hjl [Thu, 10 Aug 2017 15:29:05 +0000 (15:29 +0000)]
i386: Don't use frame pointer without stack access
When there is no stack access, there is no need to use frame pointer
even if -fno-omit-frame-pointer is used and caller's frame pointer is
unchanged.
gcc/
PR target/81736
* config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
to ...
(ix86_finalize_stack_frame_flags): This. Also clear
frame_pointer_needed if -fno-omit-frame-pointer is used without
stack access.
(ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
with ix86_finalize_stack_frame_flags.
(ix86_expand_epilogue): Likewise.
(ix86_expand_split_stack_prologue): Likewise.
* doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
dmalcolm [Thu, 10 Aug 2017 13:22:27 +0000 (13:22 +0000)]
C/C++: show pertinent open token when missing a close token
gcc/c/ChangeLog:
* c-parser.c (c_parser_error): Rename to...
(c_parser_error_richloc): ...this, making static, and adding
"richloc" parameter, passing it to the c_parse_error call,
rather than calling c_parser_set_source_position_from_token.
(c_parser_error): Reintroduce, reimplementing in terms of the
above, converting return type from void to bool.
(class token_pair): New class.
(struct matching_paren_traits): New struct.
(matching_parens): New typedef.
(struct matching_brace_traits): New struct.
(matching_braces): New typedef.
(get_matching_symbol): New function.
(c_parser_require): Add param MATCHING_LOCATION, using it to
highlight matching "opening" tokens for missing "closing" tokens.
(c_parser_skip_until_found): Likewise.
(c_parser_static_assert_declaration_no_semi): Convert explicit
parsing of CPP_OPEN_PAREN and CPP_CLOSE_PAREN to use of
class matching_parens, so that the pertinent open parenthesis is
highlighted when there are problems locating the close
parenthesis.
(c_parser_struct_or_union_specifier): Likewise.
(c_parser_typeof_specifier): Likewise.
(c_parser_alignas_specifier): Likewise.
(c_parser_simple_asm_expr): Likewise.
(c_parser_braced_init): Likewise, for matching_braces.
(c_parser_paren_condition): Likewise, for matching_parens.
(c_parser_switch_statement): Likewise.
(c_parser_for_statement): Likewise.
(c_parser_asm_statement): Likewise.
(c_parser_asm_operands): Likewise.
(c_parser_cast_expression): Likewise.
(c_parser_sizeof_expression): Likewise.
(c_parser_alignof_expression): Likewise.
(c_parser_generic_selection): Likewise.
(c_parser_postfix_expression): Likewise for cases RID_VA_ARG,
RID_OFFSETOF, RID_TYPES_COMPATIBLE_P, RID_AT_SELECTOR,
RID_AT_PROTOCOL, RID_AT_ENCODE, reindenting as necessary.
In case CPP_OPEN_PAREN, pass loc_open_paren to the
c_parser_skip_until_found call.
(c_parser_objc_class_definition): Use class matching_parens as
above.
(c_parser_objc_method_decl): Likewise.
(c_parser_objc_try_catch_finally_statement): Likewise.
(c_parser_objc_synchronized_statement): Likewise.
(c_parser_objc_at_property_declaration): Likewise.
(c_parser_oacc_wait_list): Likewise.
(c_parser_omp_var_list_parens): Likewise.
(c_parser_omp_clause_collapse): Likewise.
(c_parser_omp_clause_default): Likewise.
(c_parser_omp_clause_if): Likewise.
(c_parser_omp_clause_num_threads): Likewise.
(c_parser_omp_clause_num_tasks): Likewise.
(c_parser_omp_clause_grainsize): Likewise.
(c_parser_omp_clause_priority): Likewise.
(c_parser_omp_clause_hint): Likewise.
(c_parser_omp_clause_defaultmap): Likewise.
(c_parser_oacc_single_int_clause): Likewise.
(c_parser_omp_clause_ordered): Likewise.
