle ECF_COLD.
26383 (build_common_builtin_nodes): Mark unreachable and abort as cold.
26385 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26387 * predict.c (unlikely_executed_stmt_p): Cleanup.
26389 2017-06-09 Richard Biener <rguenther@suse.de>
26391 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
26392 model if the ref is always written to.
26394 2017-06-09 Tamar Christina <tamar.christina@arm.com>
26396 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
26398 2017-06-09 Tamar Christina <tamar.christina@arm.com>
26400 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
26403 2017-06-09 Tom de Vries <tom@codesourcery.com>
26406 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
26407 "target cannot support label values" when encountering LABEL_REF.
26409 2017-06-09 Martin Liska <mliska@suse.cz>
26411 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
26412 (gimple_gen_ic_func_profiler): Emit direct comparison
26413 of __gcov_indirect_call_callee with NULL.
26414 (gimple_gen_time_profiler): Change probability from
26415 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
26417 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26419 * profile.c (edge_gcov_counts): Turn to pointer.
26420 (compute_branch_probabilities, compute_branch_probabilities): Update.
26421 (branch_prob): Do not clear edge_gcov_count.
26422 * profile.h (edge_gcov_counts): Turn to pointer.
26423 (edge_gcov_count): Update.
26425 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26427 * gimple.h (gimple_check_failed): Mark cold.
26429 2017-06-09 Richard Biener <rguenther@suse.de>
26431 PR tree-optimization/66623
26432 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
26433 refactor check_reduction into two parts, properly computing
26434 whether we have to check reduction validity for outer loop
26437 2017-06-09 Richard Biener <rguenther@suse.de>
26439 PR tree-optimization/79483
26440 * graphite-scop-detection.c (order): New global.
26441 (get_order): Compute bb to order mapping that satisfies code
26442 generation constraints.
26443 (cmp_pbbs): New helper.
26444 (build_scops): Start domwalk at entry block, sort generated
26447 2017-06-09 Richard Biener <rguenther@suse.de>
26449 PR middle-end/81007
26450 * ipa-polymorphic-call.c
26451 (ipa_polymorphic_call_context::restrict_to_inner_class):
26452 Skip FIELD_DECLs with error_mark_node type.
26453 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
26456 2017-06-09 Martin Liska <mliska@suse.cz>
26458 * predict.c (struct branch_predictor): New struct.
26459 (test_prediction_value_range): New test.
26460 (predict_c_tests): New function.
26461 * selftest-run-tests.c (selftest::run_tests): Run the function.
26462 * selftest.h: Declare new tests.
26464 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
26467 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
26468 gen_add3_insn did not fail.
26469 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
26470 r0, construct that number in a temporary reg and add that reg to r0.
26471 If asked to put the result in r0 as well, fail.
26473 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
26475 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26476 for early expansion of vec_eqv.
26478 2017-06-08 Jakub Jelinek <jakub@redhat.com>
26480 PR middle-end/81005
26481 * ubsan.c (instrument_null): Avoid pointless code temporary.
26482 (pass_ubsan::execute): Instrument aggregate arguments of calls.
26484 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
26488 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
26491 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
26494 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26496 * predict.c (unlikely_executed_edge_p): Move ahead.
26497 (probably_never_executed_edge_p): Use it.
26499 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
26501 PR middle-end/79988
26502 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
26503 gimple_call_builtin_p call.
26505 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26507 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
26508 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
26509 rtl_check_failed_type2, rtl_check_failed_code1,
26510 rtl_check_failed_code2, rtl_check_failed_code_mode,
26511 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
26512 rtvec_check_failed_bounds, rtl_check_failed_flag,
26513 _fatal_insn_not_found, _fatal_insn): Likewise.
26514 * tree.h (tree_contains_struct_check_failed,
26515 tree_check_failed, tree_not_check_failed,
26516 tree_class_check_failed, tree_range_check_failed,
26517 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
26518 tree_vec_elt_check_failed, phi_node_elt_check_failed,
26519 tree_operand_check_failed, omp_clause_check_failed,
26520 omp_clause_operand_check_failed, omp_clause_range_check_failed):
26523 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26525 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
26526 flag_branch_probabilities.
26527 * ipa-inline.c (edge_badness): Likewise.
26528 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
26529 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
26530 * predict.c (maybe_hot_frequency_p): Likewise.
26531 (probably_never_executed): Likewise.
26532 * sched-ebb.c (schedule_ebbs): Likewise.
26533 * sched-rgn.c (find_single_block_region): Likewise.
26534 * tracer.c (tail_duplicate): Likewise.
26536 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26538 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
26539 longer requires x_flag_profile_use.
26541 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26543 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
26544 instead of flag_reorder_blocks_and_partition.
26545 * dbxout.c (dbxout_function_end): Likewise.
26546 * dwarf2out.c (gen_subprogram_die): Likewise.
26547 * haifa-sched.c (sched_create_recovery_edges): Likewise.
26548 * hw-doloop.c (reorg_loops): Likewise.
26549 * varasm.c (assemble_start_function,
26550 assemble_end_function): Likewise.
26551 (decide_function_section): Do not check for
26552 flag_reorder_blocks_and_partition.
26554 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
26556 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26558 (chkp_get_hard_register_fake_addr_expr): Ditto.
26559 (chkp_build_addr_expr): Add check for hard reg case.
26560 (chkp_parse_array_and_component_ref): Ditto.
26561 (chkp_find_bounds_1): Ditto.
26562 (chkp_process_stmt): Don't generate bounds store for
26565 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26567 * predict.c (maybe_hot_bb_p): Do not check profile status.
26568 (maybe_hot_edge_p): Likewise.
26569 (probably_never_executed): Check for zero counts even if profile
26571 (unlikely_executed_edge_p): New function.
26572 (unlikely_executed_stmt_p): New function.
26573 (unlikely_executed_bb_p): New function.
26574 (set_even_probabilities): Use unlikely predicates.
26575 (combine_predictions_for_bb): Likewise.
26576 (predict_paths_for_bb): Likewise.
26577 (predict_paths_leading_to_edge): Likewise.
26578 (determine_unlikely_bbs): New function.
26579 (estimate_bb_frequencies): Use it.
26580 (compute_function_frequency): Use zero counts even if profile is
26582 * profile-count.h: Fix typo.
26584 2017-08-08 Julia Koval <julia.koval@intel.com>
26586 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
26587 _mm512_mask_cvtsepi16_storeu_epi8,
26588 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
26589 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
26590 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
26591 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
26592 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
26593 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
26594 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
26595 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
26596 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
26597 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
26598 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
26599 __builtin_ia32_pmovuswb256mem_mask,
26600 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
26601 __builtin_ia32_pmovwb512mem_mask): New builtins.
26603 2017-08-08 Julia Koval <julia.koval@intel.com>
26605 PR target/73350,80862
26606 * config/i386/subst.md (round): Fix round pattern.
26607 * config/i386/i386.c (ix86_erase_embedded_rounding):
26608 Fix erasing rounding for the fixed pattern.
26610 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26612 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
26614 2017-06-08 Martin Liska <mliska@suse.cz>
26616 PR gcov-profile/80911
26617 * gcov.c (block_info::block_info): New constructor.
26619 2017-06-07 Carl Love <cel@us.ibm.com>
26621 * config/rs6000/rs6000-c: The return type of the following
26622 built-in functions was implemented as int not long long. Fix sign
26623 of return value for the unsigned version of vec_mulo and vec_mule.
26624 vector unsigned long long vec_bperm (vector unsigned long long,
26625 vector unsigned char)
26626 vector signed long long vec_mule (vector signed int,
26628 vector unsigned long long vec_mule (vector unsigned int,
26629 vector unsigned int)
26630 vector signed long long vec_mulo (vector signed int,
26632 vector unsigned long long vec_mulo (vector unsigned int,
26633 vector unsigned int)
26634 * doc/extend.texi: Fix the documentation for the built-in
26637 2017-06-07 Carl Love <cel@us.ibm.com>
26640 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
26643 2017-06-07 Carl Love <cel@us.ibm.com>
26645 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
26646 support, Generate doublehv for signed int/float for BE case only.
26648 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
26650 * doc/invoke.texi (mcx16): Rewrite.
26652 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26654 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
26655 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
26656 *mov<mode>_softfloat, and an anonymous splitter): Use
26657 nonimmediate_operand instead of rs6000_nonimmediate_operand.
26659 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26661 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
26663 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
26664 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
26665 (rs6000_debug_reg_global): Adjust.
26666 (rs6000_init_hard_regno_mode_ok): Adjust.
26667 (rs6000_dbx_register_number): Adjust.
26668 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
26669 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
26670 Remove SPE_ACC and SPEFSCR.
26671 (REG_ALLOC_ORDER): Ditto.
26672 (FRAME_POINTER_REGNUM): Change to 111.
26673 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
26674 (REG_CLASS_NAMES): Ditto.
26675 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
26676 (REGISTER_NAMES): Ditto.
26677 (ADDITIONAL_REG_NAMES): Ditto.
26678 (rs6000_reg_names): Ditto.
26679 * config/rs6000/rs6000.md: Renumber some register number
26682 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26684 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
26686 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
26687 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
26689 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
26690 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
26691 Delete the SPE high registers.
26692 (REG_ALLOC_ORDER): Ditto.
26693 (enum reg_class): Remove SPE_HIGH_REGS.
26694 (REG_CLASS_NAMES): Ditto.
26695 (REG_CLASS_CONTENTS): Delete the SPE high registers.
26696 (REGISTER_NAMES): Ditto.
26697 (rs6000_reg_names): Ditto.
26698 * doc/tm.texi.in: Remove SPE as example.
26699 * doc/tm.texi: Regenerate.
26701 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26703 * config/rs6000/8540.md (ppc8540_brinc): Delete.
26704 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
26705 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
26706 * config/rs6000/rs6000.md (type): Remove "brinc".
26708 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26710 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
26711 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
26712 * config/rs6000/linuxspe.h: Delete file.
26713 * config/rs6000/rs6000.md: Don't include spe.md.
26714 * config/rs6000/spe.h: Delete file.
26715 * config/rs6000/spe.md: Delete file.
26716 * config/rs6000/t-rs6000: Remove spe.md.
26718 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26720 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
26721 (reg_or_none500mem_operand): Delete.
26722 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
26723 instead of reg_or_none500mem_operand.
26725 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26727 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26728 handling of SPE flags.
26729 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
26731 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26733 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
26735 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
26736 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
26737 paired_divv2sf3): Similar.
26738 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
26739 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
26740 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
26742 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
26743 Rename the paired_* instruction patterns.
26744 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
26746 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
26747 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
26748 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
26749 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
26750 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
26751 PAIRED_VECTOR_MODE.
26752 (struct machine_function): Delete field spe_insn_chain_scanned_p.
26753 (spe_func_has_64bit_regs_p): Delete.
26754 (spe_expand_predicate_builtin): Delete.
26755 (spe_expand_evsel_builtin): Delete.
26756 (TARGET_DWARF_REGISTER_SPAN): Do not define.
26757 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
26758 (invalid_e500_subreg): Delete.
26759 (rs6000_legitimize_address): Always force_reg op2 as well, for
26760 paired single memory accesses.
26761 (rs6000_member_type_forces_blk): Delete.
26762 (rs6000_spe_function_arg): Delete.
26763 (rs6000_expand_unop_builtin): Delete SPE handling.
26764 (rs6000_expand_binop_builtin): Ditto.
26765 (spe_expand_stv_builtin): Delete.
26766 (bdesc_2arg_spe): Delete.
26767 (spe_expand_builtin): Delete.
26768 (spe_expand_predicate_builtin): Delete.
26769 (spe_expand_evsel_builtin): Delete.
26770 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
26771 (spe_init_builtins): Delete.
26772 (spe_func_has_64bit_regs_p): Delete.
26773 (savres_routine_name): Delete "info" parameter. Adjust callers.
26774 (rs6000_emit_stack_reset): Ditto.
26775 (rs6000_dwarf_register_span): Delete.
26776 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
26777 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
26778 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
26780 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
26782 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
26783 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
26784 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
26785 mulv2sf3, divv2sf3): Delete expanders.
26787 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26789 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
26791 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26793 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
26794 * config/rs6000/rs6000.c: Ditto.
26796 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26798 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
26799 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
26800 comparison_operator.
26802 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26804 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
26805 * config/rs6000/rs6000.opt: Ditto.
26806 * config/rs6000/t-rtems: Ditto.
26808 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26810 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
26811 TARGET_E500_SINGLE by 0, simplify.
26812 * config/rs6000/rs6000.c: Ditto.
26813 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
26814 (spe_build_register_parallel): Delete.
26815 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
26816 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
26817 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
26818 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
26819 (E500_CONVERT): Delete.
26820 * config/rs6000/spe.md: Remove many patterns and all define_constants.
26822 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26824 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
26825 * config/rs6000/dfp.md: Ditto.
26826 (negdd2, *negdd2_fpr): Merge.
26827 (absdd2, *absdd2_fpr): Merge.
26828 (negtd2, *negtd2_fpr): Merge.
26829 (abstd2, *abstd2_fpr): Merge.
26830 * config/rs6000/e500.h: Delete file.
26831 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
26832 TARGET_FPRS by 1 and simplify.
26833 * config/rs6000/rs6000-c.c: Ditto.
26834 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
26835 TARGET_DF_SPE by 0.
26836 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
26838 * config/rs6000/rs6000.md: Ditto.
26839 (floatdidf2, *floatdidf2_fpr): Merge.
26840 (move_from_CR_gt_bit): Delete.
26841 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
26842 (E500_CR_IOR_COMPARE): Delete.
26843 (All patterns that require !TARGET_FPRS): Delete.
26844 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
26846 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26848 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
26850 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26852 * graphds.c (add_edge): Intitialize edge's attached data.
26853 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
26854 pointer parameter. Call pointed function on each edge during
26855 graph traversing. Skip traversing the edge when the function
26857 (graphds_dfs, graphds_scc): Ditto.
26858 (for_each_edge): New parameter. Pass the new parameter to callback
26860 * graphds.h (skip_edge_callback): New function pointer type.
26861 (graphds_dfs, graphds_scc): New function pointer parameter.
26862 (graphds_edge_callback, for_each_edge): New parameter.
26864 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26866 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
26867 out code checking if runtime alias check is possible to below ...
26868 Call the new function.
26869 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
26870 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
26872 2017-06-07 Marek Polacek <polacek@redhat.com>
26875 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
26876 TYPE_OVERFLOW_WRAPS check.
26878 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26880 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
26881 if versioning is required.
26882 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
26883 peeling with the check for versioning.
26885 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26887 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
26888 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
26889 Set true to new parameter if new ssa variable is defined.
26890 (vect_gen_vector_loop_niters): Refactor. Set range information
26891 for the new vector loop bound variable.
26892 (vect_do_peeling): Ditto.
26894 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26896 * tree-affine.c (ssa.h): Include header file.
26897 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
26898 has wrapping overflow behavior.
26900 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26902 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
26904 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26906 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
26907 (tree_to_aff_combination): ... here.
26909 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26911 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
26912 reg_pressure model function.
26913 (ivopts_global_cost_for_size): Delete.
26914 (determine_set_costs, iv_ca_recount_cost): Call new model function
26915 ivopts_estimate_reg_pressure.
26917 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26919 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
26920 expensive than udiv. Remove floating point cases from mod.
26922 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26924 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
26925 Increase idiv cost.
26927 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26929 * config/aarch64/aarch64.md
26930 (copysignsf3): Fix mask generation.
26932 2017-06-07 Jakub Jelinek <jakub@redhat.com>
26934 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
26936 (class dump_manager): Add register_dumps method.
26937 * dumpfile.c: Include langhooks.h.
26938 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
26939 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
26940 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
26941 (dump_manager::dump_register): Start with 512 entries instead of 32.
26942 (dump_manager::register_dumps): New method.
26943 * toplev.c (general_init): Instead of invoking register_dumps
26944 langhook, invoke register_dumps method on the dump manager.
26945 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
26948 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
26950 * doc/md.texi: Clarify the restrictions on a define_insn condition.
26951 Say that # requires an associated define_split to exist, and that
26952 the define_split must be suitable for use after register allocation.
26954 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26956 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
26957 (compute_outgoing_frequencies): Also initialize zero counts.
26958 (find_many_sub_basic_blocks): Do not produce uninitialized profile
26959 around loops; preserve more of profile when nothing changes.
26961 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
26963 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
26965 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
26966 * config/arm/arm-cpu-cdata.h: Regenerate.
26967 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
26968 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
26969 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
26970 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
26971 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
26973 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
26975 * config/arm/t-rmprofile: Likewise.
26976 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
26978 2017-06-06 David S. Miller <davem@davemloft.net>
26981 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
26982 blockage if function uses alloca.
26984 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26986 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
26987 New "uid" fields to hold pretty-print IDs of group and ref.
26988 Memory references are now identified as <group_id>:<ref_id>
26989 instead of using [random] addresses.
26990 (dump_mem_details): Simplify, no functional change.
26991 (dump_mem_ref): Simplify and make output more concise.
26992 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
26993 (find_or_create_group): Initialize group uid.
26994 (record_ref): Initialize ref uid. Improve debug output.
26995 (prune_group_by_reuse, should_issue_prefetch_p,)
26996 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
26997 (mark_nontemporal_store, determine_loop_nest_reuse):
26998 Improve debug output.
27000 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27002 * dbgcnt.def (prefetch): New debug counter.
27003 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
27004 (schedule_prefetches): Stop issueing prefetches if debug counter
27007 2017-06-06 Tom de Vries <tom@codesourcery.com>
27009 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
27010 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
27012 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27014 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
27015 Use aarch64_reg_or_zero predicate for operand 4.
27016 (aarch64_compare_and_swap<mode> define_insn_and_split):
27017 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
27018 (aarch64_store_exclusive<mode>): Likewise for operand 2.
27020 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
27022 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
27023 (arm_compute_save_core_reg_mask): This.
27024 (thumb1_compute_save_reg_mask): Rename into ...
27025 (thumb1_compute_save_core_reg_mask): This.
27026 (arm_compute_save_reg0_reg12_mask): Adapt comment.
27027 (arm_compute_frame_layout): Likewise.
27029 2017-06-06 Richard Biener <rguenther@suse.de>
27031 PR tree-optimization/80974
27032 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
27033 keep or clear leaders SSA info.
27035 2017-06-06 Tom de Vries <tom@codesourcery.com>
27037 * config/nvptx/nvptx.c (split_mode_p): New function.
27038 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
27040 2017-06-06 Tom de Vries <tom@codesourcery.com>
27042 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
27044 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27047 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
27050 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27052 * shrink-wrap.c (handle_simple_exit): Update profile.
27053 (try_shrink_wrapping): Upate profile.
27055 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27057 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
27058 (tree_guess_outgoing_edge_probabilities): New.
27059 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
27060 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
27062 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27064 * ipa-split.c (split_function): Initialize return bb profile.
27066 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27068 * profile.c (compute_branch_probabilities): Also initialize
27069 EXIT_BLOCK profile.
27071 2017-06-06 Richard Biener <rguenther@suse.de>
27073 PR tree-optimization/80928
27074 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
27075 (vect_analyze_loop_operations): Properly guard analysis for
27077 (vect_transform_loop): Likewise.
27078 (vect_analyze_loop_2): Also reset SLP type on PHIs.
27079 (vect_model_induction_cost): Do not cost for pure SLP.
27080 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
27081 of induction in inner loop vectorization.
27082 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
27083 (vect_get_and_check_slp_defs): Handle vect_induction_def.
27084 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
27086 (vect_analyze_slp_cost_1): Cost induction.
27087 (vect_detect_hybrid_slp_stmts): Handle PHIs.
27088 (vect_get_slp_vect_defs): Likewise.
27089 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
27090 (vect_transform_stmt): Handle SLP reductions.
27091 * tree-vectorizer.h (vectorizable_induction): Adjust.
27093 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27095 * config/rs6000/rs6000.c (make_resolver_func): Update
27096 init_lowered_empty_function call.
27098 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
27100 * doc/invoke.texi: Document the -fprofile-abs-path option.
27101 * common.opt (fprofile-abs-path): New option.
27102 * gcov-io.h (gcov_write_filename): Declare.
27103 * gcov-io.c (gcov_write_filename): New function.
27104 * coverage.c (coverage_begin_function): Use gcov_write_filename.
27105 * profile.c (output_location): Likewise.
27107 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27109 * shring-wrap.c: Revert accidental commit.
27111 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
27113 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
27115 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27117 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
27119 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
27120 profile in callgraph edge.
27121 * profile-count.h (apply_probability): If THIS is 0, then result is 0
27122 (apply_scale): Likewise.
27123 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
27124 Also scale profile when inlining function with zero profile.
27125 (initialize_cfun): Update exit block profile even when it is zero.
27126 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
27127 when profile is read.
27129 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27131 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
27132 (CLONE_*): New constants to define the processors we can generate
27133 code for with the target_clone attribute.
27134 (rs6000_clone_map): New array to identify which clone processors
27135 the current program is running on.
27136 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
27137 target_clone attribute.
27138 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
27139 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
27140 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
27141 (cpu_expand_builtin): Add support for target_clone attribute.
27142 (rs6000_valid_attribute_p): Allow "default" attribute.
27143 (get_decl_name): New debug function to simplify printing the
27144 current function name in debugging statements.
27145 (rs6000_clone_priority): New functions to support the target_clone
27146 attribute, and be able to generate code to switch between ISA 2.05
27147 through ISA 3.0 (power6 through power9).
27148 (rs6000_compare_version_priority): Likewise.
27149 (rs6000_get_function_versions_dispatcher): Likewise.
27150 (make_resolver_func): Likewise.
27151 (add_condition_to_bb): Likewise.
27152 (dispatch_function_versions): Likewise.
27153 (rs6000_generate_version_dispatcher_body): Likewise.
27154 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
27155 (fusion_gpr_load_p): Fix a spacing issue.
27156 * doc/extend.texi (Common Function Attributes): Document that the
27157 PowerPC supports the target_clone attribute.
27159 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
27161 * config/arm/arm.h: explain F symbol found in description of ARM
27162 register allocation in its legend.
27164 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27166 * config/mips/frame-header-opt.c: Include profile-count.h.
27167 * config/riscv/riscv.c: Include profile-count.h
27169 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27171 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
27173 (sm_set_flag_if_changed): Add bbs field.
27174 (execute_sm_if_changed_flag_set): Pass BBS.
27175 (execute_sm): Update.
27177 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27179 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
27182 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27184 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
27185 (peephole2): New peephole2 to emit the above.
27186 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
27188 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27190 * config/aarch64/aarch64.c (define_peephole2 above
27191 *sub_<shift>_<mode>): New peephole.
27193 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27195 * config/i386/i386.c (make_resolver_func): Update.
27196 * Makefile.in: Add profile-count.h and profile-count.o
27197 * auto-profile.c (afdo_indirect_call): Update to new API.
27198 (afdo_set_bb_count): Update.
27199 (afdo_propagate_edge): Update.
27200 (afdo_propagate_circuit): Update.
27201 (afdo_calculate_branch_prob): Update.
27202 (afdo_annotate_cfg): Update.
27203 * basic-block.h: Include profile-count.h
27204 (struct edge_def): Turn count to profile_count.
27205 (struct basic_block_def): Likewie.
27206 (REG_BR_PROB_BASE): Move to profile-count.h
27207 (RDIV): Move to profile-count.h
27208 * bb-reorder.c (max_entry_count): Turn to profile_count.
27209 (find_traces): Update.
27210 (rotate_loop):Update.
27211 (connect_traces):Update.
27212 (sanitize_hot_paths):Update.
27213 * bt-load.c (migrate_btr_defs): Update.
27214 * cfg.c (RDIV): Remove.
27215 (init_flow): Use alloc_block.
27216 (alloc_block): Uninitialize count.
27217 (unchecked_make_edge): Uninitialize count.
27218 (check_bb_profile): Update.
27219 (dump_edge_info): Update.
27220 (dump_bb_info): Update.
27221 (update_bb_profile_for_threading): Update.
27222 (scale_bbs_frequencies_int): Update.
27223 (scale_bbs_frequencies_gcov_type): Update.
27224 (scale_bbs_frequencies_profile_count): New.
27225 * cfg.h (update_bb_profile_for_threading): Update.
27226 (scale_bbs_frequencies_profile_count): Declare.
27227 * cfgbuild.c (compute_outgoing_frequencies): Update.
27228 (find_many_sub_basic_blocks): Update.
27229 * cfgcleanup.c (try_forward_edges): Update.
27230 (try_crossjump_to_edge): Update.
27231 * cfgexpand.c (expand_gimple_tailcall): Update.
27232 (construct_exit_block): Update.
27233 * cfghooks.c (verify_flow_info): Update.
27234 (dump_bb_for_graph): Update.
27235 (split_edge): Update.
27236 (make_forwarder_block): Update.
27237 (duplicate_block): Update.
27238 (account_profile_record): Update.
27239 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
27240 (get_estimated_loop_iterations): Update.
27241 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
27242 (single_likely_exit): Update.
27243 * cfgloopmanip.c (scale_loop_profile): Update.
27245 (set_zero_probability): Update.
27246 (lv_adjust_loop_entry_edge): Update.
27247 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
27248 (purge_dead_edges): Update.
27249 (rtl_account_profile_record): Update.
27250 * cgraph.c (cgraph_node::create): Uninitialize count.
27251 (symbol_table::create_edge): Uninitialize count.
27252 (cgraph_update_edges_for_call_stmt_node): Update.
27253 (cgraph_edge::dump_edge_flags): Update.
27254 (cgraph_node::dump): Update.
27255 (cgraph_edge::maybe_hot_p): Update.
27256 * cgraph.h: Include profile-count.h
27257 (create_clone), create_edge, create_indirect_edge): Update.
27258 (cgraph_node): Turn count to profile_count.
27259 (cgraph_edge0: Likewise.
27260 (make_speculative, clone): Update.
27261 (create_edge): Update.
27262 (init_lowered_empty_function): Update.
27263 * cgraphclones.c (cgraph_edge::clone): Update.
27264 (duplicate_thunk_for_node): Update.
27265 (cgraph_node::create_clone): Update.
27266 * cgraphunit.c (cgraph_node::analyze): Update.
27267 (cgraph_node::expand_thunk): Update.
27268 * final.c (dump_basic_block_info): Update.
27269 * gimple-streamer-in.c (input_bb): Update.
27270 * gimple-streamer-out.c (output_bb): Update.
27271 * graphite.c (print_global_statistics): Update.
27272 (print_graphite_scop_statistics): Update.
27273 * hsa-brig.c: Include basic-block.h.
27274 * hsa-dump.c: Include basic-block.h.
27275 * hsa-gen.c (T sum_slice): Update.
27276 (convert_switch_statements):Update.
27277 * hsa-regalloc.c: Include basic-block.h.
27278 * ipa-chkp.c (chkp_produce_thunks): Update.
27279 * ipa-cp.c (struct caller_statistics): Update.
27280 (init_caller_stats): Update.
27281 (gather_caller_stats): Update.
27282 (ipcp_cloning_candidate_p): Update.
27283 (good_cloning_opportunity_p): Update.
27284 (get_info_about_necessary_edges): Update.
27285 (dump_profile_updates): Update.
27286 (update_profiling_info): Update.
27287 (update_specialized_profile): Update.
27288 (perhaps_add_new_callers): Update.
27289 (decide_about_value): Update.
27290 (ipa_cp_c_finalize): Update.
27291 * ipa-devirt.c (struct odr_type_warn_count): Update.
27292 (struct decl_warn_count): Update.
27293 (struct final_warning_record): Update.
27294 (possible_polymorphic_call_targets): Update.
27295 (ipa_devirt): Update.
27296 * ipa-fnsummary.c (redirect_to_unreachable): Update.
27297 * ipa-icf.c (sem_function::merge): Update.
27298 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
27299 * ipa-inline.c (compute_uninlined_call_time): Update.
27300 (compute_inlined_call_time): Update.
27301 (want_inline_small_function_p): Update.
27302 (want_inline_self_recursive_call_p): Update.
27303 (edge_badness): Update.
27304 (lookup_recursive_calls): Update.
27305 (recursive_inlining): Update.
27306 (inline_small_functions): Update.
27307 (dump_overall_stats): Update.
27308 (dump_inline_stats): Update.
27309 * ipa-profile.c (ipa_profile_generate_summary): Update.
27310 (ipa_propagate_frequency): Update.
27311 (ipa_profile): Update.
27312 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27313 * ipa-utils.c (ipa_merge_profiles): Update.
27314 * loop-doloop.c (doloop_modify): Update.
27315 * loop-unroll.c (report_unroll): Update.
27316 (unroll_loop_runtime_iterations): Update.
27317 * lto-cgraph.c (lto_output_edge): Update.
27318 (lto_output_node): Update.
27319 (input_node): Update.
27320 (input_edge): Update.
27321 (merge_profile_summaries): Update.
27322 * lto-streamer-in.c (input_cfg): Update.
27323 * lto-streamer-out.c (output_cfg): Update.
27324 * mcf.c (create_fixup_graph): Update.
27325 (adjust_cfg_counts): Update.
27326 (sum_edge_counts): Update.
27327 * modulo-sched.c (sms_schedule): Update.
27328 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
27329 * predict.c (maybe_hot_count_p): Update.
27330 (probably_never_executed): Update.
27331 (dump_prediction): Update.
27332 (combine_predictions_for_bb): Update.
27333 (propagate_freq): Update.
27334 (handle_missing_profiles): Update.
27335 (counts_to_freqs): Update.
27336 (rebuild_frequencies): Update.
27337 (force_edge_cold): Update.
27338 * predict.h: Include profile-count.h
27339 (maybe_hot_count_p, counts_to_freqs): UPdate.
27340 * print-rtl-function.c: Do not include cfg.h
27341 * print-rtl.c: Include basic-block.h
27342 * profile-count.c: New file.
27343 * profile-count.h: New file.
27344 * profile.c (is_edge_inconsistent): Update.
27345 (correct_negative_edge_counts): Update.
27346 (is_inconsistent): Update.
27347 (set_bb_counts): Update.
27348 (read_profile_edge_counts): Update.
27349 (compute_frequency_overlap): Update.
27350 (compute_branch_probabilities): Update; Initialize and deinitialize
27352 (branch_prob): Update.
27353 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
27354 (edge_gcov_count): New.
27355 (bb_gcov_count): New.
27356 * shrink-wrap.c (try_shrink_wrapping): Update.
27357 * tracer.c (better_p): Update.
27358 * trans-mem.c (expand_transaction): Update.
27359 (ipa_tm_insert_irr_call): Update.
27360 (ipa_tm_insert_gettmclone_call): Update.
27361 * tree-call-cdce.c: Update.
27362 * tree-cfg.c (gimple_duplicate_sese_region): Update.
27363 (gimple_duplicate_sese_tail): Update.
27364 (gimple_account_profile_record): Update.
27365 (execute_fixup_cfg): Update.
27366 * tree-inline.c (copy_bb): Update.
27367 (copy_edges_for_bb): Update.
27368 (initialize_cfun): Update.
27369 (freqs_to_counts): Update.
27370 (copy_cfg_body): Update.
27371 (expand_call_inline): Update.
27372 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
27373 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
27374 (try_unroll_loop_completely): Update.
27375 (try_peel_loop): Update.
27376 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
27377 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
27378 * tree-ssa-loop-split.c (connect_loops): Update.
27379 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
27380 * tree-ssa-reassoc.c (branch_fixup): Update.
27381 * tree-ssa-tail-merge.c (replace_block_by): Update.
27382 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
27383 (compute_path_counts): Update.
27384 (update_profile): Update.
27385 (recompute_probabilities): Update.
27386 (update_joiner_offpath_counts): Update.
27387 (estimated_freqs_path): Update.
27388 (freqs_to_counts_path): Update.
27389 (clear_counts_path): Update.
27390 (ssa_fix_duplicate_block_edges): Update.
27391 (duplicate_thread_path): Update.
27392 * tree-switch-conversion.c (case_bit_test_cmp): Update.
27393 (struct switch_conv_info): Update.
27394 * tree-tailcall.c (decrease_profile): Update.
27395 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
27396 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
27397 * value-prof.c (check_counter): Update.
27398 (gimple_divmod_fixed_value): Update.
27399 (gimple_mod_pow2): Update.
27400 (gimple_mod_subtract): Update.
27401 (gimple_ic_transform): Update.
27402 (gimple_stringop_fixed_value): Update.
27403 * value-prof.h (gimple_ic): Update.
27405 2017-06-02 Carl Love <cel@us.ibm.com>
27407 * config/rs6000/rs6000-c: Add support for built-in functions
27408 vector double vec_doublee (vector signed int);
27409 vector double vec_doublee (vector unsigned int);
27410 vector double vec_doublee (vector float);
27411 vector double vec_doubleh (vector signed int);
27412 vector double vec_doubleh (vector unsigned int);
27413 vector double vec_doubleh (vector float);
27414 vector double vec_doublel (vector signed int);
27415 vector double vec_doublel (vector unsigned int);
27416 vector double vec_doublel (vector float);
27417 vector double vec_doubleo (vector signed int);
27418 vector double vec_doubleo (vector unsigned int);
27419 vector double vec_doubleo (vector float);.
27420 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
27421 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
27423 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
27424 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
27425 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
27427 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
27428 vec_doublel, vec_doubleh.
27429 * doc/extend.texi: Update the built-in documentation file for the
27430 new built-in functions.
27432 2017-06-02 David Malcolm <dmalcolm@redhat.com>
27435 * ipa-inline-analysis.c (free_growth_caches): Set
27436 edge_removal_hook_holder to NULL after removing it.
27438 2017-06-02 Sudakshina Das <sudi.das@arm.com>
27440 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
27441 comparision with zero.
27443 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
27444 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27445 for early expansion of vec_min and vec_max builtins.
27446 (builtin_function_type): Add min/max unsigned variants to those
27447 identified as having unsigned arguments.
27449 2017-06-02 Olivier Hainque <hainque@adacore.com>
27451 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
27453 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27455 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
27456 Use VALL_F16 iterator rather than VALL.
27458 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27460 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
27461 Emit CBNZ inside loop when doing a strong exchange and comparing
27462 against zero. Generate the CC flags after the loop.
27464 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
27466 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
27467 (dl_section_ref): New.
27468 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
27469 On AIX, append an expression to subtract the size of the
27470 section length to dl_section_ref.
27472 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
27474 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27475 for early expansion of vector absolute builtins.
27477 2017-06-02 Richard Biener <rguenther@suse.de>
27479 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
27480 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
27482 2017-06-02 Richard Biener <rguenther@suse.de>
27484 PR tree-optimization/80948
27485 * tree-tailcall.c (find_tail_calls): Track stmts to move in
27486 stmt order as well.
27488 2017-06-02 Richard Biener <rguenther@suse.de>
27490 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
27492 * tree-vect-stmts.c (process_use): Do not mark backedge defs
27493 for inductions as relevant.
27495 2017-06-02 Richard Biener <rguenther@suse.de>
27497 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
27498 (vectorizable_induction): ... this. Remove dead code.
27500 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
27502 * builtins. (expand_builtin_alloca): Remove second parameter and
27503 infer its value from the first parameter instead.
27504 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
27506 2017-06-02 Jakub Jelinek <jakub@redhat.com>
27508 PR rtl-optimization/80903
27509 * loop-doloop.c (add_test): Unshare sequence.
27511 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
27513 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
27515 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
27517 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
27519 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
27520 xlogue_layout::get_instance, logue_layout::xlogue_layout,
27521 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
27522 (xlogue_layout::get_stub_rtx): Make static.
27523 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
27524 (xlogue_layout::compute_stub_managed_regs): Rename to...
27525 (xlogue_layout::count_stub_managed_regs): ...this.
27526 (xlogue_layout::is_stub_managed_reg): New function.
27527 (xlogue_layout::m_stub_names): Rename to...
27528 (xlogue_layout::s_stub_names): ...this, make static.
27529 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
27530 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
27531 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
27532 xlogue_layout::s_stub_names): Instantiate statics.
27533 (stub_managed_regs): Remove.
27534 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
27535 (disable_call_ms2sysv_xlogues): Rename to...
27536 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
27537 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
27539 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
27540 change after reload_completed.
27541 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
27543 (ix86_expand_prologue): Likewise.
27544 (ix86_expand_epilogue): Likewise.
27545 (ix86_expand_split_stack_prologue): Likewise.
27546 (ix86_compute_frame_layout): Remove frame parameter ...
27547 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
27548 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
27550 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
27551 (ix86_frame): Move from here ...
27552 * config/i386/i386.h (ix86_frame): ... to here.
27553 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
27554 complete ix86_frame data structure instead. Remove some_ld_name.
27556 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
27558 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
27559 symbols that hold a DECL_VALUE_EXPR.
27561 2017-06-01 Martin Jambor <mjambor@suse.cz>
27563 PR tree-optimization/80898
27564 * tree-sra.c (process_subtree_disqualification): Removed.
27565 (disqualify_candidate): Do not acll
27566 process_subtree_disqualification.
27567 (subtree_mark_written_and_enqueue): New function.
27568 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
27569 RHS has been disqualified and re-queue LHS if necessary. Apart
27570 from that, ignore disqualified RHS.
27572 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27574 * config/s390/s390.c (s390_emit_epilogue): Disable early return
27575 address fetch for z10 or later.
27577 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27579 * config/arc/arc.md (tst_movb): Add guard when splitting.
27581 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27583 * config/arc/arc.c (arc_can_eliminate): Test against
27584 arc_frame_pointer_needed.
27586 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27588 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
27589 to prevent store reordering.
27590 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
27591 (type): Add block type.
27592 (stack_tie): Define special instruction to be used in
27595 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27597 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
27598 constraint. It is not valid for the pattern.
27599 (noncommutative_binary_comparison): Likewise.
27601 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27603 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
27606 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27608 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
27609 be used by the reg-alloc.
27611 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27613 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
27614 reg-alloc when having mul64 or mul32x16 instructions.
27615 (mulsidi3): Likewise.
27616 (umulsidi3): Likewise.
27617 (mulsi32x16): New pattern.
27618 (mulsi64): Likewise.
27619 (mulsidi64): Likewise.
27620 (umulsidi64): Likewise.
27621 (MUL32x16_REG): Define.
27622 (mul64_600): Use MUL32x16_REG.
27623 (mac64_600): Likewise.
27624 (umul64_600): Likewise.
27625 (umac64_600): Likewise.
27627 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27629 * config/arc/arc.md (mulsi3_700): Make it commutative.
27631 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
27633 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
27635 (*sign_extendsidi2_insn): Likewise for movstosw.
27637 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
27639 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
27640 the type of the input discriminant value. Convert the
27641 discriminant value of signedness vary.
27643 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
27645 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
27646 Add to -Wall section.
27648 2017-06-01 Richard Biener <rguenther@suse.de>
27650 PR middle-end/66313
27651 * fold-const.c (fold_plusminus_mult_expr): If the factored
27652 factor may be zero use a wrapping type for the inner operation.
27653 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
27654 and handle moved defs.
27655 (process_assignment): Properly guard the unary op case. Return a
27656 tri-state indicating that moving the stmt before the call may allow
27657 to continue. Pass through to_move.
27658 (find_tail_calls): Handle moving unrelated defs before
27661 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
27664 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
27665 splitter result in the canonical way.
27667 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
27669 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
27670 also for 32bit target. Update insn attributes.
27671 (zero-extendsidi2 splitter): Allow all registers for operand 1.
27673 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
27675 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
27676 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
27677 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
27678 (_mm_maskz_min_ss): New intrinsics.
27680 2017-05-31 Martin Liska <mliska@suse.cz>
27682 * tree-vect-loop.c (vect_create_epilog_for_reduction):
27683 Change comment style to one we normally use.
27684 (vectorizable_reduction): Likewise.
27685 (vectorizable_induction): Likewise.
27686 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
27687 (vectorizable_call): Likewise.
27688 (vectorizable_simd_clone_call): Likewise.
27689 (vectorizable_conversion): Likewise.
27690 (vectorizable_assignment): Likewise.
27691 (vectorizable_shift): Likewise.
27692 (vectorizable_operation): Likewise.
27693 (vectorizable_store): Likewise.
27694 (vectorizable_load): Likewise.
27695 * tree-vectorizer.h: Likewise.
27697 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
27699 * passes.c (emergency_dump_function): New.
27700 * tree-pass.h (emergency_dump_function): Declare.
27701 * plugin.c (plugins_internal_error_function): Remove.
27702 * plugin.h (plugins_internal_error_function): Remove declaration.
27703 * toplev.c (internal_error_function): New static function. Use it...
27704 (general_init): ...here.
27706 2017-05-31 Graham Markall <graham.markall@embecosm.com>
27708 * config/arc/arc.c (arc_print_operand): Handle constant operands.
27709 (arc_rtx_costs): Add costs for new patterns.
27710 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
27711 * config/arc/predicates.md: Add _1_2_3_operand predicate.
27713 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
27715 * tree-ssa-strlen.c (get_next_strinfo): New function.
27716 (get_stridx_plus_constant): Use it.
27717 (zero_length_string): Likewise.
27718 (adjust_related_strinfos): Likewise.
27719 (adjust_last_stmt): Likewise.
27721 2017-05-31 Richard Biener <rguenther@suse.de>
27724 * config/i386/i386.c (ix86_expand_builtin): Remove assert
27725 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
27727 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
27729 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
27730 loop_vinfo argument and use of dependence distance vectors.
27731 Check instead whether the two references differ only in their
27732 initial value and assume that they have the same alignment if the
27733 difference is a multiple of the vector alignment.
27734 (vect_analyze_data_refs_alignment): Update call accordingly.
27736 2017-05-31 Martin Liska <mliska@suse.cz>
27739 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
27741 2017-05-31 Bin Cheng <bin.cheng@arm.com>
27743 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
27744 (create_intersect_range_checks): Move from ...
27745 * tree-data-ref.c (create_intersect_range_checks_index)
27746 (create_intersect_range_checks): ... to here.
27747 (create_runtime_alias_checks): New function factored from ...
27748 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
27749 here. Call above function.
27750 * tree-data-ref.h (create_runtime_alias_checks): New function.
27752 2017-05-31 Bin Cheng <bin.cheng@arm.com>
27754 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
27755 segment length for dr_b and compute it in wide_int.
27757 2017-05-31 Richard Biener <rguenther@suse.de>
27759 PR tree-optimization/80906
27760 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
27761 and pass through iv_map.
27762 (copy_bb_and_scalar_dependences): Adjust.
27763 (translate_pending_phi_nodes): Likewise.
27764 (copy_loop_close_phi_args): Handle code-generating IVs instead
27767 2017-05-30 David Malcolm <dmalcolm@redhat.com>
27769 * diagnostic-color.c (color_dict): Add "type-diff".
27770 (parse_gcc_colors): Update comment.
27771 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
27772 -fdiagnostics-show-template-tree and -fno-elide-type.
27773 (GCC_COLORS): Add type-diff to example.
27775 (-fdiagnostics-show-template-tree): New.
27776 (-fno-elide-type): New.
27777 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
27778 the pp_format_decoder callback. Call any m_format_postprocessor's
27780 (pretty_printer::pretty_printer): Initialize
27781 m_format_postprocessor.
27782 (pretty_printer::~pretty_printer): Delete any
27783 m_format_postprocessor.
27784 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
27785 (class format_postprocessor): New class.
27786 (struct pretty_printer::format_decoder): Document the new parameters.
27787 (struct pretty_printer::m_format_postprocessor): New field.
27788 * tree-diagnostic.c (default_tree_printer): Update for new
27789 bool and const char ** params.
27790 * tree-diagnostic.h (default_tree_printer): Likewise.
27792 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
27794 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
27795 (lwa_operand): Delete rs6000_gen_cell_microcode test.
27796 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27797 rs6000_gen_cell_microcode code.
27798 (rs6000_final_prescan_insn): Delete.
27799 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
27800 "warn-cell-microcode" entries.
27801 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
27802 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
27803 throughout. Change cc_reg_not_micro_cr0_operand to
27804 cc_reg_not_cr0_operand throughout.
27805 (*extendhi<mode>2_noload): Delete.
27806 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
27807 (mwarn-cell-microcode): Delete.
27808 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
27809 -mgen-cell-microcode and -mwarn-cell-microcode.
27811 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
27814 * config/i386/constraints.md (Yd): New constraint.
27816 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
27817 and (?Yd, r) alternatives. Update insn attributes.
27818 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
27819 and (?*Yd, r) alternatives. Update insn attributes.
27820 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
27822 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
27824 * gimplify.c (gimplify_modify_expr): Don't create a
27825 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
27828 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
27830 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
27832 2017-05-30 Richard Biener <rguenther@suse.de>
27834 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
27835 and reduc_def fields.
27836 (STMT_VINFO_REDUC_TYPE): New define.
27837 (STMT_VINFO_REDUC_DEF): Likewise.
27838 (vect_force_simple_reduction): Adjust prototype.
27839 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
27840 (vect_is_simple_reduction): Remove check_reduction argument.
27841 (vect_force_simple_reduction): Adjust and set
27842 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
27843 (vectorizable_reduction): Do not re-do reduction analysis
27844 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
27845 * tree-parloops.c (gather_scalar_reductions): Adjust.
27847 2017-05-30 Richard Biener <rguenther@suse.de>
27849 PR middle-end/80901
27850 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
27853 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
27855 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
27856 Introduce unknown_misalignment parameter and remove vf.
27857 (vect_peeling_hash_get_lowest_cost):
27858 Pass unknown_misalignment parameter.
27859 (vect_enhance_data_refs_alignment):
27860 Fix unsupportable data ref treatment.
27862 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27864 * tree-vect-data-refs.c (vect_get_data_access_cost):
27865 Workaround for SLP handling.
27866 (vect_enhance_data_refs_alignment):
27867 Compute costs for doing no peeling at all, compare to the best
27868 peeling costs so far and avoid peeling if cheaper.
27870 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27872 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
27873 Return peeling info and set costs to zero for unlimited cost
27875 (vect_enhance_data_refs_alignment): Also inspect all datarefs
27876 with unknown misalignment. Compute and costs for unknown
27877 misalignment, compare them to the costs for known misalignment
27878 and choose the cheapest for peeling.
27880 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27882 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
27883 (vect_get_peeling_costs_all_drs): Create function.
27884 (vect_peeling_hash_get_lowest_cost):
27885 Use vect_get_peeling_costs_all_drs.
27886 (vect_peeling_supportable): Create function.
27887 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
27889 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27891 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
27892 DR_HAS_NEGATIVE_STEP.
27893 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
27894 (vect_enhance_data_refs_alignment): Use.
27895 (vect_duplicate_ssa_name_ptr_info): Use.
27896 * tree-vectorizer.h (dr_misalignment): Use.
27897 (known_alignment_for_access_p): Use.
27899 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27902 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
27904 (has_section_name): New function.
27906 2017-05-30 Martin Liska <mliska@suse.cz>
27909 * auto-profile.c (get_function_decl_from_block): Fix
27912 2017-05-30 Richard Biener <rguenther@suse.de>
27914 PR middle-end/80876
27915 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
27917 2017-05-30 Martin Liska <mliska@suse.cz>
27919 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
27920 * dumpfile.h (struct dump_file_info): Remove ctors.
27922 2017-05-30 Martin Liska <mliska@suse.cz>
27924 * predict.def: Fix GNU coding style.
27926 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
27928 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
27929 Mark 'to' argument with ATTRIBUTE_UNUSED.
27931 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
27933 * config/xtensa/xtensa.c (xtensa_emit_call): Use
27934 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
27935 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
27938 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
27940 * doc/install.texi (Options specification): Restore entry of
27941 --enable-sjlj-exceptions.
27943 2017-05-27 Michael Eager <eager@eagercon.com>
27946 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
27948 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
27950 * config/microblaze/microblaze.h
27951 (FIXED_REGISTERS): Update in macro.
27952 (CALL_USED_REGISTERS): Update in macro.
27954 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
27956 * doc/install.texi: Add links to macOS binary distributions.
27958 2017-05-27 Jakub Jelinek <jakub@redhat.com>
27962 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27964 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
27966 2017-05-26 Martin Liska <mliska@suse.cz>
27968 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
27970 2017-05-26 Martin Liska <mliska@suse.cz>
27972 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
27973 always leading ';; '.
27974 (dump_bb_info): Likewise.
27975 (brief_dump_cfg): Likewise.
27976 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
27977 * dumpfile.c: Remove usage of TDF_VERBOSE.
27978 * dumpfile.h (enum dump_kind): Likewise.
27979 (dump_gimple_bb_header): Do not use TDF_COMMENT.
27980 * print-tree.c (debug_verbose): Remove.
27981 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
27982 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
27983 * tree-diagnostic.c (default_tree_printer): Replace
27984 TDF_DIAGNOSTIC with TDF_SLIM.
27986 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27988 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
27989 in parameter loop, rather than loop_vinfo.
27990 (create_intersect_range_checks): Ditto.
27991 (vect_create_cond_for_alias_checks): Update call to above functions.
27993 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27995 PR tree-optimization/80815
27996 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
27997 for merging runtime alias checks. Handle negative DR_STEPs.
27999 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28001 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
28003 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
28004 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
28005 out code pruning runtime alias checks.
28006 * tree-data-ref.c (prune_runtime_alias_test_list): New function
28007 factored out from above.
28008 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28010 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28012 (prune_runtime_alias_test_list): New decalaration.
28014 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28016 * tree-vect-data-refs.c (compare_tree): Rename and move ...
28017 * tree-data-ref.c (data_ref_compare_tree): ... to here.
28018 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
28019 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
28020 (operator==, comp_dr_with_seg_len_pair): Ditto.
28021 (vect_prune_runtime_alias_test_list): Ditto.
28023 2017-05-26 Martin Liska <mliska@suse.cz>
28026 * params.def: Bound partial-inlining-entry-probability param.
28028 2017-05-26 Marek Polacek <polacek@redhat.com>
28031 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
28034 2017-05-26 Richard Biener <rguenther@suse.de>
28036 PR tree-optimization/80842
28037 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
28040 2017-05-26 Richard Biener <rguenther@suse.de>
28042 PR tree-optimization/80844
28043 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
28045 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
28047 * doc/md.texi (Machine Constraints): Update x86 family
28048 machine constraints section to match 'config/i386/constraints.md'.
28050 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
28052 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
28054 2017-05-25 Nathan Sidwell <nathan@acm.org>
28056 * doc/invoke.texi (--enable-languages): Update documentation.
28058 2017-05-25 Martin Liska <mliska@suse.cz>
28060 * dumpfile.c: Add TDF_FOLDING.
28061 * dumpfile.h (enum dump_kind): Likewise.
28062 * genmatch.c (dt_simplify::gen_1): Use it.
28064 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28066 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
28068 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28070 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28071 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
28073 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28075 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
28076 * match.pd (X == C): Rewrite it here.
28077 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
28078 with_certain_nonzero_bits2): New predicates.
28079 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
28081 2017-05-24 Nathan Sidwell <nathan@acm.org>
28083 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
28086 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
28089 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
28091 * config/powerpcspe: New port. Files are copied from the rs6000
28092 port, with "rs6000" in filenames replaced by "powerpcspe".
28094 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
28096 PR rtl-optimization/80754
28097 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
28099 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
28101 * config/sparc/sparc.md (length): Return the correct value for -mflat
28102 sibcalls to match output_sibcall.
28104 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
28108 * config/rs6000/rs6000.c (struct machine_function): Add new field
28110 (rs6000_get_separate_components): Init that field, use it.
28111 (rs6000_components_for_bb): Use the field.
28113 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
28115 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
28117 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
28119 PR middle-end/80823
28120 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
28122 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28125 * config/s390/s390.c (s390_check_qrst_address): Check incoming
28126 address against address_operand predicate.
28127 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
28129 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
28131 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
28132 parameters passed indirectly.
28134 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
28136 * config/i386/i386.md (*movdi_internal): Remove SSE4
28137 alternative 18 (?r, *v). Update insn attributes.
28138 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
28139 Update insn attributes.
28140 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
28141 Update insn attributes.
28142 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
28143 alternative 1 (r, v). Remove isa attribute.
28144 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
28145 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
28146 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
28148 2017-05-23 Tom de Vries <tom@codesourcery.com>
28150 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
28153 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28155 * cgraphunit.c (symbol_table::process_new_functions): Update.
28156 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
28157 (inline_generate_summary): Rename to ...
28158 (ipa_fn_summary_generate): ... this one.
28159 (inline_read_summary): Rename to ...
28160 (ipa_fn_summary_read): ... this one.
28161 (inline_write_summary): Rename to ...
28162 (ipa_fn_summary_write): ... this one.
28163 (inline_free_summary): Rename to ...
28164 (ipa_free_fn_summary): ... this one.
28165 (pass_data_local_fn_summary, pass_local_fn_summary,
28166 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
28167 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
28168 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
28169 make_pass_ipa_fn_summary): New.
28170 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
28171 inline_write_summary, inline_free_summary): Remove.
28172 (ipa_free_fn_summary) : New.
28173 * ipa-inline.c (ipa_inline): Update.
28174 (pass_ipa_inline): Do not generate summaries.
28175 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
28177 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
28178 and add pass_ipa_fn_summary.
28179 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
28181 (make_pass_inline_parameters): Remove.
28183 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
28185 * omp-low.c (struct omp_context): Remove "default_kind" member.
28188 * omp-offload.c (execute_oacc_device_lower): Remove the
28189 parallelism dimensions function attributes for unparallelized
28190 OpenACC kernels constructs.
28192 2017-05-23 Martin Liska <mliska@suse.cz>
28194 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
28196 (cgraph_edge::make_speculative): Likewise.
28197 (cgraph_edge::resolve_speculation): Likewise.
28198 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
28199 (cgraph_node::dump): Likewise.
28200 * cgraph.h: Likewise.
28201 * cgraphunit.c (analyze_functions): Likewise.
28202 (symbol_table::compile): Likewise.
28203 * ipa-cp.c (print_all_lattices): Likewise.
28204 (determine_versionability): Likewise.
28205 (initialize_node_lattices): Likewise.
28206 (ipcp_verify_propagated_values): Likewise.
28207 (estimate_local_effects): Likewise.
28208 (update_profiling_info): Likewise.
28209 (create_specialized_node): Likewise.
28210 (perhaps_add_new_callers): Likewise.
28211 (decide_about_value): Likewise.
28212 (decide_whether_version_node): Likewise.
28213 (identify_dead_nodes): Likewise.
28214 (ipcp_store_bits_results): Likewise.
28215 * ipa-devirt.c (dump_targets): Likewise.
28216 (ipa_devirt): Likewise.
28217 * ipa-icf.c (sem_item::dump): Likewise.
28218 (sem_function::equals): Likewise.
28219 (sem_variable::equals): Likewise.
28220 (sem_item_optimizer::read_section): Likewise.
28221 (sem_item_optimizer::execute): Likewise.
28222 (congruence_class::dump): Likewise.
28223 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
28224 (dump_inline_summary): Likewise.
28225 (estimate_node_size_and_time): Likewise.
28226 (inline_analyze_function): Likewise.
28227 * ipa-inline-transform.c (inline_call): Likewise.
28228 * ipa-inline.c (report_inline_failed_reason): Likewise.
28229 (want_early_inline_function_p): Likewise.
28230 (edge_badness): Likewise.
28231 (update_edge_key): Likewise.
28232 (inline_small_functions): Likewise.
28233 * ipa-profile.c (ipa_profile): Likewise.
28234 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
28235 (ipa_make_edge_direct_to_target): Likewise.
28236 (remove_described_reference): Likewise.
28237 (ipa_impossible_devirt_target): Likewise.
28238 (propagate_controlled_uses): Likewise.
28239 (ipa_print_node_params): Likewise.
28240 (ipcp_transform_function): Likewise.
28241 * ipa-pure-const.c (pure_const_read_summary): Likewise.
28242 (propagate_pure_const): Likewise.
28243 * ipa-reference.c (generate_summary): Likewise.
28244 (read_write_all_from_decl): Likewise.
28245 (propagate): Likewise.
28246 (ipa_reference_read_optimization_summary): Likewise.
28247 * ipa-utils.c (ipa_merge_profiles): Likewise.
28248 * ipa.c (walk_polymorphic_call_targets): Likewise.
28249 (symbol_table::remove_unreachable_nodes): Likewise.
28250 (ipa_single_use): Likewise.
28251 * passes.c (execute_todo): Likewise.
28252 * predict.c (drop_profile): Likewise.
28253 * symtab.c (symtab_node::get_dump_name): New function.
28254 (symtab_node::dump_name): Likewise.
28255 (symtab_node::dump_asm_name): Likewise.
28256 (symtab_node::dump_references): Likewise.
28257 (symtab_node::dump_referring): Likewise.
28258 (symtab_node::dump_base): Likewise.
28259 (symtab_node::debug_symtab): Likewise.
28260 * tree-sra.c (convert_callers_for_node): Likewise.
28261 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
28262 * value-prof.c (init_node_map): Likewise.
28264 2017-05-23 Martin Liska <mliska@suse.cz>
28266 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
28267 and symtab_node::debug_symtab to symbol_table::debug.
28268 * cgraphunit.c (analyze_functions): Use the renamed function.
28269 (symbol_table::compile): Likewise.
28270 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
28271 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
28272 * passes.c (execute_todo): Likewise.
28273 * symtab.c (symbol_table::dump): New function.
28274 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
28276 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28278 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
28279 that nonconst implies exec.
28281 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28283 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
28284 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
28285 (inline_edge_summary_vec): Turn into ...
28286 (ipa_call_summaries): ... this one.
28287 (redirect_to_unreachable, edge_set_predicate,
28288 evaluate_properties_for_edge, inline_summary_alloc,
28289 reset_ipa_call_summary, reset_inline_summary,
28290 inline_summary_t::duplicate): Update.
28291 (inline_edge_duplication_hook): Turn to ...
28292 (ipa_call_summary_t::duplicate): ... this one.
28293 (inline_edge_removal_hook): Turn to ...
28294 (ipa_call_summary_t::remove): ... this one.
28295 (dump_inline_edge_summary): Turn to ...
28296 (dump_ipa_call_summary): ... this one.
28297 (estimate_function_body_sizes): Update.
28298 (inline_update_callee_summaries): Update.
28299 (remap_edge_change_prob): Update.
28300 (remap_edge_summaries): Update.
28301 (inline_merge_summary): Update.
28302 (do_estimate_edge_time): Update.
28303 (inline_generate_summary): Update.
28304 (inline_read_section): Update.
28305 (inline_read_summary): Update.
28306 (inline_free_summary): Update.
28307 * ipa-inline.c (can_inline_edge_p): Update.
28308 (compute_inlined_call_time): Update.
28309 (want_inline_small_function_p): Update.
28310 (edge_badness): Update.
28311 (early_inliner): Update.
28312 * ipa-inline.h (inline_edge_summary): Turn to ...
28313 (ipa_call_summary): ... this one.
28314 (ipa_call_summary_t): New class.
28315 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
28316 (ipa_call_summaries): New.
28317 (inline_edge_summary): Remove.
28318 (estimate_edge_growth): Update.
28319 * ipa-profile.c (ipa_propagate_frequency_1): Update.
28320 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28321 * ipa-split.c (execute_split_functions): Update.
28322 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
28324 2017-05-23 Tom de Vries <tom@codesourcery.com>
28326 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
28327 attributes): Document rdrand effective target.
28329 2017-05-23 Tom de Vries <tom@codesourcery.com>
28331 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
28332 attributes): Sort alphabetically.
28334 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
28336 * config/avr/genmultilib.awk: Use gsub instead of gensub.
28338 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
28341 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
28342 V2DF/V2DI splat into two separate patterns, one that handles
28343 registers, and the other that only handles memory. Drop support
28344 for splatting from a GPR on ISA 2.07 and then splitting the
28345 splat into direct move and splat.
28346 (vsx_splat_<mode>_reg): Likewise.
28347 (vsx_splat_<mode>_mem): Likewise.
28349 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
28351 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
28353 2017-05-22 Jakub Jelinek <jakub@redhat.com>
28355 PR middle-end/80809
28356 * omp-low.c (finish_taskreg_remap): New function.
28357 (finish_taskreg_scan): If unit size of ctx->record_type
28358 is non-constant, unshare the size expression and replace
28359 decls in it with possible outer var refs.
28361 PR middle-end/80809
28362 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
28363 GOVD_SHARED rather than GOVD_PRIVATE with it.
28364 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
28365 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
28367 PR middle-end/80853
28368 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
28369 as last argument to build_outer_var_ref for pointer bases of array
28370 section reductions.
28372 2017-05-19 Martin Sebor <msebor@redhat.com>
28374 * print-tree.c (print_node): Print DECL_READ_P flag.
28376 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28378 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
28379 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
28380 * cgraph.c: Likewise.
28381 * cgraphunit.c: Likewise.
28382 * gengtype.c: Likewise.
28383 * ipa-cp.c: Likewise.
28384 * ipa-devirt.c: Likewise.
28385 * ipa-icf.c: Likewise.
28386 * ipa-predicate.c: Likewise.
28387 * ipa-profile.c: Likewise.
28388 * ipa-prop.c: Likewise.
28389 * ipa-split.c: Likewise.
28391 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
28392 edge_predicate_pool, dump_inline_hints,
28393 inline_summary::account_size_time, redirect_to_unreachable,
28394 edge_set_predicate, set_hint_predicate,
28395 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
28396 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
28397 inline_summary_t::remove, remap_hint_predicate_after_duplication,
28398 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
28399 ipa_call_summary_t::remove, initialize_growth_caches,
28400 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
28401 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
28402 mark_modified, unmodified_parm_1, unmodified_parm,
28403 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
28404 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
28405 compute_bb_predicates, will_be_nonconstant_expr_predicate,
28406 will_be_nonconstant_predicate, record_modified_bb_info,
28407 get_minimal_bb, record_modified, param_change_prob,
28408 phi_result_unknown_predicate, predicate_for_phi_result,
28409 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
28410 estimate_function_body_sizes, compute_inline_parameters,
28411 compute_inline_parameters_for_curren, pass_data_inline_parameters,
28412 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
28413 inline_update_callee_summaries, remap_edge_change_prob,
28414 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
28415 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
28416 inline_analyze_function, inline_summary_t::insert,
28417 inline_generate_summary, read_ipa_call_summary, inline_read_section,
28418 inline_read_summary, write_ipa_call_summary, inline_write_summary,
28419 inline_free_summary): Move to ipa-fnsummary.h
28420 (predicate_t): Remove.
28421 * ipa-fnsummary.c: New file.
28422 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
28423 (enum inline_hints_vals, inline_hints, agg_position_info,
28424 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
28425 inline_summaries, ipa_call_summary, ipa_call_summary_t,
28426 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
28427 dump_inline_summary, dump_inline_hints, inline_generate_summary,
28428 inline_read_summary, inline_write_summary, inline_free_summary,
28429 inline_analyze_function, initialize_inline_failed,
28430 inline_merge_summary, inline_update_overall_summary,
28431 compute_inline_parameters): Move to ipa-fnsummary.h
28432 * ipa-fnsummary.h: New file.
28433 * ipa-inline-transform.h: Include ipa-inline.h.
28434 * ipa-inline.c: LIkewise.
28436 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28438 * ipa-inline.c (edge_badness): Use inlined_time instead of
28439 inline_summaries->get.
28441 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28443 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
28445 2017-05-22 Nathan Sidwell <nathan@acm.org>
28447 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
28448 (fdump-lang): Document 'raw' option.
28449 * dumpfile.h (TDI_tu): Delete.
28450 * dumpfile.c (dump_files): Remove translation-unit.
28451 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
28453 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
28455 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
28456 command option from $(AWK) call.
28457 * config/avr/genmultilib.awk: Simplify and rewrite so that it
28458 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
28459 [FORMAT]: Remove handling of variable.
28460 * config/avr/t-multilib: Regenerate.
28462 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28464 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
28466 (dump_inline_summary): Do not print self_time.
28467 (estimate_function_body_sizes): Do not set self_time.
28468 (compute_inline_parameters): Likewise.
28469 (inline_read_section, inline_write_summary): Do not stream self_time.
28470 * ipa-inline.h (inline_summary): Drop self_time.
28472 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28474 * ipa-inline-analysis.c (account_size_time): Rename to ...
28475 (inline_summary::account_size_time): ... this one.
28476 (reset_ipa_call_summary): Turn to ...
28477 (ipa_call_summary::reset): ... this one.
28478 (reset_inline_summary): Turn to ...
28479 (inline_summary::reset): ... this one.
28480 (inline_summary_t::remove): Update.
28481 (inline_summary_t::duplicate): Update.
28482 (ipa_call_summary_t::remove): Update.
28483 (dump_inline_summary): Update.
28484 (estimate_function_body_sizes): Update.
28485 (compute_inline_parameters): Update.
28486 (estimate_node_size_and_time): Update.
28487 (inline_merge_summary): Update.
28488 (inline_update_overall_summary): Update.
28489 (inline_read_section): Update.
28490 (inline_write_summary): Update.
28491 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
28492 add account_size_time and reset member functions.
28493 (ipa_call_summary): Add reset function.
28494 * ipa-predicate.h (predicate::operator &): Constify.
28496 2017-05-22 Richard Biener <rguenther@suse.de>
28498 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
28500 2017-05-19 Jason Merrill <jason@redhat.com>
28502 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
28504 2017-05-19 Marek Polacek <polacek@redhat.com>
28507 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
28508 TYPE_OVERFLOW_WRAPS checks.
28510 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
28512 * tree-core.h (enum omp_clause_default_kind): Add
28513 "OMP_CLAUSE_DEFAULT_PRESENT".
28514 * tree-pretty-print.c (dump_omp_clause): Handle it.
28515 * gimplify.c (enum gimplify_omp_var_data): Add
28516 "GOVD_MAP_FORCE_PRESENT".
28517 (gimplify_adjust_omp_clauses_1): Map it to
28518 "GOMP_MAP_FORCE_PRESENT".
28519 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
28521 * gimplify.c (oacc_default_clause): Clarify.
28523 2017-05-19 Nathan Sidwell <nathan@acm.org>
28525 LANG_HOOK_REGISTER_DUMPS
28526 * toplev.c (general_init): Call register dump lang hook.
28527 * doc/invoke.texi: Document -fdump-lang option family.
28528 * dumpfile.c (dump_files): Remove class dump here.
28529 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
28530 * dumpfile.h (tree_dump_index): Remove TDI_class.
28531 * langhooks-def.h (lhd_register_dumps): Declare.
28532 (LANG_HOOKS_REGISTER_DUMPS): Define.
28533 (LANG_HOOKS_INITIALIZER): Add it.
28534 * langhooks.c (lhd_register_dumps): Define.
28535 * langhooks.h (struct lang_hooks): Add register_dumps.
28537 2017-05-19 Nathan Sidwell <nathan@acm.org>
28539 * context.h (context::set_passes): New.
28540 * context.c (context::context): Do not create pass manager.
28541 * toplev.c (general_init): Create pass manager here.
28543 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
28545 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
28546 use this splitter if two add or or instructions would also work for
28547 the constant we want to generate.
28549 2017-05-19 Richard Biener <rguenther@suse.de>
28552 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
28553 predicate evaluation.
28555 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
28557 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
28559 * ipa-inline.c (want_inline_small_function_p): Do not cast to
28562 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
28564 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
28565 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
28566 (inline_edge_summary_vec): Turn into ...
28567 (ipa_call_summaries): ... this one.
28568 (redirect_to_unreachable, edge_set_predicate,
28569 evaluate_properties_for_edge, inline_summary_alloc,
28570 reset_ipa_call_summary, reset_inline_summary,
28571 inline_summary_t::duplicate): Update.
28572 (inline_edge_duplication_hook): Turn to ...
28573 (ipa_call_summary_t::duplicate): ... this one.
28574 (inline_edge_removal_hook): Turn to ...
28575 (ipa_call_summary_t::remove): ... this one.
28576 (dump_inline_edge_summary): Turn to ...
28577 (dump_ipa_call_summary): ... this one.
28578 (estimate_function_body_sizes): Update.
28579 (inline_update_callee_summaries): Update.
28580 (remap_edge_change_prob): Update.
28581 (remap_edge_summaries): Update.
28582 (inline_merge_summary): Update.
28583 (do_estimate_edge_time): Update.
28584 (inline_generate_summary): Update.
28585 (inline_read_section): Update.
28586 (inline_read_summary): Update.
28587 (inline_free_summary): Update.
28588 * ipa-inline.c (can_inline_edge_p): Update.
28589 (compute_inlined_call_time): Update.
28590 (want_inline_small_function_p): Update.
28591 (edge_badness): Update.
28592 (early_inliner): Update.
28593 * ipa-inline.h (inline_edge_summary): Turn to ...
28594 (ipa_call_summary): ... this one.
28595 (ipa_call_summary_t): New class.
28596 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
28597 (ipa_call_summaries): New.
28598 (inline_edge_summary): Remove.
28599 (estimate_edge_growth): Update.
28600 * ipa-profile.c (ipa_propagate_frequency_1): Update.
28601 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28602 * ipa-split.c (execute_split_functions): Update.
28603 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
28605 2017-05-19 Richard Biener <rguenther@suse.de>
28607 PR middle-end/80764
28608 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
28610 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
28612 * config/rs6000/rs6000.c (struct machine_function): Add field
28613 fpr_is_wrapped_separately.
28614 (rs6000_get_separate_components): Use 64 components. Handle the
28615 new FPR components.
28616 (rs6000_components_for_bb): Handle the FPR components.
28617 (rs6000_emit_prologue_components): Handle the FPR components.
28618 (rs6000_emit_epilogue_components): Handle the FPR components.
28619 (rs6000_set_handled_components): Handle the FPR components.
28620 (rs6000_emit_prologue): Don't output prologue code for those FPRs
28621 that are already separately shrink-wrapped.
28622 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
28623 that are already separately shrink-wrapped.
28625 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
28628 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
28631 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
28632 (define_peephole2 for Altivec d-form load): Add peepholes to catch
28633 cases where the register allocator uses a move and an offsettable
28634 memory operation to/from a FPR register on ISA 2.06/2.07.
28635 (define_peephole2 for Altivec d-form store): Likewise.
28637 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
28640 * config/i386/mmx.md (*mov<mode>_internal): Enable
28641 alternatives 11, 12, 13 and 14 also for 32bit targets.
28642 Remove alternatives 15, 16, 17 and 18.
28643 * config/i386/sse.md (vec_concatv2di): Change
28644 alternative (!x, *y) to (x, ?!*Yn).
28646 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
28648 * dumpfile.h (enum dump_kind): Remove stray comma.
28650 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
28652 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
28653 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
28654 predicate::num_conditions
28655 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
28656 (CHANGED): turn into predicate::changed.
28657 (agg_position_info): Move to ipa-predicate.h
28658 (add_condition, predicate::add_clause, predicate::operator &=,
28659 predicate::or_with, predicate::evaluate, predicate::probability,
28660 dump_condition, dump_clause, predicate::dump,
28661 predicate::remap_after_duplication, predicate::remap_after_inlining,
28662 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
28663 (evaluate_conditions_for_known_args): Update.
28664 (set_cond_stmt_execution_predicate): Update.
28665 * ipa-inline.h: Include ipa-predicate.h
28666 (condition, inline_param_summary, conditions, agg_position_info,
28667 predicate): Move to ipa-predicate.h
28668 * ipa-predicate.c: New file.
28669 * ipa-predicate.h: New file.
28671 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
28673 * final.c (leaf_function_p): Check we are not in a sequence.
28675 2017-05-18 Martin Liska <mliska@suse.cz>
28677 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
28678 * dumpfile.c (dump_register): Use new enum dump_kind.
28679 (get_dump_file_name): Likewise.
28680 (dump_enable_all): Likewise.
28681 (dump_switch_p_1): Likewise.
28682 (enable_rtl_dump_file): Remove usage of TDF_RTL.
28683 * dumpfile.h (enum dump_kind): New enum type.
28684 (struct dump_file_info): Create constructor and
28685 format fields and comments.
28686 * passes.c (pass_manager::register_one_dump_file):
28688 * statistics.c (statistics_early_init): Likewise.
28689 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
28690 TDF_TREE with TDF_SLIM.
28691 (gather_memory_references_ref): Likewise.
28693 2017-05-18 Martin Liska <mliska@suse.cz>
28695 * vec.h (struct vnull): Use it.
28697 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
28699 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
28700 (true_predicate, false_predicate, true_predicate_p,
28701 false_predicate_p): Remove.
28702 (single_cond_predicate, not_inlined_predicate): Turn to member function
28704 (add_condition): Update.
28705 (add_clause): Turn to...
28706 (predicate::add_clause): ... this one; update; allow passing NULL
28708 (and_predicates): Turn to ...
28709 (predicate::operator &=): ... this one.
28710 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
28711 (or_predicates): Turn to ...
28712 (predicate::or_with): ... this one.
28713 (evaluate_predicate): Turn to ...
28714 (predicate::evaluate): ... this one.
28715 (predicate_probability): Turn to ...
28716 (predicate::probability): ... this one.
28717 (dump_condition): Update.
28718 (dump_predicate): Turn to ...
28719 (predicate::dump): ... this one.
28720 (account_size_time): Update.
28721 (edge_set_predicate): Update.
28722 (set_hint_predicate): UPdate.
28723 (evaluate_conditions_for_known_args): Update.
28724 (evaluate_properties_for_edge): Update.
28725 (remap_predicate_after_duplication): Turn to...
28726 (predicate::remap_after_duplication): ... this one.
28727 (remap_hint_predicate_after_duplication): Update.
28728 (inline_summary_t::duplicate): UPdate.
28729 (dump_inline_edge_summary): Update.
28730 (dump_inline_summary): Update.
28731 (set_cond_stmt_execution_predicate): Update.
28732 (set_switch_stmt_execution_predicate): Update.
28733 (compute_bb_predicates): Update.
28734 (will_be_nonconstant_expr_predicate): Update.
28735 (will_be_nonconstant_predicate): Update.
28736 (phi_result_unknown_predicate): Update.
28737 (predicate_for_phi_result): Update.
28738 (array_index_predicate): Update.
28739 (estimate_function_body_sizes): Update.
28740 (estimate_node_size_and_time): Update.
28741 (estimate_ipcp_clone_size_and_time): Update.
28742 (remap_predicate): Rename to ...
28743 (predicate::remap_after_inlining): ... this one.
28744 (remap_hint_predicate): Update.
28745 (inline_merge_summary): Update.
28746 (inline_update_overall_summary): Update.
28747 (estimate_size_after_inlining): Update.
28748 (read_predicate): Rename to ...
28749 (predicate::stream_in): ... this one.
28750 (read_inline_edge_summary): Update.
28751 (write_predicate): Rename to ...
28752 (predicate::stream_out): ... this one.
28753 (write_inline_edge_summary): Update.
28754 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
28755 (clause_t): Turn to uint32_t
28756 (predicate): Turn to class; implement constructor and operators
28758 (size_time_entry): Update.
28759 (inline_summary): Update.
28760 (inline_edge_summary): Update.
28762 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
28764 * fold-const.c (fold_binary_loc): Move transformation...
28765 * match.pd (C - X CMP X): ... here.
28767 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
28769 * config/sparc/sparc.c (sparc_option_override): Set function
28770 alignment for -mcpu=niagara7 to 64 to match the I$ line.
28771 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
28773 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
28775 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
28777 2017-05-18 Marek Polacek <polacek@redhat.com>
28780 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
28781 (pass_ubsan::execute): Call gimple_assign_single_p instead of
28782 gimple_assign_load_p.
28784 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
28786 PR middle-end/80692
28787 * real.c (do_compare): Give decimal_do_compare preference over
28788 comparing just the signs.
28790 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
28792 * doc/md.texi (Canonicalization of Instructions): Describe the
28793 canonical form of instructions that inherently set a condition
28796 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
28798 PR middle-end/80775
28799 * tree-cfg.c: Move deletion of unreachable case statements to after
28800 the merging of consecutive case labels.
28802 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
28804 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
28805 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
28806 restoring of callee-saved registers.
28808 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
28810 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
28811 * config/visium/visium.c (single_set_and_flags): Likewise.
28812 * config/visium/visium.md (Substitutions): Likewise.
28814 2017-05-17 Martin Liska <mliska@suse.cz>
28816 * cfg.c: Introduce dump_flags_t type and
28817 use it instead of int type.
28819 * cfghooks.c: Likewise.
28820 * cfghooks.h (struct cfg_hooks): Likewise.
28821 * cfgrtl.c: Likewise.
28822 * cfgrtl.h: Likewise.
28823 * cgraph.c (cgraph_node::get_body): Likewise.
28824 * coretypes.h: Likewise.
28825 * domwalk.c: Likewise.
28826 * domwalk.h: Likewise.
28827 * dumpfile.c (struct dump_option_value_info): Likewise.
28828 (dump_enable_all): Likewise.
28829 (dump_switch_p_1): Likewise.
28830 (opt_info_switch_p): Likewise.
28831 * dumpfile.h (enum tree_dump_index): Likewise.
28832 (struct dump_file_info): Likewise.
28833 * genemit.c: Likewise.
28834 * generic-match-head.c: Likewise.
28835 * gengtype.c (open_base_files): Likewise.
28836 * gimple-pretty-print.c: Likewise.
28837 * gimple-pretty-print.h: Likewise.
28838 * graph.c (print_graph_cfg): Likewise.
28839 * graphite-scop-detection.c (dot_all_sese): Likewise.
28840 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
28841 * loop-unroll.c (report_unroll): Likewise.
28842 * passes.c (pass_manager::register_one_dump_file): Likewise.
28843 * print-tree.c: Likewise.
28844 * statistics.c: Likewise.
28845 * tree-cfg.c: Likewise.
28846 * tree-cfg.h: Likewise.
28847 * tree-dfa.c: Likewise.
28848 * tree-dfa.h: Likewise.
28849 * tree-dump.c (dump_function): Likewise.
28850 * tree-dump.h (struct dump_info): Likewise.
28851 * tree-pretty-print.c: Likewise.
28852 * tree-pretty-print.h: Likewise.
28853 * tree-ssa-live.c: Likewise.
28854 * tree-ssa-live.h: Likewise.
28855 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
28856 * tree-vect-loop.c: Likewise.
28857 * tree-vect-slp.c: Likewise.
28859 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
28860 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28862 PR tree-optimization/80457
28863 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
28864 of all arguments to a statement as scalar_to_vec operations.
28865 (vectorizable_call): Adjust call to vect_model_simple_cost for
28867 (vectorizable_conversion): Likewise.
28868 (vectorizable_assignment): Likewise.
28869 (vectorizable_shift): Likewise.
28870 (vectorizable_operation): Likewise.
28871 (vectorizable_comparison): Likewise.
28872 (vect_is_simple_cond): Record the def types for operands.
28873 (vectorizable_condition): Likewise, call vect_model_simple_cost.
28874 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
28875 for statement argument count.
28877 2017-05-16 Carl Love <cel@us.ibm.com>
28879 * config/rs6000/rs6000-c: Add support for built-in functions
28880 vector unsigned long long vec_bperm (vector unsigned long long,
28881 vector unsigned char)
28882 vector signed long long vec_mule (vector signed int,
28884 vector unsigned long long vec_mule (vector unsigned int,
28885 vector unsigned int)
28886 vector signed long long vec_mulo (vector signed int,
28888 vector unsigned long long vec_mulo (vector unsigned int,
28889 vector unsigned int)
28890 vector signed char vec_sldw (vector signed char,
28891 vector signed char,
28893 vector unsigned char vec_sldw (vector unsigned char,
28894 vector unsigned char,
28896 vector signed short vec_sldw (vector signed short,
28897 vector signed short,
28899 vector unsigned short vec_sldw (vector unsigned short,
28900 vector unsigned short,
28902 vector signed int vec_sldw (vector signed int,
28905 vector unsigned int vec_sldw (vector unsigned int,
28906 vector unsigned int,
28908 vector signed long long vec_sldw (vector signed long long,
28909 vector signed long long,
28911 vector unsigned long long vec_sldw (vector unsigned long long,
28912 vector unsigned long long,
28914 * config/rs6000/rs6000-c: Add support for built-in functions
28915 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
28916 * config/rs6000/altivec.h: Add defintion for vec_sldw.
28917 * doc/extend.texi: Update the built-in documentation for the
28918 new built-in functions.
28920 2017-05-16 Marek Polacek <polacek@redhat.com>
28924 * tree.c (save_expr): Don't fold the expression.
28926 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
28928 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
28929 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
28930 and (?*y,m). Update insn attributes.
28932 2017-05-16 Martin Liska <mliska@suse.cz>
28934 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
28935 flags argument of print_gimple_stmt, print_gimple_expr,
28936 print_generic_stmt and print_generic_expr.
28937 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
28938 * coretypes.h: Likewise.
28939 * except.c (dump_eh_tree): Likewise.
28940 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
28941 * gimple-pretty-print.h: Likewise.
28942 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
28943 (backprop::push_to_worklist): Likewise.
28944 (backprop::pop_from_worklist): Likewise.
28945 (backprop::process_use): Likewise.
28946 (backprop::intersect_uses): Likewise.
28947 (note_replacement): Likewise.
28948 * gimple-ssa-store-merging.c
28949 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
28950 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
28951 (pass_store_merging::execute): Likewise.
28952 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
28953 (ssa_base_cand_dump_callback): Likewise.
28954 (dump_incr_vec): Likewise.
28955 (replace_refs): Likewise.
28956 (replace_mult_candidate): Likewise.
28957 (create_add_on_incoming_edge): Likewise.
28958 (create_phi_basis): Likewise.
28959 (insert_initializers): Likewise.
28960 (all_phi_incrs_profitable): Likewise.
28961 (introduce_cast_before_cand): Likewise.
28962 (replace_one_candidate): Likewise.
28963 * gimplify.c (gimplify_expr): Likewise.
28964 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
28965 (set_rename): Likewise.
28966 (rename_uses): Likewise.
28967 (copy_loop_phi_nodes): Likewise.
28968 (add_close_phis_to_merge_points): Likewise.
28969 (copy_loop_close_phi_args): Likewise.
28970 (copy_cond_phi_args): Likewise.
28971 (graphite_copy_stmts_from_block): Likewise.
28972 (translate_pending_phi_nodes): Likewise.
28973 * graphite-poly.c (print_pdr): Likewise.
28974 (dump_gbb_cases): Likewise.
28975 (dump_gbb_conditions): Likewise.
28976 (print_scop_params): Likewise.
28977 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
28978 (build_cross_bb_scalars_use): Likewise.
28979 (gather_bbs::before_dom_children): Likewise.
28980 * hsa-dump.c (dump_hsa_immed): Likewise.
28981 * ipa-cp.c (print_ipcp_constant_value): Likewise.
28982 (get_replacement_map): Likewise.
28983 * ipa-inline-analysis.c (dump_condition): Likewise.
28984 (estimate_function_body_sizes): Likewise.
28985 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
28986 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
28987 * ipa-prop.c (ipa_dump_param): Likewise.
28988 (ipa_print_node_jump_functions_for_edge): Likewise.
28989 (ipa_modify_call_arguments): Likewise.
28990 (ipa_modify_expr): Likewise.
28991 (ipa_dump_param_adjustments): Likewise.
28992 (ipa_dump_agg_replacement_values): Likewise.
28993 (ipcp_modif_dom_walker::before_dom_children): Likewise.
28994 * ipa-pure-const.c (check_stmt): Likewise.
28995 (pass_nothrow::execute): Likewise.
28996 * ipa-split.c (execute_split_functions): Likewise.
28997 * omp-offload.c (dump_oacc_loop_part): Likewise.
28998 (dump_oacc_loop): Likewise.
28999 * trans-mem.c (tm_log_emit): Likewise.
29000 (tm_memopt_accumulate_memops): Likewise.
29001 (dump_tm_memopt_set): Likewise.
29002 (dump_tm_memopt_transform): Likewise.
29003 * tree-cfg.c (gimple_verify_flow_info): Likewise.
29004 (print_loop): Likewise.
29005 * tree-chkp-opt.c (chkp_print_addr): Likewise.
29006 (chkp_gather_checks_info): Likewise.
29007 (chkp_get_check_result): Likewise.
29008 (chkp_remove_check_if_pass): Likewise.
29009 (chkp_use_outer_bounds_if_possible): Likewise.
29010 (chkp_reduce_bounds_lifetime): Likewise.
29011 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
29012 (chkp_mark_completed_bounds): Likewise.
29013 (chkp_register_incomplete_bounds): Likewise.
29014 (chkp_mark_invalid_bounds): Likewise.
29015 (chkp_maybe_copy_and_register_bounds): Likewise.
29016 (chkp_build_returned_bound): Likewise.
29017 (chkp_get_bound_for_parm): Likewise.
29018 (chkp_build_bndldx): Likewise.
29019 (chkp_get_bounds_by_definition): Likewise.
29020 (chkp_generate_extern_var_bounds): Likewise.
29021 (chkp_get_bounds_for_decl_addr): Likewise.
29022 * tree-chrec.c (chrec_apply): Likewise.
29023 * tree-data-ref.c (dump_data_reference): Likewise.
29024 (dump_subscript): Likewise.
29025 (dump_data_dependence_relation): Likewise.
29026 (analyze_overlapping_iterations): Likewise.
29027 * tree-inline.c (expand_call_inline): Likewise.
29028 (tree_function_versioning): Likewise.
29029 * tree-into-ssa.c (dump_defs_stack): Likewise.
29030 (dump_currdefs): Likewise.
29031 (dump_names_replaced_by): Likewise.
29032 (dump_update_ssa): Likewise.
29033 (update_ssa): Likewise.
29034 * tree-object-size.c (pass_object_sizes::execute): Likewise.
29035 * tree-parloops.c (build_new_reduction): Likewise.
29036 (try_create_reduction_list): Likewise.
29037 (ref_conflicts_with_region): Likewise.
29038 (oacc_entry_exit_ok_1): Likewise.
29039 (oacc_entry_exit_single_gang): Likewise.
29040 * tree-pretty-print.h: Likewise.
29041 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
29042 (get_scalar_evolution): Likewise.
29043 (add_to_evolution): Likewise.
29044 (get_loop_exit_condition): Likewise.
29045 (analyze_evolution_in_loop): Likewise.
29046 (analyze_initial_condition): Likewise.
29047 (analyze_scalar_evolution): Likewise.
29048 (instantiate_scev): Likewise.
29049 (number_of_latch_executions): Likewise.
29050 (gather_chrec_stats): Likewise.
29051 (final_value_replacement_loop): Likewise.
29052 (scev_const_prop): Likewise.
29053 * tree-sra.c (dump_access): Likewise.
29054 (disqualify_candidate): Likewise.
29055 (create_access): Likewise.
29056 (reject): Likewise.
29057 (maybe_add_sra_candidate): Likewise.
29058 (create_access_replacement): Likewise.
29059 (analyze_access_subtree): Likewise.
29060 (analyze_all_variable_accesses): Likewise.
29061 (sra_modify_assign): Likewise.
29062 (initialize_constant_pool_replacements): Likewise.
29063 (find_param_candidates): Likewise.
29064 (decide_one_param_reduction): Likewise.
29065 (replace_removed_params_ssa_names): Likewise.
29066 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
29067 * tree-ssa-copy.c (dump_copy_of): Likewise.
29068 (copy_prop_visit_cond_stmt): Likewise.
29069 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
29070 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
29071 (record_equivalences_from_stmt): Likewise.
29072 * tree-ssa-dse.c (compute_trims): Likewise.
29073 (delete_dead_call): Likewise.
29074 (delete_dead_assignment): Likewise.
29075 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
29076 (forward_propagate_into_cond): Likewise.
29077 (pass_forwprop::execute): Likewise.
29078 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
29079 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
29081 (move_computations_worker): Likewise.
29082 (execute_sm): Likewise.
29083 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
29084 (remove_exits_and_undefined_stmts): Likewise.
29085 (remove_redundant_iv_tests): Likewise.
29086 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
29087 (adjust_iv_update_pos): Likewise.
29088 * tree-ssa-math-opts.c (bswap_replace): Likewise.
29089 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
29090 (value_replacement): Likewise.
29091 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
29092 * tree-ssa-pre.c (print_pre_expr): Likewise.
29093 (get_representative_for): Likewise.
29094 (create_expression_by_pieces): Likewise.
29095 (insert_into_preds_of_block): Likewise.
29096 (eliminate_insert): Likewise.
29097 (eliminate_dom_walker::before_dom_children): Likewise.
29098 (eliminate): Likewise.
29099 (remove_dead_inserted_code): Likewise.
29100 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
29101 * tree-ssa-reassoc.c (get_rank): Likewise.
29102 (eliminate_duplicate_pair): Likewise.
29103 (eliminate_plus_minus_pair): Likewise.
29104 (eliminate_not_pairs): Likewise.
29105 (undistribute_ops_list): Likewise.
29106 (eliminate_redundant_comparison): Likewise.
29107 (update_range_test): Likewise.
29108 (optimize_range_tests_var_bound): Likewise.
29109 (optimize_vec_cond_expr): Likewise.
29110 (rewrite_expr_tree): Likewise.
29111 (rewrite_expr_tree_parallel): Likewise.
29112 (linearize_expr): Likewise.
29113 (break_up_subtract): Likewise.
29114 (linearize_expr_tree): Likewise.
29115 (attempt_builtin_powi): Likewise.
29116 (attempt_builtin_copysign): Likewise.
29117 (transform_stmt_to_copy): Likewise.
29118 (transform_stmt_to_multiply): Likewise.
29119 (dump_ops_vector): Likewise.
29120 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
29121 (print_scc): Likewise.
29122 (set_ssa_val_to): Likewise.
29123 (visit_reference_op_store): Likewise.
29124 (visit_use): Likewise.
29125 (sccvn_dom_walker::before_dom_children): Likewise.
29126 (run_scc_vn): Likewise.
29127 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
29129 (expr_hash_elt::print): Likewise.
29130 (const_and_copies::pop_to_marker): Likewise.
29131 (const_and_copies::record_const_or_copy_raw): Likewise.
29132 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
29133 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
29134 (dump_predicates): Likewise.
29135 (find_uninit_use): Likewise.
29136 (warn_uninitialized_phi): Likewise.
29137 (pass_late_warn_uninitialized::execute): Likewise.
29138 * tree-ssa.c (verify_vssa): Likewise.
29139 (verify_ssa): Likewise.
29140 (maybe_optimize_var): Likewise.
29141 * tree-vrp.c (dump_value_range): Likewise.
29142 (dump_all_value_ranges): Likewise.
29143 (dump_asserts_for): Likewise.
29144 (register_edge_assert_for_2): Likewise.
29145 (vrp_visit_cond_stmt): Likewise.
29146 (vrp_visit_switch_stmt): Likewise.
29147 (vrp_visit_stmt): Likewise.
29148 (vrp_visit_phi_node): Likewise.
29149 (simplify_cond_using_ranges_1): Likewise.
29150 (fold_predicate_in): Likewise.
29151 (evrp_dom_walker::before_dom_children): Likewise.
29152 (evrp_dom_walker::push_value_range): Likewise.
29153 (evrp_dom_walker::pop_value_range): Likewise.
29154 (execute_early_vrp): Likewise.
29156 2017-05-16 Richard Biener <rguenther@suse.de>
29158 * dwarf2out.c (loc_list_from_tree_1): Do not create
29159 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
29161 2017-05-16 Richard Biener <rguenther@suse.de>
29163 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
29165 (note_variable_value_in_expr): If we resolved the decl ref
29166 do not push to the stack.
29168 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
29170 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
29171 operations in fast-math mode.
29172 (vaddq_f16): Likewise.
29173 (vmul_f16): Likewise.
29174 (vmulq_f16): Likewise.
29175 (vsub_f16): Likewise.
29176 (vsubq_f16): Likewise.
29177 * config/arm/neon.md (add<mode>3): New.
29179 (fma:<VH:mode>3): New. Also remove outdated comment.
29182 2017-05-16 Martin Liska <mliska@suse.cz>
29186 * ipa-devirt.c (warn_types_mismatch): Fix typo.
29187 (odr_types_equivalent_p): Likewise.
29189 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
29191 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
29193 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
29196 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
29197 non-interunit SSE move alternatives with '?'.
29198 (zero-extendsidi peephole2): New peephole to skip intermediate
29199 general register in SSE zero-extend sequence.
29201 2017-05-15 Jeff Law <law@redhat.com>
29203 * reorg.c (relax_delay_slots): Create a new variable to hold
29204 the temporary target rather than clobbering TARGET_LABEL.
29206 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
29207 missing argument to extract_bit_field call.
29208 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
29210 2017-05-15 Martin Liska <mliska@suse.cz>
29213 * gcc.c (process_command): Do not allow empty argument of -o option.
29215 2017-05-15 Renlin Li <renlin.li@arm.com>
29217 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
29218 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
29219 * config/aarch64/constraints.md (Usf): Add long call check.
29220 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
29221 (call_value): Likewise.
29222 (sibcall): Likewise.
29223 (sibcall_value): Likewise.
29225 (call_value_insn): New.
29226 (sibcall_insn): Update rtx pattern.
29227 (sibcall_value_insn): Likewise.
29228 (call_internal): Remove.
29229 (call_value_internal): Likewise.
29230 (sibcall_internal): Likewise.
29231 (sibcall_value_internal): Likewise.
29232 (call_reg): Likewise.
29233 (call_symbol): Likewise.
29234 (call_value_reg): Likewise.
29235 (call_value_symbol): Likewise.
29237 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
29240 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
29242 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
29244 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
29245 compatible with CCGOCmode and with CCZmode.
29247 2017-05-14 Martin Sebor <msebor@redhat.com>
29249 PR middle-end/77671
29250 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
29251 (gimple_fold_builtin_snprintf): Same.
29252 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
29253 (gimple_fold_builtin_snprintf): Same.
29254 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
29255 of character types.
29256 (is_call_safe): New function.
29257 (try_substitute_return_value): Call it.
29258 (try_simplify_call): New function.
29259 (pass_sprintf_length::handle_gimple_call): Call it.
29261 2017-05-14 Martin Sebor <msebor@redhat.com>
29263 PR middle-end/80669
29264 * builtins.c (expand_builtin_stpncpy): Simplify.
29266 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
29268 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
29269 * config/i386/i386.h
29270 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
29271 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
29272 (struct machine_function): Add new members call_ms2sysv,
29273 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
29274 (struct machine_frame_state): New fields sp_realigned and
29275 sp_realigned_offset.
29276 * config/i386/i386.c
29277 (enum xlogue_stub): New enum.
29278 (enum xlogue_stub_sets): New enum.
29279 (class xlogue_layout): New class.
29280 (struct ix86_frame): New fields stack_realign_allocate_offset,
29281 stack_realign_offset and outlined_save_offset. Modify comments to
29282 detail stack layout when using out-of-line stubs.
29283 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
29284 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
29285 -mcall-ms2sysv-xlogues.
29286 (stub_managed_regs): New static variable.
29287 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
29288 registers managed by out-of-line stub.
29289 (disable_call_ms2sysv_xlogues): New function.
29290 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
29291 m->call_ms2sysv when appropriate and compute frame layout for
29293 (sp_valid_at, fp_valid_at): New inline functions.
29294 (choose_basereg): New function.
29295 (choose_baseaddr): Add align parameter, use choose_basereg and modify
29297 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
29298 Use align parameter of choose_baseaddr to generated aligned SSE movs
29300 (pro_epilogue_adjust_stack): Modify to track
29301 machine_frame_state::sp_realigned.
29302 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
29303 (ix86_nsaved_sseregs): Likewise.
29304 (ix86_emit_save_regs): Likewise.
29305 (ix86_emit_save_regs_using_mov): Likewise.
29306 (ix86_emit_save_sse_regs_using_mov): Likewise.
29307 (get_scratch_register_on_entry): Likewise.
29308 (gen_frame_set): New function.
29309 (gen_frame_load): Likewise.
29310 (gen_frame_store): Likewise.
29311 (emit_outlined_ms2sysv_save): Likewise.
29312 (emit_outlined_ms2sysv_restore): Likewise.
29313 (ix86_expand_prologue): Modify stack re-alignment code and call
29314 emit_outlined_ms2sysv_save when appropriate.
29315 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
29316 parameter rtx_insn *insn, which allows the function to be used to only
29317 generate the notes.
29318 (ix86_expand_epilogue): Modify validity checks of frame and stack
29319 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
29320 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
29321 * config/i386/predicates.md
29322 (save_multiple): New predicate.
29323 (restore_multiple): Likewise.
29324 * config/i386/sse.md
29325 (save_multiple<mode>): New pattern.
29326 (save_multiple_realign<mode>): Likewise.
29327 (restore_multiple<mode>): Likewise.
29328 (restore_multiple_and_return<mode>): Likewise.
29329 (restore_multiple_leave_return<mode>): Likewise.
29330 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
29332 2017-05-14 Julia Koval <julia.koval@intel.com>
29334 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
29335 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
29336 (__builtin_ia32_xsetbv): New builtins.
29337 * config/i386/i386.c (ix86_expand_special_args_builtin):
29339 (ix86_expand_builtin): Special expand for new intrinsics.
29340 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
29341 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
29342 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
29344 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29346 * cfganal.c (inverted_post_order_compute): Change argument type
29348 * cfganal.h (inverted_post_order_compute): Adjust prototype.
29349 * df-core.c (rest_of_handle_df_initialize): Adjust.
29350 (rest_of_handle_df_finish): Likewise.
29351 (df_analyze_1): Likewise.
29352 (df_analyze): Likewise.
29353 (loop_inverted_post_order_compute): Change argument to be a vec *.
29354 (df_analyze_loop): Adjust.
29355 (df_get_n_blocks): Likewise.
29356 (df_get_postorder): Likewise.
29357 * df.h (struct df_d): Change field to be a vec.
29358 * lcm.c (compute_laterin): Adjust.
29359 (compute_available): Likewise.
29360 * lra-lives.c (lra_create_live_ranges_1): Likewise.
29361 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
29362 * tree-ssa-pre.c (compute_antic): Likewise.
29364 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29366 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
29367 (depth_first_search::depth_first_search): Change structure init
29368 function to this constructor.
29369 (depth_first_search::add_bb): Rename function to this member.
29370 (depth_first_search::execute): Likewise.
29371 (flow_dfs_compute_reverse_finish): Adjust.
29373 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29375 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
29376 (longest_simple_path): Likewise.
29377 * shrink-wrap.c (spread_components): Likewise.
29378 (disqualify_problematic_components): Likewise.
29379 (emit_common_heads_for_components): Likewise.
29380 (emit_common_tails_for_components): Likewise.
29381 (insert_prologue_epilogue_for_components): Likewise.
29383 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29385 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
29388 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29390 * df-core.c (df_set_blocks): Start using auto_bitmap.
29391 (df_compact_blocks): Likewise.
29392 * df-problems.c (df_rd_confluence_n): Likewise.
29393 * df-scan.c (df_insn_rescan_all): Likewise.
29394 (df_process_deferred_rescans): Likewise.
29395 (df_update_entry_block_defs): Likewise.
29396 (df_update_exit_block_uses): Likewise.
29397 (df_entry_block_bitmap_verify): Likewise.
29398 (df_exit_block_bitmap_verify): Likewise.
29399 (df_scan_verify): Likewise.
29400 * lra-constraints.c (lra_constraints): Likewise.
29401 (undo_optional_reloads): Likewise.
29402 (lra_undo_inheritance): Likewise.
29403 * lra-remat.c (calculate_gen_cands): Likewise.
29404 (do_remat): Likewise.
29405 * lra-spills.c (assign_spill_hard_regs): Likewise.
29406 (spill_pseudos): Likewise.
29407 * tree-ssa-pre.c (bitmap_set_and): Likewise.
29408 (bitmap_set_subtract_values): Likewise.
29410 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29412 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
29413 management with auto_bitmap.
29414 (fix_inter_tick): Likewise.
29415 (fix_recovery_deps): Likewise.
29416 * ira.c (add_store_equivs): Likewise.
29417 (find_moveable_pseudos): Likewise.
29418 (split_live_ranges_for_shrink_wrap): Likewise.
29419 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
29420 (rtx_reuse_manager::seen_def_p): Likewise.
29421 (rtx_reuse_manager::set_seen_def): Likewise.