(c_parser_omp_clause_reduction): Likewise.
(c_parser_omp_clause_schedule): Likewise.
(c_parser_omp_clause_num_teams): Likewise.
(c_parser_omp_clause_thread_limit): Likewise.
(c_parser_omp_clause_aligned): Likewise.
(c_parser_omp_clause_linear): Likewise.
(c_parser_omp_clause_safelen): Likewise.
(c_parser_omp_clause_simdlen): Likewise.
(c_parser_omp_clause_depend): Likewise.
(c_parser_omp_clause_map): Likewise.
(c_parser_omp_clause_device): Likewise.
(c_parser_omp_clause_dist_schedule): Likewise.
(c_parser_omp_clause_proc_bind): Likewise.
(c_parser_omp_clause_uniform): Likewise.
(c_parser_omp_for_loop): Likewise.
(c_parser_cilk_clause_vectorlength): Likewise.
(c_parser_cilk_clause_linear): Likewise.
(c_parser_transaction_expression): Likewise.
* c-parser.h (c_parser_require): Add param matching_location with
default UNKNOWN_LOCATION.
(c_parser_error): Convert return type from void to bool.
(c_parser_skip_until_found): Add param matching_location with
default UNKNOWN_LOCATION.
gcc/c-family/ChangeLog:
* c-common.c (c_parse_error): Add rich_location * param, using it
rather implicitly using input_location.
* c-common.h (c_parse_error): Add rich_location * param.
gcc/cp/ChangeLog:
* parser.c (cp_parser_error): Update for new param to
c_parse_error.
(class token_pair): New class.
(struct matching_paren_traits): New struct.
(matching_parens): New typedef.
(struct matching_brace_traits): New struct.
(matching_braces): New typedef.
(cp_parser_statement_expr): Convert explicit parsing of
CPP_OPEN_PAREN and CPP_CLOSE_PAREN to use of
class matching_parens, so that the pertinent open parenthesis is
highlighted when there are problems locating the close
parenthesis.
(cp_parser_primary_expression): Likewise.
(cp_parser_compound_literal_p): Remove consumption of opening
paren.
(cp_parser_postfix_expression): Convert explicit parsing of
CPP_OPEN_PAREN and CPP_CLOSE_PAREN to use matching parens, as
above. Use it to consume the opening paren previously consumed by
cp_parser_compound_literal_p.
(cp_parser_parenthesized_expression_list): Likewise.
(cp_parser_unary_expression): Likewise.
(cp_parser_new_expression): Likewise.
(cp_parser_cast_expression): Likewise.
(cp_parser_builtin_offsetof): Likewise.
(cp_parser_trait_expr): Likewise.
(cp_parser_lambda_declarator_opt): Likewise.
(cp_parser_lambda_body): Likewise, for matching_braces.
(cp_parser_compound_statement): Likewise.
(cp_parser_selection_statement): Likewise, for matching_parens.
(cp_parser_iteration_statement): Likewise.
(cp_parser_already_scoped_statement): Likewise, for
matching_braces.
(cp_parser_linkage_specification): Likewise.
(cp_parser_static_assert): Likewise, for matching_parens.
(cp_parser_decltype): Likewise.
(cp_parser_operator): Likewise.
(cp_parser_enum_specifier): Likewise.
(cp_parser_namespace_definition): Likewise.
(cp_parser_direct_declarator): Likewise.
(cp_parser_braced_list): Likewise.
(cp_parser_class_specifier_1): Likewise, for matching_braces.
(cp_parser_constant_initializer): Likewise.
(cp_parser_noexcept_specification_opt): Likewise, for
matching_parens.
(cp_parser_exception_specification_opt): Likewise.
(cp_parser_handler): Likewise.
(cp_parser_asm_specification_opt): Likewise.
(cp_parser_asm_operand_list): Likewise.
(cp_parser_gnu_attributes_opt): Likewise.
(cp_parser_std_attribute_spec): Likewise.
(cp_parser_requirement_parameter_list): Likewise.
(cp_parser_requirement_body): Likewise, for matching_braces.
(cp_parser_compound_requirement): Likewise.
(cp_parser_template_introduction): Likewise.
(cp_parser_sizeof_pack): Likewise, for matching_parens.
(cp_parser_sizeof_operand): Likewise; use it to consume the
opening paren previously consumed by cp_parser_compound_literal_p.
(get_matching_symbol): New function.
(cp_parser_required_error): Add param "matching_location". Remove
calls to cp_parser_error, instead setting a non-NULL gmsgid, and
handling it if set by calling c_parse_error, potentially with a
secondary location if matching_location was set.
(cp_parser_require): Add param "matching_location", with a default
value of UNKNOWN_LOCATION.
(cp_parser_require_keyword): Update for new param of
cp_parser_required_error.
(cp_parser_objc_encode_expression): Update to class matching_parens
as above.
(cp_parser_objc_defs_expression): Likewise.
(cp_parser_objc_protocol_expression): Likewise.
(cp_parser_objc_selector_expression): Likewise.
(cp_parser_objc_typename): Likewise.
(cp_parser_objc_superclass_or_category): Likewise.
(cp_parser_objc_try_catch_finally_statement): Likewise.
(cp_parser_objc_synchronized_statement): Likewise.
(cp_parser_objc_at_property_declaration): Likewise.
(cp_parser_oacc_single_int_clause): Likewise.
(cp_parser_oacc_shape_clause): Likewise.
(cp_parser_omp_clause_collapse): Likewise.
(cp_parser_omp_clause_default): Likewise.
(cp_parser_omp_clause_final): Likewise.
(cp_parser_omp_clause_if): Likewise.
(cp_parser_omp_clause_num_threads): Likewise.
(cp_parser_omp_clause_num_tasks): Likewise.
(cp_parser_omp_clause_grainsize): Likewise.
(cp_parser_omp_clause_priority): Likewise.
(cp_parser_omp_clause_hint): Likewise.
(cp_parser_omp_clause_defaultmap): Likewise.
(cp_parser_omp_clause_ordered): Likewise.
(cp_parser_omp_clause_schedule): Likewise.
(cp_parser_omp_clause_num_teams): Likewise.
(cp_parser_omp_clause_thread_limit): Likewise.
(cp_parser_omp_clause_aligned): Likewise.
(cp_parser_omp_clause_linear): Likewise.
(cp_parser_omp_clause_safelen): Likewise.
(cp_parser_omp_clause_simdlen): Likewise.
(cp_parser_omp_clause_depend): Likewise.
(cp_parser_omp_clause_device): Likewise.
(cp_parser_omp_clause_dist_schedule): Likewise.
(cp_parser_oacc_clause_async): Likewise.
(cp_parser_omp_critical): Likewise.
(cp_parser_omp_for_loop): Likewise.
(cp_parser_omp_sections_scope): Likewise.
(cp_parser_omp_declare_reduction_exprs): Likewise.
Update for new param to cp_parser_required_error.
(cp_parser_oacc_routine): Likewise.
(cp_parser_transaction_expression): Likewise.
(cp_parser_cilk_simd_vectorlength): Likewise.
gcc/testsuite/ChangeLog:
* c-c++-common/missing-close-symbol.c: New test case.
* c-c++-common/missing-symbol.c: New test case.
* gcc.dg/unclosed-init.c: New test case.
* g++.dg/diagnostic/unclosed-extern-c.C: New test case.
* g++.dg/diagnostic/unclosed-function.C: New test case.
* g++.dg/diagnostic/unclosed-namespace.C: New test case.
* g++.dg/diagnostic/unclosed-struct.C: New test case.
* g++.dg/parse/pragma2.C: Update to reflect movement of the
"expected identifier" error.
marxin [Thu, 10 Aug 2017 07:43:49 +0000 (07:43 +0000)]
Fix target attribute handling (PR c++/81355).