29422 * print-rtl.h (class rtx_reuse_manager): Likewise.
29424 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29426 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
29428 (migrate_btr_def): Likewise.
29429 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
29430 * df-core.c (loop_post_order_compute): Likewise.
29431 (loop_inverted_post_order_compute): Likewise.
29432 * hsa-common.h: Likewise.
29433 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
29434 * init-regs.c (initialize_uninitialized_regs): Likewise.
29435 * ipa-inline.c (resolve_noninline_speculation): Likewise.
29436 (inline_small_functions): Likewise.
29437 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
29438 * ira.c (combine_and_move_insns): Likewise.
29439 (build_insn_chain): Likewise.
29440 * loop-invariant.c (find_invariants): Likewise.
29441 * lower-subreg.c (propagate_pseudo_copies): Likewise.
29442 * predict.c (tree_predict_by_opcode): Likewise.
29443 (predict_paths_leading_to): Likewise.
29444 (predict_paths_leading_to_edge): Likewise.
29445 (estimate_loops_at_level): Likewise.
29446 (estimate_loops): Likewise.
29447 * shrink-wrap.c (try_shrink_wrapping): Likewise.
29448 (spread_components): Likewise.
29449 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
29450 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
29451 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
29452 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
29453 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
29454 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
29455 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
29456 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
29457 (mark_threaded_blocks): Likewise.
29458 (thread_through_all_blocks): Likewise.
29459 * tree-ssa.c (verify_ssa): Likewise.
29460 (execute_update_addresses_taken): Likewise.
29461 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
29463 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29465 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
29467 (post_order_compute): Likewise.
29468 (inverted_post_order_compute): Likewise.
29469 (pre_and_rev_post_order_compute_fn): Likewise.
29471 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29473 * genrecog.c (int_set::int_set): Explicitly construct our
29474 auto_vec base class.
29475 * vec.h (auto_vec::auto_vec): New constructor.
29477 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29479 * bitmap.h (class auto_bitmap): New constructor taking
29480 bitmap_obstack * argument.
29482 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29484 * bitmap.h (class auto_bitmap): Change type of m_bits to
29485 bitmap_head, and adjust ctor / dtor and member operators.
29487 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
29489 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
29490 when returned register mode doesn't match original mode.
29492 2017-05-12 Jeff Law <law@redhat.com>
29493 Jakub Jelinek <jakub@redhat.com>
29495 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
29496 we look for cc setter after the compare-elim changes.
29497 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
29498 within the vector to match what compare-elim now expects.
29499 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
29500 (xorsi3_flags, one_cmplsi2_flags): Likewise.
29502 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
29503 after the compare-elim changes.
29504 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
29505 the vector to match what compare-elim now expects.
29506 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
29507 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
29508 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
29509 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
29510 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
29512 * config/visium/visium.c (single_set_and_flags): Fix where
29513 we look for cc setter after the compare-elim changes.
29514 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
29515 with the vector to match what compare-elim now expects.
29516 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
29517 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
29518 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
29519 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
29520 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
29521 (neg<mode>2_insn_set_overflow): Likewise.
29523 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
29525 PR middle-end/79794
29526 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
29527 maybe_expand_insn call, set ops[0].target. If still set after call,
29528 set alt_rtl. Add extra arg to recursive calls.
29529 (extract_bit_field): Add alt_rtl argument. Pass to
29531 * expmed.h (extract_bit_field): Fix prototype.
29532 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
29533 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
29534 to extract_bit_field_calls.
29535 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
29536 Pass alt_rtl to extract_bit_field calls.
29537 * calls.c (store_unaligned_arguments_into_psuedos)
29538 load_register_parameters): Pass extra NULL to extract_bit_field calls.
29539 * optabs.c (maybe_legitimize_operand): Clear op->target when call
29541 * optabs.h (struct expand_operand): Add target bitfield.
29543 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
29545 * compare-elim.c (try_eliminate_compare): Canonicalize
29546 operation with embedded compare to
29547 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
29548 (set (reg) (operation)].
29550 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
29552 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
29555 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
29556 cost of adding a carry flag for ADC instruction.
29557 [case MINUS]: Ignore the cost of subtracting a carry flag
29558 for SBB instruction.
29560 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
29562 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
29564 * config/rs6000/bmiintrin.h: New file.
29565 * config/rs6000/bmi2intrin.h: New file.
29566 * config/rs6000/x86intrin.h: New file.
29568 2017-05-12 Jeff Law <law@redhat.com>
29570 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
29573 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
29575 PR middle-end/80707
29576 * tree-cfg.c: Remove cfg edges of unreachable case statements.
29578 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29580 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
29581 early expansion of vector divide builtins.
29582 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
29583 builtins identified as having unsigned arguments.
29585 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29587 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
29588 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
29589 expansion of vector logical operations (and, andc, or, xor,
29592 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29594 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
29595 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
29597 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29599 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
29600 early GIMPLE expansion of vector multiplies.
29602 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29604 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
29605 TARGET_HAVE_MOVT conditional.
29606 (movt splitter): Likewise.
29608 2017-05-12 Richard Biener <rguenther@suse.de>
29610 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
29611 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
29612 Fold all stmts not inplace.
29614 2017-05-12 Richard Biener <rguenther@suse.de>
29616 PR tree-optimization/80713
29617 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
29618 inserted_exprs bit for not removed stmts.
29620 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
29622 PR middle-end/69921
29623 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
29624 parallelized" attribute for parallelized OpenACC kernels.
29625 * omp-offload.c (execute_oacc_device_lower): Use it.
29627 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
29628 Set "oacc kernels" attribute.
29629 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
29630 parameter. Adjust all users.
29631 (oacc_fn_attrib_kernels_p): Remove function.
29632 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
29633 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
29634 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
29635 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
29636 assert "oacc kernels" attribute is set.
29638 2017-05-11 Carl Love <cel@us.ibm.com>
29640 * config/rs6000/rs6000-c: Add support for built-in functions
29641 vector unsigned char vec_popcnt (vector signed char)
29642 vector unsigned char vec_popcnt (vector unsigned char)
29643 vector unsigned short vec_popcnt (vector signed short)
29644 vector unsigned short vec_popcnt (vector unsigned short)
29645 vector unsigned int vec_popcnt (vector signed int)
29646 vector unsigned int vec_popcnt (vector unsigned int)
29647 vector unsigned long long vec_popcnt (vector signed long long)
29648 vector unsigned long long vec_popcnt (vector unsigned long long)
29649 vector signed long long vec_slo (vector signed long long,
29650 vector signed char)
29651 vector signed long long vec_slo (vector signed long long,
29652 vector unsigned char)
29653 vector unsigned long long vec_slo (vector unsigned long long,
29654 vector signed char)
29655 vector unsigned long long vec_slo (vector unsigned long long,
29656 vector unsigned char)
29657 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
29658 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
29659 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
29660 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
29661 * doc/extend.texi: Update the built-in documentation file for the
29662 new built-in functions.
29664 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
29666 * attribs.h (sorted_attr_string): Move machine independent
29667 functions for target clone support from the i386 port to common
29668 code. Rename ix86_function_versions to common_function_versions.
29669 Rename make_name to make_unique_name.
29670 (common_function_versions): Likewise.
29671 (make_unique_name): Likewise.
29672 (make_dispatcher_decl): Likewise.
29673 (is_function_default_version): Likewise.
29674 * attribs.c (attr_strcmp): Likewise.
29675 (sorted_attr_string): Likewise.
29676 (common_function_versions): Likewise.
29677 (make_unique_name): Likewise.
29678 (make_dispatcher_decl): Likewise.
29679 (is_function_default_version): Likewise.
29680 * config/i386/i386.c (attr_strcmp): Likewise.
29681 (sorted_attr_string): Likewise.
29682 (ix86_function_versions): Likewise.
29683 (make_name): Likewise.
29684 (make_dispatcher_decl): Likewise.
29685 (is_function_default_version): Likewise.
29686 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
29688 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29691 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
29692 Account for direct move costs for vec_construct of integer
29695 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
29698 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
29699 (UNSPEC_STX_ATOMIC): Ditto.
29700 (loaddi_via_sse): New insn.
29701 (storedi_via_sse): Ditto.
29702 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
29703 Update corresponding peephole2 patterns.
29704 (atomic_storedi_fpu): Ditto.
29706 2017-05-11 Julia Koval <julia.koval@intel.com>
29708 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
29709 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
29711 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
29712 (__builtin_ia32_rsqrt14ss_mask): New builtins.
29713 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
29715 2017-05-11 Nathan Sidwell <nathan@acm.org>
29717 * graphite-poly.c: Include dumpfile.h.
29719 * dumpfle.h (dump_function): Declare here ...
29720 * tree-dump.h (dump_function): ... not here.
29721 * dumpfile.c: #include tree-cfg.h.
29722 (dump_function): Move here from ...
29723 * tree-dump.c (dump_function): ... here.
29724 * gimplify.c: #include splay-tree.h, not tree-dump.h.
29725 * graphite-poly.c: Don't include tree-dump.h.
29726 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
29727 * print-tree.c: Likewise.
29728 * stor-layout.c: Likewise.
29729 * tree-nested.c: Likewise.
29731 * dumpfile.c (dump_start): Use TDF_FLAGS.
29732 (dump_enable_all): Fix TDF_KIND check thinko.
29734 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
29736 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
29737 array entries to represent two legal parameterizations of the
29738 overloaded __builtin_cmpb function, as represented by the
29739 P6_OV_BUILTIN_CMPB constant.
29740 (altivec_resolve_overloaded_builtin): Add special case handling
29741 for the __builtin_cmpb function, as represented by the
29742 P6_OV_BUILTIN_CMPB constant.
29743 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
29744 (BU_P6_64BIT_2): New macro.
29745 (BU_P6_OVERLOAD_2): New macro
29746 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
29747 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
29748 (CMPB): Add overload support to represent both 32-bit and 64-bit
29749 compare-bytes function.
29750 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
29751 support for TARGET_CMPB.
29752 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
29753 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
29754 documentation of the __builtin_cmpb overloaded built-in function.
29756 2017-05-11 Richard Biener <rguenther@suse.de>
29758 PR tree-optimization/80705
29759 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
29760 bases are not vectorizable.
29762 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29764 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
29765 when counting register pressure.
29767 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29769 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
29770 (iv_ca_more_deps): Renamed to ...
29771 (iv_ca_compare_deps): ... this.
29772 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
29774 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29776 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
29778 (determine_group_iv_costs): ... here.
29779 (find_inv_vars_cb): Record inv var if it's not recorded before.
29781 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29783 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
29784 (get_shiftadd_cost): Ditto.
29786 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29788 * tree-ssa-address.c: Include header file.
29789 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
29791 (add_to_parts): Refactor.
29792 (addr_to_parts): New parameter. Update use of move_hint_to_base.
29793 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
29796 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29798 PR tree-optimization/53090
29799 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
29801 (extract_cond_operands): Detect condition with IV on both sides
29802 and return COMP_IV_EXPR_2.
29803 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
29804 (rewrite_use_compare): Simplify by removing call to function
29805 extract_cond_operands.
29807 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29809 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
29810 (extract_cond_operands): Detect condition comparing against non-
29811 invariant bound and return appropriate enum value.
29812 (find_interesting_uses_cond): Update use of extract_cond_operands.
29813 Handle its return value accordingly.
29814 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
29816 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29818 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
29819 nonlinear iv_use computation in loop invariant sensitive way.
29821 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29823 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
29824 (find_iv_candidates): Call relate_compare_use_with_all_cands.
29826 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29828 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
29829 (dump_cand): Support iv_cand.inv_exprs.
29830 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
29832 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
29835 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29837 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
29839 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
29840 as local function. Include necessary header files.
29841 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
29843 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29845 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
29847 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29849 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
29850 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
29851 RSHIFT_EXPR and BIT_NOT_EXPR.
29853 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29855 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
29856 (adjust_setup_cost): New parameter supporting round up adjustment.
29857 (struct address_cost_data): Delete.
29858 (force_expr_to_var_cost): Don't bound cost with spill_cost.
29859 (split_address_cost, ptr_difference_cost): Delete.
29860 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
29861 (struct ainc_cost_data): New struct.
29862 (get_address_cost_ainc): New function.
29863 (get_address_cost, get_computation_cost): Reimplement.
29864 (determine_group_iv_cost_address): Record inv_expr for all uses of
29866 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
29867 (iv_ca_has_deps): Reimplemented to ...
29868 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
29870 (iv_ca_extend): Call iv_ca_more_deps.
29872 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29874 * tree-ssa-address.c (struct mem_address): Move to header file.
29875 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
29876 * tree-ssa-address.h (struct mem_address): Move from C file.
29877 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
29879 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29881 * tree-affine.h (aff_combination_type): New interface.
29882 (aff_combination_zero_p): Remove static.
29883 (aff_combination_const_p): New interface.
29884 (aff_combination_singleton_var_p): New interfaces.
29886 2017-05-11 Richard Biener <rguenther@suse.de>
29888 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
29889 Skip unreachable blocks and destinations.
29890 (eliminate): Move stmt removal and fixup ...
29891 (fini_eliminate): ... here. Skip inserted exprs.
29892 (pass_pre::execute): Move fini_pre after fini_eliminate.
29893 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
29894 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
29895 PRE to get rid of dead code that has invalid SSA form and
29896 split critical edges again.
29898 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29900 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29902 2017-05-11 Richard Biener <rguenther@suse.de>
29904 * passes.c (execute_function_todo): Verify loops if they are
29905 said to be up-to-date.
29906 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
29907 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
29909 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
29912 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
29913 handle calling assemble_external ourself.
29916 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
29917 modes with zero size. Enhance comment.
29919 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29921 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
29922 built-ins for vec_xl and vec_xst with short and char pointer
29925 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
29927 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
29928 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
29929 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
29930 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
29931 (_mm_maskz_min_round_ss): New intrinsics.
29932 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
29933 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
29934 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
29935 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
29936 (__builtin_ia32_minss_mask_round): New builtins.
29937 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29938 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29939 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
29941 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
29942 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
29944 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
29947 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
29949 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
29950 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
29951 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
29952 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
29953 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
29954 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
29955 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
29956 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29957 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
29958 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
29959 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
29960 (__builtin_ia32_mulss_mask_round): New builtins.
29961 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29962 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29963 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
29965 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
29966 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
29968 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
29971 2017-05-10 Julia Koval <julia.koval@intel.com>
29973 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
29974 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
29975 (_mm256_setr_m128i): New intrinsics.
29977 2017-05-10 Julia Koval <julia.koval@intel.com>
29979 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
29980 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
29981 (_mm_maskz_rcp14_ss): New intrinsics.
29982 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
29983 (__builtin_ia32_rcp14ss_mask): New builtins.
29984 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
29986 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
29988 PR tree-optimization/51513
29989 * tree-cfg.c (gimple_seq_unreachable_p): New function.
29990 (assert_unreachable_fallthru_edge_p): Use it.
29991 (group_case_labels_stmt): Likewise.
29992 * tree-cfg.h: Prototype it.
29993 * stmt.c: Include cfghooks.h and tree-cfg.h.
29994 (emit_case_dispatch_table) <gap_label>: New local variable.
29995 Use it to fill dispatch table gaps.
29996 Test for default_label before updating probabilities.
29997 (expand_case) <default_label>: Remove unneeded initialization.
29998 Test for unreachable default case statement and remove its edge.
29999 Set default_label accordingly.
30000 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
30002 2017-05-10 Carl Love <cel@us.ibm.com>
30004 * config/rs6000/rs6000-c: Add support for built-in functions
30005 vector signed char vec_neg (vector signed char)
30006 vector signed short int vec_neg (vector short int)
30007 vector signed int vec_neg (vector signed int)
30008 vector signed long long vec_neg (vector signed long long)
30009 vector float vec_neg (vector float)
30010 vector double vec_neg (vector double)
30011 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
30013 * config/rs6000/altivec.h: Add define for vec_neg
30014 * doc/extend.texi: Update the built-in documentation for the
30015 new built-in functions.
30017 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30019 PR tree-optimization/77644
30020 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
30022 2017-05-10 Nathan Sidwell <nathan@acm.org>
30024 * dumpfile.h (TDI_lang_all): New.
30025 (TDF_KIND): New. Renumber others
30026 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
30028 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
30030 (get_dump_file_name): Adjust suffix generation.
30031 (dump_enable_all): Use TDF_KIND.
30032 * doc/invoke.texi (-fdump-lang-all): Document.
30034 * dumpfile.h: Tabify.
30036 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
30039 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
30040 Move member access before delete.
30042 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
30044 * tree-inline.c (expand_call_inline): Split block at stmt
30047 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
30050 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
30051 are now unused after splitting mov{sf,sd}_hardfloat.
30052 (f32_lr2): Likewise.
30053 (f32_lm): Likewise.
30054 (f32_lm2): Likewise.
30055 (f32_li): Likewise.
30056 (f32_li2): Likewise.
30057 (f32_lv): Likewise.
30058 (f32_sr): Likewise.
30059 (f32_sr2): Likewise.
30060 (f32_sm): Likewise.
30061 (f32_sm2): Likewise.
30062 (f32_si): Likewise.
30063 (f32_si2): Likewise.
30064 (f32_sv): Likewise.
30065 (f32_dm): Likewise.
30066 (f32_vsx): Likewise.
30067 (f32_av): Likewise.
30068 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
30069 For movsf, order stores so the VSX stores occur before the GPR
30070 store which encourages the register allocator to use a traditional
30071 FPR instead of a GPR. For movsd, order the stores so that the GPR
30072 store comes before the VSX stores to allow the power6 to work.
30073 This is due to the power6 not having a 32-bit integer store
30074 instruction from a FPR.
30075 (movsf_hardfloat): Likewise.
30076 (movsd_hardfloat): Likewise.
30078 2017-05-09 Martin Sebor <msebor@redhat.com>
30080 PR translation/80280
30081 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
30084 PR translation/80280
30085 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
30086 data member added in r247778.
30087 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
30089 2017-05-09 Nathan Sidwell <nathan@acm.org>
30091 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
30093 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
30096 2017-05-09 Marek Polacek <polacek@redhat.com>
30098 * doc/invoke.texi: Fix typo.
30100 2017-05-09 Richard Biener <rguenther@suse.de>
30102 * tree-vrp.c (vrp_val_is_max): Adjust comment.
30103 (vrp_val_is_min): Likewise.
30104 (set_value_range_to_value): Likewise.
30105 (set_value_range_to_nonnegative): Likewise.
30106 (gimple_assign_nonzero_p): Likewise.
30107 (gimple_stmt_nonzero_p): Likewise.
30108 (vrp_int_const_binop): Likewise. Remove unreachable case.
30109 (adjust_range_with_scev): Adjust comments.
30110 (compare_range_with_value): Likewise.
30111 (extract_range_from_phi_node): Likewise.
30112 (test_for_singularity): Likewise.
30114 2017-05-09 Richard Biener <rguenther@suse.de>
30116 * tree-vrp.c (get_single_symbol): Add assert that we don't
30117 get overflowed constants as invariant part.
30118 (compare_values_warnv): Add comment before the TREE_NO_WARNING
30119 checks. Use wi::cmp instead of recursing for integer constants.
30120 (compare_values): Just ignore whether we assumed undefined
30121 overflow instead of failing the compare.
30122 (extract_range_for_var_from_comparison_expr): Add comment before the
30123 TREE_NO_WARNING sets.
30124 (test_for_singularity): Likewise.
30125 (extract_range_from_comparison): Do not disable optimization
30126 when we assumed undefined overflow.
30127 (extract_range_basic): Remove init of unused var.
30129 2017-05-09 Richard Biener <rguenther@suse.de>
30131 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
30132 (extract_range_from_multiplicative_op_1): Adjust.
30133 (extract_range_from_binary_expr_1): Use int_const_binop.
30135 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
30138 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
30139 rs6000_store_data_bypass_p in seven define_bypass directives and
30140 in several comments.
30141 * config/rs6000/rs6000-protos.h: Add prototype for
30142 rs6000_store_data_bypass_p function.
30143 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
30144 function implements slightly different (rs6000-specific) semantics
30145 than store_data_bypass_p, returning false rather than aborting
30146 with assertion error when arguments do not satisfy the
30147 requirements of store data bypass.
30148 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
30149 rs6000_store_data_bypass_p.
30151 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
30153 * config/xtensa/xtensa-protos.h
30154 (xtensa_initial_elimination_offset): New declaration.
30155 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30156 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
30157 macro definition, add case for FRAME_POINTER_REGNUM when
30158 FRAME_GROWS_DOWNWARD.
30159 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
30160 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
30161 xtensa_initial_elimination_offset.
30163 2017-05-08 Nathan Sidwell <nathan@acm.org>
30165 * doc/invoke.texi: Alphabetize -fdump options.
30167 2017-05-08 Martin Sebor <msebor@redhat.com>
30169 PR translation/80280
30170 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
30172 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
30174 * target.def (compute_frame_layout): New optional target hook.
30175 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
30176 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
30177 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
30179 * reload1.c (verify_initial_elim_offsets): Likewise.
30180 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
30181 (use_simple_return_p): Call arm_compute_frame_layout if needed.
30182 (arm_get_frame_offsets): Split up into this ...
30183 (arm_compute_frame_layout): ... and this function.
30185 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
30187 * config/aarch64/constraints.md (Usa): New constraint.
30188 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
30190 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
30192 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
30193 with_multilib_list after it has been checked.
30195 2017-05-08 Richard Biener <rguenther@suse.de>
30197 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
30198 (bitmap_set_subtract_values): Likewise.
30200 2017-05-08 Richard Biener <rguenther@suse.de>
30202 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
30203 (gimple_assign_nonzero): ... this and remove strict_overflow_p
30205 (gimple_stmt_nonzero_warnv_p): Rename to ...
30206 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
30208 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
30209 (extract_range_basic): Adjust, do not disable propagation on
30210 strict overflow sensitive simplification.
30211 (vrp_visit_cond_stmt): Likewise.
30213 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
30215 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
30216 body size unconditionally.
30218 2017-05-07 Jeff Law <law@redhat.com>
30221 2017-05-06 Jeff Law <law@redhat.com>
30222 PR tree-optimization/78496
30223 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
30226 PR tree-optimization/78496
30227 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
30228 (simplify_stmt_using_ranges): Call it.
30229 (vrp_dom_walker::before_dom_children): Extract equivalences
30230 from an ASSERT_EXPR with an equality comparison against a
30233 2017-05-06 Jeff Law <law@redhat.com>
30235 PR tree-optimization/78496
30236 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
30239 PR tree-optimization/78496
30240 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
30241 (simplify_stmt_using_ranges): Call it.
30242 (vrp_dom_walker::before_dom_children): Extract equivalences
30243 from an ASSERT_EXPR with an equality comparison against a
30246 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
30248 * lra-constraints.c (lra_copy_reg_equiv): New function.
30249 (split_reg): Use it to copy equivalence information from the
30250 original register to the spill register.
30252 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
30254 PR rtl-optimization/75964
30255 * simplify-rtx.c (simplify_const_relational_operation): Remove
30256 invalid handling of comparisons of integer ABS.
30258 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
30260 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
30261 initialize to zero.
30262 (init_regs): Remove declaration.
30263 (function_arg_advance_32): Initialize error_p as boolean variable.
30265 2017-05-05 Nathan Sidwell <nathan@acm.org>
30267 * store-motion.c (remove_reachable_equiv_notes): Reformat long
30268 lines. Use for (;;).
30270 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30272 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
30273 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
30274 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
30275 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
30276 VF=2 that require versioning.
30278 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30280 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
30283 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30285 * diagnostic.h (diagnostic_override_option_index): Convert from
30286 macro to inline function.
30288 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30290 * diagnostic.c (last_module_changed_p): New function.
30291 (set_last_module): New function.
30292 (diagnostic_report_current_module): Convert macro usage to
30293 the above functions.
30294 * diagnostic.h (diagnostic_context::last_module): Strengthen
30295 from const line_map * to const line_map_ordinary *.
30296 (diagnostic_last_module_changed): Delete macro.
30297 (diagnostic_set_last_module): Delete macro.
30299 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30301 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
30302 with diagnostic_report_diagnostic.
30303 (diagnostic_n_impl_richloc): Likewise.
30304 * diagnostic.h (report_diagnostic): Delete macro.
30305 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
30306 with diagnostic_report_diagnostic.
30307 * substring-locations.c (format_warning_va): Likewise.
30309 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30311 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
30312 save/restor of format_spec. Move option-printing code to...
30313 (print_option_information): ...this new function, and
30314 reimplement by simply printing to the pretty_printer,
30315 rather than appending to the format string.
30317 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30319 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
30320 handling logic into...
30321 (update_effective_level_from_pragmas): ...this new function.
30323 2017-05-04 Andrew Waterman <andrew@sifive.com>
30325 * config/riscv/riscv.opt (mstrict-align): New option.
30326 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
30327 (SLOW_UNALIGNED_ACCESS): Define.
30328 (riscv_slow_unaligned_access): Declare.
30329 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
30331 (riscv_slow_unaligned_access): New variable.
30332 (rocket_tune_info): Set slow_unaligned_access to true.
30333 (optimize_size_tune_info): Set slow_unaligned_access to false.
30334 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
30335 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
30336 (riscv_option_override): Set riscv_slow_unaligned_access.
30337 * doc/invoke.texi: Add -mstrict-align to RISC-V.
30339 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
30341 * config/riscv/riscv.md: Unify indentation.
30343 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
30348 * config/rs6000/rs6000.md (u code attribute): Add FIX and
30350 (extendsi<mode>2): Add support for doing sign extension via
30351 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
30352 don't have ISA 3.0 instructions.
30353 (extendsi<mode>2 splitter): Likewise.
30354 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
30355 generate the normal insns since SImode can now go in vector
30356 registers. Disallow the special UNSPECs needed for previous
30357 machines to hide SImode being used. Add new insns
30358 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
30359 (fix_trunc<mode>si2_stfiwx): Likewise.
30360 (fix_trunc<mode>si2_internal): Likewise.
30361 (fixuns_trunc<mode>si2): Likewise.
30362 (fixuns_trunc<mode>si2_stfiwx): Likewise.
30363 (fctiw<u>z_<mode>_smallint): Likewise.
30364 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
30365 of floating point to 32-bit integer from doing a direct move to
30366 the GPR registers to do a store.
30367 (fctiwz_<mode>): Break long line.
30369 2017-05-05 Bin Cheng <bin.cheng@arm.com>
30371 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
30372 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
30373 (addr_list, addr_offset_valid_p): New.
30374 (split_address_groups): Check offset validity with above function.
30375 (gt-tree-ssa-loop-ivopts.h): Include header file.
30377 2017-05-05 Nathan Sidwell <nathan@acm.org>
30379 * config.gcc (arm*-*-*): Add missing 'fi'.
30381 2017-05-05 Steve Ellcey <sellcey@cavium.com>
30383 * doc/invoke.texi (-fopt-info): Explicitly say order of options
30384 included in -fopt-info does not matter.
30385 * doc/optinfo.texi (-fopt-info): Fix description of default
30386 behavour. Explicitly say order of options included in -fopt-info
30389 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
30391 * config.gcc: Allow combinations of aprofile and rmprofile values for
30392 --with-multilib-list.
30393 * config/arm/t-multilib: New file.
30394 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
30395 variables. Remove setting of ISA and floating-point ABI in
30396 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
30397 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
30398 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
30399 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
30401 * config/arm/t-rmprofile: Likewise except for the matches changes.
30402 * doc/install.texi (--with-multilib-list): Document the combination of
30403 aprofile and rmprofile values and warn about pitfalls in doing that.
30405 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
30407 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
30408 (movdi_aarch64): Likewise.
30410 2017-05-05 Jakub Jelinek <jakub@redhat.com>
30412 PR tree-optimization/80632
30413 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
30415 (build_arrays): Initialize it for virtual phis.
30416 (fix_phi_nodes): Use it for virtual phis.
30418 PR tree-optimization/80558
30419 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
30420 [x, y] op z into [x op, y op z] for op & or | if conditions
30423 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
30424 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30427 * config/arm/arm.md (use_literal_pool): Remove.
30428 (64-bit immediate split): No longer takes cost into consideration
30429 if arm_disable_literal_pool is enabled.
30430 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
30431 used when arm_disable_literal_pool is enabled.
30432 (arm_max_const_double_inline_cost): Remove use of
30433 arm_disable_literal_pool.
30434 (push_minipool_fix): Add assert.
30435 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
30436 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
30437 (no_literal_pool_sf_immediate): New.
30439 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30441 PR tree-optimization/80613
30442 * tree-ssa-dce.c (propagate_necessity): Remove cases for
30443 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
30445 2017-05-05 Richard Biener <rguenther@suse.de>
30447 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
30449 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
30451 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
30452 of this flag from insn conditions due to removal from r247495.
30454 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
30456 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
30458 (arm_early_store_addr_dep_ptr): Likewise.
30459 * config/arm/aarch-common-protos.h
30460 (arm_early_load_addr_dep_ptr): Add prototype.
30461 (arm_early_store_addr_dep_ptr): Likewise.
30462 * config/arm/cortex-a53.md: Add new bypasses.
30464 2017-05-05 Jakub Jelinek <jakub@redhat.com>
30466 * tree.c (next_type_uid): Change type to unsigned.
30467 (type_hash_canon): Decrement back next_type_uid if
30468 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
30469 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
30472 2017-05-04 Martin Sebor <msebor@redhat.com>
30474 * builtins.c: Fix a trivial typo in a comment.
30476 PR middle-end/79234
30477 * builtins.c (check_sizes): Adjust to handle reading past the end.
30478 Avoid printing excessive upper bound of ranges. Use %E to print
30479 tree nodes instead of converting them to %wu.
30480 (expand_builtin_memchr): New function.
30481 (compute_dest_size): Rename...
30482 (compute_objsize): ...to this.
30483 (expand_builtin_memcpy): Adjust.
30484 (expand_builtin_mempcpy): Adjust.
30485 (expand_builtin_strcat): Adjust.
30486 (expand_builtin_strcpy): Adjust.
30487 (check_strncat_sizes): Adjust.
30488 (expand_builtin_strncat): Adjust.
30489 (expand_builtin_strncpy): Adjust and simplify.
30490 (expand_builtin_memset): Adjust.
30491 (expand_builtin_bzero): Adjust.
30492 (expand_builtin_memcmp): Adjust.
30493 (expand_builtin): Handle memcmp.
30494 (maybe_emit_chk_warning): Check strncat just once.
30496 2017-05-04 Martin Sebor <msebor@redhat.com>
30498 PR preprocessor/79214
30499 PR middle-end/79222
30500 PR middle-end/79223
30501 * builtins.c (check_sizes): Add inlining context and issue
30502 warnings even when -Wno-system-headers is set.
30503 (check_strncat_sizes): Same.
30504 (expand_builtin_strncat): Same.
30505 (expand_builtin_memmove): New function.
30506 (expand_builtin_stpncpy): Same.
30507 (expand_builtin): Handle memmove and stpncpy.
30509 2017-05-04 Bin Cheng <bin.cheng@arm.com>
30511 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
30512 which is not used any more.
30514 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30516 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
30518 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30520 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
30521 (cortexa53_tunings): Likewise.
30522 (cortexa57_tunings): Likewise.
30523 (cortexa72_tunings): Likewise.
30524 (cortexa73_tunings): Likewise.
30526 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30528 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
30529 Set loop alignment to 8.
30531 2017-05-04 Martin Sebor <msebor@redhat.com>
30533 PR translation/80280
30534 * builtins.c (expand_builtin_object_size): Add missing quoting to
30535 %D and like directives.
30536 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
30537 (hsa_type_for_tree_type): Same.
30538 (verify_function_arguments): Same.
30539 * symtab.c (symbol_table::change_decl_assembler_name): Same.
30540 * varasm.c (get_section): Same.
30543 2017-05-04 Martin Sebor <msebor@redhat.com>
30545 PR translation/80280
30546 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
30548 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30550 * config/aarch64/aarch64.c (generic_addrcost_table):
30551 Change HI/TI mode setting.
30553 2017-05-04 Martin Jambor <mjambor@suse.cz>
30555 PR tree-optimization/80622
30556 * tree-sra.c (comes_initialized_p): New function.
30557 (build_accesses_from_assign): Only set write lazily when
30558 comes_initialized_p is false.
30559 (analyze_access_subtree): Use comes_initialized_p.
30560 (propagate_subaccesses_across_link): Assert !comes_initialized_p
30561 instead of testing for PARM_DECL.
30563 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30565 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
30566 constraint on operand 0 to allow more general addressing modes.
30567 Adjust output template.
30568 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
30570 * config/aarch64/aarch64-protos.h
30571 (aarch64_address_valid_for_prefetch_p): Declare prototype.
30572 * config/aarch64/constraints.md (Dp): New address constraint.
30573 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
30576 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
30578 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
30579 update use of estimate_ipcp_clone_size_and_time.
30580 (estimate_local_effects): Update use of
30581 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
30582 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
30583 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
30584 Return nonspecialized time.
30586 2017-05-04 Richard Biener <rguenther@suse.de>
30588 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
30589 for the last VUSE which def dominates the PHI. Directly call
30591 (get_continuation_for_phi_1): Remove.
30593 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
30595 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
30596 to explain the use of truncating division. Cap the number of
30597 iterations to the maximum given by nb_iterations_upper_bound,
30600 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
30602 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
30603 * configure: Regenerate.
30604 * config.in: Regenerate.
30605 * config/i386/driver-mingw32.c: new file.
30606 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
30607 * config.host: Link driver-mingw32.o on MinGW host.
30608 * doc/install.texi: Document new --enable-mingw-wildcard configure
30611 2017-05-04 Marek Polacek <polacek@redhat.com>
30613 PR tree-optimization/80612
30614 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
30616 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30617 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
30619 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
30620 (movt splitter): Likewise.
30621 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
30622 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
30623 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
30624 block for Thumb-1 with MOVT.
30625 (thumb2_legitimate_address_p): Move code block ...
30626 (can_avoid_literal_pool_for_label_p): ... into this new function.
30627 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
30629 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
30630 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
30631 "M-profile targets with the MOVT instruction".
30633 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30635 * config/arm/arm-builtins.c (arm_init_builtins): Rename
30636 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
30637 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
30639 2017-05-04 Martin Liska <mliska@suse.cz>
30641 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
30642 variable cond_code.
30644 2017-05-04 Richard Biener <rguenther@suse.de>
30646 * tree.c (array_at_struct_end_p): Handle arrays at struct
30647 end with flexarrays more conservatively. Refactor and treat
30648 arrays of arrays or aggregates more strict. Fix
30649 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
30650 * tree.c (array_at_struct_end_p): Adjust prototype.
30651 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
30652 * gimple-fold.c (get_range_strlen): Likewise.
30653 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
30655 2017-05-04 Richard Biener <rguenther@suse.de>
30657 PR tree-optimization/31130
30658 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
30660 (supports_overflow_infinity): Likewise.
30661 (is_negative_overflow_infinity): Likewise.
30662 (is_positive_overflow_infinity): Likewise.
30663 (is_overflow_infinity): Likewise.
30664 (stmt_overflow_infinity): Likewise.
30665 (overflow_infinity_range_p): Likewise.
30666 (usable_range_p): Remove as always returning true.
30667 (make_overflow_infinity): Remove.
30668 (negative_overflow_infinity): Likewise.
30669 (positive_overflow_infinity): Likewise.
30670 (avoid_overflow_infinity): Likewise.
30671 (set_value_range): Adjust accordingly.
30672 (set_value_range_to_nonnegative): Likewise, remove now unused
30673 overflow_infinity arg.
30674 (vrp_operand_equal_p): Adjust.
30675 (update_value_range): Likewise.
30676 (range_int_cst_singleton_p): Likewise.
30677 (operand_less_p): Likewise.
30678 (compare_values_warnv): Likewise.
30679 (extract_range_for_var_from_comparison_expr): Likewise.
30680 (vrp_int_const_binop): Likewise.
30681 (zero_nonzero_bits_from_vr): Likewise.
30682 (extract_range_from_multiplicative_op_1): Likewise.
30683 (extract_range_from_binary_expr_1): Likewise.
30684 (extract_range_from_unary_expr): Likewise.
30685 (extract_range_from_comparison): Likewise.
30686 (extract_range_basic): Likewise.
30687 (adjust_range_with_scev): Likewise.
30688 (compare_ranges): Likewise.
30689 (compare_range_with_value): Likewise.
30690 (dump_value_range): Likewise.
30691 (test_for_singularity): Likewise, remove strict_overflow_p parameter
30693 (simplify_cond_using_ranges): Adjust.
30695 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
30697 * brig-builtins.def: Added a builtin for class_f64.
30698 * builtin-types.def: Added a builtin type needed by class_f64.
30700 2017-05-03 Jason Merrill <jason@redhat.com>
30702 * timevar.def: Add TV_CONSTEXPR.
30704 2017-05-03 David Malcolm <dmalcolm@redhat.com>
30706 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
30708 2017-05-03 Martin Jambor <mjambor@suse.cz>
30710 * ipa-prop.c (ipa_update_after_lto_read): Removed.
30711 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
30712 * ipa-cp.c (ipcp_propagate_stage): Do not call
30713 ipa_update_after_lto_read.
30714 * ipa-inline.c (ipa_inline): Likewise.
30716 2017-05-03 Martin Jambor <mjambor@suse.cz>
30718 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
30719 tag. Added a default constructor and a destructor.
30720 (ipa_edge_args_sum_t): New class;
30721 (ipa_edge_args_sum): Declare.
30722 (ipa_edge_args_vector): Remove declaration.
30723 (IPA_EDGE_REF): Use ipa_edge_args_sum.
30724 (ipa_free_edge_args_substructures): Remove declaration.
30725 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
30726 (ipa_edge_args_info_available_for_edge_p): Likewise.
30727 * ipa-prop.c (ipa_edge_args_vector): Removed.
30728 (edge_removal_hook_holder): Likewise.
30729 (edge_duplication_hook_holder): Likewise.
30730 (ipa_edge_args_sum): New variable.
30731 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
30732 ipa_edge_args_vector.
30733 (ipa_free_edge_args_substructures): Likewise.
30734 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
30735 ipa_edge_args_vector.
30736 (ipa_edge_removal_hook): Turned into method
30737 ipa_edge_args_sum_t::remove.
30738 (ipa_edge_duplication_hook): Turned into method
30739 ipa_edge_args_sum_t::duplicate.
30740 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
30741 registering edge hooks.
30742 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
30743 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
30744 ipa_edge_args_sum instead of ipa_edge_args_vector.
30745 * ipa-profile.c (ipa_profile): Likewise.
30747 2017-05-03 Martin Jambor <mjambor@suse.cz>
30749 * symbol-summary.h (function_summary): New method exists.
30750 (function_summary::symtab_removal): Deallocate through release.
30751 (call_summary): New class.
30752 (gt_ggc_mx): New overload.
30753 (gt_pch_nx): Likewise.
30754 (gt_pch_nx): Likewise.
30756 2017-05-03 Jeff Law <law@redhat.com>
30758 PR tree-optimization/78496
30759 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
30760 from simplify_cond_using_ranges. Split off code to walk
30761 backwards through casts into ...
30762 (simplify_cond_using_ranges_2): New function.
30763 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
30764 (execute_vrp): After identifying jump threads, call
30765 simplify_cond_using_ranges_2.
30767 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
30770 * ipa-inline.h (inline_summary): Add ctor.
30771 (create_ggc): Do not use ggc_cleared_alloc.
30773 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
30774 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
30776 * gcc.c (handle_braces): Support escaping in switch matching
30778 * doc/invoke.texi (Spec Files): Document it.
30779 Remove superfluous @code markup in items.
30781 2017-05-03 David Malcolm <dmalcolm@redhat.com>
30783 * diagnostic-show-locus.c (struct column_range): New struct.
30784 (get_affected_columns): New function.
30785 (get_printed_columns): New function.
30786 (struct correction): New struct.
30787 (correction::ensure_capacity): New function.
30788 (correction::ensure_terminated): New function.
30789 (struct line_corrections): New struct.
30790 (line_corrections::~line_corrections): New dtor.
30791 (line_corrections::add_hint): New function.
30792 (layout::print_trailing_fixits): Reimplement in terms of the new
30794 (selftest::test_overlapped_fixit_printing): New function.
30795 (selftest::diagnostic_show_locus_c_tests): Call it.
30797 2017-05-03 Nathan Sidwell <nathan@acm.org>
30799 Canonicalize canonical type hashing
30800 * tree.h (type_hash_canon_hash): Declare.
30801 * tree.c (type_hash_list, attribute_hash_list): Move into
30802 type_hash_canon_hash.
30803 (build_type_attribute_qual_variant): Break out hash code calc into
30804 type_hash_canon_hash.
30805 (type_hash_canon_hash): New. Generic type hash computation.
30806 (build_range_type_1, build_array_type_1, build_function_type,
30807 build_method_type_directly, build_offset_type, build_complex_type,
30808 make_vector_type): Call it.
30810 2017-05-03 Richard Biener <rguenther@suse.de>
30812 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
30813 When all DRs have unknown misaligned do not always peel
30814 when there is a store but apply the same costing model as if
30815 there were only loads.
30817 2017-05-03 Richard Biener <rguenther@suse.de>
30820 PR tree-optimization/80492
30821 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30822 compare_base_decls returning dont-know properly.
30824 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
30826 * config/arm/iterators.md (CCSI): New mode iterator.
30827 (arch): New mode attribute.
30828 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
30829 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
30830 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
30831 code iterator for success result mode.
30832 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
30833 the corresponding new insn generators.
30835 2017-05-03 Bin Cheng <bin.cheng@arm.com>
30838 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30839 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30841 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
30843 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
30844 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
30845 (DDR_A): Wrap DDR argument in brackets.
30846 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
30847 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
30848 (DDR_REVERSED_P): Likewise.
30850 2017-05-03 Jakub Jelinek <jakub@redhat.com>
30852 PR tree-optimization/79472
30853 * tree-switch-conversion.c (struct switch_conv_info): Add
30854 contiguous_range and default_case_nonstandard fields.
30855 (collect_switch_conv_info): Compute contiguous_range and
30856 default_case_nonstandard fields, don't clear final_bb if
30857 contiguous_range and only the default case doesn't have the required
30859 (check_all_empty_except_final): Set default_case_nonstandard instead
30860 of failing if contiguous_range and the default case doesn't have empty
30862 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
30863 and only the default case doesn't have the required constants. Skip
30865 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
30866 if default_case_nonstandard.
30867 (build_constructors): Build constant 1 just once. Assert that default
30868 values aren't inserted in between cases if contiguous_range. Skip
30870 (build_arrays): Skip virtual phis.
30871 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
30872 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
30873 Handle virtual phis.
30874 (gen_inbound_check): Handle default_case_nonstandard case.
30875 (process_switch): Adjust check_final_bb caller. Call
30876 gather_default_values with the first non-default case instead of
30877 default case if default_case_nonstandard.
30879 2017-05-02 Nathan Sidwell <nathan@acm.org>
30881 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
30882 check. Fix formatting.
30884 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
30886 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
30887 errors when comparing specialized and unspecialized times.
30889 2017-05-02 David Malcolm <dmalcolm@redhat.com>
30891 * diagnostic-show-locus.c
30892 (layout::should_print_annotation_line_p): Make private.
30893 (layout::print_annotation_line): Make private.
30894 (layout::annotation_line_showed_range_p): Make private.
30895 (layout::show_ruler): Make private.
30896 (layout::print_source_line): Make private. Pass in line and
30897 line_width, rather than calling location_get_source_line. Drop
30899 (layout::print_leading_fixits): New method.
30900 (layout::print_any_fixits): Rename to...
30901 (layout::print_trailing_fixits): ...this, and make private.
30902 Don't print newline fixits.
30903 (diagnostic_show_locus): Move logic for printing one row into...
30904 (layout::print_line): ...this new function. Move the
30905 location_get_source_line call and error-handling from
30906 print_source_line to here. Call print_leading_fixits, and rename
30907 print_any_fixits to print_trailing_fixits.
30908 (selftest::test_fixit_insert_containing_newline): Update now that
30909 newlines are partially supported.
30910 (selftest::test_fixit_insert_containing_newline_2): New test.
30911 (selftest::test_fixit_replace_containing_newline): Update comments.
30912 (selftest::diagnostic_show_locus_c_tests): Call the new test.
30913 * edit-context.c (class added_line): New class.
30914 (class edited_line): Describe newline handling in comment.
30915 (edited_line::actually_edited_p): New method.
30916 (edited_line::print_content): Delete redundant decl.
30917 (edited_line::m_predecessors): New field.
30918 (edited_file::print_content): Call edited_line::print_content.
30919 (edited_file::print_diff): Update to support newlines.
30920 (edited_file::print_diff_hunk): Likewise.
30921 (edited_file::print_run_of_changed_lines): New function.
30922 (edited_file::print_diff_line): Convert to...
30923 (print_diff_line): ...this.
30924 (edited_file::get_effective_line_count): New function.
30925 (edited_line::edited_line): Initialize new field m_predecessors.
30926 (edited_line::~edited_line): Clean up m_predecessors.
30927 (edited_line::apply_fixit): Handle newlines.
30928 (edited_line::get_effective_line_count): New function.
30929 (edited_line::print_content): New function.
30930 (edited_line::print_diff_lines): New function.
30931 (selftest::test_applying_fixits_insert_containing_newline): New
30933 (selftest::test_applying_fixits_replace_containing_newline): New
30935 (selftest::insert_line): New function.
30936 (selftest::test_applying_fixits_multiple_lines): Add example of
30938 (selftest::edit_context_c_tests): Call the new tests.
30940 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30942 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
30943 parameter cand. Update dump information.
30944 (get_computation_cost): Update uses.
30946 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30948 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
30949 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
30950 (get_computation_at, rewrite_use_address): Update use of
30951 get_computation_aff.
30953 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30955 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
30956 (get_computation): Delete.
30957 (get_computation_cost): Implement like get_computation_cost_at.
30958 Use get_computation_at.
30959 (get_computation_cost_at): Delete.
30960 (rewrite_use_nonlinear_expr): Use get_computation_at.
30961 (rewrite_use_compare, remove_unused_ivs): Ditto.
30963 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30965 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
30967 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30969 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
30970 (ivopts_global_cost_for_size): Rename parameter and update uses.
30971 (iv_ca_recount_cost): Update uses.
30972 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
30973 candidates seperately in n_invs and n_cands.
30974 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
30976 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30978 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
30979 (find_inv_vars_cb): New.
30980 (find_depends): Renamed to ...
30981 (find_inv_vars): ... this.
30982 (add_candidate_1, force_var_cost): Call find_inv_vars.
30983 (split_address_cost, determine_group_iv_cost_cond): Ditto.
30985 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30987 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
30988 inv_vars. Add inv_exprs.
30989 (struct iv_cand): Rename depends_on to inv_vars.
30990 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
30991 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
30992 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
30993 (dump_cand): Dump inv_vars.
30994 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
30995 (record_invariant, find_depends, add_candidate_1): Ditto.
30996 (set_group_iv_cost, force_var_cost): Ditto.
30997 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
30998 (get_computation_cost_at, get_computation_cost): Ditto.
30999 (determine_group_iv_cost_generic): Ditto.
31000 (determine_group_iv_cost_address): Ditto.
31001 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
31002 (determine_group_iv_costs): Ditto.
31003 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
31004 (iv_ca_set_remove_invariants): Renamed to ...
31005 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
31006 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
31007 (iv_ca_set_add_invariants): Renamed to ...
31008 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
31009 (iv_ca_set_cp): Use iv_ca_set_add_invs.
31010 (iv_ca_has_deps): Support inv_vars and inv_exprs.
31011 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
31012 (create_new_ivs): Remove useless dump.
31014 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31016 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
31018 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
31019 (iv_ca_set_no_cp, create_new_iv): Ditto.
31021 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31023 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31025 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
31027 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
31028 function tree_check2.
31030 2017-05-02 Martin Liska <mliska@suse.cz>
31032 * doc/gcov.texi: Add missing preposition.
31033 * gcov.c (function_info::function_info): Properly fill up
31034 all member variables.
31036 2017-05-02 Tamar Christina <tamar.christina@arm.com>
31038 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
31040 2017-05-02 Tamar Christina <tamar.christina@arm.com>
31042 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
31044 2017-05-02 Martin Liska <mliska@suse.cz>
31047 * lto-streamer-in.c (lto_read_tree_1): Remove
31048 LTO_STREAMER_DEBUG.
31049 * lto-streamer.c (struct tree_hash_entry): Likewise.
31050 (struct tree_entry_hasher): Likewise.
31051 (tree_entry_hasher::hash): Likewise.
31052 (tree_entry_hasher::equal): Likewise.
31053 (lto_streamer_init): Likewise.
31054 (lto_orig_address_map): Likewise.
31055 (lto_orig_address_get): Likewise.
31056 (lto_orig_address_remove): Likewise.
31057 * lto-streamer.h: Likewise.
31058 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
31059 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
31061 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
31063 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
31064 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
31065 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
31066 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
31067 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
31068 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
31069 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
31070 (mm_maskz_sub_ss): New intrinsics.
31071 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31072 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31073 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
31074 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
31075 (__builtin_ia32_subss_mask_round): New builtins.
31076 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31077 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31078 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
31080 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
31081 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31083 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31086 2017-05-02 Martin Jambor <mjambor@suse.cz>
31088 PR tree-optimization/78687
31089 * tree-sra.c (access): New field parent.
31090 (process_subtree_disqualification): New function.
31091 (disqualify_candidate): Call it.
31092 (build_accesses_from_assign): Reset write flag if creating an
31094 (build_access_subtree): Fill in parent field and also prpagate
31095 down grp_write flag.
31096 (create_artificial_child_access): New parameter set_grp_write, set
31097 grp_write to its value.
31098 (propagate_subaccesses_across_link): Also propagate grp_write flag
31100 (propagate_all_subaccesses): Push the closest parent back to work
31101 queue if add_access_to_work_queue returned true.
31103 2017-05-02 Richard Biener <rguenther@suse.de>
31105 * common.opt (fstrict-overflow): Alias negative to fwrapv.
31106 * doc/invoke.texi (fstrict-overflow): Remove all traces of
31107 -fstrict-overflow documentation.
31108 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
31109 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
31110 flag_strict_overflow.
31111 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
31112 * lto-opts.c (lto_write_options): Do not stream it.
31113 * lto-wrapper.c (merge_and_complain): Do not handle it.
31114 * opts.c (default_options_table): Do not set -fstrict-overflow.
31115 (finish_options): Likewise do not clear it when sanitizing.
31116 * simplify-rtx.c (simplify_const_relational_operation): Do not
31117 test flag_strict_overflow.
31119 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
31121 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
31122 using enabled attribute.
31123 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
31124 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
31125 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
31126 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
31127 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
31128 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
31129 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
31130 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
31131 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
31132 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
31134 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
31136 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
31138 2017-05-02 Richard Biener <rguenther@suse.de>
31140 PR tree-optimization/80591
31142 2017-04-10 Richard Biener <rguenther@suse.de>
31144 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31147 2017-05-02 Richard Biener <rguenther@suse.de>
31149 PR tree-optimization/80549
31150 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
31151 (cleanup_tree_cfg_noloop): Create forwarders to known loop
31152 headers if they do not have a preheader.
31154 2017-05-02 Martin Liska <mliska@suse.cz>
31157 * common.opt: Fix typo.
31158 * doc/invoke.texi: Likewise.
31160 2017-05-01 Jan Beulich <jbeulich@suse.com>
31162 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
31163 swapping, add (x,x,m,x,n) alternative.
31165 2017-05-01 Nathan Sidwell <nathan@acm.org>
31167 * calls.c (combine_pending_stack_adjustment_and_call): Remove
31168 unnecessary unadjusted_alignment check.
31170 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
31173 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
31175 * gimplify.c (gimplify_cilk_detach): New function.
31176 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
31177 * tree-core.h: Document EXPR_CILK_SPAWN.
31178 * tree.h (EXPR_CILK_SPAWN): Define.
31180 2017-05-01 David Malcolm <dmalcolm@redhat.com>
31182 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
31183 to use new fixit_hint representation, using the "replace" logic.
31184 (get_line_span_for_fixit_hint): Likewise.
31185 (layout::print_any_fixits): Likewise.
31186 (selftest::test_one_liner_many_fixits): Rename to...
31187 (selftest::test_one_liner_many_fixits_1): ...this, and update
31188 comment and expected output to reflect that the multiple fix-it
31189 hints are now consolidated into one insertion.
31190 (selftest::test_one_liner_many_fixits_2): New test.
31191 (selftest::test_diagnostic_show_locus_one_liner): Update for
31193 (selftest::test_fixit_consolidation): Update for fix-it API
31195 * diagnostic.c (print_parseable_fixits): Likewise.
31196 * edit-context.c (edited_line::m_line_events): Convert from
31197 auto_vec <line_event *> to auto_vec <line_event>.
31198 (class line_event): Convert from abstract base class to a concrete
31199 class, taking over the role of replace_event.
31200 (class insert_event): Delete.
31201 (class replace_event): Rename to class line_event. Convert to
31203 (edit_context::add_fixits): Reimplement.
31204 (edit_context::apply_insert): Delete.
31205 (edit_context::apply_replace): Rename to...
31206 (edit_context::apply_fixit): ...this. Convert to half-open range.
31207 (edited_file::apply_insert): Delete.
31208 (edited_file::apply_replace): Rename to...
31209 (edited_file::apply_fixit): ...this.
31210 (edited_line::~edited_line): Drop deletion of events.
31211 (edited_line::apply_insert): Delete.
31212 (edited_line::apply_replace): Rename to...
31213 (edited_line::apply_fixit): ...this. Convert to half-open range.
31214 Update for change to type of m_line_events.
31215 * edit-context.h (edit_context::apply_insert): Delete.
31216 (edit_context::apply_replace): Rename to...
31217 (edit_context::apply_fixit): ...this.
31219 2017-05-01 Martin Sebor <msebor@redhat.com>
31221 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
31224 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
31227 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
31228 __get_cpuid_max returns 0.
31229 (__get_cpuid_count): Ditto.
31231 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
31233 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
31234 replacement expression is another instance of one of its arguments.
31236 2017-05-01 Jakub Jelinek <jakub@redhat.com>
31239 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
31240 check for stack push/pop autoinc.
31241 * config/i386/i386.c (ix86_agi_dependent): Return false
31242 if the only reason why modified_in_p returned true is that
31243 addr is SP based and set_insn is a push or pop.
31245 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
31247 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
31250 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
31253 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
31254 (account_size_time): Use two predicates - exec_pred and
31256 (evaluate_conditions_for_known_args): Compute both clause and
31258 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
31259 (inline_summary_t::duplicate): Update.
31260 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
31262 (compute_inline_parameters): Likewise.
31263 (estimate_edge_size_and_time): Update caluclation of time.
31264 (estimate_node_size_and_time): Compute both time and nonspecialized
31266 (estimate_ipcp_clone_size_and_time): Update.
31267 (inline_merge_summary): Update.
31268 (do_estimate_edge_time): Update.
31269 (do_estimate_edge_size): Update.
31270 (do_estimate_edge_hints): Update.
31271 (inline_read_section, inline_write_summary): Stream both new predicates.
31272 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
31274 (compute_inlined_call_time): Cleanup.
31275 (big_speedup_p): Update.
31276 (edge_badness): Update.
31277 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
31278 (size_time_entry): Replace predicate by exec_predicate and
31279 nonconst_predicate.
31280 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
31281 (estimate_edge_time): Return also nonspec_time.
31282 (reset_edge_growth_cache): Update.
31284 2017-04-29 Jakub Jelinek <jakub@redhat.com>
31286 PR rtl-optimization/80491
31287 * ifcvt.c (noce_process_if_block): When looking for x setter
31288 with missing else_bb, don't check only the insn right before
31289 cond_earliest, but look for the last insn that x is modified in
31290 within the same bb.
31292 PR rtl-optimization/80491
31293 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
31295 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
31297 PR tree-optimization/80487
31298 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
31300 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31302 PR tree-optimization/79697
31303 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
31304 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
31305 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
31307 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
31308 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
31310 2017-04-28 Martin Sebor <msebor@redhat.com>
31312 PR tree-optimization/80523
31313 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
31314 (init_target_to_host_charmap, target_to_host, target_strtol10): New
31316 (maybe_warn, format_directive, parse_directive): Use new functions.
31317 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
31319 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
31321 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
31323 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
31325 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
31326 target_header_dir): Set correctly.
31327 * configure: Regenerated.
31328 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31329 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31330 instead of SYSTEM_HEADER_DIR.
31332 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
31334 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
31335 (estimate_local_effects): Likewise.
31336 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
31337 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
31338 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
31339 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
31340 do_estimate_edge_time, estimate_edge_time): Likewise.
31341 * ipa-inline-analysis.c (estimate_node_size_and_time,
31342 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
31343 (estimate_time_after_inlining): Remove.
31345 2017-04-28 Martin Liska <mliska@suse.cz>
31347 * doc/gcov.texi: Enhance documentation of gcov.
31349 2017-04-28 Martin Liska <mliska@suse.cz>
31351 * doc/gcov.texi: Sort options in alphabetic order.
31352 * doc/gcov-dump.texi: Likewise.
31353 * doc/gcov-tool.texi: Likewise.
31354 * gcov.c (print_usage): Likewise.
31355 * gcov-dump.c (print_usage): Likewise.
31356 * gcov-tool.c (print_merge_usage_message): Likewise.
31357 (print_rewrite_usage_message): Likewise.
31358 (print_overlap_usage_message): Likewise.
31360 2017-04-28 Martin Liska <mliska@suse.cz>
31362 PR gcov-profile/53915
31363 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
31365 2017-04-28 Martin Liska <mliska@suse.cz>
31367 PR gcov-profile/79891
31368 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
31369 is marked by compiler as living on a line.
31370 (get_cycles_count): Remove usage of the union.
31371 (output_intermediate_file): Likewise.
31372 (find_source): Fix GNU coding style.
31373 (accumulate_line_counts): Remove old non-all block mode.
31374 (output_lines): Remove usage of the union.
31375 * profile.c (output_location): Include all BBs, even if
31376 belonging to a same line (and file) as a previous BB.
31378 2017-04-28 Martin Liska <mliska@suse.cz>
31380 * gcov.c (process_args): Handle new argument 'w'.
31381 (read_graph_file): Assign ID to BBs.
31382 (output_branch_count): Display BB # if verbose flag is set.
31383 (output_lines): Likewise for arcs.
31384 (print_usage): Add '--verbose' option help.
31385 * doc/gcov.texi: Document --verbose (-w) option.
31387 2017-04-28 Martin Liska <mliska@suse.cz>
31389 * gcov.c (struct block_location_info): New struct.
31390 (process_file): Fill up the new structure.
31391 (read_graph_file): Replace usage of encoding by the newly added
31393 (add_line_counts): Likewise.
31394 (accumulate_line_counts): Remove usage of the union.
31395 (function_info::function_info): New function.
31396 (function_info::~function_info): Likewise.
31397 (process_file): Call delete instead of release_function.
31398 (release_function): Release the function.
31399 (release_structures): Call delete instead of release_function.
31400 (solve_flow_graph): Replace usage of num_blocks.
31401 (find_exception_blocks): Likewise.
31402 (output_lines): Fix GNU coding style.
31404 2017-04-28 Martin Liska <mliska@suse.cz>
31407 * coverage.c (coverage_remove_note_file): New function.
31408 * coverage.h: Declare the function.
31409 * toplev.c (finalize): Clean if an error has been seen.
31411 2017-04-28 Martin Liska <mliska@suse.cz>
31413 PR gcov-profile/80031
31414 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
31415 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
31416 * gcov.c (read_graph_file): Read just number of blocks.
31417 * profile.c (branch_prob): Do not stream 0 flags per a basic
31420 2017-04-28 Martin Liska <mliska@suse.cz>
31422 * gcov-dump.c (tag_*): Add new argument to declarations.
31423 (dump_gcov_file): Likewise.
31424 (tag_blocks): Add and use new argument depth.
31425 (tag_arcs): Likewise.
31426 (tag_lines): Likewise.
31427 (tag_counters): Likewise.
31428 (tag_summary): Likewise.
31429 (dump_working_sets): Use depth to do a proper indentation.
31431 2017-04-28 Jakub Jelinek <jakub@redhat.com>
31434 * cgraph.h (symtab_node::debug_symtab): No longer inline.
31435 * symtab.c (symtab_node::debug_symtab): Move definition here.
31437 2017-04-28 Richard Biener <rguenther@suse.de>
31439 * lto-streamer.h (LTO_major_version): Bump to 7.
31441 2017-04-28 Richard Biener <rguenther@suse.de>
31443 * tree-vrp.c (assert_info): New struct.
31444 (add_assert_info): New helper.
31445 (register_edge_assert_for_2): Refactor to add asserts to a vector
31447 (register_edge_assert_for_1): Likewise.
31448 (register_edge_assert_for): Likewise.
31449 (finish_register_edge_assert_for): New helper actually registering
31450 asserts where live on edge.
31451 (find_conditional_asserts): Adjust.
31452 (find_switch_asserts): Likewise.
31453 (evrp_dom_walker::try_find_new_range): Generalize.
31454 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
31456 2017-04-27 Marek Polacek <polacek@redhat.com>
31459 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
31460 arg10 and arg11 to itype.
31462 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
31464 * doc/extend.texi (Object Size Checking): Improve grammar.
31466 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
31469 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
31470 that the logic for permitting reciprocal estimates matches that
31473 2017-04-27 Jakub Jelinek <jakub@redhat.com>
31476 * tree.c (type_cache_hasher::equal): Only compare
31477 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
31478 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
31479 non-aggregate element types.
31480 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
31481 about the flag on ARRAY_TYPEs in the comment, formatting fix.
31483 2017-04-27 Richard Biener <rguenther@suse.de>
31485 PR middle-end/80533
31486 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
31487 stripping ARRAY_REFs from MEM_EXPR make sure we're not
31488 keeping a reference to a trailing array.
31490 2017-04-27 Richard Biener <rguenther@suse.de>
31492 PR middle-end/80539
31493 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
31494 being in loop-closed SSA form conservatively.
31495 (chrec_fold_multiply_poly_poly): Likewise.
31497 2017-04-27 Tamar Christina <tamar.christina@arm.com>
31499 PR middle-end/79665
31500 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
31501 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
31503 2017-04-27 Jakub Jelinek <jakub@redhat.com>
31506 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
31507 (aarch64_function_arg_alignment): Return unsigned int again, but still
31508 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
31509 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
31510 Don't emit -Wpsabi note.
31511 (aarch64_function_arg_boundary): Likewise.
31512 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
31515 2017-04-26 Nathan Sidwell <nathan@acm.org>
31517 * tree.h (crc32_unsigned_n): Declare.
31518 (crc32_unsigned, crc32_unsigned): Make inline.
31519 * tree.c (crc32_unsigned_bits): Replace with ...
31520 (crc32_unsigned_n): ... this.
31521 (crc32_unsigned, crc32_byte): Remove.
31522 (crc32_string): Remove unnecessary braces.
31524 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
31526 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
31527 * ipa-inline-analysis.c (MAX_TIME): Remove.
31528 (account_size_time): Use sreal for time.
31529 (dump_inline_summary): Update.
31530 (estimate_function_body_sizes): Update.
31531 (estimate_edge_size_and_time): Update.
31532 (estimate_calls_size_and_time): Update.
31533 (estimate_node_size_and_time): Update.
31534 (inline_merge_summary): Update.
31535 (inline_update_overall_summary): Update.
31536 (estimate_time_after_inlining): Update.
31537 (inline_read_section): Update.
31538 (inline_write_summary): Update.
31539 * ipa-inline.c (compute_uninlined_call_time): Update.
31540 (compute_inlined_call_time): Update.
31541 (recursive_inlining): Update.
31542 (inline_small_functions): Update.
31543 (dump_overall_stats): Update.
31544 * ipa-inline.h: Include sreal.h.
31545 (size_time_entry): Turn time to sreal.
31546 (inline_summary): Turn self_time nad time to sreal.
31548 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
31550 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
31552 (sreal::stream_out, sreal::stream_in): New.
31553 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
31555 2017-04-25 Jakub Jelinek <jakub@redhat.com>
31557 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
31560 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
31563 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
31565 (dimode_scalar_chain::compute_convert_gain): Ditto.
31566 (dimode_scalar_chain::make_vector_copies): Ditto.
31567 (dimode_scalar_chain::convert_reg): Ditto.
31568 (dimode_scalar_chain::convert_insn): Ditto.
31569 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
31570 (VI248_AVX512BW_1): New mode iterator.
31571 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
31572 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
31575 2017-04-25 Martin Sebor <msebor@redhat.com>
31577 PR tree-optimization/80497
31578 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
31579 constants are representable in HOST_WIDE_INT.
31580 (parse_directive): Ditto.
31582 2017-04-25 Martin Sebor <msebor@redhat.com>
31585 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
31586 (new_zero_array): Adjust signature.
31587 (dom_info::dom_init): Used unsigned rather that size_t.
31588 (dom_info::dom_info): Same.
31590 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
31591 Jakub Jelinek <jakub@redhat.com>
31594 * config/arm/arm.c: Include gimple.h.
31595 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
31596 returns negative, increment ncrn only if it returned positive.
31597 (arm_needs_doubleword_align): Return int instead of bool,
31598 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
31599 members, but if there is any such non-FIELD_DECL
31600 > PARM_BOUNDARY aligned decl, return -1 instead of false.
31601 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
31602 returns negative, increment nregs only if it returned positive.
31603 (arm_setup_incoming_varargs): Likewise.
31604 (arm_function_arg_boundary): Emit -Wpsabi note if
31605 arm_needs_doubleword_align returns negative, return
31606 DOUBLEWORD_ALIGNMENT only if it returned positive.
31608 2017-04-25 Marek Polacek <polacek@redhat.com>
31611 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
31612 first argument to type.
31614 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
31617 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
31618 type checks to test for compatibility instead of equality.
31620 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
31621 Jakub Jelinek <jakub@redhat.com>
31624 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
31626 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
31627 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
31628 the alignment computation, but return their maximum in warn_alignment.
31629 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
31630 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
31632 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
31633 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
31636 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31638 * config/arc/simdext.md (dmpyh): Fix typo.
31640 2017-04-25 Richard Biener <rguenther@suse.de>
31642 PR tree-optimization/80492
31643 * alias.c (compare_base_decls): Handle registers with asm
31644 specification conservatively.
31645 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31646 compare_base_decls returning dont-know properly.
31648 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31650 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
31651 (legitimate_offset_address_p): New function.
31652 (arc_legitimate_address_p): Use above function.
31654 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31656 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
31658 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31660 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
31661 ACCH registers whenever they are available.
31663 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31665 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
31666 double regs fix when not used.
31668 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31670 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
31672 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
31673 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
31675 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31677 * config/arc/arc.c (arc_output_addsi): Check for h-register class
31678 when emitting short ADD instructions.
31680 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31682 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
31684 (cmpsi_cc_c_insn): Likewise.
31685 (cbranchsi4_scratch): Compute proper instruction length using
31686 compact_hreg_operand.
31687 * config/arc/predicates.md (compact_hreg_operand): New predicate.
31689 2017-04-25 Richard Biener <rguenther@suse.de>
31691 PR middle-end/80509
31692 * passes.c (pass_manager::pass_manager): Initialize
31693 m_name_to_pass_map.
31695 2017-04-25 Richard Biener <rguenther@suse.de>
31697 PR tree-optimization/79201
31698 * tree-ssa-sink.c (statement_sink_location): Handle calls.
31700 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31703 * config/s390/vector.md: Split MEM->GPR vector moves for
31704 non-s_operand addresses.
31706 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31709 * config/s390/predicates.md (reload_const_wide_int_operand): New
31711 * config/s390/s390.md ("movti"): Remove d/P alternative.
31712 ("movti_bigconst"): New pattern definition.
31714 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31717 * s390-protos.h (s390_expand_cs_hqi): Removed.
31718 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
31719 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
31720 modes as well as CCZ1mode and CCZmode.
31721 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
31722 signature of s390_emit_compare_and_swap.
31723 (s390_expand_cs_hqi): Likewise, make static.
31724 (s390_expand_cs_tdsi): Generate an explicit compare before trying
31725 compare-and-swap, in some cases.
31726 (s390_expand_cs): Wrapper function.
31727 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
31729 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
31730 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
31731 patterns for small and large integers. Forbid symref memory operands.
31732 Move expander to s390.c. Require cc register.
31733 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
31734 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
31735 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
31736 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
31737 symref memory operands. Remove CC mode and call s390_match_ccmode
31739 ("atomic_exchange<mode>"): Allow and implement all integer modes.
31741 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31743 * config/s390/s390.md (define_peephole2): New peephole to help
31744 combining the load-and-test pattern with volatile memory.
31746 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31748 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
31749 with CCZmode for TARGET_Z196.
31751 2017-04-25 Jakub Jelinek <jakub@redhat.com>
31753 PR rtl-optimization/80501
31754 * combine.c (make_compound_operation_int): Set subreg_code to SET
31755 even for AND with mask of the sign bit of mode.
31757 PR rtl-optimization/80500
31758 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
31759 sum's initial value.
31761 2017-04-25 Julian Brown <julian@codesourcery.com>
31762 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31764 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
31766 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
31768 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
31770 2017-04-25 Julian Brown <julian@codesourcery.com>
31771 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31773 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
31774 (thunderx2t99_sha): New Reservation.
31776 2017-04-25 Julian Brown <julian@codesourcery.com>
31777 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31779 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
31780 type for 1-element load.
31782 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
31784 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
31786 2017-04-24 Martin Jambor <mjambor@suse.cz>
31788 PR tree-optimization/80293
31789 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
31790 char arrays not totally scalarizable if it is false.
31791 (analyze_all_variable_accesses): Pass correct value in the new
31792 parameter. Add a statistics counter.
31794 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
31796 PR middle-end/79931
31797 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
31799 2017-04-24 Richard Biener <rguenther@suse.de>
31801 PR tree-optimization/80494
31802 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
31803 out for complex types.
31805 2017-04-24 Richard Biener <rguenther@suse.de>
31807 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
31808 * tree-ssa-sccvn.c (print_scc): Print SCC size.
31809 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
31810 (DFS): Adjust and never fail.
31811 (sccvn_dom_walker::fail): Remove.
31812 (sccvn_dom_walker::before_dom_children): Adjust.
31813 (run_scc_vn): Likewise and never fail.
31814 * tree-ssa-pre.c (pass_pre::execute): Adjust.
31815 (pass_fre::execute): Likewise.
31817 2017-04-24 Richard Biener <rguenther@suse.de>
31819 PR tree-optimization/79725
31820 * tree-ssa-sink.c (statement_sink_location): Return whether
31821 failure reason was zero uses. Move that check later.
31822 (sink_code_in_bb): Deal with zero uses by removing the stmt
31825 2017-04-24 Richard Biener <rguenther@suse.de>
31828 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
31829 pointer-based references.
31831 2017-04-24 Richard Biener <rguenther@suse.de>
31834 * pass_manager.h (pass_manager::operator new): Remove.
31835 (pass_manager::operator delete): Likewise.
31836 * passes.c (pass_manager::operator new): Remove.
31837 (pass_manager::operator delete): Likewise.
31838 (pass_manager::pass_manager): Zero individual pass members.
31840 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
31843 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
31844 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
31845 Check "XEXP (src, 1)" operand here.
31846 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
31847 Check "XEXP (src, 1)" operand here.
31848 (dimode_scalar_chain::make_vector_copies): Detect count register
31849 of a shift instruction. Zero extend count register from QImode
31850 to DImode to satisfy vector shift pattern count operand predicate.
31851 Substitute vector shift count operand with a DImode copy.
31852 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
31855 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
31857 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
31858 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
31859 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
31860 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
31861 (UNSPEC_NOREX_MEM): Remove definition.
31863 2017-04-21 Richard Biener <rguenther@suse.de>
31865 PR tree-optimization/79547
31866 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
31867 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
31868 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
31869 without any constraints.
31871 2017-04-21 Richard Biener <rguenther@suse.de>
31873 PR tree-optimization/78847
31874 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
31876 2017-04-21 Richard Biener <rguenther@suse.de>
31878 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
31879 (build_distinct_type_copy): Likewise.
31880 (build_variant_type_copy): Likewise.
31881 * tree.c (build_qualified_type): Pass down mem-stat info.
31882 (build_distinct_type_copy): Likewise.
31883 (build_variant_type_copy): Likewise.
31885 2017-04-21 Richard Biener <rguenther@suse.de>
31887 PR tree-optimization/80237
31888 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
31890 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
31891 for a simplified result.
31893 2016-04-21 Richard Biener <rguenther@suse.de>
31895 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
31896 sth as strict as a simple_iv but a chrec without symbols and an
31897 operand defined in the loop we are peeling (and not some subloop).
31898 (propagate_constants_for_unrolling): Propagate all constants.
31900 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
31903 * config/i386/i386.c (print_reg): Remove assert for disalowed
31904 regno values, call output_operand_lossage instead.
31906 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
31909 * config/i386/constraints.md (Yc): New register constraint.
31910 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
31911 Use Yc constraint for alternative 2 of operand 0. Remove
31912 preferred_for_speed attribute.
31914 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
31916 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
31917 lastprivate clauses in SIMT case.
31919 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
31921 * doc/invoke.texi (-Wextra-semi): Document new warning option.
31923 2017-04-20 Richard Biener <rguenther@suse.de>
31925 PR tree-optimization/57796
31926 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
31927 as N scalar stores.
31928 (vect_model_load_cost): Cost gathers as N scalar loads.
31930 2017-04-20 Richard Biener <rguenther@suse.de>
31932 * ggc-page.c (ggc_allocated_p): Rename to ...
31933 (safe_lookup_page_table_entry): ... this and return the lookup
31935 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
31937 2017-04-20 Richard Biener <rguenther@suse.de>
31939 PR tree-optimization/80453
31940 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
31941 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
31942 from the conditions.
31943 (vn_phi_eq): Pass them down.
31944 (vn_phi_lookup): Record them.
31945 (vn_phi_insert): Likewise.
31947 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
31949 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
31950 uninitialized variable warning to avoid buffer overrun.
31952 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
31955 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
31956 is suppressed for '{ 0 }' in C.
31958 2017-04-20 Jakub Jelinek <jakub@redhat.com>
31960 * BASE-VER: Set to 8.0.0.
31962 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
31964 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
31965 priority .init_array and .fini_array section with SECTION_NOTYPE
31968 2017-04-20 Jakub Jelinek <jakub@redhat.com>
31970 PR middle-end/80423
31971 * tree.h (build_array_type): Add typeless_storage default argument.
31972 * tree.c (type_cache_hasher::equal): Also compare
31973 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
31974 (build_array_type): Add typeless_storage argument, set
31975 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
31977 (build_nonshared_array_type): Adjust build_array_type_1 caller.
31978 (build_array_type): Likewise. Add typeless_storage argument.
31980 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
31981 Jakub Jelinek <jakub@redhat.com>
31983 PR tree-optimization/80426
31984 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
31985 operation on symbolic operands, also compute the overflow for the
31986 invariant part when the operation degenerates into a negation.
31988 2017-04-19 Jakub Jelinek <jakub@redhat.com>
31991 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
31992 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
31995 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
31997 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
32000 * config/avr/avr.c (tree.h): Include it.
32001 (cgraph.h): Include it.
32002 (avr_encode_section_info): Don't warn for uninitialized progmem
32003 variable if it's just an alias.
32005 2017-04-19 Richard Biener <rguenther@suse.de>
32008 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
32009 when needed by AutoPGO.
32011 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
32014 * doc/lto.texi: Remove an extra 'that'.
32016 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
32018 PR rtl-optimization/80429
32019 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
32020 are only used in debug insns.
32022 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32023 Vladimir Makarov <vmakarov@redhat.com>
32025 * config/sparc/predicates.md (input_operand): Add comment. Return
32026 true for any memory operand when LRA is in progress.
32027 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
32029 2017-04-18 Jeff Law <law@redhat.com>
32032 * mips.md ({return,simple_return}_internal): Do not overwrite
32035 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32037 PR tree-optimization/80443
32038 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
32039 instead of adding 1, subtract -1 and similarly instead of subtracting
32042 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
32044 PR rtl-optimization/80357
32045 * haifa-sched.c (tmp_bitmap): New variable.
32046 (model_recompute): Handle duplicate use records.
32047 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
32048 (free_global_sched_pressure_data): Free it.
32050 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
32053 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
32054 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32055 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32056 instead of SYSTEM_HEADER_DIR.
32058 2017-04-18 Jeff Law <law@redhat.com>
32060 PR middle-end/80422
32061 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
32062 predecessors after walking up the insn chain.
32064 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32067 * dwarf2out.c (modified_type_die): Try harder not to emit internal
32068 sizetype type into debug info.
32070 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32073 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
32074 unneeded test for TARGET_UPPER_REGS_SF.
32075 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
32077 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32080 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
32081 instead of gsi_after_labels.
32083 2017-04-18 Jeff Law <law@redhat.com>
32085 * regcprop.c (maybe_mode_change): Avoid creating copies of the
32089 2017-04-13 Jeff Law <law@redhat.com>
32090 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
32091 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
32093 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
32096 * config/avr/avr.c (intl.h): Include it.
32097 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
32099 2017-04-18 Martin Liska <mliska@suse.cz>
32101 PR gcov-profile/78783
32102 * gcov-tool.c (gcov_output_files): Validate that destination
32103 file is either removed by the tool or by a user.
32105 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
32106 Guy Benyei <guybe@mellanox.com>
32108 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
32109 block, and do not negate it, the stored id is already negative.
32111 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
32113 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
32115 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32118 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
32119 masks of options that should be turned off if the VSX vector
32120 options are turned off.
32121 (OTHER_P8_VECTOR_MASKS): Likewise.
32122 (OTHER_VSX_VECTOR_MASKS): Likewise.
32123 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
32124 rs6000_disable_incompatible_switches to validate no type switches
32126 (rs6000_incompatible_switch): New function to disallow turning on
32127 other vector options if -mno-vsx, -mno-power8-vector, or
32128 -mno-power9-vector are specified.
32130 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32132 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
32134 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32136 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
32137 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
32138 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
32139 (ARG_POINTER_CFA_OFFSET): Likewise.
32141 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32143 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
32144 conditions to take advantage of various optimizations.
32146 2017-04-13 Jeff Law <law@redhat.com>
32148 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
32149 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
32150 (zero_extendsidi2_dext): Likewise.
32152 2017-04-13 Jakub Jelinek <jakub@redhat.com>
32155 * fold-const.c (fold_ternary_loc): Revert
32156 use op0 instead of fold_convert_loc (loc, type, arg0) part of
32159 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
32161 PR rtl-optimization/80343
32162 * lra-remat.c (update_scratch_ops): Assign original hard reg to
32163 new scratch pseudo.
32165 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
32168 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
32169 to ubsan_encode_value.
32171 2017-04-13 Jeff Law <law@redhat.com>
32173 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
32174 appearing in DEBUG_INSNs.
32176 2017-04-13 Martin Liska <mliska@suse.cz>
32178 PR gcov-profile/80413
32179 * gcov-io.c (gcov_write_string): Copy to buffer just when
32180 allocated size is greater than zero.
32182 2017-04-13 Jakub Jelinek <jakub@redhat.com>
32185 * dwarf2out.c (decls_for_scope): Ignore declarations of
32186 current_function_decl in BLOCK_NONLOCALIZED_VARS.
32188 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
32191 * ipa-visibility.c (non_local_p): Fix typos.
32192 (localize_node): When localizing symbol in same comdat group,
32193 dissolve the group only when we know external symbols are going
32195 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
32197 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32199 PR tree-optimization/79390
32200 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
32201 order does not result in usable sequence, retry with reversed operand
32207 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
32208 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
32209 op0 instead of fold_convert_loc (loc, type, arg0).
32211 2017-04-12 Jeff Law <law@redhat.com>
32213 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
32214 has a delay slot in the generated code.
32216 * config/cris/cris.md (cris_preferred_reload_class): Return
32217 GENNONACR_REGS rather than GENERAL_REGS.
32219 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32222 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
32223 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
32224 signedness of the result type.
32226 2017-04-12 Richard Biener <rguenther@suse.de>
32227 Jeff Law <law@redhat.com>
32229 PR tree-optimization/80359
32230 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
32231 trim stores to TARGET_MEM_REFs.
32233 2017-04-12 Richard Biener <rguenther@suse.de>
32235 PR tree-optimization/79390
32236 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
32237 threading case even more.
32239 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
32242 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
32243 for quad_address_p for TImode, instead of just not indexed_address.
32245 2017-04-12 Richard Biener <rguenther@suse.de>
32246 Bernd Edlinger <bernd.edlinger@hotmail.de>
32248 PR middle-end/79671
32249 * alias.c (component_uses_parent_alias_set_from): Handle
32250 TYPE_TYPELESS_STORAGE.
32251 (get_alias_set): Likewise.
32252 * tree-core.h (tree_type_common): Add typeless_storage flag.
32253 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
32254 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
32255 for types containing members with TYPE_TYPELESS_STORAGE.
32256 (place_field): Likewise.
32257 (layout_type): Likewise for ARRAY_TYPE.
32258 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
32259 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
32260 TYPE_TYPELESS_STORAGE.
32261 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
32263 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32266 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32267 first argument to type.
32269 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32273 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
32274 CONST0_RTX (mode) rather than const0_rtx where appropriate.
32275 (rs6000_expand_binop_builtin): Likewise.
32276 (rs6000_expand_ternop_builtin): Likewise; also add missing
32277 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
32278 vshasigma built-ins.
32279 * doc/extend.texi: Document that vec_xxpermdi's third argument
32280 must be a constant.
32282 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
32284 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
32285 Use shift_const cost parameter when calculating gain of STV shifts.
32287 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
32289 PR rtl-optimization/70478
32290 * lra-constraints.c (process_alt_operands): Check memory for
32291 disfavoring memory insn operand.
32293 2017-04-11 Jakub Jelinek <jakub@redhat.com>
32295 PR middle-end/80100
32296 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
32297 left shift in unsigned HOST_WIDE_INT type.
32299 PR rtl-optimization/80385
32300 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
32301 (not (neg X)) into (plus X -1) for complex or non-integral modes.
32304 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
32305 if they have any depend clauses.
32307 2017-04-11 Martin Liska <mliska@suse.cz>
32310 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
32311 * ipa-split.c (split_function): Create a local comdat symbol
32312 if caller is in a comdat group.
32314 2017-04-11 Martin Liska <mliska@suse.cz>
32317 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
32320 2017-04-11 Martin Sebor <msebor@redhat.com>
32322 PR middle-end/80364
32323 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
32324 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
32326 (directive::set_width, directive::set_precision, format_character):
32328 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
32331 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
32334 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
32335 conflict, set target->arch_name instead of target->cpu_name.
32337 2017-04-11 Richard Biener <rguenther@suse.de>
32339 PR tree-optimization/80374
32340 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
32341 build_zero_cst, remove fold_convertible_p check again.
32343 2017-04-11 Martin Liska <mliska@suse.cz>
32346 * ubsan.c (instrument_object_size): Do not instrument register
32349 2017-04-11 Jakub Jelinek <jakub@redhat.com>
32352 * config/i386/i386-builtin-types.def
32353 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
32354 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
32355 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
32356 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
32357 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
32358 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
32359 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
32360 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
32361 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
32362 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
32363 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
32364 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
32365 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
32366 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
32367 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
32368 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
32369 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
32370 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
32371 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
32372 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
32373 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
32374 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
32375 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
32376 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
32377 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
32378 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
32379 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
32380 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
32381 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
32382 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
32383 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
32384 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
32385 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
32386 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
32387 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
32388 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
32389 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
32390 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
32391 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
32392 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
32393 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
32394 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
32395 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
32396 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
32397 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
32398 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
32400 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
32401 flag to second_arg_count, handle 4 argument function type _COUNT
32402 aliases, handle second_arg_count on second argument rather than last.
32404 2017-04-10 Jeff Law <law@redhat.com>
32406 PR tree-optimization/80374
32407 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
32408 record anything if we can not convert integer_zero_node to the
32411 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
32414 * config/rs6000/rs6000.c (rs6000_option_override_internal):
32415 Enhance special handling given to the TARGET_P9_MINMAX option in
32416 relation to certain other options.
32418 2017-04-10 Bin Cheng <bin.cheng@arm.com>
32420 PR tree-optimization/80153
32421 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
32422 remove POINTER_PLUS_EXPR's base part directly, rather than through
32425 2017-04-10 Richard Biener <rguenther@suse.de>
32426 Bin Cheng <bin.cheng@arm.com>
32428 PR tree-optimization/80153
32429 * tree-affine.c (aff_combination_to_tree): Get base pointer from
32430 the first element of pointer type aff_tree. Build result expr in
32432 (add_elt_to_tree): Convert to type unconditionally. Remove other
32433 fold_convert calls.
32434 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
32435 (rewrite_use_nonlinear_expr): Check invariant using iv information.
32437 2017-04-10 Richard Biener <rguenther@suse.de>
32439 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32442 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
32444 PR rtl-optimization/70478
32445 * lra-constraints.c (curr_small_class_check): New.
32446 (update_and_check_small_class_inputs): New.
32447 (process_alt_operands): Update curr_small_class_check. Disfavor
32448 alternative insn memory operands. Check available regs for small
32451 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
32454 * config/mips/mips.opt (-mvirt): Update description.
32455 * doc/invoke.texi (-mvirt): Likewise.
32457 2017-04-10 Richard Biener <rguenther@suse.de>
32459 PR middle-end/80362
32460 * fold-const.c (fold_binary_loc): Look at unstripped ops when
32461 looking for NEGATE_EXPR in -A / -B to A / B folding.
32463 2017-04-10 Martin Liska <mliska@suse.cz>
32465 PR gcov-profile/80224
32466 * gcov.c (print_usage): Fix usage string.
32467 (get_gcov_intermediate_filename): Remove.
32468 (output_gcov_file): Use both for normal and intermediate format.
32469 (generate_results): Do not initialize special file for
32470 intermediate format.
32472 2017-04-10 Richard Biener <rguenther@suse.de>
32474 PR tree-optimization/80304
32475 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
32478 2017-04-10 Nathan Sidwell <nathan@acm.org>
32481 * config/rs6000/rs6000.c (rs6000_vector_type): New.
32482 (rs6000_init_builtins): Use it.
32484 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32486 * config/arm/arm.md (<mrc>): Add mode to SET source.
32487 (<mrrc>): Likewise.
32489 2017-04-10 Richard Biener <rguenther@suse.de>
32491 PR middle-end/80344
32492 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
32494 2017-04-10 Jakub Jelinek <jakub@redhat.com>
32497 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
32498 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
32499 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
32500 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
32501 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
32502 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
32503 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
32504 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
32505 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
32506 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
32507 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
32508 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
32509 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
32510 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
32511 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
32512 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
32513 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
32514 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
32515 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
32516 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
32517 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
32518 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
32519 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
32521 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
32523 PR rtl-optimization/70478
32524 * lra-constraints.c: Reverse the last patch.
32526 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
32528 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
32529 Add comment for WCHAR_T.
32531 2017-04-08 Martin Liska <mliska@suse.cz>
32534 2017-04-07 Martin Liska <mliska@suse.cz>
32537 * ipa-split.c (split_function): Add function part to a same comdat
32540 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32543 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
32545 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
32547 * rs6000/rs6000.c (vec_load_pendulum): Rename...
32548 (vec_pairing): ...to this.
32549 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
32550 (rs6000_sched_init): Adjust for name change.
32551 (struct rs6000_sched_context): Likewise.
32552 (rs6000_init_sched_context): Likewise.
32553 (rs6000_set_sched_context): Likewise.
32555 2017-04-07 Jakub Jelinek <jakub@redhat.com>
32561 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
32563 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
32564 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
32565 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
32567 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
32569 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
32571 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
32573 PR rtl-optimization/70703
32574 * ira-color.c (update_conflict_hard_regno_costs): Use
32575 int64_t instead of HOST_WIDE_INT.
32577 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
32579 PR rtl-optimization/70478
32580 * lra-constraints.c (process_alt_operands): Disfavor alternative
32581 insn memory operands.
32583 2017-04-07 Jeff Law <law@redhat.com>
32585 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
32586 CALL and NOTE_INSN_CALL_ARG_LOCATION.
32588 2017-04-07 Martin Liska <mliska@suse.cz>
32591 * config/aarch64/aarch64.c (aarch64_process_target_attr):
32592 Show error message instead of an ICE.
32594 2017-04-07 Martin Liska <mliska@suse.cz>
32597 * ipa-split.c (split_function): Add function part to a same comdat
32600 2017-04-07 Richard Biener <rguenther@suse.de>
32602 PR middle-end/80341
32603 * tree.c (get_unwidened): Also handle ! for_type case for
32605 * convert.c (do_narrow): Split out from ...
32606 (convert_to_integer_1): ... here. Do not pass final truncation
32607 type to get_unwidened for TRUNC_DIV_EXPR.
32609 2017-04-07 Richard Biener <rguenther@suse.de>
32611 * tree-affine.c (wide_int_ext_for_comb): Take type rather
32613 (aff_combination_const): Adjust.
32614 (aff_combination_scale): Likewise.
32615 (aff_combination_add_elt): Likewise.
32616 (aff_combination_add_cst): Likewise.
32617 (aff_combination_convert): Likewise.
32618 (add_elt_to_tree): Likewise. Remove unused argument.
32619 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
32621 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
32623 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
32625 * config/arm/arm.c (arm_default_short_enums): Use
32626 ARM_DEFAULT_SHORT_ENUMS.
32627 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
32629 2017-04-06 Jakub Jelinek <jakub@redhat.com>
32632 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
32633 members with redundant out-of-class redeclaration.
32635 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
32638 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
32639 * config/i386/i386.md (*zero_extendsidi2):
32640 Add (?*x,*x) and (?*v,*v) alternatives.
32642 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
32645 * config/i386/i386.c (ix86_expand_builtin)
32646 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
32647 mode from insn data. Convert operands to insn operand mode.
32648 Copy operands that don't satisfy insn predicate to a register.
32650 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
32652 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
32655 2017-04-06 Richard Biener <rguenther@suse.de>
32657 PR tree-optimization/80334
32658 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
32659 preserve alignment of accesses.
32661 2017-04-06 Richard Biener <rguenther@suse.de>
32663 PR tree-optimization/80262
32664 * tree-sra.c (build_ref_for_offset): Preserve address-space
32666 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
32667 Drop useless address-space information on MEM_REF offsets.
32669 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
32671 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
32673 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
32675 PR rtl-optimization/70703
32676 * ira-color.c (update_conflict_hard_regno_costs): Use
32677 HOST_WIDE_INT instead of long.
32679 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
32682 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
32683 not defined for x86_64 target. Add -mmmx target option when __SSE2__
32685 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
32686 for x86_64 target. Handle -m3dnowa option.
32688 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
32690 PR rtl-optimization/70703
32691 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
32692 (update_conflict_hard_regno_costs): Use long instead of unsigned
32693 arithmetic for cost calculation.
32695 2017-04-05 Jakub Jelinek <jakub@redhat.com>
32696 Bernd Edlinger <bernd.edlinger@hotmail.de>
32699 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
32702 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
32705 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
32706 ptr_mode with Pmode throughout.
32707 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
32708 into probe_stack_range and use DImode.
32710 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
32713 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
32714 call_eh_return is true.
32716 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32718 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
32719 Initialize last_match_fntype_index.
32721 2017-04-05 Jakub Jelinek <jakub@redhat.com>
32724 * tree-nvr.c: Include internal-fn.h.
32725 (pass_return_slot::execute): Ignore internal calls without
32728 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32729 Richard Biener <rguenther@suse.de>
32732 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
32733 captures used multiple times, except for the last use.
32734 * generic-match-head.c: Include gimplify.h.
32736 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32738 PR tree-optimization/79390
32739 * target.h (struct noce_if_info): Declare.
32740 * targhooks.h (default_noce_conversion_profitable_p): Declare.
32741 * target.def (noce_conversion_profitable_p): New target hook.
32742 * ifcvt.h (struct noce_if_info): New type, moved from ...
32743 * ifcvt.c (struct noce_if_info): ... here.
32744 (noce_conversion_profitable_p): Renamed to ...
32745 (default_noce_conversion_profitable_p): ... this. No longer
32747 (noce_try_store_flag_constants, noce_try_addcc,
32748 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
32749 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
32750 instead of noce_conversion_profitable_p.
32751 * config/i386/i386.c: Include ifcvt.h.
32752 (ix86_option_override_internal): Don't override
32753 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
32754 (ix86_noce_conversion_profitable_p): New function.
32755 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
32756 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
32757 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
32758 * doc/tm.texi: Regenerated.
32760 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32762 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
32765 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32768 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
32769 instructions for small multiply cores.
32771 2017-04-04 Jeff Law <law@redhat.com>
32773 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
32775 (mips_expand_vec_perm_const): Initialize elements in orig_perm
32776 that are not set by the loop over the elements.
32778 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32781 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
32782 int mode, convert_modes it to mode as unsigned, otherwise use
32783 lowpart_subreg to mode rather than SImode.
32784 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
32785 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
32786 Use DImode instead of SImode for the shift count operand.
32787 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
32790 2017-04-04 Richard Biener <rguenther@suse.de>
32792 PR middle-end/80281
32793 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
32794 arithmetic done for the negate or the plus. Simplify.
32795 (A - (-B) -> A + B): Likewise.
32796 * fold-const.c (split_tree): Make sure to not negate pointers.
32798 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
32800 PR rtl-optimization/60818
32801 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
32802 a compare of comparisons with the thing compared if this results
32803 in a different machine mode.
32805 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
32807 * alias.c (base_alias_check): Fix typo in comment.
32808 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
32809 * cgraphunit.c (symbol_table::compile): Likewise.
32810 * collect2.c (maybe_run_lto_and_relink): Likewise.
32811 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
32812 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
32813 * config/avr/avr.c (avr_map_op_t): Likewise.
32814 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
32815 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
32816 * config/epiphany/epiphany.md (movcc): Likewise.
32817 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
32818 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
32820 * config/mips/mips.c (mips_save_restore_reg): Likewise.
32821 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
32822 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
32823 * config/sh/sh.c (sh_rtx_costs): Likewise.
32824 * fold-const.c (fold_truth_andor): Likewise.
32825 * genautomata.c (collapse_flag): Likewise.
32826 * gengtype.h (struct type::u::s): Likewise.
32827 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
32828 * input.c (FORMAT_AMOUNT): Likewise.
32829 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
32830 (known_aggs_to_agg_replacement_list): Likewise.
32831 * ipa-inline-analysis.c: Likewise.
32832 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
32833 * ipa-polymorphic-call.c
32834 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
32835 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
32836 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
32838 * modulo-sched.c (apply_reg_moves): Likewise.
32839 * omp-expand.c (build_omp_regions_1): Likewise.
32840 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
32841 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
32842 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
32843 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
32844 * value-prof.c: Likewise.
32845 * var-tracking.c (val_reset): Likewise.
32847 2017-04-03 Richard Biener <rguenther@suse.de>
32849 PR tree-optimization/80275
32850 * fold-const.c (split_address_to_core_and_offset): Handle
32853 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
32855 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
32856 descriptors is at least equal to that of functions.
32858 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
32860 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
32862 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
32865 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
32866 (mov<IMOD4:mode>): New expander.
32867 (*mov<IMOD4:mode>_internal): New insn and split pattern.
32869 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
32871 PR rtl-optimization/79405
32872 * fwprop.c (propagations_left): New variable.
32873 (forward_propagate_into): Decrement it.
32874 (fwprop_init): Initialize it.
32875 (fw_prop): If the variable has reached zero, stop propagating.
32876 (fwprop_addr): Ditto.
32878 2017-03-31 Jakub Jelinek <jakub@redhat.com>
32881 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
32882 a FUNCTION_DECL, pass it as decl instead of origin to
32885 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
32887 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
32890 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
32893 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
32894 TARGET_VSX_SMALL_INTEGER.
32896 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32898 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
32899 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
32901 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
32903 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
32904 extraction from odd-numbered MSA register.
32906 2017-03-31 Jakub Jelinek <jakub@redhat.com>
32908 PR middle-end/80173
32909 * expmed.c (store_bit_field_1): Don't attempt to create
32910 a word subreg out of hard registers wider than word if they
32911 have HARD_REGNO_NREGS of 1 for their mode.
32913 PR middle-end/80163
32914 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
32915 conversions to integer types wider than word and pointer.
32918 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
32919 (rtx_equal_for_cselib_p): Pass 0 to it.
32920 * cselib.c (cselib_hasher::equal): Likewise.
32921 (rtx_equal_for_cselib_1): Add depth argument. If depth
32922 is 128, don't look up VALUE locs and punt. Increment
32923 depth in recursive calls when walking VALUE locs.
32925 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
32927 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
32928 (make_gcov_file_name): Use the canonical path name for generating
32930 (read_line): Fix handling of files with ascii null bytes.
32932 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
32934 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
32935 to initialise a vector register instead
32936 of using a const_int.
32938 2017-03-30 Jakub Jelinek <jakub@redhat.com>
32940 PR translation/80189
32941 * gimplify.c (omp_default_clause): Use %qs instead of %s in
32942 diagnostic messages.
32944 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
32947 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
32948 (dfp_diex_<mode>): Update mode of operand 1.
32949 * doc/extend.texi (dxex, dxexq): Document change to return type.
32950 (diex, diexq): Document change to argument type.
32952 2017-03-30 Martin Jambor <mjambor@suse.cz>
32955 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
32956 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
32957 it reflects the signature changes performed at the callee side.
32958 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
32959 to cgraph_build_function_type_skip_args.
32960 (build_function_decl_skip_args): Adjust call to the above function.
32962 2017-03-30 Jakub Jelinek <jakub@redhat.com>
32965 * config/i386/sse.md
32966 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
32967 register as dest whenever it is a MEM not rtx_equal_p to the
32968 corresponding dup operand, and when forcing into reg move the
32969 reg into the memory afterwards.
32970 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
32971 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
32972 for the force_reg mode.
32973 (avx512vl_vextractf128<mode>): Use register as dest either
32974 always when a MEM, or when it is a MEM not rtx_equal_p to the
32975 corresponding dup operand, or even not when it is a CONST_VECTOR
32976 depending on the mode and lo vs. hi.
32977 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
32979 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
32980 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
32981 Likewise. Require that operands[2] is even.
32982 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
32983 Remove extraneous parens. Require that operands[2] is a multiple
32985 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
32986 operands[0] is a MEM if <mask_applied>, the predicates/constraints
32987 disallow memory then.
32989 2017-03-30 Richard Biener <rguenther@suse.de>
32991 PR tree-optimization/77498
32992 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
32993 to non-constants over backedges.
32995 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
32997 PR rtl-optimization/80233
32998 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
32999 as last_combined_insn. Do not test for BARRIER_P separately.
33001 2017-03-29 Andreas Schwab <schwab@suse.de>
33004 * calls.c (prepare_call_address): Convert funexp to Pmode before
33005 copying to temp reg.
33007 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33009 PR tree-optimization/80158
33010 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
33011 Handle possible future case of more than one alternate
33013 (replace_rhs_if_not_dup): Likewise.
33014 (replace_one_candidate): Likewise.
33016 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
33018 PR rtl-optimization/80193
33019 * ira.c (ira): Do not check allocation for LRA.
33021 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
33023 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
33024 (nvptx_output_simt_exit): Declare.
33025 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
33026 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
33027 (init_softstack_frame): Move initialization of crtl->is_leaf to...
33028 (nvptx_declare_function_name): ...here. Emit declaration of local
33029 memory space buffer for omp_simt_enter insn.
33030 (nvptx_output_unisimt_switch): New.
33031 (nvptx_output_softstack_switch): New.
33032 (nvptx_output_simt_enter): New.
33033 (nvptx_output_simt_exit): New.
33034 * config/nvptx/nvptx.h (struct machine_function): New fields
33035 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
33036 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
33037 (UNSPECV_SIMT_EXIT): Ditto.
33038 (omp_simt_enter_insn): New insn.
33039 (omp_simt_enter): New expansion.
33040 (omp_simt_exit): New insn.