2017-08-10 Martin Liska <mliska@suse.cz>
PR c++/81355
* c-attribs.c (handle_target_attribute):
Report warning for an empty string argument of target attribute.
2017-08-10 Martin Liska <mliska@suse.cz>
dmalcolm [Thu, 10 Aug 2017 00:18:19 +0000 (00:18 +0000)]
jit: add gcc_jit_type_get_vector
gcc/jit/ChangeLog:
* docs/cp/topics/types.rst (Vector types): New section.
* docs/topics/compatibility.rst (LIBGCCJIT_ABI_8): New tag.
* docs/topics/types.rst (gcc_jit_context_get_type): Fix typo in
example.
(Vector types): New section.
* docs/_build/texinfo/libgccjit.texi: Regenerate.
* jit-playback.c (gcc::jit::playback::type::get_vector): New
method.
* jit-playback.h (gcc::jit::playback::type::get_vector): New
method.
* jit-recording.c: In namespace gcc::jit::recording::
(type::get_vector): New method.
(memento_of_get_aligned::write_reproducer): Fix typo
in leading comment.
(memento_of_get_vector::replay_into): New method.
(memento_of_get_vector::make_debug_string): New method.
(memento_of_get_vector::write_reproducer): New method.
* jit-recording.h: In namespace gcc::jit::recording::
(type::get_vector): New
method.
(class memento_of_get_vector): New class.
* libgccjit++.h (gccjit::type::get_vector): New method.
* libgccjit.c (gcc_jit_type_get_vector): New public entrypoint.
* libgccjit.h (LIBGCCJIT_HAVE_gcc_jit_type_get_vector): New
define.
(gcc_jit_type_get_vector): New decl.
* libgccjit.map (LIBGCCJIT_ABI_8): New ABI tag.
gcc/testsuite/ChangeLog:
* jit.dg/all-non-failing-tests.h: Add note about
test-vector-types.cc.
* jit.dg/test-error-gcc_jit_type_get_vector-bad-type.c: New test
case.
* jit.dg/test-error-gcc_jit_type_get_vector-non-power-of-two.c:
New test case.
* jit.dg/test-vector-types.cc: New test case.
redi [Wed, 9 Aug 2017 21:38:05 +0000 (21:38 +0000)]
Fix test for __has_unique_object_representations support in Clang
* include/std/type_traits (_GLIBCXX_NO_BUILTIN_HAS_UNIQ_OBJ_REP):
Replace with _GLIBCXX_HAVE_BUILTIN_HAS_UNIQ_OBJ_REP and use
__is_identifier to set it.
willschm [Wed, 9 Aug 2017 21:28:39 +0000 (21:28 +0000)]
[gcc]
2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
* config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
to indicate when early gimple folding has been disabled.
(rs6000_gimple_fold_builtin): Add debug content.
(rs6000_invalid_builtin): Fix whitespace.
(rs6000_expand_builtin): Fix whitespace.
* config/rs6000/rs6000.opt: Add option for -mfold-gimple.
ian [Wed, 9 Aug 2017 21:24:00 +0000 (21:24 +0000)]
compiler: fix buglet in lshift type determination
It's possible to construct an lshift expression using unsafe.Sizeof
that is technically a compile-time constant but can't be evaluated
without going through backend methods. In this case, insure that
Type::make_non_abstract_type is called on the numeric operand of the
shift (as opposed to leaving as abstract), to avoid an assert later on
in the compiler flow.
segher [Wed, 9 Aug 2017 21:08:33 +0000 (21:08 +0000)]
rs6000: Use SAVE_MULTIPLE only if we restore what it saves (PR80938)
We can have SAVE_MULTIPLE while we do not have REST_MULTIPLE. If the
inline restore does not restore all registers, the CFI for the save
and restore can conflict if things are shrink-wrapped.
We could restore all registers that are saved (not ideal), or emit
the CFI notes to say we did (which will work fine, but is also not
so great); instead, let's not save the registers that are unused.