33041 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
33043 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
33044 (expand_GOMP_SIMT_ENTER_ALLOC): New.
33045 (expand_GOMP_SIMT_EXIT): New.
33046 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
33047 (GOMP_SIMT_ENTER_ALLOC): Ditto.
33048 (GOMP_SIMT_EXIT): Ditto.
33049 * target-insns.def (omp_simt_enter): New insn.
33050 (omp_simt_exit): Ditto.
33051 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
33053 (lower_rec_simd_input_clauses): Implement SIMT privatization.
33054 (lower_rec_input_clauses): Likewise.
33055 (lower_lastprivate_clauses): Handle SIMT privatization.
33057 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
33058 (ompdevlow_adjust_simt_enter): New.
33059 (find_simtpriv_var_op): New.
33060 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
33061 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
33063 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
33064 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
33065 (copy_decl_for_dup_finish): Ditto.
33067 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
33069 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
33072 * config/i386/i386.c (ix86_option_override_internal): Always
33073 allow -mpreferred-stack-boundary=3 for 64-bit targets.
33075 2017-03-28 Bin Cheng <bin.cheng@arm.com>
33077 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
33079 2017-03-28 Bin Cheng <bin.cheng@arm.com>
33081 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
33082 mark new edge's irreducible flag accordign to it.
33083 (vect_do_peeling): Check loop preheader edge's irreducible flag
33084 and pass it to function slpeel_add_loop_guard.
33086 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
33088 PR tree-optimization/80218
33089 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33090 Update block frequencies and counts.
33092 2017-03-28 Richard Biener <rguenther@suse.de>
33094 PR tree-optimization/78644
33095 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
33096 of a simplification result we may not use it at all.
33098 2017-03-28 Richard Biener <rguenther@suse.de>
33101 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
33102 without arguments, generate default definition of a SSA name.
33104 2017-03-28 Richard Biener <rguenther@suse.de>
33106 PR middle-end/80222
33107 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
33108 TYPE_REF_CAN_ALIAS_ALL references.
33109 * fold-const.c (fold_indirect_ref_1): Likewise.
33111 2017-03-28 Martin Liska <mliska@suse.cz>
33114 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
33115 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
33117 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
33118 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
33120 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
33121 (EXTRA_SPECS): Define.
33122 (SUBTARGET_EXTRA_SPECS): Likewise.
33123 (SUBTARGET_CPP_SPEC): Likewise.
33124 * config/arc/elf.h (EXTRA_SPECS): Renamed to
33125 SUBTARGET_EXTRA_SPECS.
33126 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
33128 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
33130 * config/arc/simdext.md (vst64_insn): Update pattern.
33131 (vld32wh_insn): Likewise.
33132 (vld32wl_insn): Likewise.
33133 (vld64_insn): Likewise.
33134 (vld32_insn): Likewise.
33136 2017-03-28 Marek Polacek <polacek@redhat.com>
33139 * fold-const.c (fold_comparison): Use protected_set_expr_location
33140 instead of SET_EXPR_LOCATION.
33142 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
33144 * tree.c (add_expr): Avoid name lookup warning.
33146 2017-03-27 Jeff Law <law@redhat.com>
33148 PR tree-optimization/80216
33149 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
33150 function name. Limit recursion depth.
33151 (record_temporary_equivalences): Corresponding changes.
33153 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
33155 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
33158 2017-03-27 Jakub Jelinek <jakub@redhat.com>
33161 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
33163 * cfgcleanup.c (reg_note_cfa_p): New array.
33164 (insns_have_identical_cfa_notes): New function.
33165 (old_insns_match_p): Don't cross-jump in between /f
33166 and non-/f instructions. If both i1 and i2 are frame related,
33167 verify all CFA notes, their order and content.
33169 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
33172 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
33173 HImode and SImode with zero extend to DImode to one insn.
33174 (bswap<mode>2_extenddi): Likewise.
33175 (bswapsi2_extenddi): Likewise.
33176 (bswaphi2_extendsi): Likewise.
33177 (bswaphi2): Combine bswap HImode and SImode into one insn.
33178 Separate memory insns from swapping register.
33179 (bswapsi2): Likewise.
33180 (bswap<mode>2): Likewise.
33181 (bswaphi2_internal): Delete, no longer used.
33182 (bswapsi2_internal): Likewise.
33183 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
33184 store, and gpr<-gpr swap insns.
33185 (bswap<mode>2_store): Likewise.
33186 (bswaphi2_reg): Register only splitter, combine with the splitter.
33187 (bswaphi2 splitter): Likewise.
33188 (bswapsi2_reg): Likewise.
33189 (bswapsi2 splitter): Likewise.
33190 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
33191 the insns into load, store, and register/register insns.
33192 (bswapdi2_ldbrx): Likewise.
33193 (bswapdi2_load): Likewise.
33194 (bswapdi2_store): Likewise.
33195 (bswapdi2_reg): Likewise.
33197 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
33199 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
33200 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
33202 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
33205 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
33207 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33208 special handling for target option conflicts between dform
33209 options (-mpower9-dform, -mpower9-dform-vector,
33210 -mpower9-dform-scalar) and -mno-direct-move.
33212 2017-03-27 Richard Biener <rguenther@suse.de>
33214 PR tree-optimization/80181
33215 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
33217 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33219 * config/arc/predicates.md (move_double_src_operand): Replace the
33220 call to move_double_src_operand with a call to address_operand.
33222 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33224 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
33225 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
33226 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
33228 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33230 * config/arc/predicates.md (long_immediate_loadstore_operand):
33231 Consider scaled addresses cases.
33233 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33235 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
33236 restored when in interrupt.
33237 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
33238 doesn't have delay slot.
33240 2017-03-27 Richard Biener <rguenther@suse.de>
33243 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
33244 inlined thunk clones.
33246 2017-03-27 Jakub Jelinek <jakub@redhat.com>
33249 * asan.c (instrument_derefs): Copy over last operand from
33250 original COMPONENT_REF to the new COMPONENT_REF with
33251 DECL_BIT_FIELD_REPRESENTATIVE.
33252 * ubsan.c (instrument_object_size): Likewise.
33254 2017-03-27 Richard Biener <rguenther@suse.de>
33256 PR tree-optimization/80170
33257 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
33258 sure DR/SCEV didnt fold in constants we do not see when looking
33259 at the reference base alignment.
33261 2017-03-27 Richard Biener <rguenther@suse.de>
33263 PR middle-end/80171
33264 * gimple-fold.c (fold_ctor_reference): Properly guard against
33265 NULL return value from canonicalize_constructor_val.
33267 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
33270 * config/i386/i386.c (ix86_expand_builtin)
33271 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
33272 flags reg setting and flags reg using instructions.
33273 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
33274 clobbering instructions to zero extend op2.
33276 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
33278 * doc/install.texi (Configuration) <--with-aix-soname>:
33279 Update link to AIX ld.
33281 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
33283 PR rtl-optimization/80160
33284 PR rtl-optimization/80159
33285 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
33286 reg_alternate_class into account.
33288 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
33291 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
33292 to consider in curr_insn_transform.
33294 2017-03-24 Jakub Jelinek <jakub@redhat.com>
33296 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
33297 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
33298 and emit_mode_inner.
33300 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33302 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
33303 argument to the overloaded builtin variants. Use the new flag to
33304 deprecate certain builtin variants.
33305 * config/s390/s390-builtin-types.def: Add new builtin types.
33306 * config/s390/s390-builtins.h: Support new flags field for
33307 overloaded builtins.
33308 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
33309 (s390_macro_to_expand): Enable vector float data type.
33310 (s390_cpu_cpp_builtins_internal): Indicate support of the new
33311 builtins by incrementing the __VEC__ version number.
33312 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
33314 (s390_resolve_overloaded_builtin): Emit error messages depending
33315 on the builtin flags.
33316 * config/s390/s390.c (s390_expand_builtin): Support additional
33317 flags argument. Change error message to match the messages
33318 emitted in s390-c.c.
33319 * config/s390/s390.md: New UNSPEC_* constants.
33320 (op_type): Add new instruction types.
33321 * config/s390/vecintrin.h: Add new builtins and test data class
33323 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
33324 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
33325 (VEC_INEXACT, VEC_NOINEXACT): New constants.
33326 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
33327 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
33328 ("vec_mergel<mode>"): V_HW -> VEC_HW.
33330 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
33331 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
33332 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
33333 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
33335 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
33336 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
33337 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
33338 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
33340 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
33341 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
33342 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
33343 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
33344 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
33345 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
33346 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
33348 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
33349 ("vec_scatter_element<V_HW_4:mode>_DI")
33350 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
33351 ("vec_fpint<mode>", "vflls")
33352 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
33353 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
33354 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
33355 ("*vec_cmphe<mode>_cc"): ... these.
33357 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
33358 mode constant instead of magic value.
33360 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33362 * config/s390/s390.c (s390_expand_vec_compare): Support other
33363 vector floating point modes than just V2DF.
33364 (s390_expand_vcond): Likewise.
33365 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
33366 (s390_cannot_change_mode_class): Prevent mode changes between TF
33367 and V1TF in vector registers.
33368 * config/s390/s390.md (DF, SF): New mode attributes.
33369 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
33370 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
33371 SFmode support for VRs.
33372 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
33374 (VFT, VF_HW): New mode iterators.
33375 (vw, sdx): New mode attributes.
33376 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
33377 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
33378 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
33379 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
33380 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
33381 also the new vector floating point modes. Renaming to ...
33383 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
33384 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
33385 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
33386 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
33387 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
33388 ("vec_unordered<mode>"): ... these.
33390 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
33391 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
33392 ("*vec_extendv2df"): New insn definitions.
33394 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33396 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
33397 ("mulditi3_2", "*muldi3_sign"): New patterns.
33398 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
33399 rename the pattern definition.
33401 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33403 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
33405 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
33407 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33409 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
33410 instruction if possible.
33411 * config/s390/vector.md (vec_halfnumelts): New mode
33413 ("*vec_vllezlf<mode>"): New pattern.
33415 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33417 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
33418 ("popcountv4si2", "popcountv2di2"): Rename to ...
33419 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
33420 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
33422 ("popcount<mode>2_vxe"): New pattern.
33424 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33426 * common/config/s390/s390-common.c (processor_flags_table): Add
33428 * config.gcc: Add arch12.
33429 * config/s390/driver-native.c (s390_host_detect_local_cpu):
33430 Default to arch12 for unknown CPU model numbers.
33431 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
33432 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
33433 PROCESSOR_max sanity check.
33434 * config/s390/s390-opts.h (enum processor_type): Add
33436 * config/s390/s390.c (processor_table): Add arch12.
33437 (s390_expand_builtin): Add check for B_VXE flag.
33438 (s390_issue_rate): Add PROCESSOR_ARCH12.
33439 (s390_get_sched_attrmask): Likewise.
33440 (s390_get_unit_mask): Likewise.
33441 (s390_sched_score): Enable z13 scheduling for arch12.
33442 (s390_sched_reorder): Likewise.
33443 (s390_sched_variable_issue): Likewise.
33444 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
33446 (s390_tune_attr): Use z13 scheduling also for arch12.
33447 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
33448 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
33449 (TARGET_VXE_P): New macros.
33450 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
33451 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
33452 * config/s390/s390.opt: Add arch12 as processor_type.
33454 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33456 * config/s390/s390.md
33457 ("fixuns_truncdddi2", "fixuns_trunctddi2")
33458 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
33459 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
33461 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
33462 Rename expanders to ...
33464 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
33465 ("fixuns_truncdddi2_emu"): ... these.
33467 ("fixuns_trunc<mode>si2_emu"): New expander.
33469 ("*fixuns_truncdfdi2_z13"): Rename to ...
33470 ("*fixuns_truncdfdi2_vx"): ... this.
33472 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33474 * config/s390/2964.md: Remove the single element vector compare
33475 instructions which are no longer used.
33476 * config/s390/s390.c (s390_select_ccmode): Remove handling of
33478 (s390_canonicalize_comparison): Remove handling of DFmode
33480 (s390_expand_vec_compare_scalar): Remove function.
33481 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
33482 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
33484 ("*cmp<mode>_ccs"): Add wfcdb instruction.
33486 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33488 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
33490 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
33491 will anyway by matched by mov<mode>_64dfp.
33493 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33495 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
33496 vlef/vstef. Add missing operand to vleif.
33498 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33500 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
33501 pair for all vector types with 64 bit elements.
33502 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
33503 * config/s390/vector.md (V_HW_64): ... here.
33504 (V_128_NOSINGLE): New mode iterator.
33505 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
33506 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
33507 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
33508 ("*vec_load_pairv2di"): Change to ...
33509 ("*vec_load_pair<mode>"): ... this one.
33511 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33513 * config/s390/constraints.md: Add comments.
33514 (jKK): Reject element sizes > 8 bytes.
33515 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
33517 * config/s390/s390.md: Add the s_operand checks formerly in
33518 s390_split_ok_p to various splitters where they are still
33520 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
33521 for 128 bit vectors. Plus two splitters.
33523 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33525 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
33528 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33531 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
33532 error if the boundary argument is not constant.
33534 2017-03-24 Jakub Jelinek <jakub@redhat.com>
33536 PR rtl-optimization/80112
33537 * loop-doloop.c (doloop_condition_get): Don't check condition
33538 if cmp isn't SET with IF_THEN_ELSE src.
33540 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33542 PR tree-optimization/80158
33543 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
33544 replacing a candidate statement, also replace it for the
33545 candidate's alternate interpretation.
33546 (replace_rhs_if_not_dup): Likewise.
33547 (replace_one_candidate): Likewise.
33549 2017-03-24 Richard Biener <rguenther@suse.de>
33551 PR tree-optimization/80167
33552 * graphite-isl-ast-to-gimple.c
33553 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
33555 (translate_isl_ast_to_gimple::get_rename): Likewise.
33557 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33559 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
33560 handling of certain combinations of target options, including the
33561 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
33562 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
33564 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33567 * config/arm/arm.md (*load_multiple): Add reload_completed to
33568 matching condition.
33570 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33571 Richard Biener <rguenth@suse.de>
33573 PR tree-optimization/79908
33574 PR tree-optimization/80136
33575 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
33576 been cast away, gimplify_and_add suffices.
33578 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
33580 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
33582 2017-03-23 Richard Biener <rguenther@suse.de>
33584 PR tree-optimization/80032
33585 * gimplify.c (gimple_push_cleanup): Forced unconditional
33586 cleanups still have to go to the conditional_cleanups
33589 2017-03-22 Jakub Jelinek <jakub@redhat.com>
33591 PR tree-optimization/80072
33592 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
33594 (next_operand_entry_id): Change type to unsigned int.
33595 (sort_by_operand_rank): Make sure to return the right return value
33596 even if unsigned fields are bigger than INT_MAX.
33597 (struct oecount): Change cnt and id type to unsigned int.
33598 (oecount_hasher::equal): Formatting fix.
33599 (oecount_cmp): Make sure to return the right return value
33600 even if unsigned fields are bigger than INT_MAX.
33601 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
33604 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
33605 TREE_READONLY on result if writing it more than once.
33608 * doc/invoke.texi (-fsanitize=thread): Document that with
33609 -fnon-call-exceptions atomics are not able to throw
33613 * tsan.c: Include tree-eh.h.
33614 (instrument_builtin_call): Call maybe_clean_eh_stmt or
33615 maybe_clean_or_replace_eh_stmt where needed.
33616 (instrument_memory_accesses): Add cfg_changed argument.
33617 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
33618 if it returned true.
33619 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
33621 PR rtl-optimization/63191
33622 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
33623 wrapper function, moved the whole old content into ...
33624 (ix86_delegitimize_address_1): ... this. New inline function.
33625 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
33626 true as last argument instead of ix86_delegitimize_address.
33628 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
33630 * config/aarch64/aarch64.c (generic_branch_cost): Copy
33631 cortexa57_branch_cost.
33633 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
33635 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
33637 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33640 * doc/md.texi (Constraints): Document wA constraint.
33641 * config/rs6000/constraints.md (wA): New.
33642 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
33643 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
33644 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
33645 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
33647 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
33650 * gimplify.c (is_oacc_declared): New function.
33651 (oacc_default_clause): Use it to set default flags for acc declared
33652 variables inside parallel regions.
33653 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
33654 declared variables.
33655 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
33656 declare attribute to any decl as necessary.
33658 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
33661 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
33662 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
33663 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
33664 (arm_arch_lpae): This.
33665 * config/arm/arm.c (arm_arch7ve): Rename into ...
33666 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
33667 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
33670 2017-03-22 Martin Liska <mliska@suse.cz>
33673 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
33674 error message instead of an ICE.
33676 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33678 * doc/extend.texi (6.11 Additional Floating Types): Revise.
33680 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
33682 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
33684 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33687 2017-03-21 Martin Sebor <msebor@redhat.com>
33689 * doc/extend.texi: Use "cannot" instead of "can't."
33690 * doc/hostconfig.texi: Same.
33691 * doc/install.texi: Same.
33692 * doc/invoke.texi: Same.
33693 * doc/loop.texi: Same.
33694 * doc/md.texi: Same.
33695 * doc/objc.texi: Same.
33696 * doc/rtl.texi: Same.
33697 * doc/tm.texi: Same.
33698 * doc/tm.texi.in: Same.
33699 * doc/trouble.texi: Same.
33701 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
33704 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
33705 (collect_checksum_attributes): Set it.
33706 (die_checksum_ordered): Use it.
33708 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33710 PR tree-optimization/79908
33711 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
33712 change: For a VA_ARG whose LHS has been cast away, use
33713 force_gimple_operand to construct the side effects.
33715 2017-03-21 David Malcolm <dmalcolm@redhat.com>
33717 PR translation/80001
33718 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
33719 more amenable to translation.
33720 (oacc_loop_auto_partitions): Likewise.
33722 2017-03-21 Marek Polacek <polacek@redhat.com>
33723 Martin Sebor <msebor@redhat.com>
33725 PR tree-optimization/80109
33726 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
33727 on INTEGRAL_TYPE_P.
33729 2017-03-21 Jakub Jelinek <jakub@redhat.com>
33730 Segher Boessenkool <segher@kernel.crashing.org>
33733 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
33734 check reg_used_between_p between insn and one of succ or succ2
33735 depending on if succ is artificial insn not inserted into insn
33738 2017-03-21 Martin Liska <mliska@suse.cz>
33740 PR gcov-profile/80081
33741 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
33742 * doc/gcc.texi: Include gcov-dump stuff.
33743 * doc/gcov-dump.texi: New file.
33745 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
33747 PR rtl-optimization/79150
33748 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
33749 conditional jump, if the jump is the last insn of the loop.
33751 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33752 Richard Biener <rguenth@suse.de>
33754 PR tree-optimization/79908
33755 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
33756 been cast away, use force_gimple_operand to construct the side
33759 2017-03-21 Martin Liska <mliska@suse.cz>
33761 PR libfortran/79956
33762 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
33765 2017-03-21 Brad Spengler <spender@grsecurity.net>
33768 * plugin.c (htab_hash_plugin): New function.
33769 (add_new_plugin): Use it and adjust.
33770 (parse_plugin_arg_opt): Adjust.
33771 (init_one_plugin): Likewise.
33773 2017-03-21 Richard Biener <rguenther@suse.de>
33775 PR tree-optimization/80032
33776 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
33777 if set force the cleanup to happen unconditionally.
33778 (gimplify_target_expr): Push inserted clobbers with force_uncond
33779 to avoid them being removed by control-dependent DCE.
33781 2017-03-21 Richard Biener <rguenther@suse.de>
33783 PR tree-optimization/80122
33784 * tree-inline.c (copy_bb): Do not expans va-arg packs or
33785 va_arg_pack_len when the inlined call stmt requires pack
33787 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
33789 2017-03-21 Jakub Jelinek <jakub@redhat.com>
33792 * tsan.c (instrument_builtin_call): If the memory model argument
33793 is not a constant, assume it is valid.
33796 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
33799 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
33801 PR rtl-optimization/79910
33802 * combine.c (can_combine_p): Do not allow combining an I0 or I1
33803 if its dest is used by an insn before I2 (other than the combined
33804 insns themselves, which are properly handled already).
33806 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
33809 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
33811 * combine.c (record_used_regs): New static function.
33812 (try_combine): Handle situations where there is an additional
33813 instruction between I2 and I3 which needs to have a LOG_LINK
33817 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
33819 * combine.c (try_combine): Delete redundant i1 test. Call
33820 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
33822 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33825 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
33826 alternatives 13/14.
33828 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33830 PR tree-optimization/80054
33831 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
33832 the optimization if a PHI or any of its arguments is not dominated
33833 by the candidate's basis. Use gphi* rather than gimple* as
33835 (replace_profitable_candidates): Clean up a gimple* variable that
33836 should be a gphi* variable.
33838 2017-03-20 Martin Sebor <msebor@redhat.com>
33841 * doc/extend.texi (attribute constructor): Document present limitation.
33843 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
33846 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
33847 __POWER9_VECTOR__ #ifdef control, change template definition to
33848 use Power9-specific built-in function.
33849 (vec_any_eq): Likewise.
33850 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
33851 to control outcomes from this test.
33852 (vector_ae_<mode>p): For VEC_F modes, likewise.
33854 2017-03-20 Ian Lance Taylor <iant@google.com>
33856 * config/i386/i386.c (ix86_function_regparm): Save an extra
33857 register for -fsplit-stack with DECL_STATIC_CHAIN.
33859 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
33862 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
33863 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
33865 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
33867 * config/riscv/riscv.c (riscv_print_operand): Use "fence
33869 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
33872 2017-03-20 Marek Polacek <polacek@redhat.com>
33875 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
33877 2017-03-20 Richard Biener <rguenther@suse.de>
33879 PR tree-optimization/80113
33880 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
33881 allocate extra SSA name for PHI def.
33882 (add_close_phis_to_outer_loops): Likewise.
33883 (add_close_phis_to_merge_points): Likewise.
33884 (copy_loop_close_phi_args): Likewise.
33885 (copy_cond_phi_nodes): Likewise.
33887 2017-03-20 Martin Liska <mliska@suse.cz>
33889 PR middle-end/79753
33890 * tree-chkp.c (chkp_build_returned_bound): Do not build
33891 returned bounds for a LHS that's not a BOUNDED_P type.
33893 2017-03-20 Martin Liska <mliska@suse.cz>
33897 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
33898 COMPLEX_CST and VECTOR_CST.
33900 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33903 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
33904 target operand. A new splitter adds the clobber statement in case
33905 the target operand is dead anyway.
33907 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
33909 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
33910 to age-old versions of binutils and glibc.
33912 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
33914 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
33916 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
33918 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
33920 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
33922 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
33923 requirement for binutils 2.13.
33925 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
33927 * combine.c (try_combine): Delete redundant i1 test. Call
33928 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
33930 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
33932 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
33933 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
33935 <riscv64-*-elf>: Re-arrange section
33936 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
33937 <riscv32-*-linux>: Likewise.
33938 <riscv64-*-elf>: Likewise
33939 <riscv64-*-linux>: Likewise.
33941 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
33944 * aarch64.opt(verbose-cost-dump): Fix typo.
33946 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
33949 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
33950 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
33952 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
33954 * reload.c (find_reloads): When reloading a nonoffsettable address,
33955 use RELOAD_OTHER for it and its address reloads.
33957 PR rtl-optimization/79910
33958 * combine.c (record_used_regs): New static function.
33959 (try_combine): Handle situations where there is an additional
33960 instruction between I2 and I3 which needs to have a LOG_LINK
33963 2017-03-17 Jeff Law <law@redhat.com>
33965 PR tree-optimization/71437
33966 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
33967 conditional in the hash table first.
33968 (vrp_dom_walker::before_dom_children): Extract condition from
33969 ASSERT_EXPR. Record condition, its inverion and any implied
33970 conditions as well.
33972 2017-03-17 Marek Polacek <polacek@redhat.com>
33973 Markus Trippelsdorf <markus@trippelsdorf.de>
33975 PR tree-optimization/80079
33976 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
33979 2017-03-17 Richard Biener <rguenther@suse.de>
33981 PR middle-end/80075
33982 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
33983 Properly verify the LHS before the RHS possibly claims to be
33985 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
33988 2017-03-17 Martin Jambor <mjambor@suse.cz>
33990 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
33991 (List of -O2 options): Likewise.
33992 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
33995 2017-03-17 Tom de Vries <tom@codesourcery.com>
33997 * gcov-dump.c (print_usage): Print bug_report_url.
33999 2017-03-17 Richard Biener <rguenther@suse.de>
34001 PR middle-end/80050
34002 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
34003 (parser::peek): Likewise.
34005 2017-03-17 Richard Biener <rguenther@suse.de>
34007 PR tree-optimization/80048
34008 * sese.c (free_sese_info): Properly release rename_map and
34009 copied_bb_map elements.
34011 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
34013 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
34014 Add linked-list forward and backlinks. Insert on
34015 construction, remove on destruction.
34016 (class pass_store_merging): Add m_stores_head field.
34017 (pass_store_merging::terminate_and_process_all_chains):
34018 Iterate over m_stores_head list.
34019 (pass_store_merging::terminate_all_aliasing_chains):
34021 (pass_store_merging::execute): Check for debug stmts first.
34022 Push new chains onto the m_stores_head stack.
34024 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
34027 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
34028 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
34029 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
34031 2017-03-16 Jeff Law <law@redhat.com>
34033 PR tree-optimization/71437
34034 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
34035 member function. Implementation moved into after_dom_children
34036 member function and into the threader's thread_outgoing_edges
34038 (dom_opt_dom_walker::after_dom_children): Simplify by moving
34039 some code into new thread_outgoing_edges.
34040 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
34041 definition. Simplify marker handling (do it here). Assume we always
34042 have the available expression and the const/copies tables.
34043 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
34045 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
34046 * tree-vrp.c (equiv_stack): No longer file scoped.
34047 (vrp_dom_walker): New class.
34048 (vrp_dom_walker::before_dom_children): New member function.
34049 (vrp_dom_walker::after_dom_children): Likewise.
34050 (identify_jump_threads): Setup domwalker. Use it rather than
34051 walking edges in a random order by hand. Simplify setup/finalization.
34052 (finalize_jump_threads): Remove.
34053 (vrp_finalize): Do not call identify_jump_threads here.
34054 (execute_vrp): Do it here instead and call thread_through_all_blocks
34057 PR tree-optimization/71437
34058 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
34060 (simplify_stmt_for_jump_threading): Add basic_block argument. All
34062 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
34063 (dom_opt_dom_walker::thread_across_edge): Remove
34064 handle_dominating_asserts argument. All callers changed.
34065 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
34066 changes. Remove calls to lhs_of_dominating_assert. Other
34067 uses of handle_dominating_asserts turn into unconditional code
34068 (simplify_control_stmt_condition_1): Likewise.
34069 (simplify_control_stmt_condition): Likewise.
34070 (thread_through_normal_block, thread_across_edge): Likewise.
34071 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
34072 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
34073 object if it is not an SSA_NAME.
34074 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
34075 before calling into the VRP specific simplifiers.
34076 (identify_jump_threads): Remove handle_dominating_asserts
34079 2017-03-16 Jakub Jelinek <jakub@redhat.com>
34082 * tree-diagnostic.c (default_tree_printer): No longer static.
34083 * tree-diagnostic.h (default_tree_printer): New prototype.
34085 2017-03-16 Tamar Christina <tamar.christina@arm.com>
34087 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
34088 Change ins into fmov.
34090 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34092 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
34093 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
34094 Use h_con constraint for operand 1.
34095 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
34096 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
34098 2017-03-15 Jeff Law <law@redhat.com>
34100 PR tree-optimization/71437
34101 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
34102 (record_temporary_equivalences): Use it.
34104 PR tree-optimization/71437
34105 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
34106 tree-ssa-scopedtables.
34107 (lookup_avail_expr, build_and_record_new_cond): Likewise.
34108 (record_conditions, record_cond, vuse_eq): Likewise.
34109 (record_edge_info): Adjust to API tweak of record_conditions.
34110 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
34111 (record_temporary_equivalences, optimize_stmt): Likewise.
34112 (eliminate_redundant_computations): Likewise.
34113 (record_equivalences_from_stmt): Likewise.
34114 * tree-ssa-scopedtables.c: Include options.h and params.h.
34115 (vuse_eq): New function, moved from tree-ssa-dom.c
34116 (build_and_record_new_cond): Likewise.
34117 (record_conditions): Likewise. Accept vector of conditions rather
34118 than edge_equivalence structure for first argument.
34119 for the first argument.
34120 (avail_exprs_stack::lookup_avail_expr): New member function, moved
34121 from tree-ssa-dom.c.
34122 (avail_exprs_stack::record_cond): Likewise.
34123 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
34124 from tree-ssa-dom.c.
34125 (avail_exprs_stack): Add new member functions lookup_avail_expr
34127 (record_conditions): Declare.
34129 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
34132 * lra-constraints.c (process_alt_operands): Increase reject for
34133 reloading an input/output operand.
34135 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
34138 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
34139 insns to convert from signed/unsigned char/short to IEEE 128-bit
34141 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
34143 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
34146 * config/i386/i386.c (ix86_vector_duplicate_value): Create
34147 subreg of inner mode for values already in registers.
34149 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
34151 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
34152 iteration reg is used after the loop.
34154 2017-03-14 Martin Sebor <msebor@redhat.com>
34156 PR tree-optimization/79800
34157 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
34158 precision in negative-positive range.
34159 (format_floating): Call non-const overload with adjusted precision.
34161 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34164 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
34167 2017-03-14 Martin Sebor <msebor@redhat.com>
34169 PR middle-end/80020
34170 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
34171 * builtins.def (aligned_alloc): Use it.
34174 * Makefile.in (GTFILES): Add calls.c.
34175 * calls.c: Include "gt-calls.h".
34177 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
34179 PR rtl-optimization/79728
34180 * regs.h (struct target_regs): New field
34181 x_contains_allocatable_regs_of_mode.
34182 (contains_allocatable_regs_of_mode): New macro.
34183 * reginfo.c (init_reg_sets_1): Initialize it, and change
34184 contains_reg_of_mode so it includes global regs as well.
34185 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
34186 rather than contains_regs_of_mode.
34188 2017-03-14 Martin Liska <mliska@suse.cz>
34190 * doc/invoke.texi: Document options that can't be combined with
34191 -fcheck-pointer-bounds.
34193 2017-03-14 Martin Liska <mliska@suse.cz>
34195 PR middle-end/79831
34196 * doc/invoke.texi (-Wchkp): Document the option.
34198 2017-03-14 Martin Liska <mliska@suse.cz>
34200 * Makefile.in: Install gcov-dump.
34202 2017-03-14 Martin Liska <mliska@suse.cz>
34204 * multiple_target.c (expand_target_clones): Bail out for
34205 an invalid attribute.
34207 2017-03-14 Richard Biener <rguenther@suse.de>
34209 * alias.c (struct alias_set_entry): Pack properly.
34210 * cfgloop.h (struct loop): Likewise.
34211 * cse.c (struct set): Likewise.
34212 * ipa-utils.c (struct searchc_env): Likewise.
34213 * loop-invariant.c (struct invariant): Likewise.
34214 * lra-remat.c (struct cand): Likewise.
34215 * recog.c (struct change_t): Likewise.
34216 * rtl.h (struct address_info): Likewise.
34217 * symbol-summary.h (function_summary): Likewise.
34218 * tree-loop-distribution.c (struct partition): Likewise.
34219 * tree-object-size.c (struct object_size_info): Likewise.
34220 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
34221 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
34222 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
34223 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
34224 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
34225 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
34226 (struct _stmt_vec_info): Likewise.
34228 2017-03-14 Martin Liska <mliska@suse.cz>
34231 * multiple_target.c (create_dispatcher_calls): Check that
34232 a target can create a function dispatcher.
34234 2017-03-14 Martin Liska <mliska@suse.cz>
34237 * multiple_target.c (expand_target_clones): Drop local.local
34238 flag for default implementation.
34240 2017-03-14 Richard Biener <rguenther@suse.de>
34242 PR tree-optimization/80030
34243 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
34245 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
34247 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
34248 gcc_fallthrough() instead of __attribute__((fallthrough));
34250 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
34252 * doc/gcc.texi: Remove "up" link to (DIR).
34253 * doc/gccint.texi: Ditto.
34255 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
34257 * doc/install.texi (Specific) <avr>: Remove reference to
34260 2017-03-13 Jeff Law <law@redhat.com>
34262 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
34263 attribute rather than comments.
34265 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
34266 match_scratch operand is highest.
34268 2017-03-13 Martin Liska <mliska@suse.cz>
34270 PR middle-end/78339
34271 * ipa-pure-const.c (warn_function_noreturn): If the declarations
34272 is a CHKP clone, use original declaration.
34274 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34276 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
34277 (arc_conditional_register_usage): Use a different allocation order
34278 when optimizing for size.
34279 * common/config/arc/arc-common.c (arc_option_optimization_table):
34280 Section anchors default on when optimizing for size.
34282 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34284 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
34286 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34288 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
34289 * config/arc/arc.md (cpu_facility): Add cd variant.
34290 (*movqi_insn): Add code density variant.
34291 (*movhi_insn): Likewise.
34292 (*movqi_insn): Likewise.
34293 (*addsi3_mixed): Likewise.
34294 (subsi3_insn): Likewise.
34296 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34298 * config/arc/arc.md (movsi_cond_exec): Update constraint.
34300 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34302 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
34303 expressions with MINUS and UNARY ops.
34305 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34308 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
34310 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
34311 between vec_select and vector argument.
34312 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
34313 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
34314 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
34315 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
34316 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
34317 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
34319 2017-03-13 Richard Biener <rguenther@suse.de>
34322 * params.def (vect-max-peeling-for-alignment): Fix typo.
34324 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34326 * doc/install.texi (Specific) <mips-*-*>: Remove description of
34327 issue that only occurred with binutils below 2.18.
34329 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34331 * doc/install.texi (Specific) <cris-axis-elf>: No longer
34332 refer to binutils 2.11/2.12 minimum.
34334 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34336 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
34337 ftp.kernel.org and simplify binutils requirement.
34339 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
34341 * doc/invoke.texi (Warning Options): Fix spelling of link-time
34343 (Optimize Options): Ditto. Also remove redundancy.
34345 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34347 PR translation/79848
34348 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
34350 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
34351 to G_ to avoid double translation.
34353 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34355 PR translation/79923
34356 * auto-profile.c (get_combined_location): Convert leading
34357 character of diagnostics to lower case and remove trailing period.
34358 (read_profile): Likewise for various diagnostics.
34359 * config/arm/arm.c (arm_option_override): Remove trailing period
34360 from various diagnostics.
34361 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
34362 (msp430_expand_delay_cycles): Likewise.
34364 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34367 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
34368 full command-line argument, rather than just "str".
34369 (aarch64_validate_march): Likewise.
34370 (aarch64_validate_mtune): Likewise.
34372 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
34374 PR rtl-optimization/78911
34375 * lra-assigns.c (must_not_spill_p): New function.
34376 (spill_for): Use it.
34378 2017-03-10 Jakub Jelinek <jakub@redhat.com>
34380 PR tree-optimization/79981
34381 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
34382 ATOMIC_COMPARE_EXCHANGE ifn result.
34383 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
34384 IFN_ATOMIC_COMPARE_EXCHANGE.
34386 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34389 * opts.c (parse_sanitizer_options): Add missing question mark to
34390 "did you mean" message.
34392 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34394 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
34396 (VMULEUH_UNS): Likewise.
34397 (VMULOUB_UNS): Likewise.
34398 (VMULOUH_UNS): Likewise.
34399 * config/rs6000/rs6000.c (builtin_function_type): Remove
34400 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
34402 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34405 * read-rtl-function.c (function_reader::read_rtx_operand): Update
34406 x with result of extra_parsing_for_operand_code_0.
34407 (function_reader::extra_parsing_for_operand_code_0): Convert
34408 return type from void to rtx, returning x. When reading
34409 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
34410 larger size containing struct block_symbol.
34412 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
34414 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
34415 -mfloat128-hardware without -m64.
34417 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
34420 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
34421 entries to the case statement that marks unsigned arguments to
34422 overloaded functions.
34424 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34426 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
34427 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
34429 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
34432 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
34433 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
34435 2017-03-10 Martin Liska <mliska@suse.cz>
34439 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
34440 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
34443 2017-03-10 Olivier Hainque <hainque@adacore.com>
34445 * tree-switch-conversion (array_value_type): Start by resetting
34446 candidate type to it's main variant.
34448 2017-03-10 Jakub Jelinek <jakub@redhat.com>
34450 PR rtl-optimization/79909
34451 * combine.c (try_combine): Use simplify_replace_rtx on individual
34452 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
34453 of the whole CALL_INSN_FUNCTION_USAGE.
34455 PR tree-optimization/79972
34456 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
34457 get_range_info on SSA_NAMEs. Formatting fixes.
34459 2017-03-10 Richard Biener <rguenther@suse.de>
34460 Jakub Jelinek <jakub@redhat.com>
34462 PR tree-optimization/77975
34463 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
34464 edge to be constant.
34465 (get_val_for): For constant x return it. Formatting fix.
34466 (loop_niter_by_eval): Avoid pointless looping if the next iteration
34467 would use the same bases as the current one.
34469 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34471 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
34472 instead of vec_select for V1TImode.
34473 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
34475 (VSX_LE_128): Add V1TI to this mode iterator.
34476 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
34477 (*vsx_le_perm_store_<mode>): Likewise.
34478 (pre-reload splitter for VSX stores): Likewise.
34479 (post-reload splitter for VSX stores): Likewise.
34480 (*vsx_xxpermdi2_le_<mode>): Likewise.
34481 (*vsx_lxvd2x2_le_<mode>): Likewise.
34482 (*vsx_stxvd2x2_le_<mode>): Likewise.
34484 2017-03-09 Michael Eager <eager@eagercon.com>
34486 Correct failures with --enable-checking=yes,rtl.
34488 * config/microblaze/microblaze.c (microblaze_expand_shift):
34489 Replace GET_CODE test with CONST_INT_P and INTVAL test with
34490 test for const0_rtx.
34491 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
34492 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
34494 2017-03-09 Richard Biener <rguenther@suse.de>
34496 PR tree-optimization/79977
34497 * graphite-scop-detection.c (scop_detection::merge_sese):
34498 Handle the case of extra exits to blocks dominating the entry.
34500 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
34502 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
34505 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
34507 PR rtl-optimization/79949
34508 * lra-constraints.c (process_alt_operands): Check memory when
34509 trying to predict a cycle. Print about the overall increase.
34511 2017-03-09 Richard Biener <rguenther@suse.de>
34513 PR middle-end/79971
34514 * gimple-expr.c (useless_type_conversion_p): Preserve
34515 TYPE_SATURATING for fixed-point types.
34517 2017-03-09 Richard Biener <rguenther@suse.de>
34520 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
34521 alignment of BLKmode params.
34523 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34526 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
34527 (VALL_NO_V2Q): Likewise.
34529 * config/aarch64/aarch64-simd.md
34530 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
34532 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
34533 VALL_NO_V2Q mode iterator.
34534 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
34536 2017-03-09 Martin Liska <mliska@suse.cz>
34538 PR tree-optimization/79631
34539 * tree-chkp-opt.c (chkp_is_constant_addr): Call
34540 tree_int_cst_sign_bit just for INTEGER constants.
34542 2017-03-09 Martin Liska <mliska@suse.cz>
34546 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
34549 2017-03-09 Marek Polacek <polacek@redhat.com>
34552 * tree.c (inchash::add_expr): Handle TREE_VEC.
34554 2017-03-09 Martin Liska <mliska@suse.cz>
34557 (chkp_narrow_size_and_offset): New function.
34558 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
34559 (void chkp_parse_bit_field_ref): New function.
34560 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
34561 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
34563 2017-03-09 Martin Liska <mliska@suse.cz>
34566 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
34567 (chkp_find_bounds_1): Remove gcc_unreachable.
34569 2017-03-09 Jakub Jelinek <jakub@redhat.com>
34572 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
34573 BUILT_IN_SYNC*, determine the access type from the size suffix and
34574 always build a MEM_REF with that type. Handle forgotten
34575 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
34578 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
34579 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
34580 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
34581 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
34582 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
34583 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
34584 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
34585 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
34586 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
34587 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
34588 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
34589 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
34590 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
34591 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
34592 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
34593 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
34594 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
34595 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
34596 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
34597 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
34598 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
34599 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
34600 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
34601 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
34602 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
34603 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
34604 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
34605 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
34606 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
34607 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
34608 definitions outside of __OPTIMIZE__ guarded section.
34611 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
34612 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
34613 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
34614 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
34617 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34619 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
34620 ("vfenez<mode>"): Add missing constraints.
34622 2017-03-08 Martin Sebor <msebor@redhat.com>
34625 * config/nds32/nds32.c (nds32_option_override):
34626 Fix misspelled diagnostic.
34628 2017-03-08 Jakub Jelinek <jakub@redhat.com>
34631 * gimplify.c (gimplify_omp_for): Replace index var in outer
34632 taskloop statement with an artificial variable and add
34633 OMP_CLAUSE_PRIVATE clause for it.
34635 2017-03-08 Richard Biener <rguenther@suse.de>
34637 PR tree-optimization/79955
34638 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
34639 for accesses that are completely outside of the variable.
34641 2017-03-08 Andrew Haley <aph@redhat.com>
34643 PR tree-optimization/79943
34644 * tree-ssa-loop-split.c (compute_new_first_bound): When
34645 calculating the new upper bound, (END-BEG) should be added, not
34648 2017-03-08 Jakub Jelinek <jakub@redhat.com>
34650 * config/avr/avr.md (setmemhi): Make sure match_dup
34651 operand number comes before match_scratch.
34653 2017-03-08 Richard Biener <rguenther@suse.de>
34655 PR tree-optimization/79920
34656 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
34657 with ncopies == 1 to ...
34658 (vect_transform_slp_perm_load): ... here. Properly compute
34659 all element loads by iterating VF times over the group. Do
34660 not handle ncopies (computed in a broken way) in
34661 vect_create_mask_and_perm.
34663 2017-03-08 Jakub Jelinek <jakub@redhat.com>
34666 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
34667 is a uniform vector, use uniform_vector_p return value instead of
34668 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
34670 2017-03-07 Marek Polacek <polacek@redhat.com>
34672 PR middle-end/79809
34673 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
34674 (alloca_call_type): Likewise.
34676 2017-03-07 Martin Liska <mliska@suse.cz>
34678 * gcov.c (process_args): Put comment to correct location.
34680 2017-03-07 Martin Liska <mliska@suse.cz>
34682 PR middle-end/68270
34683 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
34684 Use array_at_struct_end_p instead of DECL_CHAIN (field).
34685 (chkp_narrow_bounds_for_field): Likewise.
34686 (chkp_parse_array_and_component_ref): Pass one more argument to
34689 2017-03-07 Richard Biener <rguenther@suse.de>
34691 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
34694 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
34696 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
34697 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
34699 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34702 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
34703 to end of description.
34704 (PARAM_MAX_STORES_TO_MERGE): Likewise.
34706 2017-03-07 Jakub Jelinek <jakub@redhat.com>
34708 PR rtl-optimization/79901
34709 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
34711 (*avx512f_<code><mode>3<mask_name>): ... this.
34712 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
34713 iterator instead of VI8_AVX2_AVX512BW.
34715 PR rtl-optimization/79901
34716 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
34717 min/max expander, expand it using expand_vec_cond_expr.
34720 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
34723 2017-03-06 Jakub Jelinek <jakub@redhat.com>
34726 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
34727 to void * for PCH reasons.
34728 * dwarf2out.c (output_loc_operands, output_die): Cast
34729 v.val_vec.array to unsigned char *.
34731 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
34734 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
34737 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
34739 PR rtl-optimization/79571
34740 * lra-constraints.c (process_alt_operands): Calculate static
34741 reject and subtract it from overall when only addresses will be
34744 2017-03-06 Julia Koval <julia.koval@intel.com>
34747 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
34748 incoming stack boundary to 128 for 64-bit targets.
34750 2017-03-06 Richard Biener <rguenther@suse.de>
34752 PR tree-optimization/79894
34753 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
34754 to NULL after folding it.
34756 2017-03-06 Richard Biener <rguenther@suse.de>
34758 PR tree-optimization/79824
34759 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
34760 check disabling peeling for gaps.
34762 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
34764 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
34765 attributes): Document gettimeofday.
34767 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
34769 * config/s390/s390.c (s390_option_override_internal): Set
34770 PARAM_MIN_VECT_LOOP_BOUND
34772 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
34774 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
34775 * config/s390/s390.md: Likewise.
34777 2017-03-06 Jakub Jelinek <jakub@redhat.com>
34780 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
34781 (<avx2_avx512>_perm<mode>): Rename to ...
34782 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
34784 (<avx512>_perm<mode>_mask): Rename to ...
34785 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
34787 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
34788 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
34789 instead of VI8F_256_512.
34790 (avx512f_perm<mode>): New define_expand.
34791 (avx512f_perm<mode>_mask): Likewise.
34792 (avx512f_perm<mode>_1<mask_name>): New define_insn.
34793 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
34795 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34797 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
34798 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
34800 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
34802 2017-03-06 Martin Liska <mliska@suse.cz>
34805 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
34806 when having a SSA NAME w/o VAR_DECL assigned to it.
34808 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34810 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
34811 msa_dpsub_<su>_d): Fix MODE for vec_select.
34813 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34815 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
34817 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
34819 2017-03-06 Richard Biener <rguenther@suse.de>
34821 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
34822 * plugin.c (register_plugin_info): Likewise.
34823 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
34825 2017-03-05 Jakub Jelinek <jakub@redhat.com>
34827 * config/i386/sse.md (sse_storehps, sse_storelps,
34828 avx_<castmode><avxsizesuffix>_<castmode>,
34829 avx512f_<castmode><avxsizesuffix>_<castmode>,
34830 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
34831 in condition that at least one operand is not a MEM.
34833 2017-03-03 Jakub Jelinek <jakub@redhat.com>
34835 PR middle-end/79805
34836 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
34837 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
34839 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
34840 gimple_call_nothrow_p flag based on whether original builtin can throw.
34841 If it can, emit following stmts on the fallthrough edge.
34842 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
34843 don't create new bb if inserting just debug stmts on the edge, try to
34844 insert them on the fallthru bb or just reset debug stmts.
34846 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
34849 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
34850 restore recog_data (including the operand rtxes inside it) around
34851 the call to get_insn_template.
34853 2017-03-03 Martin Sebor <msebor@redhat.com>
34855 PR tree-optimization/79699
34856 * context.c (context::~context): Free MPFR caches to avoid
34857 a memory leak on program exit.
34859 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34861 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
34862 Use wide_int::ulow () instead of .elt (0).
34864 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
34866 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
34867 (*pushxf): Limit oF constraint to 32bit targets and add oC
34868 constraint for 64bit targets.
34869 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
34870 (*pushdf): Change rmF constraint to rmC.
34872 2017-03-03 Martin Liska <mliska@suse.cz>
34874 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
34875 Remove unused variable.
34877 2017-03-03 Jakub Jelinek <jakub@redhat.com>
34880 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
34881 is a memory operand, increase num_memory.
34882 (ix86_expand_args_builtin): Likewise.
34884 2017-03-03 Jan Hubicka <jh@suse.cz>
34887 * ipa-devirt.c (maybe_record_node): Properly handle
34888 __cxa_pure_virtual visibility.
34890 2017-03-03 Martin Liska <mliska@suse.cz>
34892 PR tree-optimization/79803
34893 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
34895 (pass_loop_prefetch::execute): Disabled optimization if an
34896 assumption about L1 cache size is not met.
34898 2017-03-03 Martin Liska <mliska@suse.cz>
34900 PR rtl-optimization/79574
34901 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
34902 (hash_scan_set): Likewise.
34903 (dump_hash_table): Likewise.
34904 (hoist_code): Likewise.
34906 2017-03-03 Richard Biener <rguenther@suse.de>
34908 * fixed-value.c (fixed_from_string): Restore use of elt (1)
34909 in place of uhigh ().
34910 (fixed_convert_from_real): Likewise.
34912 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
34915 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
34917 2017-03-03 Richard Biener <rguenther@suse.de>
34919 PR middle-end/79818
34920 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
34921 TYPE_OVERFLOW_UNDEFINED check.
34923 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34925 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
34927 (vector_ae_<mode>_p): Likewise.
34928 (vector_nez_<mode>_p): Likewise.
34929 (vector_ne_v2di_p): Likewise.
34930 (vector_ae_v2di_p): Likewise.
34931 (vector_ne_<mode>_p): Likewise.
34932 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
34934 (vsx_tsqrt<mode>2_fe): Likewise.
34936 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
34939 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
34941 2017-03-02 Jakub Jelinek <jakub@redhat.com>
34943 PR rtl-optimization/79780
34944 * cprop.c (one_cprop_pass): When second and further conditional trap
34945 in a single basic block is turned into an unconditional trap, turn it
34946 into a deleted note to avoid RTL verification failures.
34948 2017-03-02 Richard Biener <rguenther@suse.de>
34950 * fold-const.c (const_binop): Use ulow () instead of elt (0).
34952 2017-03-02 Richard Biener <rguenther@suse.de>
34954 PR tree-optimization/79345
34956 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
34957 param and abort the walk, returning -1 if it is hit.
34958 (walk_aliased_vdefs): Take a limit param and pass it on.
34959 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
34960 defaulting to 0 and return a signed int.
34961 * tree-ssa-uninit.c (struct check_defs_data): New struct.
34962 (check_defs): New helper.
34963 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
34964 about uninitialized memory.
34965 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
34966 bogus uninitialized warning.
34967 (fixed_convert_from_real): Likewise.
34969 2017-03-02 Bin Cheng <bin.cheng@arm.com>
34971 PR tree-optimization/66768
34972 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
34973 iv_use if base object can't be determined.
34975 2017-03-02 Jakub Jelinek <jakub@redhat.com>
34977 PR tree-optimization/79345
34978 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
34979 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
34980 (get_pattern_stats): Initialize it.
34981 * genemit.c (gen_expand): Verify match_scratch numbers come after
34982 match_operand/match_dup numbers.
34983 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
34984 match_scratch numbers.
34985 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
34987 * config/s390/s390.md (trunctdsd2): Likewise.
34989 2017-03-02 Richard Biener <rguenther@suse.de>
34991 * wide-int.h (wide_int_storage::operator=): Implement in terms
34994 2017-03-02 Richard Biener <rguenther@suse.de>
34996 PR tree-optimization/79777
34997 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
34998 the to insert expression to sth existing.
35000 2017-03-01 Martin Sebor <msebor@redhat.com>
35002 PR middle-end/79692
35003 * gimple-ssa-sprintf.c
35004 (directive::known_width_and_precision): New function.
35005 (format_integer): Use it.
35006 (get_mpfr_format_length): Consider the full range of precision
35007 when computing %g output with the # flag. Set the likely byte
35008 count to 3 rather than 1 when precision is indeterminate.
35009 (format_floating): Correct the lower bound of precision.
35011 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35013 * doc/invoke.texi: Document default code model for 64-bit Linux.
35015 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35018 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
35019 udiv rather than div since input pattern is unsigned.
35021 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
35023 * config/i386/i386.c (print_reg): Warn for values of
35024 unsupported size in integer register.
35026 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
35029 * config/rs6000/predicates.md (current_file_function_operand): Do
35030 not allow self calls to be local if the function is replaceable.
35032 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
35035 * config/rs6000/altivec.h (vec_ctz and others): Change the
35036 preprocessor macro that controls conditional compilation from
35037 _ARCH_PWR9 to __POWER9_VECTOR__.
35038 (vec_all_ne): Change parameterization of __altivec_scalar_pred
35039 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
35040 control (instead of _ARCH_PWR9 control) so that template
35041 definition uses power9-specific function.
35042 (vec_any_eq): Likewise.
35043 (vec_all_ne): Change macro definition to use a power9-specific
35044 expansion under #ifdef __POWER9_VECTOR__ control (instead of
35045 _ARCH_PWR9 control).
35046 (vec_any_eq) Likewise.
35047 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
35048 expansion for CMPNEF to remove support for xvcmpnesp instruction.
35049 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
35050 support for xvcmpnedp instruction.
35051 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
35052 macro expansion so that Power9 implementation of vec_all_ne does
35053 not use the AltiVec predicate framework.
35054 (VCMPNEH_P): Likewise.
35055 (VCMPNEW_P): Likewise.
35056 (VCMPNED_P): Likewise.
35057 (VCMPNEFP_P): Likewise.
35058 (VCMPNEDP_P): Likewise.
35059 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
35060 implementation of vec_any_eq to not use AltiVec predicate
35062 (VCMPAEH_P): Likewise.
35063 (VCMPAEW_P): Likewise.
35064 (VCMPAED_P): Likewise.
35065 (VCMPAEFP_P): Likewise.
35066 (VCMPAEDP_P): Likewise.
35067 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
35068 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
35069 not use the AltiVec predicate framework.
35070 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
35071 of vec_any_eq to not use AltiVec predicate framework.
35072 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35073 support for predefined __POWER9_VECTOR__ macro to indicate that
35074 Power9 instruction selection is enabled.
35075 (altivec_overloaded_builtins): Remove extraneous
35076 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
35077 function argument types RS6000_BTI_bool_V16QI and
35078 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
35079 entry for overloaded function argument types RS6000_BTI_bool_V4SI
35080 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
35081 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
35082 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
35083 Power9 for implementations of vec_cmpne. Change the signature for
35084 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
35085 (representing vec_all_ne) to remove the previously described first
35086 argument of type RS6000_BTI_INTSI, as this was an artifact of
35087 reliance on the AltiVec predicate framework, which is no longer
35088 used in the implementation of these functions. Add
35089 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
35090 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
35091 since, unlike the AltiVec predicate framework implementation, we
35092 do not share function descriptors between vec_alle and vec_anyeq.
35093 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
35094 set of modes that receive special treatment even when
35095 TARGET_P9_VECTOR is true. The special treatment emits code that
35096 does not depend on Power9 instructions.
35097 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
35098 define_expand to not rely on AltiVec predicate framework.
35099 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
35101 (vector_ne_v2di_p): Change this define_expand to not rely on
35102 AltiVec predicate framework.
35103 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
35105 (vector_ne_<mode>_p): Change this define_expand to not rely on
35106 AltiVec predicate framework.
35107 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
35109 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
35110 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
35111 define_insn pattern.
35112 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
35113 define_insn pattern because the xvcmpne<VSs>. instruction is not
35115 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
35116 instruction is not supported.
35118 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35120 * config/nvptx/nvptx.c: Include intl.h.
35122 2017-03-01 Martin Jambor <mjambor@suse.cz>
35125 * ipa-prop.h (ipa_bits): Removed field known.
35126 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
35127 to pointers. Adjusted their comments to warn about their sharing.
35128 (ipcp_transformation_summary): Change bits to a vector of pointers.
35129 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
35130 (ipa_get_ipa_bits_for_value): Declare.
35131 * tree-vrp.h (value_range): Mark as GTY((for_user)).
35132 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
35133 (ipa_bits_hash_table): Likewise.
35134 (ipa_vr_ggc_hash_traits): Likewise.
35135 (ipa_vr_hash_table): Likewise.
35136 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
35137 being pointers and vr_known being removed.
35138 (ipa_set_jf_unknown): Likewise.
35139 (ipa_get_ipa_bits_for_value): New function.
35140 (ipa_set_jfunc_bits): Likewise.
35141 (ipa_get_value_range): New overloaded functions.
35142 (ipa_set_jfunc_vr): Likewise.
35143 (ipa_compute_jump_functions_for_edge): Use the above functions to
35144 construct bits and vr parts of jump functions.
35145 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
35146 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
35148 (ipcp_grow_transformations_if_necessary): Also allocate
35149 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
35151 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
35152 them. Fix too long lines.
35153 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
35154 vr_known being removed.
35155 (ipa_read_jump_function): Use new setter functions to construct bits
35156 and vr parts of jump functions or set them to NULL.
35157 (write_ipcp_transformation_info): Adjust for bits being pointers.
35158 (read_ipcp_transformation_info): Likewise.
35159 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
35161 Include gt-ipa-prop.h.
35162 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
35164 (ipcp_store_bits_results): Likewise.
35165 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
35166 Do not write to existing jump functions but use a temporary instead.
35168 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35171 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
35172 attempt to use its first operand as BIT_FIELD_REF base.
35174 2017-03-01 Richard Biener <rguenther@suse.de>
35176 PR middle-end/79721
35177 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
35178 interpolating formula in wrapping arithmetic.
35179 (chrec_apply): Convert chrec_evaluate return value to wanted type.
35181 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35183 PR tree-optimization/79734
35184 * tree-vect-generic.c (expand_vector_condition): Optimize
35185 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
35186 Handle VEC_COND_EXPR where comparison has different inner width from
35187 type's inner width.
35189 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
35191 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
35192 markup, and similar issues. Remove @opindex entries for things
35193 that aren't options. Add missing -mmpy-option entries.
35195 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35197 PR tree-optimization/79737
35198 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
35199 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
35200 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
35201 instead of byte_size. Formatting fix.
35202 (shift_bytes_in_array_right): Formatting fix.
35204 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
35207 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
35208 condition on optimize for the leaf function test.
35210 2017-02-28 Martin Liska <mliska@suse.cz>
35213 * read-rtl-function.c (function_reader::handle_unknown_directive):
35214 Bail out when one uses -flto.
35216 2017-02-28 Martin Liska <mliska@suse.cz>
35218 * common.opt: Replace space with tabular for options of <number>
35220 * config/i386/i386.opt: Show <number> value for
35221 -mlarge-data-threshold.
35222 * opts.c (print_filtered_help): Do not display number in hexadecimal
35225 2017-02-28 Martin Liska <mliska@suse.cz>
35227 * common.opt: Fix --help=option -Q for options which are of
35230 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
35232 * config/i386/i386.c (print_reg): Error out for values
35233 of 8-bit size in invalid integer register.
35235 2017-02-28 Martin Sebor <msebor@redhat.com>
35237 PR tree-optimization/79691
35238 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
35240 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35243 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
35244 gcc_unreachable with output_operand_lossage.
35246 2017-02-28 Richard Biener <rguenther@suse.de>
35248 PR tree-optimization/79740
35249 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
35251 (visit_nary_op): Insert the nary into the hashtable if we
35252 pattern-matched sth.
35253 * tree-ssa-pre.c (eliminate_insert): Robustify.
35255 2017-02-28 Richard Biener <rguenther@suse.de>
35257 PR middle-end/79731
35258 * fold-const.c (decode_field_reference): Reject out-of-bound
35261 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35263 * config/i386/i386.c: Include intl.h.
35264 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
35265 instead of just cond ? "..." : "...".
35266 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
35267 * coverage.c (read_counts_file): Likewise.
35268 * omp-offload.c: Include intl.h.
35269 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
35270 of just cond ? "..." : "...".
35271 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
35272 of just cond ? "..." : "...".
35274 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
35277 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
35279 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
35280 'tune for' CPU name.
35281 * config/arm/arm-cpu-data.h: Regenerated.
35283 2017-02-28 Richard Biener <rguenther@suse.de>
35285 PR tree-optimization/79732
35286 * tree-inline.c (expand_call_inline): Do not shadow var.
35288 2017-02-28 Richard Biener <rguenther@suse.de>
35290 PR tree-optimization/79723
35291 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
35292 address-space properly.
35294 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
35296 * doc/optinfo.texi (Optimization groups): Fix option used for
35298 * doc/invoke.texi (-fopt-info): Document "omp".
35299 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
35300 (OPTGROUP_ALL): Add OPTGROUP_OMP.
35301 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
35302 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
35303 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
35305 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
35307 * dumpfile.c (optgroup_options): Instead of "openmp", associate
35308 OPTGROUP_OMP with "omp".
35310 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
35313 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
35314 for arithmetic shift of unsigned V2DI.
35316 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
35318 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
35319 arc/linux.h headers.
35320 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
35321 (LINK_SPEC): Likewise.
35322 (ARC_TLS_EXTRA_START_SPEC): Likewise.
35323 (EXTRA_SPECS): Likewise.
35324 (STARTFILE_SPEC): Likewise.
35325 (ENDFILE_SPEC): Likewise.
35326 (LIB_SPEC): Likewise.
35327 (TARGET_SDATA_DEFAULT): Likewise.
35328 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
35329 (MULTILIB_DEFAULTS): Likewise.
35330 (DWARF2_UNWIND_INFO): Likewise.
35331 * config/arc/big.h: New file.
35332 * config/arc/elf.h: Likewise.
35333 * config/arc/linux.h: Likewise.
35334 * config/arc/t-uClibc: Remove.
35336 2017-02-27 Bin Cheng <bin.cheng@arm.com>
35338 PR tree-optimization/77536
35339 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
35340 (tree_transform_and_unroll_loop): Use above function to compute the
35341 estimated niter of unrolled loop and use it when scaling profile.
35342 Also use count info rather than frequency if it's non-zero.
35343 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
35344 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
35345 (vect_transform_loop): Call above function.
35347 2017-02-27 Richard Biener <rguenther@suse.de>
35349 PR tree-optimization/45397
35350 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
35351 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
35352 (visit_nary_op): Add pattern matching for CSEing sign-changed
35353 or truncated operations with wider ones.
35355 2017-02-27 Richard Biener <rguenther@suse.de>
35357 PR tree-optimization/79690
35358 * tree-vect-stmts.c (vectorizable_store): Use vector type
35359 built from the DR with address-space.
35361 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
35363 * doc/invoke.texi (Optimize Options): Refine the description
35364 of asan-use-after-return.
35366 2017-02-25 Alan Modra <amodra@gmail.com>
35368 PR rtl-optimization/79584
35369 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
35370 base, not ad->base_term, the reg within base. Remove assertion
35371 that ad->base == ad->base_term. Replace gen_int_mode using
35372 bogus mode with const0_rtx.
35374 2017-02-25 Jakub Jelinek <jakub@redhat.com>
35376 PR middle-end/79396
35377 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
35378 FMA_EXPR like tcc_binary or tcc_unary.
35380 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
35383 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
35385 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
35386 (output_loc_operands): Handle DW_OP_call_ref and
35387 DW_OP_GNU_variable_value.
35388 (struct variable_value_struct): New type.
35389 (struct variable_value_hasher): Likewise.
35390 (variable_value_hash): New variable.
35391 (string_types): Remove.
35392 (copy_loc_descr): New function.
35393 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
35394 (prepend_loc_descr_to_each): New function.
35395 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
35396 instead of add_loc_descr_to_each if the first argument is single
35397 location list and the second has multiple.
35398 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
35399 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
35400 when looking for variable value which doesn't have other location info.
35401 (loc_list_from_tree): Formatting fix.
35402 (gen_array_type_die): Simplify DW_AT_string_length handling.
35403 (adjust_string_types): Remove.
35404 (gen_subprogram_die): Don't call adjust_string_types nor test/set
35405 string_types. Call resolve_variable_values.
35406 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
35407 (resolve_addr_in_expr): Likewise. Add A argument.
35408 (copy_deref_exprloc): Remove deref argument. Adjust for the
35409 original expression being DW_OP_GNU_variable_value with optionally
35410 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
35411 optionally after it.
35412 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
35413 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
35414 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
35415 (variable_value_hasher::hash, variable_value_hasher::equal): New
35417 (resolve_variable_value_in_expr, resolve_variable_value,
35418 resolve_variable_values, note_variable_value_in_expr,
35419 note_variable_value): New functions.
35420 (dwarf2out_early_finish): Call note_variable_value on all toplevel
35423 2017-02-24 Jakub Jelinek <jakub@redhat.com>
35426 * opts.h (handle_generated_option): Add GENERATED_P argument.
35427 * opts-common.c (handle_option): Adjust function comment.
35428 (handle_generated_option): Add GENERATED_P argument, pass it to
35430 (control_warning_option): Pass false to handle_generated_option
35432 * opts.c (maybe_default_option): Pass true to handle_generated_option
35434 * optc-gen.awk: Likewise.
35436 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35438 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
35439 a REG, look at the REG it is a SUBREG of.
35440 (splitter for cmpeqsi_t): Ditto.
35442 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35444 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
35445 the special USEs with the pattern of the insn, not the insn itself.
35447 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
35450 * doc/invoke.texi: Document -mload-store-pairs.
35452 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35453 Sandra Loosemore <sandra@codesourcery.com>
35455 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
35456 argument isn't a CONST_INT.
35457 (nios2_alternate_compare_const): Assert op is a CONST_INT.
35458 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
35459 (nios2_validate_compare): Bypass alternate compare logic if *op2
35460 is not a CONST_INT.
35461 (ldstwm_operation_p): Return false if first_base is not a REG or
35462 if first_offset is not a CONST_INT.
35464 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35466 * config/cris/cris.md: Use correct operand in a define_peephole2.
35468 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35470 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
35472 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35474 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
35475 this_insn if it is an INSN or JUMP_INSN.
35476 (force_offsettable): Look at base, not at addr.
35477 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
35478 on things that aren't necessarily CONST_INTs.
35480 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
35482 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
35483 -mfpmath=sse is the default also for x86-32 targets with SSE2
35484 instruction set when @option{-ffast-math} is enabled
35486 2017-02-24 Jeff Law <law@redhat.com>
35488 PR rtl-optimizatoin/79286
35489 * ira.c (update_equiv_regs): Drop may_trap_p exception to
35492 2017-02-24 Richard Biener <rguenther@suse.de>
35494 PR tree-optimization/79389
35495 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
35498 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
35500 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
35501 function comment to reflect reality.
35502 (loop_exits_before_overflow): Fix typo in function description.
35504 2017-02-24 Richard Biener <rguenther@suse.de>
35506 PR tree-optimization/79389
35507 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
35508 properly that a threading opportunity exists. Detect conditional
35509 copy/constant propagation opportunities.
35511 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
35513 * config/visium/visium.md (type): Add trap.
35514 (b): New mode attribute.
35515 (*btst): Rename into...
35516 (*btst<mode>): ...this and adjust.
35517 (*cbranchsi4_btst_insn): Rename into...
35518 (*cbranch<mode>4_btst_insn): ...this and adjust.
35519 (trap): New define_insn.
35521 2017-02-23 Jakub Jelinek <jakub@redhat.com>
35523 PR tree-optimization/79389
35524 * ifcvt.c (struct noce_if_info): Add rev_cond field.
35525 (noce_reversed_cond_code): New function.
35526 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
35527 reversed_comparison_code. Formatting fix.
35528 (noce_try_store_flag): Test rev_cond != NULL in addition to
35529 reversed_comparison_code.
35530 (noce_try_store_flag_constants): Likewise.
35531 (noce_try_store_flag_mask): Likewise.
35532 (noce_try_addcc): Use rev_cond if non-NULL instead of
35533 reversed_comparison_code.
35534 (noce_try_cmove_arith): Likewise. Formatting fixes.
35535 (noce_try_minmax, noce_try_abs): Clear rev_cond.
35536 (noce_find_if_block): Initialize rev_cond.
35537 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
35538 instead of false as last argument never attempt to reverse it
35541 2017-02-23 Bin Cheng <bin.cheng@arm.com>
35543 PR tree-optimization/79663
35544 * tree-predcom.c (combine_chains): Process refs in reverse order
35545 only for ZERO length chains, and add explaining comment.
35547 2017-02-23 Jeff Law <law@redhat.com>
35549 PR tree-optimization/79578
35550 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
35551 in call to operand_equal_p.
35553 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
35556 * config/i386/cpuid.h: Fix another undefined behavior.
35558 2017-02-23 Richard Biener <rguenther@suse.de>
35560 PR tree-optimization/79683
35561 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
35562 vector types for data-refs.
35564 2017-02-23 Martin Liska <mliska@suse.cz>
35566 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
35568 2017-02-23 Jakub Jelinek <jakub@redhat.com>
35570 PR middle-end/79665
35571 * internal-fn.c (get_range_pos_neg): Moved to ...
35572 * tree.c (get_range_pos_neg): ... here. No longer static.
35573 * tree.h (get_range_pos_neg): New prototype.
35574 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
35575 are known to be in between 0 and signed maximum inclusive, try to
35576 expand both unsigned and signed divmod and use the cheaper one from
35579 2017-02-22 Jeff Law <law@redhat.com>
35581 PR tree-optimization/79578
35582 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
35583 to compare base operands.
35585 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
35588 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
35589 gpc_reg_operand instead of fpr_reg_operand.
35591 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
35593 * config/mips/mips.c (mips_return_in_memory): Force FP
35594 vector types to be returned in memory for o32 ABI.
35596 2017-02-22 Jakub Jelinek <jakub@redhat.com>
35598 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
35599 instead of DW_TAG_member for static data member declarations and don't
35600 set no_linkage_name for static inline data members.
35601 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
35604 2017-02-22 Martin Liska <mliska@suse.cz>
35606 * doc/invoke.texi: Replace inequality signs with square brackets
35609 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
35612 * lra-constraints.c (simplify_operand_subreg): Handle
35613 WORD_REGISTER_OPERATIONS targets.
35615 2017-02-22 Jakub Jelinek <jakub@redhat.com>
35618 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
35619 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
35620 elimination by swapping fld*.
35622 2017-02-22 Richard Biener <rguenther@suse.de>
35624 PR tree-optimization/79673
35625 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
35626 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
35627 irrelevant address-space qualifiers and avoiding a
35628 ADDR_SPACE_CONVERT_EXPR from fold_convert.
35630 2017-02-22 Richard Biener <rguenther@suse.de>
35632 PR tree-optimization/79666
35633 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
35634 to not symbolically negate if that may introduce undefined
35637 2017-02-22 Martin Liska <mliska@suse.cz>
35640 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
35641 * data-streamer-out.c (streamer_write_gcov_count_stream):
35643 * value-prof.c (stream_out_histogram_value): Make assert more
35644 precise based on type of counter.
35646 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
35649 * config/i386/i386.md (standard_x87sse_constant_load splitter):
35650 Use nonimmediate_operand instead of memory_operand for operand 1.
35651 (float-extend standard_x87sse_constant_load splitter): Ditto.
35653 2017-02-21 Jeff Law <law@redhat.com>
35655 PR tree-optimization/79621
35656 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
35657 blocks with edges to themselves.
35659 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35662 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
35663 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
35664 Use gimple_call_builtin_p.
35667 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
35668 on temporarily removed DEBUG_INSNs.
35670 PR tree-optimization/79649
35671 * tree-loop-distribution.c (classify_partition): Give up on
35672 non-generic address space loads/stores.
35674 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
35676 * doc/loop.texi (Loop manipulation): Remove nonexistent
35677 tree_ssa_loop_version from the documentation.
35678 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
35680 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35683 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
35684 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
35685 * config/rs6000/rs6000.c: Include except.h.
35686 (rs6000_expand_split_stack_prologue): Call
35687 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
35689 2017-02-21 Martin Jambor <mjambor@suse.cz>
35692 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
35693 have been analyzed.
35695 2017-02-21 Martin Jambor <mjambor@suse.cz>
35697 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
35698 for backward compatibility only.
35699 * doc/invoke.texi (Option Summary): Remove all references to
35700 -fipa-cp-alignment.
35702 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
35706 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35708 * lra-constraints.c (curr_insn_transform): Handle
35709 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
35711 2017-02-21 Martin Liska <mliska@suse.cz>
35713 * config/i386/i386.opt: Replace -masm-dialect with -masm.
35715 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
35717 PR translation/79638
35718 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
35720 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
35723 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
35724 (arm_function_ok_for_sibcall): Return false for an indirect call by
35725 descriptor if all the argument registers are used.
35726 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
35727 alignment of the function.
35729 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35731 PR tree-optimization/61441
35732 * simplify-rtx.c (simplify_const_unary_operation): For
35733 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
35734 the sNaN unmodified.
35736 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
35738 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
35739 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
35740 instead of SYSTEM_HEADER_DIR.
35742 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
35743 Martin Liška <mliska@suse.cz>
35745 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
35746 Fix typos and grammar, use active voice, and clarify.
35748 2017-02-20 Marek Polacek <polacek@redhat.com>
35750 PR middle-end/79537
35751 * gimplify.c (gimplify_expr): Handle unused *&&L;.
35754 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
35756 2017-02-20 Jakub Jelinek <jakub@redhat.com>
35759 * config/i386/i386.c (ix86_expand_builtin): Handle
35760 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
35761 ix86_builtins_isa[fcode].isa as a requirement of those
35762 flags and any other flag in the bitmask.
35763 (ix86_init_mmx_sse_builtins): Use 0 instead of
35764 ~OPTION_MASK_ISA_64BIT as mask.
35765 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
35766 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
35767 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
35768 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
35770 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35773 * lra-constraints.c (split_reg): Check requested split mode
35774 is supported by the register.
35776 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35778 * lra-constraints.c (simplify_operand_subreg): Remove early
35781 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35784 * lra-constraints.c (curr_insn_transform): Tighten condition
35785 for converting SUBREG reloads from OP_OUT to OP_INOUT.
35787 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35790 * lra-constraints.c (curr_insn_transform): Handle
35791 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
35793 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
35796 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
35798 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
35800 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
35803 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
35806 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
35808 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
35811 2017-02-18 Jakub Jelinek <jakub@redhat.com>
35813 * final.c (last_columnnum, override_columnnum): New variables.
35814 (final_start_function): Set last_columnnum, pass it to begin_prologue
35815 hook and pass 0 to dwarf2out_begin_prologue.
35816 (final_scan_insn): Update override_columnnum. Pass last_columnnum
35817 to source_line debug hook.
35818 (notice_source_line): Compute last_columnnum and for debug_column_info
35819 return true on column changes.
35820 * debug.h (struct gcc_debug_hooks): Add column argument to
35821 source_line and begin_prologue hooks.
35822 (debug_nothing_int_charstar_int_bool): Remove prototype.
35823 (debug_nothing_int_int_charstar,
35824 debug_nothing_int_int_charstar_int_bool): New prototypes.
35825 (dwarf2out_begin_prologue): Add column argument.
35826 * debug.c (do_nothing_debug_hooks): Adjust source_line and
35827 begin_prologue hooks.
35828 (debug_nothing_int_charstar_int_bool): Remove.
35829 (debug_nothing_int_int_charstar,
35830 debug_nothing_int_int_charstar_int_bool): New functions.
35831 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
35832 through to dwarf2out_source_line.
35833 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
35834 (dwarf2out_source_line): Add column argument, emit it if requested.
35835 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
35837 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
35838 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
35839 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
35840 through to dwarf2out_begin_prologue.
35841 (vmsdbgout_source_line): Add column argument, pass it through to
35842 dwarf2out_source_line.
35843 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
35844 dbxout_source_line caller.
35845 (dbxout_source_line): Add column argument.
35847 * common.opt (gno-column-info, gcolumn-info): New options.
35848 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
35849 (check_die): Also test for multiple DW_AT_decl_column attributes.
35850 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
35851 DW_AT_decl_column if requested.
35852 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
35854 (gen_variable_die): Likewise.
35855 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
35856 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
35859 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
35860 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
35861 (ix86_handle_option): Handle OPT_m3dnowa.
35862 * doc/invoke.texi (-m3dnowa): Document.
35863 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
35864 -m3dnowa instead of -m3dnow -march=athlon.
35867 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
35868 instead of gcc_assert for K, r and R code checks. Formatting fixes.
35870 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35873 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
35874 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
35875 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
35876 generator for vsx_xxpermdi_<mode>_be.
35877 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
35878 force big-endian semantics.
35879 (vsx_xxpermdi_<mode>_be): New define_expand with same
35880 implementation as previous version of vsx_xxpermdi_<mode>.
35882 2017-02-17 Jakub Jelinek <jakub@redhat.com>
35884 PR tree-optimization/79327
35885 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
35886 variable, its initialization and use.
35888 2017-02-17 Julia Koval <julia.koval@intel.com>
35890 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
35891 (OPTION_MASK_ISA_PKU_UNSET): New.
35892 (ix86_handle_option): Handle -mrdpid.
35893 * config/i386/cpuid.h (bit_RDPID): New.
35894 * config/i386/driver-i386.c (host_detect_local_cpu):
35895 Detect RDPID feature.
35896 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
35897 * config/i386/i386-c.c (ix86_target_macros_internal):
35899 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
35900 (ix86_valid_target_attribute_inner_p): Add "rdpid".
35901 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
35902 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
35903 * config/i386/i386.md (define_insn "rdpid"): New.
35904 * config/i386/i386.opt Add -mrdpid.
35905 * config/i386/immintrin.h (_rdpid_u32): New.
35907 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
35909 PR rtl-optimization/79541
35910 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
35911 instead of transforming it into USE.
35913 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
35915 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
35916 If HONOR_SNANS (SFmode) force the input to a register.
35917 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
35918 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
35919 an frsp or similar insn.
35921 2017-02-17 Martin Liska <mliska@suse.cz>
35923 PR rtl-optimization/79577
35924 * params.def (selsched-max-sched-times): Increase minimum to 1.
35926 2017-02-17 Martin Liska <mliska@suse.cz>
35928 PR rtl-optimization/79574
35929 * gcse.c (want_to_gcse_p): Prevent integer overflow.
35931 2017-02-17 Martin Liska <mliska@suse.cz>
35933 PR tree-optimization/79529
35934 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
35935 ssa_defined_default_def_p to handle cases which are implicitly
35937 * tree-ssa.c (ssa_defined_default_def_p): New function.
35938 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
35939 which are implicitly defined.
35940 * tree-ssa.h (ssa_defined_default_def_p): Declare.
35942 2017-02-17 Richard Biener <rguenther@suse.de>
35944 PR middle-end/79576
35945 * params.def (max-ssa-name-query-depth): Limit to 10.
35947 2017-02-17 Richard Biener <rguenther@suse.de>
35949 PR tree-optimization/79552
35950 * tree-ssa-structalias.c (visit_loadstore): Properly verify
35953 2017-02-17 Richard Biener <rguenther@suse.de>
35956 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
35958 2017-02-17 Marek Polacek <polacek@redhat.com>
35960 PR middle-end/79536
35961 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
35962 (fold_negate_expr): New wrapper.
35964 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
35966 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
35967 Correct terminology and de-emphasize pre-standard behavior.
35969 2017-02-16 Alan Modra <amodra@gmail.com>
35971 PR rtl-optimization/79286
35972 * ira.c (def_dominates_uses): New function.
35973 (update_equiv_regs): Don't create an equivalence for insns that
35974 may trap where the register def does not dominate the use.
35976 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
35978 PR rtl-optimization/78127
35979 * lra.c (lra): Call lra_eliminate before finish the loop after
35982 2017-02-16 Richard Biener <rguenther@suse.de>
35984 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
35986 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
35987 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
35988 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
35989 (isl_val_int_from_wi): New function.
35990 (extract_affine_gmp): Rename to ...
35991 (extract_affine_wi): ... this, take a widest_int.
35992 (extract_affine_int): Just wrap extract_affine_wi.
35993 (add_param_constraints): Use isl_val_int_from_wi.
35994 (add_loop_constraints): Likewise, and extract_affine_wi.
35996 2017-02-15 Jeff Law <law@redhat.com>
35998 PR middle-end/79521
35999 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
36000 ira_init_register_move_cost_if_necessary.
36002 2017-02-15 Martin Sebor <msebor@redhat.com>
36004 PR middle-end/32003
36005 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
36006 removed in a prior commit.
36008 2017-02-15 Bin Cheng <bin.cheng@arm.com>
36010 PR tree-optimization/79347
36011 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
36012 counters during peeling.
36014 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
36016 * Makefile.in (site.exp): Remove "set ISLVER".
36018 2017-02-15 Jakub Jelinek <jakub@redhat.com>
36021 * real.c (real_from_integer): Call real_convert even for decimal.
36023 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
36026 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
36028 2017-02-14 Andrew Pinski <apinski@cavium.com>
36030 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
36031 cores and change the partno/implementer to be correct.
36032 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
36033 the 'B" as the implementer.
36034 * config/aarch64/aarch64-tune.md: Regenerate.
36036 2017-02-14 Carl Love <cel@us.ibm.com>
36038 * config/rs6000/rs6000.c: Add case statement entry to make the
36039 xvcvuxdsp built-in argument unsigned.
36040 * config/rs6000/vsx.md: Fix the source and return operand types so they
36041 match the instruction definitions from the ISA document. Fix typo
36042 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
36045 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
36048 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
36049 member early_clobber_alts.
36050 * lra-lives.c (reg_early_clobber_p): New.
36051 (process_bb_lives): Use it.
36052 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
36053 (debug_operand_data): Initialize early_clobber_alts.
36054 (setup_operand_alternative): Set up early_clobber_alts.
36055 (collect_non_operand_hard_regs): Ditto. Pass early clobber
36056 alternatives to new_insn_reg.
36057 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
36059 (lra_update_insn_regno_info): Pass the new arg.
36061 2017-02-14 Jakub Jelinek <jakub@redhat.com>
36063 PR middle-end/79505
36064 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
36065 (new_oacc_loop_raw): Don't clear already cleared fields.
36068 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
36069 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
36070 _mm512_prefetch_i64gather_ps): New inline functions and macros.
36072 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
36075 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
36077 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
36080 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
36081 the extra instruction to the right place to store 128-bit constant
36084 2017-02-14 Martin Sebor <msebor@redhat.com>
36086 PR middle-end/79448
36087 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
36088 warning for strings of unknown length.
36090 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
36092 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
36094 2017-02-14 Jeff Law <law@redhat.com>
36097 * ira-costs.c (scan_one_insn): Initialize register move costs
36098 for pseudos seen in USE/CLOBBER insns.
36100 PR tree-optimization/79095
36101 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
36102 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
36103 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
36104 if the operands are known to be not equal, then the resulting range
36106 (intersect_ranges): If the new range is ~[0,0] and the old range is
36107 wide, then prefer ~[0,0].
36108 * tree-vrp.c (overflow_comparison_p_1): New function.
36109 (overflow_comparison_p): New function.
36110 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
36111 if NAME is used in an overflow test.
36112 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
36113 overflow check that can be expressed as an equality test, then adjust
36114 ops to be that equality test.
36116 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36118 * config/s390/s390-builtin-types.def: Remove flags argument.
36119 * config/s390/s390.c (s390_init_builtins): Likewise.
36121 2017-02-14 Martin Liska <mliska@suse.cz>
36123 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
36124 vector. Fix trailing white spaces.
36126 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
36128 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
36131 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36133 PR rtl-optimization/68664
36134 * config/arm/arm.c (arm_sched_can_speculate_insn):
36135 New function. Declare prototype.
36136 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
36138 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36140 PR rtl-optimization/68664
36141 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
36143 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
36145 2017-02-14 Amit Pawar <amit.pawar@amd.com>
36147 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
36148 max skip bytes for function, loop and jump.
36150 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36152 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
36153 ABS_EXPR for gimple dump.
36155 2017-02-14 Jakub Jelinek <jakub@redhat.com>
36158 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
36160 PR tree-optimization/79408
36161 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
36162 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
36163 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
36164 also if rhs1 is INTEGER_CST.
36166 2017-02-14 Richard Biener <rguenther@suse.de>
36168 PR middle-end/79432
36169 * tree-into-ssa.c (insert_phi_nodes): When the function can
36170 have abnormal edges rewrite SSA names with broken use-def
36171 dominance out of SSA and register them for PHI insertion.
36173 2017-02-13 Martin Sebor <msebor@redhat.com>
36175 PR middle-end/79496
36176 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
36177 clearing info.nowrite flag when snprintf size argument is a range.
36179 2017-02-13 Jakub Jelinek <jakub@redhat.com>
36181 * cprop.c (cprop_jump): Add missing space in string literal.
36182 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
36183 (get_constraint_for_component_ref): Likewise.
36184 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
36185 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
36186 * lra-constraints.c (process_alt_operands): Likewise.
36187 * ipa-inline.c (inline_small_functions): Likewise.
36188 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
36189 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
36190 * trans-mem.c (diagnose_tm_1_op): Likewise.
36191 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
36192 (grid_parallel_clauses_gridifiable): Likewise.
36194 * config/nvptx/mkoffload.c (process): Add space in between
36197 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
36198 "MOD4_SSE_REGS" and "ALL_REGS".
36200 * spellcheck.c (test_data): Add , in between "foo" and "food".
36202 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36205 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
36206 boundary crossing check and subsequent code generation agree.
36208 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36210 * config/aarch64/aarch64.c (has_memory_op): Delete.
36211 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
36214 2017-02-13 Jakub Jelinek <jakub@redhat.com>
36216 PR rtl-optimization/79388
36217 PR rtl-optimization/79450
36218 * combine.c (distribute_notes): When removing TEM_INSN for which
36219 corresponding dest has last value recorded, invalidate that last
36222 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36224 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
36225 of explicit '@'. Add missing assembly comment marker on branch costs
36228 2017-02-13 Nathan Sidwell <nathan@acm.org>
36230 * gengtype-lex.l (<in_struct>): Add '/'.
36232 2017-02-13 Martin Liska <mliska@suse.cz>
36235 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
36237 2017-02-13 Richard Biener <rguenther@suse.de>
36239 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
36241 * configure: Re-generate.
36242 * config.in: Likewise.
36243 * graphite-dependences.c: Simplify as if
36244 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
36245 * graphite-isl-ast-to-gimple.c: Likewise.
36246 * graphite-optimize-isl.c: Likewise.
36247 * graphite-poly.c: Likewise.
36248 * graphite-sese-to-poly.c: Likewise.
36249 * graphite.h: Likewise.
36250 * toplev.c: Include isl/version.h and use isl_version () for
36251 printing the ISL version.
36252 * doc/install.texi: Update ISL requirement.
36254 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
36256 * doc/standards.texi (Standards): Update reference to
36259 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
36261 * doc/extend.texi (Named Address Spaces): sourceware.org now
36263 * doc/install.texi (Binaries): Ditto.
36266 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
36268 * doc/cpp.texi: Replace "stringify"/"stringification" with C
36269 standard terminology "stringize"/"stringizing" throughout.
36270 * doc/cppinternals.texi: Likewise.
36272 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
36274 * doc/extend.texi: Fix some spelling mistakes and typos.
36275 * doc/invoke.texi: Likewise.
36277 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
36280 * params.def (inline-min-speedup) Change from 10 to 8.
36282 2017-02-11 Jakub Jelinek <jakub@redhat.com>
36284 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
36287 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
36290 * ipa-inline-analysis.c (get_minimal_bb): New function.
36291 (record_modified): Use it.
36292 (remap_edge_change_prob): Handle also ancestor functions.
36294 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
36296 * doc/contrib.texi (Contributors): Remove broken link into
36297 the Mauve CVS repository.
36299 2017-02-11 Jakub Jelinek <jakub@redhat.com>
36301 PR middle-end/79454
36302 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
36303 result computation whenever lhs doesn't have vector mode, not
36304 just when it has BLKmode.
36306 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
36308 * doc/makefile.texi (profiledbootstrap): Refer to the
36309 installation instructions only in textual form.
36311 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36314 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
36316 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
36318 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
36319 (Specific): Update mingw-w64 reference.
36321 (Specific): Remove broken link to Renesas RX processor.
36323 2017-02-10 Richard Biener <rguenther@suse.de>
36325 * toplev.c (process_options): Do not mention obsolete graphite
36326 options when printing sorry message about missing graphite support.
36327 Mention -floop-nest-optimize.
36329 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
36331 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
36332 (vtst_p16): Likewise.
36333 (vtstq_p8): Likewise.
36334 (vtstq_p16): Likewise.
36336 (vtstq_p64): Likewise.
36337 * config/arm/arm_neon.h (vgetq_lane_p64): New.
36338 (vset_lane_p64): New.
36339 (vsetq_lane_p64): New.
36341 2017-02-10 Jakub Jelinek <jakub@redhat.com>
36343 PR tree-optimization/79411
36344 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
36345 stmt operands are SSA_NAMEs used in abnormal phis.
36346 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
36349 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
36352 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
36355 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
36357 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
36359 2017-02-09 Jakub Jelinek <jakub@redhat.com>
36361 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
36365 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
36366 not arbitrary TREE_CONSTANT.
36369 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
36370 "omp declare target link" attribute unless is_global_var.
36371 * omp-offload.c (find_link_var_op): Likewise.
36373 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
36374 Chung-Lin Tang <cltang@codesourcery.com>
36376 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
36378 (gimplify_adjust_omp_clauses): Don't delete TILE.
36379 (gimplify_omp_for): Deal with TILE.
36380 * internal-fn.c (expand_GOACC_TILE): New function.
36381 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
36383 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
36384 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
36386 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
36387 avoid DIV for outermost collapse var.
36388 (expand_oacc_for): Insert tile element loop as needed. Adjust.
36389 Remove out of date comments, fix whitespace.
36390 * omp-general.c (omp_extract_for_data): Deal with tiling.
36391 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
36392 adjust OLF_DIM_BASE value.
36393 (struct omp_for_data): Add tiling field.
36394 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
36395 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
36396 for auto loops. Remove default auto determining, moved to
36397 oacc_loop_fixed_partitions.
36398 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
36399 stmts, add e_mask field.
36400 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
36401 (oacc_thread_numbers): Use oacc_dim_call.
36402 (oacc_xform_tile): New.
36403 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
36404 (finish_oacc_loop): Adjust for ifns vector.
36405 (oacc_loop_discover_walk): Append loop abstraction sites to list,
36406 add case for GOACC_TILE fns.
36407 (oacc_loop_xform_loop): Delete.
36408 (oacc_loop_process): Iterate over call list directly, and add
36409 handling for GOACC_TILE fns.
36410 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
36412 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
36413 vector partitioning to outer loops. Assign 2 partitions to loops
36414 when available. Add TILE handling.
36415 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
36416 (execite_oacc_device_lower): Process GOACC_TILE fns,
36417 ignore unknown specs.
36418 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
36419 * tree.c (omp_clause_num_ops): Adjust TILE ops.
36420 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
36422 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
36424 * configure.ac (ACX_BUGURL): Update.
36425 * configure: Regenerate.
36427 2017-02-09 Richard Biener <rguenther@suse.de>
36429 PR tree-optimization/69823
36430 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
36431 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
36433 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
36435 * config/arc/arc-c.def: Add __NPS400__ definition.
36436 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
36437 (TARGET_NPS400): Define.
36439 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
36441 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
36443 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
36444 pointer, arch_info.
36445 (arc_cpu_types): Fill the arch_info field with a pointer into the
36446 arc_arch_types table.
36447 (arc_selected_cpu): Declare.
36448 * config/arc/arc.c (arc_selected_cpu): Make global.
36449 (arc_selected_arch): Delete.
36450 (arc_base_cpu): Delete.
36451 (arc_override_options): Remove references to deleted variables,
36452 update access to arch information.
36453 (ARC_OPT): Update access to arch information.
36454 (ARC_OPTX): Likewise.
36455 * config/arc/arc.h (arc_base_cpu): Remove declaration.
36456 (TARGET_ARC600): Update access to arch information.
36457 (TARGET_ARC601): Likewise.
36458 (TARGET_ARC700): Likewise.
36459 (TARGET_EM): Likewise.
36460 (TARGET_HS): Likewise.
36461 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
36464 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
36467 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
36468 condition/operands for integer GE/LE/GEU/LEU operations.
36470 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
36472 PR translation/79397
36473 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
36476 2017-02-08 Martin Jambor <mjambor@suse.cz>
36479 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
36480 whether allocation happened.
36481 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
36482 nothing was allocated.
36484 2017-02-08 Jakub Jelinek <jakub@redhat.com>
36486 PR tree-optimization/79408
36487 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
36488 constant, but SSA_NAME with a known integer range, use the minimum
36489 of that range instead of op1 to determine if modulo can be replaced
36490 with its first operand.
36492 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36494 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
36496 2017-02-08 Richard Biener <rguenther@suse.de>
36498 PR tree-optimization/71824
36499 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
36500 Check all loops contained in the merged region.
36502 2017-02-07 Andrew Pinski <apinski@cavium.com>
36504 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
36506 2017-02-07 Andrew Pinski <apinski@cavium.com>
36508 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
36509 (thunderxt88): Likewise.
36510 (thunderxt81): Disable LSE and change v8.1 to v8.
36511 (thunderxt83): Likewise.
36513 2017-02-07 Jakub Jelinek <jakub@redhat.com>
36514 Richard Biener <rguenther@suse.de>
36516 PR middle-end/79399
36517 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
36518 type from int to size_t.
36519 * ira-costs.c (struct_costs_size): Change type from int to size_t.
36521 2017-02-07 Jakub Jelinek <jakub@redhat.com>
36523 PR rtl-optimization/79386
36524 * cprop.c (bypass_conditional_jumps): Initialize
36525 bypass_last_basic_block already before splitting bbs after
36526 unconditional traps...
36527 (bypass_conditional_jumps): ... rather than here.
36530 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
36531 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
36532 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
36533 fix -masm=intel patterns.
36535 2017-02-07 Richard Biener <rguenther@suse.de>
36537 PR tree-optimization/79256
36538 PR middle-end/79278
36539 * builtins.c (get_object_alignment_2): Use min_align_of_type
36540 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
36541 and ADJUST_FIELD_ALIGN.
36543 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
36545 * doc/tm.texi: Regenerate.
36546 * stor-layout.c (layout_decl): Adjust.
36547 (update_alignment_for_field): Likewise.
36548 (place_field): Likewise.
36549 (min_align_of_type): Likewise.
36550 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
36551 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
36552 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
36553 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
36554 * config/frv/frv.c (frv_adjust_field_align): Likewise.
36555 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
36556 * config/i386/i386.c (x86_field_alignment): Likewise.
36557 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
36558 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
36559 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
36560 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
36561 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
36562 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
36566 2017-01-30 Richard Biener <rguenther@suse.de>
36568 PR tree-optimization/79256
36569 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
36570 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
36573 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
36575 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
36576 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
36577 builtins to SImode and emit a zero-extend, if necessary.
36579 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
36581 * docs/invoke.texi (RISC-V Options): Alphabetize.
36583 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
36585 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
36588 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
36590 * config/riscv/riscv.c: New file.
36591 * common/config/riscv/riscv-common.c: Likewise.
36592 * config.gcc: Likewise.
36593 * config/riscv/constraints.md: Likewise.
36594 * config/riscv/elf.h: Likewise.
36595 * config/riscv/generic.md: Likewise.
36596 * config/riscv/linux.h: Likewise.
36597 * config/riscv/multilib-generator: Likewise.
36598 * config/riscv/peephole.md: Likewise.
36599 * config/riscv/pic.md: Likewise.
36600 * config/riscv/predicates.md: Likewise.
36601 * config/riscv/riscv-builtins.c: Likewise.
36602 * config/riscv/riscv-c.c: Likewise.
36603 * config/riscv/riscv-ftypes.def: Likewise.
36604 * config/riscv/riscv-modes.def: Likewise.
36605 * config/riscv/riscv-opts.h: Likewise.
36606 * config/riscv/riscv-protos.h: Likewise.
36607 * config/riscv/riscv.h: Likewise.
36608 * config/riscv/riscv.md: Likewise.
36609 * config/riscv/riscv.opt: Likewise.
36610 * config/riscv/sync.md: Likewise.
36611 * config/riscv/t-elf-multilib: Likewise.
36612 * config/riscv/t-linux: Likewise.
36613 * config/riscv/t-linux-multilib: Likewise.
36614 * config/riscv/t-riscv: Likewise.
36615 * configure.ac: Likewise.
36616 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
36617 Waterman as RISC-V maintainers.
36618 * doc/install.texi: Add RISC-V entries.
36619 * doc/invoke.texi: Add RISC-V options section.
36620 * doc/md.texi: Add RISC-V constraints section.
36621 * configure: Regenerated.
36623 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
36626 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
36627 false values to be constant vectors with all 0 or all 1 bits set.
36628 (vcondu<mode><mode>): Likewise.
36629 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
36631 (fpmask_comparison_operator): Update comment.
36632 (vecint_comparison_operator): New predicate.
36633 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
36634 vector conditionals when the true and false values are constant
36635 vectors with all 0 bits or all 1 bits set.
36637 2017-02-06 Martin Sebor <msebor@redhat.com>
36639 PR tree-optimization/79376
36640 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
36642 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
36644 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
36645 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
36646 to simplify split condition.
36648 2017-02-06 Jakub Jelinek <jakub@redhat.com>
36650 * omp-expand.c (oxpand_omp_atomic_fetch_op,
36651 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
36654 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
36656 PR rtl-optimization/68664
36657 * target.def (can_speculate_insn): New hook.
36658 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
36659 * doc/tm.texi: Regenerate.
36660 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
36661 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
36662 (rs6000_sched_can_speculate_insn): New function.
36664 2017-02-06 Jakub Jelinek <jakub@redhat.com>
36666 PR tree-optimization/79284
36667 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
36668 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
36669 vectorizable_mask_load_store, vectorizable_operation,
36670 vect_is_simple_cond, get_same_sized_vectype): Use it instead
36671 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
36672 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
36673 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
36674 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
36675 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
36676 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
36677 is_gimple_assign (stmt). Replace another such test with
36678 is_gimple_assign (stmt).
36680 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
36683 * config/avr/avr.c (rtl-iter.h): Include it.
36684 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
36685 (avr_legitimate_combined_insn): ...and implementation.
36687 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
36689 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
36690 * config/s390/s390.c (s390_const_operand_ok)
36691 (s390_canonicalize_comparison, s390_extract_part)
36692 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
36693 (s390_contiguous_bitmask_p, s390_rtx_costs)
36694 (legitimize_pic_address): Likewise.
36695 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
36696 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
36697 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
36698 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
36699 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
36701 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
36703 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
36704 REGNO($0) == REGNO($1).
36706 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36708 * config/s390/linux.h(SIZE_TYPE): Add comment.
36710 2017-02-06 Julian Brown <julian@codesourcery.com>
36711 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
36712 Virendra Pathak <virendra.pathak@broadcom.com>
36714 * config/aarch64/aarch64-cores.def: Change the scheduler
36716 * config/aarch64/aarch64.md: Include thunderx2t99.md.
36717 * config/aarch64/thunderx2t99.md: New file.
36719 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36721 * doc/standards.texi (Go Language): Update link to language
36724 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
36726 * tree-eh.c (lower_resx): Sanitize profile.
36727 (cleanup_empty_eh_move_lp): Likewise.
36729 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
36732 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
36734 * cfgloopmanip.h (loop_version): Update prototype.
36735 * modulo-sched.c (sms_schedule): Update call of loop_version.
36736 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
36737 * tree-parloops.c (gen_parallel_loop): Likewise.
36738 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
36739 * tree-ssa-loop-split.c (split_loop): Likewise.
36740 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
36741 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
36743 2017-02-05 Martin Liska <mliska@suse.cz>
36746 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
36748 (print_operand_address): Initialize a struct to zero.
36750 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36752 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
36753 garbage collector only in textual form.
36755 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36757 * doc/extend.texi (x86 specific memory model extensions for
36758 transactional memory): Simplify a phrase.
36760 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
36763 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
36764 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
36765 (atomic_storedi_1): Likewise.
36767 2017-02-04 Jakub Jelinek <jakub@redhat.com>
36769 PR tree-optimization/79338
36770 * tree-parloops.c (gather_scalar_reductions): Don't call
36771 vect_analyze_loop_form for loop->inner before destroying loop's
36774 2017-02-03 Martin Sebor <msebor@redhat.com>
36776 PR tree-optimization/79327
36777 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
36778 when precision has resulted in leading zeros.
36779 (format_integer): Adjust the likely counter to assume an unknown
36780 argument that may be zero is non-zero.
36782 2017-02-03 Jason Merrill <jason@redhat.com>
36785 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
36786 avoid copying non-taken branch.
36788 2017-02-03 Jakub Jelinek <jakub@redhat.com>
36790 PR tree-optimization/79340
36791 * tree-vect-loop.c (vectorizable_reduction): Release
36792 vec_defs elements after safe_splicing them into other vectors.
36795 PR tree-optimization/79327
36796 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
36797 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
36799 (format_integer): Use wide_int_to_tree instead of build_int_cst
36800 + to_?hwi. If argmin is NULL, just set argmin and argmax to
36801 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
36802 of shortest and longest sequence.
36804 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
36806 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
36807 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
36809 2017-02-03 Walter Lee <walt@tilera.com>
36812 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
36813 after initial stackframe link reg save.