PR target/80938
* config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
SAVE_MULTIPLE if not all the registers that saves, should be saved.
redi [Wed, 9 Aug 2017 17:52:10 +0000 (17:52 +0000)]
PR libstdc++/81751 don't call fflush(NULL)
PR libstdc++/79820
PR libstdc++/81751
* config/io/basic_file_stdio.cc (sys_open(FILE*, ios_base::openmode)):
Call fflush on the stream instead of calling sync() while _M_cfile is
null. Restore original value of errno.
* testsuite/ext/stdio_filebuf/char/79820.cc: New.
* testsuite/ext/stdio_filebuf/char/81751.cc: New.
ian [Wed, 9 Aug 2017 17:15:02 +0000 (17:15 +0000)]
compiler: handle >32bit exponent in Ldexp
Libgo's implementation of math.Ldexp declared the libc "ldexp" as
taking an 'int' exponent argument, which is not quite right for 64-bit
platforms (exp arg is always int32); this could yield incorrect
results for exponent values outside the range of Minint32/Maxint32.
Fix by upating the type for the libc version of ldexp, and adding
guards to screen for out-of-range exponents.
mpolacek [Wed, 9 Aug 2017 11:28:22 +0000 (11:28 +0000)]
PR c/81233
* c-typeck.c (pedwarn_init): Make the function take a variable list.
Call emit_diagnostic_valist instead of pedwarn.
(convert_for_assignment): Unroll the PEDWARN_FOR_ASSIGNMENT macro.
Print the relevant types in diagnostics.
* diagnostic-core.h (emit_diagnostic_valist): Add declaration.
* diagnostic.c (emit_diagnostic): Add a comment.
(emit_diagnostic_valist): New function.
mpolacek [Wed, 9 Aug 2017 08:51:20 +0000 (08:51 +0000)]
PR c/81417
* c-array-notation.c (fix_builtin_array_notation_fn): Update calls to
build_conditional_expr.
* c-parser.c (c_parser_conditional_expression): Create locations for
EXP1 and EXP2 from their source ranges. Pass the locations down to
build_conditional_expr.
* c-tree.h (build_conditional_expr): Update declaration.
* c-typeck.c (build_conditional_expr): Add location_t parameters.
For -Wsign-compare, also print the types.
* input.c (make_location): New overload.
* input.h (make_location): Declare.
* objc-next-runtime-abi-02.c (build_v2_build_objc_method_call): Update
a call to build_conditional_expr.
hjl [Tue, 8 Aug 2017 22:06:21 +0000 (22:06 +0000)]
PR driver/81523: Make -static override -pie
-static and -pie together behave differently depending on whether GCC is
configured with --enable-default-pie. On x86, "-static -pie" fails to
create executable when --enable-default-pie isn't used, but creates a
static executable when --enable-default-pie is used. This patch makes
-static completely override -pie to create a static executable, regardless
if --enable-default-pie is used to configure GCC.
gcc/
PR driver/81523
* gcc.c (NO_PIE_SPEC): Delete.
(PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
exclusion..
(LINK_PIE_SPEC): ..to here.
(LINK_COMMAND_SPEC): Support -no-pie.
* config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
chain of crtbegin*.o selection, update for PIE_SPEC changes and
format.
(GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
* config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
(ENDFILE_CRTEND_SPEC): Similarly.
rguenth [Tue, 8 Aug 2017 12:51:20 +0000 (12:51 +0000)]
2017-08-08 Richard Biener <rguenther@suse.de>
PR middle-end/81719
* tree-ssa-loop-niter.c: Include tree-dfa.h.
(expand_simple_operations): Also look through ADDR_EXPRs with
MEM_REF bases treating them as POINTER_PLUS_EXPR.
rguenth [Tue, 8 Aug 2017 12:49:39 +0000 (12:49 +0000)]
2017-08-08 Richard Biener <rguenther@suse.de>
PR tree-optimization/81723
* tree-vect-slp.c (struct bst_traits): New hash traits.