36814 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
36816 2017-02-03 Jakub Jelinek <jakub@redhat.com>
36819 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
36820 wu for stxssp alternative.
36822 2017-02-03 Martin Sebor <msebor@redhat.com>
36824 PR tree-optimization/79352
36825 * gimple-fold.c (get_range_strlen): Add argument.
36826 (get_range_strlen): Change return type to bool.
36827 (get_maxval_strlen): Pass in a dummy argument.
36828 * gimple-fold.h (get_range_strlen): Change return type to bool.
36829 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
36830 * tree.h (array_at_struct_end_p): Add argument.
36831 * tree.c (array_at_struct_end_p): Handle it.
36833 2017-02-03 Martin Liska <mliska@suse.cz>
36836 * multiple_target.c (create_dispatcher_calls): Redirect edge
36837 from a caller of a dispatcher.
36838 (expand_target_clones): Make the clones local.
36839 (ipa_target_clone): Do both target clones and resolvers.
36840 (ipa_dispatcher_calls): Remove the pass.
36841 (pass_dispatcher_calls::gate): Likewise.
36842 (make_pass_dispatcher_calls): Likewise.
36843 * passes.def (pass_target_clone): Put as very first IPA early
36846 2017-02-03 Martin Liska <mliska@suse.cz>
36848 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
36849 in case of a function with ifunc attribute.
36851 2017-02-03 Martin Liska <mliska@suse.cz>
36853 * cgraph.c (cgraph_node::dump): Dump function version info.
36854 * symtab.c (symtab_node::dump_base): Add missing new line.
36856 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
36858 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
36859 (ifcombine_ifandif): Use it.
36861 2017-02-03 Martin Liska <mliska@suse.cz>
36863 * doc/invoke.texi: Document default value for
36864 use-after-scope-direct-emission-threshold.
36866 2017-02-03 Martin Liska <mliska@suse.cz>
36868 PR tree-optimization/79339
36869 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
36870 (format_floating): Likewise.
36872 2017-02-03 Martin Liska <mliska@suse.cz>
36875 * ipa-prop.c (ipa_node_params_t::insert): Remove current
36877 (ipa_node_params_t::remove): Likewise.
36878 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
36879 initialization from removed ipa_node_params_t::insert.
36880 (ipa_node_params::~ipa_node_params): Move from removed
36881 ipa_node_params_t::release.
36882 * symbol-summary.h (symbol_summary::m_released): New member.
36883 Do not release a summary twice. Do not allow to call finalizer
36884 for types of a summary that live in GGC memory.
36886 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
36888 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
36891 2017-02-02 Martin Sebor <msebor@redhat.com>
36893 PR middle-end/79275
36894 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
36895 (format_string): Tighten up the range of output for non-constant
36896 strings and correct the expected range for wide non-constant strings.
36898 2017-02-02 Martin Sebor <msebor@redhat.com>
36900 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
36902 PR middle-end/32003
36903 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
36905 (-fdump-tree-@var): Add to index and document how to come up
36906 with pass-specific option and dump file names.
36907 (-fdump-passes): Clarify where to look for output.
36909 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
36911 PR middle-end/77445
36912 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
36913 statistics of the analyzed path; allow threading for speed when
36914 any of BBs along the path are optimized for speed.
36916 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
36918 PR middle-end/78468
36919 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
36920 settings of the virtual registers.
36923 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
36925 * explow.c (get_dynamic_stack_size): Take known alignment of stack
36926 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
36929 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36931 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
36932 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
36934 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36936 * config/s390/s390.md: Add missing comments with the expanded
36938 * config/s390/vector.md: Likewise.
36939 * config/s390/vx-builtins.md: Likewise.
36941 2017-02-02 Jakub Jelinek <jakub@redhat.com>
36944 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
36945 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
36946 conditions on a single line.
36948 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36950 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
36951 __S390_VX__ to __VX__.
36953 2017-02-01 Andrew Pinski <apinski@cavium.com>
36955 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
36956 stmt_info to record_stmt_cost.
36957 (vect_get_known_peeling_cost): Pass stmt_info if known to
36959 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
36960 cpu_vector_cost field into
36961 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
36962 field into vec_int_stmt_cost and vec_fp_stmt_cost.
36963 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
36964 splitting of scalar_stmt_cost and vec_stmt_cost.
36965 (thunderx_vector_cost): Likewise.
36966 (cortexa57_vector_cost): LIkewise.
36967 (exynosm1_vector_cost): Likewise.
36968 (xgene1_vector_cost): Likewise.
36969 (thunderx2t99_vector_cost): Improve after the splitting of the two
36971 (aarch64_builtin_vectorization_cost): Update for the splitting of
36972 scalar_stmt_cost and vec_stmt_cost.
36974 2017-02-01 Torvald Riegel <triegel@redhat.com>
36975 Richard Henderson <rth@redhat.com>
36977 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
36978 conditional on existance of a fast atomic load.
36979 * optabs-query.c (can_atomic_load_p): New function.
36980 * optabs-query.h (can_atomic_load_p): Declare it.
36981 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
36982 no fast atomic load is available for the particular size of access.
36983 (expand_atomic_compare_and_swap): Likewise.
36984 (expand_atomic_load): Likewise.
36985 (expand_atomic_store): Likewise.
36986 (expand_atomic_fetch_op): Likewise.
36987 * testsuite/lib/target-supports.exp
36988 (check_effective_target_sync_int_128): Remove x86 because it provides
36989 no fast atomic load.
36990 (check_effective_target_sync_int_128_runtime): Likewise.
36992 2017-02-01 Richard Biener <rguenther@suse.de>
36994 * graphite.c: Include tree-vectorizer.h for find_loop_location.
36995 (graphite_transform_loops): Provide opt-info for optimized nests.
36996 * tree-parloop.c (parallelize_loops): Provide opt-info for
36997 parallelized loops.
36999 2017-02-01 Richard Biener <rguenther@suse.de>
37001 PR middle-end/79315
37002 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
37003 was not set before.
37005 2017-02-01 Richard Biener <rguenther@suse.de>
37007 PR tree-optimization/71824
37008 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37009 Verify the loops are valid in the merged SESE region.
37010 (scop_detection::can_represent_loop_1): Check analyzing the
37011 evolution of the number of iterations in the region succeeds.
37013 2017-01-31 Ian Lance Taylor <iant@golang.org>
37015 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
37016 REG_ARGS_SIZE note to 32-bit push insns and call insn.
37018 2017-01-31 David Malcolm <dmalcolm@redhat.com>
37020 PR preprocessor/79210
37021 * input.c (get_substring_ranges_for_loc): Replace line_width
37022 assertion with error-handling.
37024 2017-01-31 Richard Biener <rguenther@suse.de>
37026 PR tree-optimization/77318
37027 * graphite-sese-to-poly.c (extract_affine): Fix assert.
37028 (create_pw_aff_from_tree): Take loop parameter.
37029 (add_condition_to_pbb): Pass loop of the condition to
37030 create_pw_aff_from_tree.
37032 2017-01-31 Jakub Jelinek <jakub@redhat.com>
37034 * config/s390/s390.c (s390_asan_shadow_offset): New function.
37035 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
37037 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
37041 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
37043 (convert_int_to_float128): Likewise.
37044 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
37045 (convert_int_to_float128): Likewise.
37046 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
37047 (UNSPEC_IEEE128_CONVERT): Likewise.
37048 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
37049 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
37050 Use local variables for IBM extended format.
37051 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
37052 (fix_trunc<mode>si2_fprs): Likewise.
37053 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
37054 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
37055 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
37056 to know that we can now have integers of all sizes in vector
37058 (fix<uns>_<mode>di2_hw): Likewise.
37059 (float<uns>_<mode>si2_hw): Likewise.
37060 (fix_<mode>si2_hw): Likewise.
37061 (fixuns_<mode>si2_hw): Likewise.
37062 (float<uns>_<mode>di2_hw): Likewise.
37063 (float_<mode>di2_hw): Likewise.
37064 (float_<mode>si2_hw): Likewise.
37065 (floatuns_<mode>di2_hw): Likewise.
37066 (floatuns_<mode>si2_hw): Likewise.
37067 (xscvqp<su>wz_<mode>): Delete, no longer used.
37068 (xscvqp<su>dz_<mode>): Likewise.
37069 (xscv<su>dqp_<mode>): Likewise.
37070 (ieee128_mfvsrd_64bit): Likewise.
37071 (ieee128_mfvsrd_32bit): Likewise.
37072 (ieee128_mfvsrwz): Likewise.
37073 (ieee128_mtvsrw): Likewise.
37074 (ieee128_mtvsrd_64bit): Likewise.
37075 (ieee128_mtvsrd_32bit): Likewise.
37077 2017-01-31 Martin Liska <mliska@suse.cz>
37080 * ipa-prop.c (ipa_free_all_node_params): Call release method
37081 instead of ~sumbol_summary to not to trigger double times
37084 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
37086 PR tree-optimization/71691
37087 * bitmap.h (class auto_bitmap): New.
37088 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
37089 is_maybe_undefined instead of ssa_undefined_value_p.
37091 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37093 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37094 __S390_ARCH_LEVEL__ to __ARCH__.
37096 2017-01-31 Jakub Jelinek <jakub@redhat.com>
37098 PR tree-optimization/79267
37099 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
37100 if should_remove_lhs_p is true.
37102 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
37105 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
37106 (add_alignment_attribute): New.
37107 (base_type_die): Add alignment attribute.
37108 (subrange_type_die): Likewise.
37109 (modified_type_die): Likewise.
37110 (gen_array_type_die): Likewise.
37111 (gen_descr_array_type_die: Likewise.
37112 (gen_enumeration_type_die): Likewise.
37113 (gen_subprogram_die): Likewise.
37114 (gen_variable_die): Likewise.
37115 (gen_field_die): Likewise.
37116 (gen_ptr_to_mbr_type_die): Likewise.
37117 (gen_struct_or_union_type_die): Likewise.
37118 (gen_subroutine_type_die): Likewise.
37119 (gen_typedef_die): Likewise.
37120 (base_type_cmp): Compare alignment attribute.
37122 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37125 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
37126 (setb_unsigned) New pattern for setb with CCUNS.
37127 * config/rs6000/rs6000.c (expand_block_compare): Use a different
37128 subfc./subfe sequence to avoid overflow problems. Generate a
37129 shorter sequence with cmpld/setb for power9.
37130 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
37131 for generating subfc. instruction.
37132 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
37133 now uses this instruction.
37135 2017-01-30 Ian Lance Taylor <iant@google.com>
37138 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
37139 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
37141 2017-01-30 Martin Sebor <msebor@redhat.com>
37143 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
37144 Move constant to the right of a relational operator.
37145 (get_mpfr_format_length, format_character, format_string): Ditto.
37146 (should_warn_p, maybe_warn): Same.
37148 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
37150 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
37153 * asan.c (get_translation_unit_decl): Remove function.
37154 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
37156 2017-01-30 Martin Liska <mliska@suse.cz>
37158 PR gcov-profile/79259
37159 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
37160 -fprofile-generate.
37162 2017-01-30 Martin Liska <mliska@suse.cz>
37165 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
37166 Initialize variables with NULL value.
37168 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
37171 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
37173 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
37175 2017-01-30 Richard Biener <rguenther@suse.de>
37177 PR tree-optimization/79276
37178 * tree-vrp.c (process_assert_insertions): Properly adjust common
37179 when removing a duplicate.
37181 2017-01-30 Richard Biener <rguenther@suse.de>
37183 PR tree-optimization/79256
37184 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37185 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37187 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
37189 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
37192 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
37193 ("*r<noxa>sbg_<mode>_sll_bitmask")
37194 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
37195 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
37196 Use contiguous_bitmask_nowrap_operand.
37198 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37201 * config/rs6000/altivec.h (vec_xl): Revise #define.
37202 (vec_xst): Likewise.
37204 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
37206 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
37208 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
37210 PR rtl-optimization/79194
37211 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
37212 traps before call to bypass_conditional_jumps.
37214 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
37216 PR tree-optimization/71374
37217 * lra-constraints.c (check_conflict_input_operands): New.
37218 (match_reload): Use it.
37220 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
37223 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
37224 account to calculate conflict_set.
37226 2017-01-27 Bin Cheng <bin.cheng@arm.com>
37228 PR rtl-optimization/78559
37229 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
37230 other_insn in combine.
37232 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
37234 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
37235 uint16_type_node for BT_UINT16.
37237 2017-01-27 David Malcolm <dmalcolm@redhat.com>
37239 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
37240 "RTL Tests" to menu.
37241 (GIMPLE Tests): New node.
37242 (RTL Tests): New node.
37244 2017-01-27 Richard Biener <rguenther@suse.de>
37246 PR tree-optimization/79245
37247 * tree-loop-distribution.c (distribute_loop): Apply cost
37248 modeling also to detected patterns.
37250 2017-01-27 Richard Biener <rguenther@suse.de>
37252 PR tree-optimization/71433
37253 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
37254 (compare_assert_loc): New function.
37255 (process_assert_insertions): Sort and optimize assert locations
37256 to remove duplicates and push down identical assertions on
37257 edges to their destination block.
37259 2017-01-27 Richard Biener <rguenther@suse.de>
37261 PR tree-optimization/79244
37262 * tree-vrp.c (remove_range_assertions): Forcefully propagate
37263 out SSA names even if abnormal.
37265 2017-01-27 Jakub Jelinek <jakub@redhat.com>
37267 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
37268 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
37269 instead of MPFR_RNDN.
37271 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
37274 * arm.c (arm_option_override): Don't call build_target_option_node
37275 until after doing all option overrides.
37276 (arm_valid_target_attribute_tree): Likewise.
37278 2017-01-27 Martin Liska <mliska@suse.cz>
37280 * doc/invoke.texi (-fprofile-arcs): Document profiling support
37281 for {cd}tors and C++ {cd}tors.
37283 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
37285 * config/s390/s390.md ("*setmem_long_and")
37286 ("*setmem_long_and_31z"): Use zero_extend instead of and.
37288 2017-01-26 Martin Sebor <msebor@redhat.com>
37290 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
37293 2017-01-26 Martin Sebor <msebor@redhat.com>
37295 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
37296 HAVE_DFmode before using XFmode or DFmode.
37297 (parse_directive): Avoid using the z length modifier to avoid
37298 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
37300 PR middle-end/78703
37301 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
37302 to accept adjustment as an array.
37303 (get_int_range): New function.
37304 (struct directive): Make width and prec arrays.
37305 (directive::set_width, directive::set_precision): Call get_int_range.
37306 (format_integer, format_floating): Handle width and precision ranges.
37307 (format_string, parse_directive): Same.
37309 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37312 * dwarf2out.c (generate_skeleton_bottom_up): For children with
37313 comdat_type_p set, just clone them, but keep the children in the
37317 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
37318 which have direct callers with -fvar-tracking-assignments enabled
37320 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
37321 inside of type units.
37323 2017-01-26 Martin Sebor <msebor@redhat.com>
37325 PR middle-end/78703
37326 * gimple-ssa-sprintf.c (struct result_range): Add likely and
37328 (struct format_result): Replace number_chars, number_chars_min,
37329 and number_chars_max with a single member of struct result_range.
37331 (format_result::operator+=): Adjust.
37332 (struct fmtresult): Remove bounded. Handle likely and unlikely
37334 (fmtresult::adjust_for_width_or_precision): New function.
37335 (fmtresult:type_max_digits): New function.
37336 (bytes_remaining): Handle likely and unlikely counters.
37337 (min_bytes_remaining): Remove.
37338 (format_percent): Simplify.
37339 (format_integer, format_floating): Set likely and unlikely counters.
37340 (get_string_length, format_character, format_string): Same.
37341 (format_plain, should_warn_p): New function.
37342 (maybe_warn): Call should_warn_p. Update diagnostic messages
37343 and handle those for all directives, including plain strings.
37344 (format_directive): Handle likely and unlikely counters.
37345 Remove unnecessary quoting from diagnostics. Add an informational
37347 (add_bytes): Remove.
37348 (pass_sprintf_length::compute_format_length): Simplify.
37349 (try_substitute_return_value): Handle likely and unlikely counters.
37351 2017-01-26 Carl Love <cel@us.ibm.com>
37353 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
37354 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
37356 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
37359 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
37360 endianess for subregs into account.
37361 * lra-constraints.c (lra_constraints): Do risky transformations
37362 always on the first iteration.
37363 * lra-lives.c (check_pseudos_live_through_calls): Add arg
37364 last_call_used_reg_set.
37365 (process_bb_lives): Define and use last_call_used_reg_set.
37366 * lra.c (lra): Always continue after lra_constraints on the first
37369 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
37371 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
37373 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
37375 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37377 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
37378 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
37379 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
37380 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
37381 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
37382 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
37383 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
37384 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
37385 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
37387 2017-01-26 Marek Polacek <polacek@redhat.com>
37390 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
37391 for the third operand.
37393 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37395 PR middle-end/79236
37396 * omp-low.c (struct omp_context): Add simt_stmt field.
37397 (scan_omp_for): Return omp_context *.
37398 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
37399 context to the _simt_ SIMD stmt.
37400 (lower_omp_for): For combined SIMD with sibling _simt_
37401 SIMD, make sure to use the same decls in _looptemp_
37402 clauses as in the sibling.
37404 2017-01-26 David Sherwood <david.sherwood@arm.com>
37406 PR middle-end/79212
37407 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
37410 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37413 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
37414 emit fld b; fld a; if possible.
37416 * brig-builtins.def: Update copyright years.
37417 * config/arm/arm_acle_builtins.def: Update copyright years.
37419 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
37422 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
37423 constraint instead of o for the stxsd instruction.
37425 2017-01-25 Carl Love <cel@us.ibm.com>
37427 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
37428 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
37430 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
37432 * doc/invoke.texi (C++ Dialect Options): Fix typo.
37434 2017-01-25 Richard Biener <rguenther@suse.de>
37436 PR tree-optimization/69264
37437 * target.def (vector_alignment_reachable): Improve documentation.
37438 * doc/tm.texi: Regenerate.
37439 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
37441 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
37442 earlier changes with respect to TYPE_USER_ALIGN.
37443 (vector_alignment_reachable_p): Likewise. Improve dumping.
37445 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37448 * config/arm/arm.md (xordi3): Force constant operand into a register
37451 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37453 * doc/invoke.texi (-fstore-merging): Correct default optimization
37454 levels at which it is enabled.
37455 (-O): Move -fstore-merging from list to...
37458 2017-01-25 Richard Biener <rguenther@suse.de>
37461 * omp-expand.c: Include debug.h.
37462 (expand_omp_taskreg): Make sure to generate early debug before
37463 outlining anything from a function.
37464 (expand_omp_target): Likewise.
37465 (grid_expand_target_grid_body): Likewise.
37467 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
37470 * asan.c (get_translation_unit_decl): New function.
37471 (asan_add_global): Extract modules file name from globals
37472 TRANSLATION_UNIT_DECL name.
37474 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
37477 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
37478 for long calls with APCS frame and VFP.
37480 2017-01-24 David Malcolm <dmalcolm@redhat.com>
37482 * cfg.c (original_copy_tables_initialized_p): New function.
37483 * cfg.h (original_copy_tables_initialized_p): New decl.
37484 * cfgrtl.c (relink_block_chain): Guard the call to
37485 free_original_copy_tables with a call to
37486 original_copy_tables_initialized_p.
37487 * cgraph.h (symtab_node::native_rtl_p): New decl.
37488 * cgraphunit.c (symtab_node::native_rtl_p): New function.
37489 (symtab_node::needed_p): Don't assert for early assembly output
37490 for __RTL functions.
37491 (cgraph_node::finalize_function): Set "force_output" for __RTL
37493 (cgraph_node::analyze): Bail out early for __RTL functions.
37494 (analyze_functions): Update assertion to support __RTL functions.
37495 (cgraph_node::expand): Bail out early for __RTL functions.
37496 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
37498 * function.h (struct function): Update comment for field
37500 * gimple-expr.c: Include "tree-pass.h".
37501 (gimple_has_body_p): Return false for __RTL functions.
37502 * Makefile.in (OBJS): Add run-rtl-passes.o.
37503 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
37505 (gcc::pass_manager::get_clean_slate): New accessor.
37506 * passes.c: Include "insn-addr.h".
37507 (should_skip_pass_p): Add logging. Update logic for running
37508 "expand" to be compatible with both __GIMPLE and __RTL. Guard
37509 property-provider override so it is only done for gimple passes.
37511 (skip_pass): New function.
37512 (execute_one_pass): Call skip_pass when skipping passes.
37513 * read-md.c (md_reader::read_char): Support filtering
37514 the input to a subset of line numbers.
37515 (md_reader::md_reader): Initialize fields
37516 m_first_line and m_last_line.
37517 (md_reader::read_file_fragment): New function.
37518 * read-md.h (md_reader::read_file_fragment): New decl.
37519 (md_reader::m_first_line): New field.
37520 (md_reader::m_last_line): New field.
37521 * read-rtl-function.c (function_reader::create_function): Only
37522 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
37523 curr_properties. Set DECL_INITIAL to a dummy block.
37524 (read_rtl_function_body_from_file_range): New function.
37525 * read-rtl-function.h (read_rtl_function_body_from_file_range):
37527 * run-rtl-passes.c: New file.
37528 * run-rtl-passes.h: New file.
37530 2017-01-24 Jeff Law <law@redhat.com>
37532 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
37535 2017-01-24 Bin Cheng <bin.cheng@arm.com>
37537 PR tree-optimization/79159
37538 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
37539 (record_nonwrapping_iv): Improve boundary using above function if no
37540 value range information.
37542 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
37543 Martin Jambor <mjambor@suse.cz>
37545 * brig-builtins.def: New file.
37546 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
37547 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
37548 (DEF_HSAIL_SAT_BUILTIN): Likewise.
37549 (DEF_HSAIL_INTR_BUILTIN): Likewise.
37550 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
37551 * builtin-types.def (BT_INT8): New.
37552 (BT_INT16): Likewise.
37553 (BT_UINT8): Likewise.
37554 (BT_UINT16): Likewise.
37555 (BT_FN_ULONG): Likewise.
37556 (BT_FN_UINT_INT): Likewise.
37557 (BT_FN_UINT_ULONG): Likewise.
37558 (BT_FN_UINT_LONG): Likewise.
37559 (BT_FN_UINT_PTR): Likewise.
37560 (BT_FN_ULONG_PTR): Likewise.
37561 (BT_FN_INT8_FLOAT): Likewise.
37562 (BT_FN_INT16_FLOAT): Likewise.
37563 (BT_FN_UINT32_FLOAT): Likewise.
37564 (BT_FN_UINT16_FLOAT): Likewise.
37565 (BT_FN_UINT8_FLOAT): Likewise.
37566 (BT_FN_UINT64_FLOAT): Likewise.
37567 (BT_FN_UINT16_UINT32): Likewise.
37568 (BT_FN_UINT32_UINT16): Likewise.
37569 (BT_FN_UINT16_UINT16_UINT16): Likewise.
37570 (BT_FN_INT_PTR_INT): Likewise.
37571 (BT_FN_UINT_PTR_UINT): Likewise.
37572 (BT_FN_LONG_PTR_LONG): Likewise.
37573 (BT_FN_ULONG_PTR_ULONG): Likewise.
37574 (BT_FN_VOID_UINT64_UINT64): Likewise.
37575 (BT_FN_UINT8_UINT8_UINT8): Likewise.
37576 (BT_FN_INT8_INT8_INT8): Likewise.
37577 (BT_FN_INT16_INT16_INT16): Likewise.
37578 (BT_FN_INT_INT_INT): Likewise.
37579 (BT_FN_UINT_FLOAT_UINT): Likewise.
37580 (BT_FN_FLOAT_UINT_UINT): Likewise.
37581 (BT_FN_ULONG_UINT_UINT): Likewise.
37582 (BT_FN_ULONG_UINT_PTR): Likewise.
37583 (BT_FN_ULONG_ULONG_ULONG): Likewise.
37584 (BT_FN_UINT_UINT_UINT): Likewise.
37585 (BT_FN_VOID_UINT_PTR): Likewise.
37586 (BT_FN_UINT_UINT_PTR: Likewise.
37587 (BT_FN_UINT32_UINT64_PTR): Likewise.
37588 (BT_FN_INT_INT_UINT_UINT): Likewise.
37589 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
37590 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
37591 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
37592 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
37593 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
37594 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
37595 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
37596 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
37597 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
37598 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
37599 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
37600 * doc/frontends.texi: List BRIG FE.
37601 * doc/install.texi (Testing): Add BRIG tesring requirements.
37602 * doc/invoke.texi (Overall Options): Mention BRIG.
37603 * doc/standards.texi (Standards): Doucment BRIG HSA version.
37605 2017-01-24 Richard Biener <rguenther@suse.de>
37607 PR translation/79208
37608 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
37610 2017-01-24 Martin Jambor <mjambor@suse.cz>
37613 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
37614 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
37615 and known_contexts.
37617 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
37619 PR middle-end/79123
37620 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
37621 casts from signed to unsigned really don't have a range.
37623 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
37625 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
37626 GMP_RNDx for compatiblity.
37628 2017-01-24 Martin Liska <mliska@suse.cz>
37631 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
37632 that would prevent us to call alloca with -1 as argument.
37634 2017-01-24 Jakub Jelinek <jakub@redhat.com>
37636 * dwarf2out.c (output_compilation_unit_header, output_file_names):
37637 Avoid -Wformat-security warning.
37639 2017-01-23 Andrew Pinski <apinski@cavium.com>
37641 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
37644 2017-01-23 Martin Sebor <msebor@redhat.com>
37646 PR middle-end/78703
37647 * gimple-ssa-sprintf.c (warn_level): New global.
37648 (format_integer): Use it here and throughout the rest of the file.
37649 Use the same switch to compute sign as base.
37650 (maybe_warn): New function.
37651 (format_directive): Factor out warnings into maybe_warn.
37652 Add debugging output. Use warn_level.
37653 (add_bytes): Use warn_level.
37654 (pass_sprintf_length::compute_format_length): Add debugging output.
37655 (try_substitute_return_value): Same.
37656 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
37658 PR middle-end/78703
37659 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
37660 (struct fmtresult, format_integer, format_floating): Adjust.
37661 (fmtresult::fmtresult): Set max correctly in two argument ctor.
37662 (get_string_length, format_string,format_directive): Same.
37663 (pass_sprintf_length::compute_format_length): Same.
37664 (try_substitute_return_value): Simplify slightly.
37666 PR middle-end/78703
37667 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
37668 (fmtresult::operator+=): Outlined.
37669 (struct fmtresult): Add ctors.
37670 (struct conversion_spec): Rename...
37671 (struct directive): ...to this. Add and remove data members.
37672 (directive::set_width, directive::set_precision): New functions.
37673 (format_percent): Use fmtresult ctor.
37674 (get_width_and_precision): Remove.
37675 (format_integer): Make naming changes. Avoid computing width and
37677 (format_floating): Same. Adjust indentation.
37678 (format_character, format_none): New functions.
37679 (format_string): Moved character handling to format_character.
37680 (format_directive): Remove arguments, change return type.
37681 (parse_directive): New function.
37682 (pass_sprintf_length::compute_format_length): Move directive
37683 parsing to parse_directive.
37685 2017-01-23 Jakub Jelinek <jakub@redhat.com>
37687 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
37688 (assign_assembler_name_if_needed): ... this.
37689 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
37690 (assign_assembler_name_if_needed): ... this.
37691 (free_lang_data_in_cgraph): Adjust callers.
37692 * cgraphunit.c (cgraph_node::analyze): Likewise.
37693 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
37696 2017-01-23 Richard Biener <rguenther@suse.de>
37698 PR tree-optimization/79088
37699 PR tree-optimization/79188
37700 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
37701 resetting loop bounds after last path deletion. Reset loop
37702 bounds of the target loop, make code match the comments.
37703 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
37704 Make sure loops need no fixups.
37706 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
37708 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
37709 exponent support with double type for first argument.
37710 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
37711 type returned by __builtin_vec_extract_sig,
37712 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
37713 functions from "vector int" to "vector unsigned int" or from
37714 "vector long long int" to "vector unsigned long long int".
37715 Changed type returned by __builtin_vec_extract_exp,
37716 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
37717 functions from "vector int" to "vector unsigned int" or from
37718 "vector long long int" to "vector unsigned long long int".
37719 Changed return type of __builtin_vec_test_data_class,
37720 __builtin_vec_test_data_class_sp, and
37721 __builtin_vec_test_data_class_dp from "vector int" to
37722 "vector bool int" or from "vector long long int" to "vector bool
37723 long long int" and changed second argument type from "unsigned
37724 int" to "int". Added new overloaded function forms "vector float
37725 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
37726 "vector float __builtin_vec_insert_exp_sp (vector float, vector
37727 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
37728 double, vector unsigned long long int)" and "vector double
37729 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
37730 long int)". Changed return type of
37731 __builtin_scalar_test_data_class and
37732 __builtin_scalar_test_data_class_sp and
37733 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
37734 int" and changed second argument from "unsigned int" to "int".
37735 Changed type returned by __builtin_scalar_test_neg,
37736 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
37737 from "int" to "bool int". Added new overloaded function form
37738 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
37739 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
37740 exponent double-precision with floating point first argument.
37741 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
37742 documentation of scalar_test_data_class, scalar_test_neg,
37743 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
37744 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
37745 vec_test_data_class built-in functions to reflect refinements in
37746 their type signatures.
37748 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
37750 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
37752 (aarch64_elf_asm_destructor): Likewise.
37754 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
37756 PR rtl-optimization/78634
37757 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
37758 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
37759 * ifcvt.c (noce_try_cmove): Add missing cost check.
37761 PR rtl-optimization/71724
37762 * combine.c (if_then_else_cond): Look for situations where it is
37763 beneficial to undo the work of one of the recursive calls.
37765 2017-01-23 Bin Cheng <bin.cheng@arm.com>
37767 PR tree-optimization/70754
37768 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
37769 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
37770 combined stmt before it if not NULL.
37771 (combine_chains): Process refs reversely and compute dominance point
37774 2017-01-23 Martin Liska <mliska@suse.cz>
37776 PR tree-optimization/79196
37777 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
37778 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
37780 (strlen_optimize_stmt): Call the renamed function.
37782 2017-01-23 Michael Matz <matz@suse.de>
37784 PR tree-optimization/78384
37785 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
37787 2017-01-23 Richard Biener <rguenther@suse.de>
37789 PR tree-optimization/79186
37790 * tree-vrp.c (register_new_assert_for): Make sure we've seen
37791 both incoming edges before moving an assert.
37793 2017-01-23 Martin Jambor <mjambor@suse.cz>
37795 * ipa-prop.c (load_from_param_1): Removed.
37796 (load_from_unmodified_param): Bits from load_from_param_1 put back
37798 (load_from_param): Removed.
37799 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
37800 with stmt. Reverted back to use of load_from_unmodified_param.
37802 2017-01-23 Martin Jambor <mjambor@suse.cz>
37805 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
37806 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
37807 field a pointer to garbage collected vector, mark lattices and
37808 ipcp_orig_node with GTY((skip)).
37809 (ipa_get_param_count): Adjust to descriptors being a pointer.
37810 (ipa_get_param): Likewise.
37811 (ipa_get_type): Likewise.
37812 (ipa_get_param_move_cost): Likewise.
37813 (ipa_set_param_used): Likewise.
37814 (ipa_get_controlled_uses): Likewise.
37815 (ipa_set_controlled_uses): Likewise.
37816 (ipa_is_param_used): Likewise.
37817 (ipa_node_params_t): Move into garbage collector. New methods insert
37819 (ipa_node_params_sum): Annotate wth GTY(()).
37820 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
37822 (ipa_load_from_parm_agg): Adjust declaration.
37823 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
37824 * ipa-profile.c (ipa_profile): Likewise.
37825 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
37826 (ipa_populate_param_decls): Make descriptors parameter garbage
37828 (ipa_dump_param): Adjust to descriptors being a pointer.
37829 (ipa_alloc_node_params): Likewise.
37830 (ipa_initialize_node_params): Likewise.
37831 (load_from_param_1): Make descriptors parameter garbage collected.
37832 (load_from_unmodified_param): Likewise.
37833 (load_from_param): Likewise.
37834 (ipa_load_from_parm_agg): Likewise.
37835 (ipa_node_params::~ipa_node_params): Removed.
37836 (ipa_free_all_node_params): Remove call to delete operator.
37837 (ipa_node_params_t::insert): New.
37838 (ipa_node_params_t::remove): Likewise.
37839 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
37840 copy known_csts and known_contexts vectors.
37841 (ipa_read_node_info): Adjust to descriptors being a pointer.
37842 (ipcp_modif_dom_walker): Make m_descriptors field garbage
37844 (ipcp_transform_function): Make descriptors variable garbage
37847 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
37849 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
37850 * config/i386/avx512dqintrin.h: Ditto.
37851 * config/i386/avx512fintrin.h: Ditto.
37852 * config/i386/i386.c: Handle new builtins.
37853 * config/i386/i386-builtin.def: Add new builtins.
37854 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
37855 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
37857 2017-01-23 Jakub Jelinek <jakub@redhat.com>
37858 Martin Liska <mliska@suse.cz>
37860 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
37861 * asan.c (asan_expand_poison_ifn): Support stores and use
37862 appropriate ASAN report function.
37863 * internal-fn.c (expand_ASAN_POISON_USE): New function.
37864 * internal-fn.def (ASAN_POISON_USE): Declare.
37865 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
37866 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
37867 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
37868 ASAN_POISON calls w/o LHS.
37869 * tree-ssa.c (execute_update_addresses_taken): Create clobber
37870 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
37871 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
37872 * gimplify.c (asan_poison_variables): Add attribute
37873 use_after_scope_memory to variables that really needs to live
37875 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
37876 having the attribute.
37878 2017-01-23 Martin Liska <mliska@suse.cz>
37880 * asan.c (create_asan_shadow_var): New function.
37881 (asan_expand_poison_ifn): Likewise.
37882 * asan.h (asan_expand_poison_ifn): New declaration.
37883 * internal-fn.c (expand_ASAN_POISON): Likewise.
37884 * internal-fn.def (ASAN_POISON): New builtin.
37885 * sanopt.c (pass_sanopt::execute): Expand
37886 asan_expand_poison_ifn.
37887 * tree-inline.c (copy_decl_for_dup_finish): Make function
37889 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
37890 * tree-ssa.c (is_asan_mark_p): New function.
37891 (execute_update_addresses_taken): Rewrite local variables
37892 (identified just by use-after-scope as addressable) into SSA.
37894 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
37896 * doc/install.texi (Specific): opensource.apple.com uses https
37897 now. Remove trailing slash.
37899 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
37901 * README.Portability: Remove note on an Irix compatibility issue.
37903 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
37905 * gcov.c (INCLUDE_ALGORITHM): Define.
37906 (INCLUDE_VECTOR): Define.
37907 No longer include <vector> and <algorithm> directly.
37909 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
37911 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
37913 * doc/invoke.texi (Code Gen Options): Ditto.
37915 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
37918 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
37920 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
37922 rtl-optimization/79125
37923 * cprop.c (local_cprop_pass): Handle cases where we make an
37924 unconditional trap.
37926 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
37930 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
37931 read from, for big endian.
37933 2017-01-20 Jiong Wang <jiong.wang@arm.com>
37935 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
37936 register pauth builtins for LP64 only.
37938 2017-01-20 Marek Polacek <polacek@redhat.com>
37941 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
37944 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
37946 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
37948 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
37949 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
37950 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
37952 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37955 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
37956 in big-endian mode when they are not a single duplicated value.
37958 2017-01-20 Richard Biener <rguenther@suse.de>
37960 * BASE-VER: Bump to 7.0.1.
37962 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
37964 * omp-low.c (omplow_simd_context): New struct. Use it...
37965 (lower_rec_simd_input_clauses): ...here and...
37966 (lower_rec_input_clauses): ...here to hold common data. Adjust all
37967 references to idx, lane, max_vf, is_simt.
37969 2017-01-20 Graham Markall <graham.markall@embecosm.com>
37971 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
37974 2017-01-20 Martin Jambor <mjambor@suse.cz>
37976 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
37977 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
37979 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
37980 (GTFILES): Rename hsa.c to hsa-common.c.
37981 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
37982 * hsa-dump.c: Likewise.
37983 * hsa-gen.c: Likewise.
37984 * hsa-regalloc.c: Likewise.
37985 * ipa-hsa.c: Likewise.
37986 * omp-expand.c: Likewise.
37987 * omp-low.c: Likewise.
37988 * toplev.c: Likewise.
37990 2017-01-20 Marek Polacek <polacek@redhat.com>
37993 * doc/invoke.texi: Document -Wduplicated-branches.
37994 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
37995 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
37996 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
37997 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
37998 return 0 only when not OEP_LEXICOGRAPHIC.
37999 (fold_build_cleanup_point_expr): Use the expression
38000 location when building CLEANUP_POINT_EXPR.
38001 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
38002 * tree.c (add_expr): Handle error_mark_node.
38004 2017-01-20 Martin Liska <mliska@suse.cz>
38007 * tree-profile.c (init_ic_make_global_vars): Do not call
38009 (gimple_init_gcov_profiler): Likewise.
38011 2017-01-20 Martin Liska <mliska@suse.cz>
38014 * cgraph.h (maybe_create_reference): Remove argument and
38016 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
38018 * ipa-cp.c (create_specialized_node): Likewise.
38019 * symtab.c (symtab_node::maybe_create_reference): Handle
38020 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
38022 2017-01-20 Martin Liska <mliska@suse.cz>
38024 * read-rtl-function.c (function_reader::create_function): Use
38025 build_decl instread of build_decl_stat.
38027 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
38029 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
38030 * config/i386/avx512dqintrin.h: Ditto.
38031 * config/i386/avx512fintrin.h: Ditto.
38032 * config/i386/i386-builtin-types.def: Add new types.
38033 * config/i386/i386.c: Handle new types.
38034 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
38035 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
38036 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
38037 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
38038 (__builtin_ia32_kshiftridi): New.
38039 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
38041 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
38045 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
38046 define to rs6000_init_stack_protect_guard.
38047 (rs6000_init_stack_protect_guard): New function.
38049 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
38050 Yunqiang Su <yunqiang.su@imgtec.com>
38052 * config.gcc (supported_defaults): Add madd4.
38053 (with_madd4): Add validation.
38054 (all_defaults): Add madd4.
38055 * config/mips/mips.opt (mmadd4): New option.
38056 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
38058 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
38060 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
38061 (ISA_HAS_FUSED_MADD4): Likewise.
38062 * doc/invoke.texi (-mmadd4): Document the new option.
38063 * doc/install.texi (--with-madd4): Document the new option.
38065 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38067 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
38068 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
38069 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
38070 (aarch64_init_pauth_hint_builtins): New.
38071 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
38072 (aarch64_expand_builtin): Expand new builtins.
38074 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38076 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
38077 * combine-stack-adj.c (no_unhandled_cfa): Handle
38078 REG_CFA_TOGGLE_RA_MANGLE.
38079 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
38080 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
38081 info for return address signing.
38082 (aarch64_expand_epilogue): Likewise.
38084 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38086 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
38087 * config/aarch64/aarch64-protos.h
38088 (aarch64_return_address_signing_enabled): New declaration.
38089 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
38091 (aarch64_expand_prologue): Sign return address before it's pushed onto
38093 (aarch64_expand_epilogue): Authenticate return address fetched from
38095 (aarch64_override_options): Sanity check for ILP32 and ISA level.
38096 (aarch64_attributes): New function attributes for "sign-return-address".
38097 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
38098 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
38099 ("*do_return"): Generate combined instructions according to key index.
38100 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
38101 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
38103 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
38104 * config/aarch64/aarch64.opt (msign-return-address=): New.
38105 * doc/extend.texi (AArch64 Function Attributes): Documents
38106 "sign-return-address=".
38107 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
38109 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
38111 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
38112 overall option summary.
38114 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38116 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
38117 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
38118 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
38119 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
38121 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
38123 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
38124 -mpower9-minmax by default for -mcpu=power9.
38125 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
38126 128-bit floating point.
38128 2017-01-20 Alan Modra <amodra@gmail.com>
38130 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
38131 optimizing for size.
38133 2017-01-20 Alan Modra <amodra@gmail.com>
38136 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
38137 for strcmp and strncmp from corresponding builtin decl.
38139 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
38141 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
38142 instead of i386/rtems-64.h.
38143 * config/i386/rtems-64.h: Remove.
38145 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
38149 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
38151 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
38153 2017-01-19 Tamar Christina <tamar.christina@arm.com>
38155 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
38156 Change int to HOST_WIDE_INT.
38157 * config/aarch64/aarch64-protos.h
38158 (aarch64_simd_gen_const_vector_dup): Likewise.
38159 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
38161 2017-01-19 David Malcolm <dmalcolm@redhat.com>
38163 * langhooks-def.h (lhd_type_for_size): New decl.
38164 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
38165 * langhooks.c (lhd_type_for_size): New function, taken from
38168 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
38170 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
38171 define_bypass for CR latency.
38172 (power9-cracked-alu): Update bypass latency and remove power9-branch.
38173 (power9-alu2): Add define_bypass for CR latency.
38175 (power9-mul): Update insn latency.
38176 (power9-mul-compare): Update insn latency, bypass latency and remove
38179 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38181 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
38183 * config/aarch64/aarch64.md
38184 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
38185 aarch64_nopcrelative_literal_loads.
38186 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
38188 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
38190 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
38191 TARGET_LOONGSON_3A.
38192 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
38194 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
38197 * config.gcc (supported_defaults): Add lxc1-sxc1.
38198 (with_lxc1_sxc1): Add validation.
38199 (all_defaults): Add lxc1-sxc1.
38200 * config/mips/mips.opt (mlxc1-sxc1): New option.
38201 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
38203 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
38204 __mips_no_lxc1_sxc1.
38205 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
38206 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
38207 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
38209 2017-01-19 Richard Biener <rguenther@suse.de>
38211 PR tree-optimization/72488
38212 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
38213 sure to restore SSA info.
38214 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
38216 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
38218 PR rtl-optimization/79121
38219 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
38220 of the inner type when shifting an extended value.
38222 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
38225 * symtab.c (symtab_node::equal_address_to): Fix comparing of
38226 interposable aliases.
38228 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
38231 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
38232 Use the evmergelohi instruction.
38233 (mov_si<mode>_e500_subreg4_2_le): Likewise.
38234 (mov_sitf_e500_subreg8_2_be): Likewise.
38235 (mov_sitf_e500_subreg12_2_le): Likewise.
38236 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
38237 (mov_si<mode>_e500_subreg4_2_be): Likewise.
38238 (mov_sitf_e500_subreg8_2_le): Likewise.
38239 (mov_sitf_e500_subreg12_2_be): Likewise.
38241 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38243 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
38244 attribute from vecsimple to vecperm.
38245 (altivec_vbpermq2): Likewise.
38247 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38250 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
38252 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38253 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
38254 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
38255 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
38256 case where N arg is SIZE_MAX.
38257 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
38258 (cmpstrsi): Add pattern.
38260 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
38262 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
38263 __builtin_vec_revb builtins.
38264 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
38265 built-in functions to support generation of the ISA 3.0 XXBR<x>
38266 vector byte reverse instructions.
38267 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
38268 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
38269 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
38270 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
38271 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
38272 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
38273 (P9V_BUILTIN_VEC_REVB): Likewise.
38274 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
38275 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
38276 (p9_xxbrq_v16qi): Likewise.
38277 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
38278 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
38279 (p9_xxbrh_v8hi): Likewise.
38280 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
38281 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
38282 vec_revb built-in functions.
38284 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
38286 PR rtl-optimization/78952
38287 * config/i386/i386.md (any_extract): New code iterator.
38288 (*insvqi_2): Use any_extract for source operand.
38289 (*insvqi_3): Use any_shiftrt for source operand.
38291 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
38293 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
38295 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
38297 2017-01-18 Matthias Klose <doko@ubuntu.com>
38299 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
38301 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38303 * config/rs6000/altivec.h (vec_bperm): Change #define.
38304 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
38305 (altivec_vbpermq2): New define_insn.
38306 (altivec_vbpermd): Likewise.
38307 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
38308 function interface.
38309 (VBPERMD): Likewise.
38310 (VBPERM): New polymorphic function interface.
38311 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
38312 Add entries for P9V_BUILTIN_VEC_VBPERM.
38313 * doc/extend.texi: Add interfaces for vec_bperm.
38315 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38317 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
38318 first letter of error messages.
38319 (s390_resolve_overloaded_builtin): Likewise.
38320 * config/s390/s390.c (s390_expand_builtin): Likewise.
38321 (s390_invalid_arg_for_unprototyped_fn): Likewise.
38322 (s390_valid_target_attribute_inner_p): Likewise.
38323 * config/s390/s390.md ("tabort"): Likewise.
38325 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
38327 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
38328 (ISA_AVOID_DIV_HILO): New macro.
38329 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
38330 (ISA_HAS_DDIV): Likewise.
38332 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38334 * doc/invoke.texi (fabi-version): Correct number of occurrences.
38336 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38338 * doc/invoke.texi (fabi-version): Spelling fix.
38340 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38343 * doc/invoke.texi (fabi-version): Mention mangling fix for
38346 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38349 * doc/invoke.texi (fabi-version): Document discriminator mangling.
38351 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
38354 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
38355 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
38357 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
38359 (stack_protect_test): Ditto.
38360 * config/rs6000/rs6000.opt (mstack-protector-guard=,
38361 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
38363 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
38364 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
38365 -mstack-protector-guard-offset=.
38366 (RS/6000 and PowerPC Options): Ditto.
38368 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
38370 * config/i386/i386.h (MASK_CLASS_P): New define.
38371 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
38372 there are no registers from different register sets also when
38373 mask registers are used. Update function comment.
38374 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
38375 to (*k/*r) and (*k/*km) alternatives.
38377 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
38379 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
38380 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
38381 (EH_RETURN_HANDLER_RTX): New define.
38382 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
38383 Force frame pointer in EH return functions.
38384 (aarch64_expand_epilogue): Add barrier for eh_return.
38385 (aarch64_final_eh_return_addr): Remove.
38386 (aarch64_eh_return_handler_rtx): New function.
38387 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
38389 (aarch64_eh_return_handler_rtx): New prototype.
38391 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38393 * config/rs6000/altivec.h (vec_rlmi): New #define.
38394 (vec_vrlnm): Likewise.
38395 (vec_rlnm): Likewise.
38396 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
38397 (UNSPEC_VRLNM): Likewise.
38398 (VIlong): New mode iterator.
38399 (altivec_vrl<VI_char>mi): New define_insn.
38400 (altivec_vrl<VI_char>nm): Likewise.
38401 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
38403 (VRLDNM): Likewise.
38404 (RLNM): New polymorphic function entry.
38405 (VRLWMI): New monomorphic function entry.
38406 (VRLDMI): Likewise.
38407 (RLMI): New polymorphic function entry.
38408 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
38409 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
38410 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
38413 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38416 * dwarf2out.c (field_byte_offset): Restore the
38417 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
38418 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
38419 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
38422 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
38425 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
38427 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38430 * dwarf2out.c (add_data_member_location_attribute): For constant
38431 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
38432 instead of DW_AT_data_member_location, DW_AT_bit_offset and
38433 DW_AT_byte_size attributes.
38435 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
38437 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
38438 after forcing to constant memory when the code model is medium.
38440 2017-01-17 Julia Koval <julia.koval@intel.com>
38443 * config/i386/avx512fintrin.h
38444 (_mm512_i32gather_ps): Change __addr type to void const*.
38445 (_mm512_mask_i32gather_ps): Ditto.
38446 (_mm512_i32gather_pd): Ditto.
38447 (_mm512_mask_i32gather_pd): Ditto.
38448 (_mm512_i64gather_ps): Ditto.
38449 (_mm512_mask_i64gather_ps): Ditto.
38450 (_mm512_i64gather_pd): Ditto.
38451 (_mm512_mask_i64gather_pd): Ditto.
38452 (_mm512_i32gather_epi32): Ditto.
38453 (_mm512_mask_i32gather_epi32): Ditto.
38454 (_mm512_i32gather_epi64): Ditto.
38455 (_mm512_mask_i32gather_epi64): Ditto.
38456 (_mm512_i64gather_epi32): Ditto.
38457 (_mm512_mask_i64gather_epi32): Ditto.
38458 (_mm512_i64gather_epi64): Ditto.
38459 (_mm512_mask_i64gather_epi64): Ditto.
38460 (_mm512_i32scatter_ps): Change __addr type to void*.
38461 (_mm512_mask_i32scatter_ps): Ditto.
38462 (_mm512_i32scatter_pd): Ditto.
38463 (_mm512_mask_i32scatter_pd): Ditto.
38464 (_mm512_i64scatter_ps): Ditto.
38465 (_mm512_mask_i64scatter_ps): Ditto.
38466 (_mm512_i64scatter_pd): Ditto.
38467 (_mm512_mask_i64scatter_pd): Ditto.
38468 (_mm512_i32scatter_epi32): Ditto.
38469 (_mm512_mask_i32scatter_epi32): Ditto.
38470 (_mm512_i32scatter_epi64): Ditto.
38471 (_mm512_mask_i32scatter_epi64): Ditto.
38472 (_mm512_i64scatter_epi32): Ditto.
38473 (_mm512_mask_i64scatter_epi32): Ditto.
38474 (_mm512_i64scatter_epi64): Ditto.
38475 (_mm512_mask_i64scatter_epi64): Ditto.
38476 * config/i386/avx512pfintrin.h
38477 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
38478 (_mm512_mask_prefetch_i32gather_ps): Ditto.
38479 (_mm512_mask_prefetch_i64gather_pd): Ditto.
38480 (_mm512_mask_prefetch_i64gather_ps): Ditto.
38481 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
38482 (_mm512_prefetch_i32scatter_ps): Ditto.
38483 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
38484 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
38485 (_mm512_prefetch_i64scatter_pd): Ditto.
38486 (_mm512_prefetch_i64scatter_ps): Ditto.
38487 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
38488 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
38489 * config/i386/avx512vlintrin.h
38490 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
38491 (_mm_mmask_i32gather_ps): Ditto.
38492 (_mm256_mmask_i32gather_pd): Ditto.
38493 (_mm_mmask_i32gather_pd): Ditto.
38494 (_mm256_mmask_i64gather_ps): Ditto.
38495 (_mm_mmask_i64gather_ps): Ditto.
38496 (_mm256_mmask_i64gather_pd): Ditto.
38497 (_mm_mmask_i64gather_pd): Ditto.
38498 (_mm256_mmask_i32gather_epi32): Ditto.
38499 (_mm_mmask_i32gather_epi32): Ditto.
38500 (_mm256_mmask_i32gather_epi64): Ditto.
38501 (_mm_mmask_i32gather_epi64): Ditto.
38502 (_mm256_mmask_i64gather_epi32): Ditto.
38503 (_mm_mmask_i64gather_epi32): Ditto.
38504 (_mm256_mmask_i64gather_epi64): Ditto.
38505 (_mm_mmask_i64gather_epi64): Ditto.
38506 (_mm256_i32scatter_ps): Change __addr type to void*.
38507 (_mm256_mask_i32scatter_ps): Ditto.
38508 (_mm_i32scatter_ps): Ditto.
38509 (_mm_mask_i32scatter_ps): Ditto.
38510 (_mm256_i32scatter_pd): Ditto.
38511 (_mm256_mask_i32scatter_pd): Ditto.
38512 (_mm_i32scatter_pd): Ditto.
38513 (_mm_mask_i32scatter_pd): Ditto.
38514 (_mm256_i64scatter_ps): Ditto.
38515 (_mm256_mask_i64scatter_ps): Ditto.
38516 (_mm_i64scatter_ps): Ditto.
38517 (_mm_mask_i64scatter_ps): Ditto.
38518 (_mm256_i64scatter_pd): Ditto.
38519 (_mm256_mask_i64scatter_pd): Ditto.
38520 (_mm_i64scatter_pd): Ditto.
38521 (_mm_mask_i64scatter_pd): Ditto.
38522 (_mm256_i32scatter_epi32): Ditto.
38523 (_mm256_mask_i32scatter_epi32): Ditto.
38524 (_mm_i32scatter_epi32): Ditto.
38525 (_mm_mask_i32scatter_epi32): Ditto.
38526 (_mm256_i32scatter_epi64): Ditto.
38527 (_mm256_mask_i32scatter_epi64): Ditto.
38528 (_mm_i32scatter_epi64): Ditto.
38529 (_mm_mask_i32scatter_epi64): Ditto.
38530 (_mm256_i64scatter_epi32): Ditto.
38531 (_mm256_mask_i64scatter_epi32): Ditto.
38532 (_mm_i64scatter_epi32): Ditto.
38533 (_mm_mask_i64scatter_epi32): Ditto.
38534 (_mm256_i64scatter_epi64): Ditto.
38535 (_mm256_mask_i64scatter_epi64): Ditto.
38536 (_mm_i64scatter_epi64): Ditto.
38537 (_mm_mask_i64scatter_epi64): Ditto.
38538 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
38539 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
38540 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
38541 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
38542 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
38543 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
38544 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
38545 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
38546 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
38547 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
38548 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
38549 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
38550 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
38551 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
38552 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
38553 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
38554 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
38555 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
38556 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
38557 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
38558 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
38559 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
38560 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
38561 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
38562 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
38563 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
38564 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
38565 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
38566 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
38567 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
38568 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
38569 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
38570 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
38571 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
38572 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
38573 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
38574 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
38575 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
38576 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
38577 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
38578 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
38579 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
38580 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
38581 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
38582 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
38583 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
38584 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
38585 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
38586 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
38587 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
38588 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
38589 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
38590 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
38591 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
38592 definitions accordingly.
38594 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
38595 Kuan-Lin Chen <kuanlinchentw@gmail.com>
38598 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
38601 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
38604 * ira-conflicts.c (ira_build_conflicts): Update total conflict
38605 hard regs for inner regno.
38607 2017-01-17 Martin Liska <mliska@suse.cz>
38610 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
38611 assumption and add comment.
38613 2017-01-17 Nathan Sidwell <nathan@acm.org>
38615 * ipa-visibility.c (localize_node): New function, broken out of ...
38616 (function_and_variable_visibility): ... here. Call it.
38618 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
38620 PR middle-end/77445
38621 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
38622 correctly set frequency of oudgoing edge.
38623 (duplicate_thread_path): Fix profile updating.
38625 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38628 * configure.ac: Add GCC_BASE_VER.
38629 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
38630 version from BASE-VER file.
38631 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
38632 (gcc.o): Depend on $(BASEVER).
38633 * common.opt (dumpfullversion): New option.
38634 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
38635 * doc/invoke.texi: Document -dumpfullversion.
38636 * doc/install.texi: Document --with-gcc-major-version-only.
38637 * configure: Regenerated.
38639 2017-01-17 Richard Biener <rguenther@suse.de>
38641 PR tree-optimization/71433
38642 * tree-vrp.c (register_new_assert_for): Merge same asserts
38643 on all incoming edges.
38644 (process_assert_insertions_for): Handle insertions at the
38647 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
38649 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
38650 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
38652 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
38655 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
38658 2017-01-17 Alan Modra <amodra@gmail.com>
38661 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
38662 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
38663 symbolic stack limit when pic.
38665 2017-01-16 Martin Sebor <msebor@redhat.com>
38667 PR tree-optimization/78608
38668 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
38670 2017-01-16 Jeff Law <law@redhat.com>
38673 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
38674 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
38675 for several include directories that may be relative to sysroot.
38676 * config/i386/x-mingw32 (gplus_includedir): Define.
38677 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
38678 (native_system_includedir): Likewise.
38679 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
38680 override if TARGET_SYSTEM_ROOT is defined.
38681 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
38683 PR tree-optimization/79090
38684 PR tree-optimization/33562
38685 PR tree-optimization/61912
38686 PR tree-optimization/77485
38687 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
38688 and computed trims into the dump file.
38690 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
38692 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
38694 2017-01-16 Jakub Jelinek <jakub@redhat.com>
38697 * gimplify.c (gimplify_init_constructor): If want_value and
38698 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
38702 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
38703 sequence. Formatting fixes.
38704 (doloop_optimize): Formatting fixes.
38707 * gcc.c (debug_level_greater_than_spec_func): New function.
38708 (static_spec_functions): Add debug-level-gt spec function.
38709 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
38711 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
38712 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
38713 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
38714 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
38715 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
38716 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
38718 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
38720 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
38721 QImode fixups to general and mask registers only.
38723 2017-01-16 Carl Love <cel@us.ibm.com>
38725 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
38726 for built-in functions
38727 vector signed char vec_nabs (vector signed char)
38728 vector signed short vec_nabs (vector signed short)
38729 vector signed int vec_nabs (vector signed int)
38730 vector signed long long vec_nabs (vector signed long long)
38731 vector float vec_nabs (vector float)
38732 vector double vec_nabs (vector double)
38733 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
38735 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
38736 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
38737 * doc/extend.texi: Update the documentation file for the new built-in
38740 2017-01-16 Martin Sebor <msebor@redhat.com>
38742 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
38745 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38747 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
38748 UNSPEC_VSX__XXSPLTD to require special splat handling.
38750 2017-01-16 David Malcolm <dmalcolm@redhat.com>
38753 * system.h: Poison strndup.
38755 2017-01-16 Alan Modra <amodra@gmail.com>
38758 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
38761 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
38763 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
38765 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
38767 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
38768 call recog here. Assert that INSN_CODE (insn) is non-negative.
38770 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
38773 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
38775 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
38776 in the currently scheduled RTL fragment.
38778 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
38780 PR rtl-optimization/78751
38781 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
38784 2017-01-14 Jeff Law <law@redhat.com>
38786 PR tree-optimization/79090
38787 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
38788 variable length stores.
38789 (compute_trims): Delete dead assignment to *trim_tail.
38790 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
38793 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
38795 PR rtl-optimization/78626
38796 PR rtl-optimization/78727
38797 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
38798 of a block, and split such blocks after everything else is finished.
38800 2017-01-14 Alan Modra <amodra@gmail.com>
38803 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
38804 target legitimate_combined_insn.
38805 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
38806 (rs6000_legitimate_combined_insn): New function.
38807 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
38809 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
38810 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
38811 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
38813 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
38815 * doc/frontends.texi (G++ and GCC): Remove references to Java.
38817 2017-01-13 Jeff Law <law@redhat.com>
38819 PR tree-optimization/33562
38820 PR tree-optimization/61912
38821 PR tree-optimization/77485
38822 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
38824 (delete_dead_assignment): Likewise.
38825 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
38826 statement to delete_dead_call and delete_dead_assignment.
38828 2017-01-13 David Malcolm <dmalcolm@redhat.com>
38831 * substring-locations.c (format_warning_va): Strengthen case 1 so
38832 that both endpoints of the substring must be within the format
38833 range for just the substring to be printed.
38835 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
38837 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
38838 * config/i386/i386.c (ix86_target_string): Add missing options
38839 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
38840 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
38841 flags_other and ix86_target_other to flags2_other. Display unknown
38843 (ix86_valid_target_attribute_inner_p): Add missing options and
38844 reorder options by implied ISAs, as in ix86_target_string.
38846 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
38848 * hash-table.h (hash_table::too_empty_p): New function.
38849 (hash_table::expand): Use it.
38850 (hash_table::traverse): Likewise.
38851 (hash_table::empty_slot): Use sizeof (value_type) instead of
38852 sizeof (PTR) to convert bytes to elements. Shrink the table
38853 if the current size is excessive for the current number of
38856 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
38858 * ira-costs.c (record_reg_classes): Break from the inner loop
38859 early once alt_fail is known to be true. Update outer loop
38860 handling accordingly.
38862 2017-01-13 Jeff Law <law@redhat.com>
38864 * tree-ssa-dse.c (decrement_count): New function.
38865 (increment_start_addr, maybe_trim_memstar_call): Likewise.
38866 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
38867 when we know the partially dead statement is a mem* function.
38869 PR tree-optimization/61912
38870 PR tree-optimization/77485
38871 * tree-ssa-dse.c: Include expr.h.
38872 (maybe_trim_constructor_store): New function.
38873 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
38875 PR tree-optimization/33562
38876 PR tree-optimization/61912
38877 PR tree-optimization/77485
38878 * doc/invoke.texi: Document new dse-max-object-size param.
38879 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
38880 * tree-ssa-dse.c: Include params.h.
38881 (dse_store_status): New enum.
38882 (initialize_ao_ref_for_dse): New, partially extracted from
38884 (valid_ao_ref_for_dse, normalize_ref): New.
38885 (setup_live_bytes_from_ref, compute_trims): Likewise.
38886 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
38887 (maybe_trim_partially_dead_store): Likewise.
38888 (maybe_trim_complex_store): Likewise.
38889 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
38890 Track what bytes live from the original store. Return tri-state
38891 for dead, partially dead or live.
38892 (dse_dom_walker): Add constructor, destructor and new private members.
38893 (delete_dead_call, delete_dead_assignment): New extracted from
38895 (dse_optimize_stmt): Make a member of dse_dom_walker.
38896 Use initialize_ao_ref_for_dse.
38898 PR tree-optimization/33562
38899 PR tree-optimization/61912
38900 PR tree-optimization/77485
38901 * sbitmap.h (bitmap_count_bits): Prototype.
38902 (bitmap_clear_range, bitmap_set_range): Likewise.
38903 * sbitmap.c (bitmap_clear_range): New function.
38904 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
38906 2017-01-13 Martin Liska <mliska@suse.cz>
38909 * function.c (set_cfun): Add new argument force.
38910 * function.h (set_cfun): Likewise.
38911 * ipa-inline-transform.c (inline_call): Use the function when
38912 strict alising from is dropped for function we inline to.
38914 2017-01-13 Richard Biener <rguenther@suse.de>
38916 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
38917 for dumping GIMPLE INTEGER_CSTs.
38919 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
38921 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
38922 to 201112L since C++17.
38924 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
38927 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
38928 if -fsanitize=kernel-address is present.
38930 2017-01-13 Richard Biener <rguenther@suse.de>
38932 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
38933 as _Literal ( type ) number in case usual suffixes do not
38934 preserve all information.
38936 2017-01-13 Richard Biener <rguenther@suse.de>
38938 PR tree-optimization/77283
38939 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
38940 and ssa-iterators.h.
38941 (is_feasible_trace): Implement a cost model based on joiner
38944 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
38947 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
38948 char or short to __float128/_Float128 directly.
38950 2017-01-12 Martin Sebor <msebor@redhat.com>
38952 to -Wformat-overflow.
38953 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
38954 (min_bytes_remaining): Same.
38955 (get_string_length): Same.
38956 (format_string): Same.
38957 (format_directive): Same.
38959 (pass_sprintf_length::handle_gimple_call): Same.
38961 2017-01-12 Jakub Jelinek <jakub@redhat.com>
38963 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
38964 info.nowrite calls with no lhs that can't throw. Return bool
38965 whether gsi_remove has been called or not.
38966 (pass_sprintf_length::handle_gimple_call): Return bool whether
38967 try_substitute_return_value called gsi_remove. Formatting fix.
38968 (pass_sprintf_length::execute): Don't use gsi_remove if
38969 handle_gimple_call returned true.
38972 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
38973 be removed due to side-effects, don't remove following barrier nor
38974 turn the successor edge into fallthru edge.
38976 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38979 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
38980 element-reversing loads and stores as not swappable.
38982 2017-01-12 Nathan Sidwell <nathan@acm.org>
38983 Nicolai Stange <nicstange@gmail.com>
38985 * combine.c (try_combine): Don't ignore result of overlap checking
38986 loop. Combine overlap & asm check into single loop.
38988 2017-01-12 Richard Biener <rguenther@suse.de>
38990 * tree-pretty-print.c (dump_generic_node): Provide -gimple
38991 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
38993 2017-01-12 Richard Biener <rguenther@suse.de>
38995 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
38996 and TS_TARGET_OPTION directly derive from TS_BASE.
38997 * tree-core.h (tree_optimization_option): Derive from tree_base.
38998 (tree_target_option): Likewise.
39000 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
39002 * config/i386/i386.c (memory_address_length): Increase len
39003 only when rip_relative_addr_p returns false.
39005 2017-01-11 Julia Koval <julia.koval@intel.com>
39007 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
39008 (OPTION_MASK_ISA_SGX_SET): New.
39009 (ix86_handle_option): Handle OPT_msgx.
39010 * config.gcc: Added sgxintrin.h.
39011 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
39012 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
39013 * config/i386/i386.c (ix86_target_string): Add -msgx.
39015 (ix86_option_override_internal): Handle new options.
39016 (ix86_valid_target_attribute_inner_p): Add sgx.
39017 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
39018 * config/i386/i386.opt: Add msgx.
39019 * config/i386/sgxintrin.h: New file.
39020 * config/i386/x86intrin.h: Add sgxintrin.h.
39022 2017-01-11 Jakub Jelinek <jakub@redhat.com>
39025 * fold-const.c (maybe_nonzero_address): Return 1 for function
39027 (tree_single_nonzero_warnv_p): Don't handle function local objects
39031 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
39034 2017-01-11 David Malcolm <dmalcolm@redhat.com>
39037 * opts.c: Include "spellcheck.h"
39038 (struct string_fragment): New struct.
39039 (struct edit_distance_traits<const string_fragment &>): New
39041 (get_closest_sanitizer_option): New function.
39042 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
39044 2017-01-11 Jakub Jelinek <jakub@redhat.com>
39046 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
39048 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
39049 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
39050 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
39051 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
39052 for initial die_offset if dwarf_split_debug_info.
39053 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
39054 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
39056 (output_skeleton_debug_sections): Formatting fix. Use
39057 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
39058 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
39060 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
39062 * config/arm/cortex-a53.md: Add bypasses for
39063 cortex_a53_r2f_cvt.
39064 (cortex_a53_r2f): Only use for transfers.
39065 (cortex_a53_f2r): Likewise.
39066 (cortex_a53_r2f_cvt): Add reservation for conversions.
39067 (cortex_a53_f2r_cvt): Likewise.
39069 2017-01-11 Tamar Christina <tamar.christina@arm.com>
39071 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
39072 to all inlined functions, change static to extern.
39074 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
39077 * config/arm/arm.c (legitimize_pic_address): Handle reference to
39079 (arm_assemble_integer): Likewise.
39081 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
39083 * config.gcc: Use new awk script to check CPU, FPU and architecture
39084 parameters for --with-... options.
39085 * config/arm/parsecpu.awk: New file
39086 * config/arm/arm-cpus.in: New file.
39087 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
39089 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
39091 * config/arm/t-arm: Update dependency rules.
39092 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
39093 of processing .def files.
39094 * config/arm/genopt.sh: Deleted.
39095 * config/arm/gentune.sh: Deleted.
39096 * config/arm/arm-cores.def: Deleted.
39097 * config/arm/arm-arches.def: Deleted.
39098 * config/arm/arm-fpus.def: Deleted.
39099 * config/arm/arm-tune.md: Regenerated.
39100 * config/arm/arm-tables.opt: Regenerated.
39101 * config/arm/arm-cpu.h: New generated file.
39102 * config/arm/arm-cpu-data.h: New generated file.
39103 * config/arm/arm-cpu-cdata.h: New generated file.
39105 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
39108 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
39110 (input_varpool_node): Unpack dynamically_initialized bit.
39112 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
39114 PR rtl-optimization/79032
39115 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
39116 the alignment of the adjusted memory reference against that of MODE,
39117 instead of the alignment of the original memory reference.
39119 2017-01-11 Martin Jambor <mjambor@suse.cz>
39121 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
39123 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
39124 decorated functions.
39126 2017-01-11 Richard Biener <rguenther@suse.de>
39128 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
39129 set range/nonnull info for PHI results. Do not set it on
39130 stmts marked for removal.
39132 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
39134 * expr.c (store_field): In the bitfield case, fetch the return value
39135 from the registers before applying a single big-endian adjustment.
39136 Always do a final load for a BLKmode value not larger than a word.
39138 2017-01-10 David Malcolm <dmalcolm@redhat.com>
39141 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
39142 that we correctly handle column numbers greater than
39143 LINE_MAP_MAX_COLUMN_NUMBER.
39145 2017-01-10 Martin Sebor <msebor@redhat.com>
39147 PR middle-end/78245
39148 * gimple-ssa-sprintf.c (get_destination_size): Call
39149 {init,fini}object_sizes.
39150 * tree-object-size.c (addr_object_size): Adjust.
39151 (pass_through_call): Adjust.
39152 (pass_object_sizes::execute): Adjust.
39153 * tree-object-size.h (fini_object_sizes): Declare.
39155 2017-01-10 Martin Sebor <msebor@redhat.com>
39157 PR tree-optimization/78775
39158 * builtins.c (get_size_range): Move...
39159 * calls.c: ...to here.
39160 (alloc_max_size): Accept zero argument.
39161 (operand_signed_p): Remove.
39162 (maybe_warn_alloc_args_overflow): Call get_size_range.
39163 * calls.h (get_size_range): Declare.
39165 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
39167 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
39168 from TI's devices.csv file as of September 2016.
39169 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
39171 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
39173 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
39174 * doc/invoke.texi: Likewise.
39175 * doc/md.texi: Likewise.
39176 * doc/objc.texi: Likewise.
39178 2017-01-10 Joshua Conner <joshconner@google.com>
39180 * config/arm/fuchsia-elf.h: New file.
39181 * config/fuchsia.h: New file.
39182 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
39183 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
39185 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
39187 2016-01-10 Richard Biener <rguenther@suse.de>
39189 PR tree-optimization/79034
39190 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
39191 Propagate out degenerate PHIs in the joiner.
39193 2017-01-10 Martin Liska <mliska@suse.cz>
39195 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
39196 (sort_congruence_classes_by_decl_uid): Likewise.
39197 (sort_congruence_class_groups_by_decl_uid): Likewise.
39198 (sem_item_optimizer::merge_classes): Sort class, groups in these
39199 classes and members in the groups by DECL_UID of declarations.
39200 This would make merge operations stable.
39202 2017-01-10 Martin Liska <mliska@suse.cz>
39204 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
39205 usage of m_classes_vec.
39206 (sem_item_optimizer::~sem_item_optimizer): Likewise.
39207 (sem_item_optimizer::get_group_by_hash): Likewise.
39208 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
39209 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
39210 (sem_item_optimizer::verify_classes): Likewise.
39211 (sem_item_optimizer::process_cong_reduction): Likewise.
39212 (sem_item_optimizer::dump_cong_classes): Likewise.
39213 (sem_item_optimizer::merge_classes): Likewise.
39214 * ipa-icf.h (congruence_class_hash): Rename from
39215 congruence_class_group_hash. Remove declaration of m_classes_vec.
39217 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
39219 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
39220 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
39221 * config.gcc: Add avx512vpopcntdqintrin.h.
39222 * config/i386/avx512vpopcntdqintrin.h: New.
39223 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
39224 * config/i386/i386-builtin-types.def: Add new types.
39225 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
39226 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
39227 __builtin_ia32_vpopcountq_v8di_mask): New.
39228 * config/i386/i386-c.c (ix86_target_macros_internal): Define
39229 __AVX512VPOPCNTDQ__.
39230 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
39231 (PTA_AVX512VPOPCNTDQ): Define.
39232 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
39233 TARGET_AVX512VPOPCNTDQ_P): Define.
39234 * config/i386/i386.opt: Add mavx512vpopcntdq.
39235 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
39236 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
39238 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39240 PR middle-end/77484
39241 * predict.def (PRED_CALL): Set to 67.
39243 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
39245 * expr.c (store_field): In the bitfield case, if the value comes from
39246 a function call and is of an aggregate type returned in registers, do
39247 not modify the field mode; extract the value in all cases if the mode
39248 is BLKmode and the size is not larger than a word.
39250 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
39253 * config/i386/cpuid.h: Fix undefined behavior.
39255 2017-01-04 Jeff Law <law@redhat.com>
39257 PR tree-optimization/79007
39258 PR tree-optimization/67955
39259 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
39260 conservative for pt.null when flag_non_call_exceptions is on.
39262 2017-01-09 Jakub Jelinek <jakub@redhat.com>
39264 PR translation/79019
39265 PR translation/79020
39266 * params.def (PARAM_INLINE_MIN_SPEEDUP,
39267 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
39268 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
39270 * config/avr/avr.opt (maccumulate-args): Likewise.
39271 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
39272 * common.opt (freport-bug): Likewise.
39273 * cif-code.def (CIF_FINAL_ERROR): Likewise.
39274 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
39275 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
39276 translatable string.
39277 * config/i386/i386.c (function_value_32): Likewise.
39278 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
39279 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
39281 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
39282 * common/config/msp430/msp430-common.c (msp430_handle_option):
39284 * symtab.c (symtab_node::verify_base): Likewise.
39285 * opts.c (set_debug_level): Likewise.
39286 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
39287 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
39288 missing whitespace to translatable strings.
39289 * config/avr/avr.md (bswapsi2): Fix typo in comment.
39290 * config/sh/superh.h: Likewise.
39291 * config/i386/xopintrin.h: Likewise.
39292 * config/i386/znver1.md: Likewise.
39293 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
39294 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
39295 * double-int.h (struct double_int): Likewise.
39296 * double-int.c (div_and_round_double): Likewise.
39297 * wide-int.cc: Likewise.
39298 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
39299 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
39300 * cfgcleanup.c (crossjumps_occured): Renamed to ...
39301 (crossjumps_occurred): ... this.
39302 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
39305 PR tree-optimization/78899
39306 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
39307 returning bool return struct loop *, NULL for failure and the new
39309 (versionable_outer_loop_p): Don't version outer loop if it has
39310 dont_vectorized bit set.
39311 (tree_if_conversion): When versioning outer loop, ensure
39312 tree_if_conversion is performed also on the inner loop of the
39313 non-vectorizable outer loop copy.
39314 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
39315 LOOP_VECTORIZED in inner loop of the scalar outer loop and
39316 prevent vectorization of it.
39317 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
39318 the outer loop vectorization of the non-scalar version is attempted
39319 before vectorization of the inner loop in scalar version. If
39320 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
39321 vectorization of its inner loop.
39322 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
39323 has 2 inner loops, rename also on edges from bb whose single pred
39324 is outer_loop->header. Fix typo in function comment.
39326 2017-01-09 Martin Sebor <msebor@redhat.com>
39329 * asan.c (asan_emit_stack_protection): Increase local buffer size
39330 to avoid snprintf truncation warning.
39332 2017-01-09 Andrew Pinski <apinski@cavium.com>
39334 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
39335 to reference thunderx2t99 for the tuning structure
39336 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
39338 (thunderx2t99_extra_costs): This.
39339 * config/aarch64/aarch64-tune.md: Regenerate.
39340 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
39341 (vulcan_addrcost_table): This.
39342 (vulcan_regmove_cost): Rename to ...
39343 (thunderx2t99_regmove_cost): This.
39344 (vulcan_vector_cost): Rename to ...
39345 (thunderx2t99_vector_cost): this.
39346 (vulcan_branch_cost): Rename to ...
39347 (thunderx2t99_branch_cost): This.
39348 (vulcan_tunings): Rename to ...
39349 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
39350 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
39352 2017-01-09 Martin Jambor <mjambor@suse.cz>
39356 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
39357 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
39358 (propagate_vr_accross_jump_function): Use the above function for all
39359 value range computations for pass-through jump functions and type
39360 converasion from explicit value range values.
39361 (ipcp_propagate_stage): Do not attempt to deduce types of formal
39362 parameters from TYPE_ARG_TYPES.
39363 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
39364 (ipa_write_node_info): Stream type of the actual argument.
39365 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
39367 2017-01-09 Martin Liska <mliska@suse.cz>
39370 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
39371 (lookup_compiler): Do not show error message with have_E.
39373 2017-01-09 Jakub Jelinek <jakub@redhat.com>
39375 PR tree-optimization/78938
39376 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
39377 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
39378 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
39379 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
39382 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39384 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
39387 2017-01-09 Richard Biener <rguenther@suse.de>
39389 PR tree-optimization/78997
39390 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
39391 name condition properly.
39393 2017-01-09 Richard Biener <rguenther@suse.de>
39396 * dwarf2out.c (is_cxx): New overload with context.
39397 (is_naming_typedef_decl): Use it.
39399 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39401 * invoke.texi (Option Summary): Correct spacing in option lists
39402 and add line breaks to fix over-long lines.
39404 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39406 PR middle-end/17660
39408 * extend.texi (Common Variable Attributes): Add xref to GCC
39409 Internals manual to explain mode attribute keywords.
39411 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39414 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
39415 and Preprocessor Options.
39416 (Options for Linking): Document -pthread here....
39417 (RS/6000 and PowerPC Options): ...not here.
39418 (Solaris 2 Options): ...or here.
39419 * doc/cppopts.texi: Document -pthread.
39421 2017-01-08 Martin Sebor <msebor@redhat.com>
39423 PR middle-end/77708
39424 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
39425 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
39426 New member functions.
39427 (format_directive): Used them.
39429 (pass_sprintf_length::handle_gimple_call): Same.
39430 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
39431 to avoid truncation for any argument.
39432 (extract_affine_mul): Same.
39433 * tree.c (get_file_function_name): Same.
39435 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39437 PR middle-end/77484
39438 * predict.def (PRED_INDIR_CALL): Set to 86.
39440 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39442 PR preprocessor/54124
39443 * doc/cppopts.texi: Reformat -d subtable to list the full name
39444 of the options. Add cross-reference to the docs for the general
39445 compiler -d options.
39446 * doc/invoke.texi (Developer Options): Add cross-reference to the
39447 preprocessor-specific -d option documentation.
39449 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39451 PR preprocessor/13498
39452 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
39453 redudant material, and reflect new command-line options.
39454 (System Headers): Likewise.
39456 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39458 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
39459 -isystem, and -idirafter. Copy-edit.
39460 * doc/cppopts.texi: Copy-edit. Remove contradiction about
39461 default for -ftrack-macro-expansion. Delete obsolete and
39462 badly-formatted implementation details about -fdebug-cpp output.
39463 * doc/cppwarnopts.texi: Copy-edit.
39465 2017-01-07 David Malcolm <dmalcolm@redhat.com>
39468 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
39469 that the transition from a max line width >= 1<<10 to narrower
39470 lines works correctly.
39472 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
39474 * doc/options.texi (PerFunction): New.
39475 * opt-functions.awk (switch_flags): Map both Optimization and
39476 PerFunction to CL_OPTIMIZATION.
39477 * opth-gen.awk: Test for PerFunction flag along with
39479 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
39480 it only when the latter is present. Skip those that don't in
39481 the hash function generator.
39482 * common.opt (fvar-tracking): Mark as PerFunction instead of
39484 (fvar-tracking-assignments): Likewise.
39485 (fvar-tracking-assignments-toggle): Likewise.
39486 (fvar-tracking-uninit): Likewise.
39488 2017-01-07 Jakub Jelinek <jakub@redhat.com>
39490 PR translation/79018
39491 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
39494 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
39497 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
39498 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
39499 TARGET_LONG_JUMP_TABLE_OFFSETS.
39500 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
39501 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
39502 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
39503 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
39504 * config/m68k/m68k.md (tablejump expander): Likewise.
39505 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
39506 TARGET_LONG_JUMP_TABLE_OFFSETS.
39507 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
39508 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
39510 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
39511 David Holsgrove <david.holsgrove@xilinx.com>
39513 * common/config/microblaze/microblaze-common.c
39514 (TARGET_EXCEPT_UNWIND_INFO): Remove.
39515 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
39517 * config/microblaze/microblaze.c (microblaze_must_save_register)
39518 (microblaze_expand_epilogue, microblaze_return_addr): Handle
39520 (microblaze_eh_return): New function.
39521 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
39522 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
39523 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
39524 * config/microblaze/microblaze.md (eh_return): New pattern.
39526 2017-01-06 Jakub Jelinek <jakub@redhat.com>
39528 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
39529 GCC_DIAGNOSTIC_STRINGIFY): Define.
39531 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
39533 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39535 * config/arm/arm.md (<mcrr>): New.
39537 * config/arm/arm.c (arm_arch5te): New.
39538 (arm_option_override): Set arm_arch5te.
39539 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
39541 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
39542 (arm_mcrr_qualifiers): ... this. New.
39543 (MRRC_QUALIFIERS): Define to...
39544 (arm_mrrc_qualifiers): ... this. New.
39545 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
39547 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
39548 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
39549 (MRRCI, mrrc, MRRC): New.
39550 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
39551 VUNSPEC_MRRC2): New.
39553 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39555 * config/arm/arm.md (<mcr>): New.
39557 * config/arm/arm.c (arm_coproc_builtin_available): Add
39558 support for mcr, mrc, mcr2 and mrc2.
39559 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
39560 (arm_mcr_qualifiers): ... this. New.
39561 (MRC_QUALIFIERS): Define to ...
39562 (arm_mrc_qualifiers): ... this. New.
39563 (MCR_QUALIFIERS): Define to ...
39564 (arm_mcr_qualifiers): ... this. New.
39565 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
39567 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
39568 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
39569 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
39570 VUNSPEC_MRC2): New.
39572 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39574 * config/arm/arm.md (*ldc): New.
39578 * config/arm/arm.c (arm_coproc_builtin_available): Add
39579 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
39580 (arm_coproc_ldc_stc_legitimate_address): New.
39581 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
39582 'qualifier_const_pointer'.
39583 (LDC_QUALIFIERS): Define to...
39584 (arm_ldc_qualifiers): ... this. New.
39585 (STC_QUALIFIERS): Define to...
39586 (arm_stc_qualifiers): ... this. New.
39587 * config/arm/arm-protos.h
39588 (arm_coproc_ldc_stc_legitimate_address): New.
39589 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
39590 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
39591 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
39592 stc2, stcl, stc2l): New.
39593 * config/arm/constraints.md (Uz): New.
39594 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
39595 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
39596 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
39597 VUNSPEC_STC2L): New.
39599 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39601 * config/arm/arm.md (<cdp>): New.
39602 * config/arm/arm.c (neon_const_bounds): Rename this ...
39603 (arm_const_bounds): ... this.
39604 (arm_coproc_builtin_available): New.
39605 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
39606 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
39607 (CDP_QUALIFIERS): Define to...
39608 (arm_cdp_qualifiers): ... this. New.
39610 (arm_expand_builtin_args): Add case for 6 arguments.
39611 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
39612 (arm_const_bounds): ... this.
39613 (arm_coproc_builtin_available): New.
39614 * config/arm/arm_acle.h (__arm_cdp): New.
39616 * config/arm/arm_acle_builtins.def (cdp): New.
39618 * config/arm/iterators.md (CDPI,CDP,cdp): New.
39619 * config/arm/neon.md: Rename all 'neon_const_bounds' to
39620 'arm_const_bounds'.
39621 * config/arm/types.md (coproc): New.
39622 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
39623 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
39624 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
39625 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
39627 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39629 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
39630 (UBINOP_QUALIFIERS): New.
39632 (acle_builtin_data): New. Change comment.
39633 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
39634 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
39635 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
39636 arm_acle_builtins.def.
39637 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
39638 (arm_init_acle_builtins): New.
39639 (CRC32_BUILTIN): Remove.
39640 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
39641 crc32cb, crc32ch and crc32cw.
39642 (arm_init_crc32_builtins): Remove.
39643 (arm_init_builtins): Use arm_init_acle_builtins rather
39644 than arm_init_crc32_builtins.
39645 (arm_expand_acle_builtin): New.
39646 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
39647 * config/arm/arm_acle_builtins.def: New.
39649 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39651 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
39652 (arm_builtin_datum): ... this.
39653 (arm_init_neon_builtin): Rename to ...
39654 (arm_init_builtin): ... this. Add a new parameters PREFIX
39655 and USE_SIG_IN_NAME.
39656 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
39657 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
39658 'arm_builtin_datum'.
39659 (arm_init_vfp_builtins): Likewise.
39660 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
39661 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
39662 (arm_expand_neon_args): Rename to ...
39663 (arm_expand_builtin_args): ... this. Rename builtin_arg
39664 enum values and differentiate between ARG_BUILTIN_MEMORY
39665 and ARG_BUILTIN_NEON_MEMORY.
39666 (arm_expand_neon_builtin_1): Rename to ...
39667 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
39668 values, arm_expand_builtin_args and add bool parameter NEON.
39669 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
39670 (arm_expand_vfp_builtin): Likewise.
39671 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
39673 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39675 PR middle-end/77484
39676 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
39677 * predict.c (tree_estimate_probability_bb): Reverse direction of
39678 polymorphic call predictor.
39680 2017-01-06 David Malcolm <dmalcolm@redhat.com>
39682 * passes.c (execute_one_pass): Split out pass-skipping logic into...
39683 (determine_pass_name_match): ...this new function and...
39684 (should_skip_pass_p): ...this new function.
39686 2017-01-06 Nathan Sidwell <nathan@acm.org>
39688 * ipa-visibility.c (function_and_variable_visibility): Reformat
39689 comments and long lines. Remove extrneous if.
39690 * symtab.c (symtab_node::make_decl_local): Fix code format.
39691 (symtab_node::set_section_for_node): Fix comment typo.
39693 2017-01-06 Martin Liska <mliska@suse.cz>
39696 * lra-constraints.c: Rename invariant to lra_invariant.
39697 * predict.c (set_even_probabilities): Initialize e to NULL.
39699 2017-01-05 Martin Sebor <msebor@redhat.com>
39701 PR tree-optimization/78910
39702 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
39703 (format_integer): Correct off-by-one error in the handling
39704 of precision with negative numbers in signed conversions..
39706 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
39708 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
39710 2017-01-05 Jakub Jelinek <jakub@redhat.com>
39712 PR tree-optimization/71016
39713 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
39714 factor_out_conditional_conversion. Formatting fix.
39715 (factor_out_conditional_conversion): Add cond_stmt argument.
39716 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
39717 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
39720 2017-01-05 David Malcolm <dmalcolm@redhat.com>
39722 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
39723 read-rtl-function.o, and selftest-rtl.o.
39724 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
39725 (selftest::aarch64_test_loading_full_dump): New function.
39726 (selftest::aarch64_run_selftests): New function.
39727 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
39728 selftest::aarch64_run_selftests.
39729 * config/i386/i386.c
39730 (selftest::ix86_test_loading_dump_fragment_1): New function.
39731 (selftest::ix86_test_loading_call_insn): New function.
39732 (selftest::ix86_test_loading_full_dump): New function.
39733 (selftest::ix86_test_loading_unspec): New function.
39734 (selftest::ix86_run_selftests): Call the new functions.
39735 * emit-rtl.c (maybe_set_max_label_num): New function.
39736 * emit-rtl.h (maybe_set_max_label_num): New decl.
39737 * function.c (instantiate_decls): Guard call to
39738 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
39739 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
39741 * gensupport.c (gen_reader::gen_reader): Pass "false"
39742 for new "compact" param of rtx_reader.
39743 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
39744 rather than an empty string for NULL strings.
39745 * read-md.c: Potentially include config.h rather than bconfig.h.
39746 Wrap include of errors.h with #ifdef GENERATOR_FILE.
39747 (have_error): New global, copied from errors.c.
39748 (md_reader::read_name): Rename to...
39749 (md_reader::read_name_1): ...this, adding "out_loc" param,
39750 and converting "missing name or number" to returning false, rather
39752 (md_reader::read_name): Reimplement in terms of read_name_1.
39753 (md_reader::read_name_or_nil): New function.
39754 (md_reader::read_string): Handle "(nil)" by returning NULL.
39755 (md_reader::md_reader): Add new param "compact".
39756 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
39757 (md_reader::read_file): New method.
39758 * read-md.h (md_reader::md_reader): Add new param "compact".
39759 (md_reader::read_file): New method.
39760 (md_reader::is_compact): New accessor.
39761 (md_reader::read_name): Convert return type from void to file_location.
39762 (md_reader::read_name_or_nil): New decl.
39763 (md_reader::read_name_1): New decl.
39764 (md_reader::m_compact): New field.
39765 (noop_reader::noop_reader): Pass "false" for new "compact" param
39767 (rtx_reader::rtx_reader): Add new "compact" param.
39768 (rtx_reader::read_rtx_operand): Make virtual and convert return
39769 type from void to rtx.
39770 (rtx_reader::read_until): New decl.
39771 (rtx_reader::handle_any_trailing_information): New virtual function.
39772 (rtx_reader::postprocess): New virtual function.
39773 (rtx_reader::finalize_string): New virtual function.
39774 (rtx_reader::m_in_call_function_usage): New field.
39775 (rtx_reader::m_reuse_rtx_by_id): New field.
39776 * read-rtl-function.c: New file.
39777 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
39778 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
39779 (selftest::verify_three_block_rtl_cfg): New decl.
39780 * read-rtl-function.h: New file.
39781 * read-rtl.c: Potentially include config.h rather than bconfig.h.
39782 For host, include function.h, memmodel.h, and emit-rtl.h.
39783 (one_time_initialization): New function.
39784 (struct compact_insn_name): New struct.
39785 (compact_insn_names): New array.
39786 (find_code): Handle insn codes in compact dumps.
39787 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
39788 (bind_subst_iter_and_attr): Likewise.
39789 (add_condition_to_string): Likewise.
39790 (add_condition_to_rtx): Likewise.
39791 (apply_attribute_uses): Likewise.
39792 (add_current_iterators): Likewise.
39793 (apply_iterators): Likewise.
39794 (initialize_iterators): Guard usage of apply_subst_iterator with
39795 #ifdef GENERATOR_FILE.
39796 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
39797 (md_reader::read_mapping): Likewise.
39798 (add_define_attr_for_define_subst): Likewise.
39799 (add_define_subst_attr): Likewise.
39800 (read_subst_mapping): Likewise.
39801 (check_code_iterator): Likewise.
39802 (rtx_reader::read_rtx): Likewise. Move one-time initialization
39804 (one_time_initialization): New function.
39805 (rtx_reader::read_until): New method.
39806 (read_flags): New function.
39807 (parse_reg_note_name): New function.
39808 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
39809 Handle reuse_rtx ids.
39810 Wrap iterator lookup within #ifdef GENERATOR_FILE.
39811 Add parsing support for RTL dumps, mirroring the special-cases in
39812 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
39813 values, and calling handle_any_trailing_information.
39814 (rtx_reader::read_rtx_operand): Convert return type from void
39815 to rtx, returning return_rtx. Handle case 'e'. Call
39816 finalize_string on XSTR and XTMPL fields.
39817 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
39818 "(nil)" values were omitted. Call the postprocess vfunc on the
39820 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
39821 class ctor. Initialize m_in_call_function_usage. Call
39822 one_time_initialization.
39823 * rtl-tests.c (selftest::test_uncond_jump): Call
39824 set_new_first_and_last_insn.
39825 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
39826 * selftest-rtl.c: New file.
39827 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
39828 (selftest::get_insn_by_uid): New decl.
39829 * selftest-run-tests.c (selftest::run_tests): Call
39830 read_rtl_function_c_tests.
39831 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
39832 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
39835 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
39837 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
39838 operands in a special way. Assert that pos+len <= mode precision.
39840 2017-01-05 Jakub Jelinek <jakub@redhat.com>
39842 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
39843 3 argument Alias with unlimited for the negative form.
39844 (fno-vect-cost-model): Removed.
39846 2017-01-05 Martin Liska <mliska@suse.cz>
39848 * hsa-gen.c (gen_hsa_divmod): New function.
39849 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
39851 2017-01-05 Martin Liska <mliska@suse.cz>
39854 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
39857 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39859 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
39860 small constant length operands.
39862 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39864 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
39865 between loop iterations.
39867 2017-01-05 Martin Liska <mliska@suse.cz>
39870 * gimplify.c (gimplify_decl_expr): Compare to
39871 asan_poisoned_variables instread of checking flags.
39872 (gimplify_target_expr): Likewise.
39873 (gimplify_expr): Likewise.
39874 (gimplify_function_tree): Conditionally initialize
39875 asan_poisoned_variables.
39877 2017-01-04 Jeff Law <law@redhat.com>
39879 PR tree-optimizatin/78812
39880 * rtl.h (contains_mem_rtx_p): Prototype.
39881 * ifcvt.c (containts_mem_rtx_p): Move from here to...
39882 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
39883 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
39884 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
39885 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
39887 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39889 * input.c (assert_char_at_range): Default-initialize actual_range.
39891 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39893 * df-scan.c (df_ref_create_structure): Make regno unsigned,
39894 to match the caller.
39896 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39898 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
39899 insns after final jump in test to emit dummy move.
39901 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39903 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
39904 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
39906 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39908 * multiple_target.c (create_dispatcher_calls): Init e_next.
39909 * tree-ssa-loop-split.c (split_loop): Init border.
39910 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
39913 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
39918 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
39919 (altivec_register_operand): Do not return true if the operand
39920 contains a SUBREG mixing SImode and SFmode.
39921 (vsx_register_operand): Likewise.
39922 (vsx_reg_sfsubreg_ok): New predicate.
39923 (vfloat_operand): Do not return true if the operand contains a
39924 SUBREG mixing SImode and SFmode.
39925 (vint_operand): Likewise.
39926 (vlogical_operand): Likewise.
39927 (gpc_reg_operand): Likewise.
39928 (int_reg_operand): Likewise.
39929 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
39930 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
39931 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
39933 (rs6000_emit_move_si_sf_subreg): New helper function.
39934 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
39935 fixup SUBREGs involving SImode and SFmode.
39936 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
39937 numbers for the new peephole2 optimization.
39938 (peephole2 for SFmode unions): New peephole2 to optimize cases in
39939 the GLIBC math library that do AND/IOR/XOR operations on single
39940 precision floating point.
39941 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
39942 target macros to say whether we need to avoid SUBREGs mixing
39944 (TARGET_ALLOW_SF_SUBREG): Likewise.
39945 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
39946 (UNSPEC_SI_FROM_SF): Likewise.
39947 (iorxor): Change spacing.
39948 (and_ior_xor): New iterator for AND, IOR, and XOR.
39949 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
39950 (movdi_from_sf_zero_ext): Likewise.
39951 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
39952 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
39953 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
39954 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
39955 (fms<mode>4): Likewise.
39956 (fnma<mode>4): Likewise.
39957 (fnms<mode>4): Likewise.
39958 (nfma<mode>4): Likewise.
39959 (nfms<mode>4): Likewise.
39961 2017-01-04 Marek Polacek <polacek@redhat.com>
39964 * doc/invoke.texi: Document -Wpointer-compare.
39966 2017-01-04 Jakub Jelinek <jakub@redhat.com>
39968 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
39971 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
39972 descriptions for -gdwarf-5 and emit them as uleb128 instead of
39975 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
39978 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
39979 documentation of the powerpc_popcntb_ok attribute.
39980 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
39981 code to issue warning messages if a requested CPU configuration is
39982 not supported by the binary (assembler and loader) toolchain.
39983 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
39984 made to define a built-in function that has been disabled.
39985 (paired_init_builtins): Add assertion to prevent ICE if attempt is
39986 made to define a built-in function that has been disabled.
39987 (altivec_init_builtins): Add comment explaining why definition
39988 of the DST built-in functions is not preceded by an assertion
39989 check. Add assertions to prevent ICE if attempts are made to
39990 define an altivec predicate or an abs* built-in function that has
39992 (htm_init_builtins): Add comment explaining why definition of the
39993 htm built-in functions is not preceded by an assertion check.
39995 2017-01-04 Jeff Law <law@redhat.com>
39997 PR tree-optimizatin/67955
39998 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
39999 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
40000 the points-to solution does not include pt_null. Use DECL_PT_UID
40003 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
40005 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
40006 Use gen_int_mode instead of gen_lopwart for const_int operands.
40008 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40010 PR tree-optimization/71563
40011 * match.pd: Simplify X << Y into X if Y is known to be 0 or
40012 out of range value - has low bits known to be zero.
40014 2017-01-04 Alan Modra <amodra@gmail.com>
40016 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
40017 * configure: Regenerate.
40018 * config.in: Regenerate.
40020 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40023 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
40024 a substring of the message, but strcmp with the whole message. Ifdef
40025 ENABLE_NLS, translate the message first using dgettext.
40027 2017-01-03 Jeff Law <law@redhat.com>
40029 PR tree-optimizatin/78856
40030 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
40031 (mark_threaded_blocks): Remove code to truncate thread paths that
40032 cross multiple loop headers. Instead invalidate the cached loop
40033 iteration information and handle case of a thread path walking
40034 into an irreducible region.
40036 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
40039 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
40040 assertions. Add support for doing the signbit if the IEEE 128-bit
40041 floating point value is in a GPR.
40042 * config/rs6000/rs6000.md (Fsignbit): Delete.
40043 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
40044 Update the length attribute if the value is in a GPR.
40045 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
40046 the sign or zero extension instruction, since the value is always 0/1.
40047 (signbit<mode>2_dm2): Delete using <Fsignbit>.
40050 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
40051 extracting SImode to a GPR register so that we can generate a
40052 store, limit the vector to be in a traditional Altivec register
40053 for the vextuwrx instruction.
40055 2017-01-03 Ian Lance Taylor <iant@google.com>
40057 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
40059 2017-01-03 Martin Sebor <msebor@redhat.com>
40061 PR tree-optimization/78696
40062 * gimple-ssa-sprintf.c (format_floating): Correct handling of
40063 precision. Use MPFR for %f for greater fidelity. Correct handling
40065 (pass_sprintf_length::compute_format_length): Set width and precision
40066 specified by asrerisk to void_node for vararg functions.
40067 (try_substitute_return_value): Adjust dump output.
40069 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
40071 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
40073 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
40075 * doc/invoke.texi (SPARC options): Document -mlra as the default.
40076 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
40077 -mlra/-mno-lra was passed to the compiler.
40079 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
40081 PR rtl-optimization/65618
40082 * emit-rtl.c (try_split): Move initialization of "before" and
40083 "after" to just before the call to emit_insn_after_setloc.
40085 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
40087 * doc/md.texi (Standard Names): Remove reference to Java frontend.
40089 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
40091 * dwarf2out.c (gen_enumeration_type_die): When
40092 -gno-strict-dwarf, add a DW_AT_encoding attribute.
40094 2017-01-03 Jakub Jelinek <jakub@redhat.com>
40096 PR tree-optimization/78965
40097 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
40098 Change first argument from const call_info & to call_info &. For %n
40099 set info.nowrite to false.
40101 PR middle-end/78901
40102 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
40103 possibly throwing calls.
40105 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
40106 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
40107 and fns handling, rather than in a separate case SSA_NAME.
40109 2017-01-02 Jeff Law <law@redhat.com>
40111 * config/darwin-driver.c (darwin_driver_init): Const-correctness
40112 fixes for first_period and second_period variables.
40114 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
40117 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
40118 (*insvqi_1): New insn pattern.
40119 (*insvqi_1_mem_rex64): Ditto.
40120 (*insvqi_2): Ditto.
40121 (*insvqi_3): Rename from *insvqi.
40123 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
40125 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
40127 * doc/cfg.texi (Edges): Remove reference to Java.
40128 (Maintaining the CFG): Ditto.
40130 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40132 PR middle-end/77674
40133 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
40134 transparent aliases.
40136 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40138 PR middle-end/77484
40139 * predict.def (PRED_CALL): Update hitrate.
40140 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
40141 * predict.c (tree_estimate_probability_bb): Split CALL predictor
40142 into direct/indirect/polymorphic variants.
40144 2017-01-01 Jakub Jelinek <jakub@redhat.com>
40146 Update copyright years.
40148 * gcc.c (process_command): Update copyright notice dates.
40149 * gcov-dump.c (print_version): Ditto.
40150 * gcov.c (print_version): Ditto.
40151 * gcov-tool.c (print_version): Ditto.
40152 * gengtype.c (create_file): Ditto.
40153 * doc/cpp.texi: Bump @copying's copyright year.
40154 * doc/cppinternals.texi: Ditto.
40155 * doc/gcc.texi: Ditto.
40156 * doc/gccint.texi: Ditto.
40157 * doc/gcov.texi: Ditto.
40158 * doc/install.texi: Ditto.
40159 * doc/invoke.texi: Ditto.
40161 Copyright (C) 2017 Free Software Foundation, Inc.
40163 Copying and distribution of this file, with or without modification,
40164 are permitted in any medium without royalty provided the copyright
40165 notice and this notice are preserved.