(bst_fail): New global.
(vect_build_slp_tree_2): New worker, split out from ...
(vect_build_slp_tree): ... this now wrapping it with using
bst_fail set to cache SLP tree build fails. Properly handle
max_tree_size.
(vect_analyze_slp_instance): Allocate and free bst_fail.
PR tree-opt/81696
* ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
LABEL_DECLs that can be from a different function.
2017-08-08 Martin Liska <mliska@suse.cz>
PR tree-opt/81696
* gcc.dg/ipa/pr81696.c: New test.
meissner [Mon, 7 Aug 2017 23:51:27 +0000 (23:51 +0000)]
[gcc]
2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
PR target/81593
* config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
constraints since the -mupper-regs-* switches have been
eliminated.
(vsx_concat_<mode>_1): New combiner insns to recognize inserting
into a vector from a double word element that was extracted from
another vector, and eliminate extra XXPERMDI instructions.
(vsx_concat_<mode>_2): Likewise.
(vsx_concat_<mode>_3): Likewise.
(vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
concat to allow optimizing inserts from previous extracts.
[gcc/testsuite]
2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
PR target/81593
* gcc.target/powerpc/vec-setup.h: New tests to test various
combinations of setting up vectors of 2 double word elements.
* gcc.target/powerpc/vec-setup-long.c: Likewise.
* gcc.target/powerpc/vec-setup-double.c: Likewise.
* gcc.target/powerpc/vec-setup-be-long.c: Likewise.
* gcc.target/powerpc/vec-setup-be-double.c: Likewise.
* gcc.target/powerpc/vsx-extract-6.c: New tests for optimzing
vector inserts from vector extracts.
* gcc.target/powerpc/vsx-extract-7.c: Likewise.
marxin [Mon, 7 Aug 2017 17:12:52 +0000 (17:12 +0000)]
Add missing header file attribs.h to couple of targets.
2017-08-07 Martin Liska <mliska@suse.cz>
* config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
* config/rl78/rl78.c: Add include of attribs.h.
* config/sh/sh.c: Likewise.
* config/v850/v850.c: Likewise.
wilco [Mon, 7 Aug 2017 14:17:09 +0000 (14:17 +0000)]
[AArch64] Use gen_frame_mem for callee-saves
The frame code uses a mixture of gen_rtx_MEM and gen_frame_mem for
callee-saves. Callee-saves never alias with local variables, so using
gen_frame_mem is best.
hjl [Mon, 7 Aug 2017 11:49:10 +0000 (11:49 +0000)]
i386: Don't use frame pointer without stack access
When there is no stack access, there is no need to use frame pointer
even if -fno-omit-frame-pointer is used.
gcc/
PR target/81736
* config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
to ...
(ix86_finalize_stack_frame_flags): This. Also clear
frame_pointer_needed if -fno-omit-frame-pointer is used without
stack access.
(ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
with ix86_finalize_stack_frame_flags.
(ix86_expand_epilogue): Likewise.
(ix86_expand_split_stack_prologue): Likewise.
hjl [Mon, 7 Aug 2017 11:47:22 +0000 (11:47 +0000)]
i386: Set priority to P_AES for Westmere
The difference between Nehalem and Westmere is AES. We should set
priority to P_AES for Westmere, not P_PROC_SSE4_2 which is for Nehalem.
Otherwise, we will pick Nehalem implementation on Westmere. Tested on
Westmere.
PR target/81743
* config/i386/i386.c (get_builtin_code_for_version): Set priority
to P_AES for Westmere.
marxin [Mon, 7 Aug 2017 08:37:07 +0000 (08:37 +0000)]
Canonicalize names of attributes.
2017-08-07 Martin Liska <mliska@suse.cz>
* attribs.h (canonicalize_attr_name): New function.
(cmp_attribs): Move from c-format.c and adjusted.
(is_attribute_p): Moved from tree.h.
* tree-inline.c: Add new includes.
* tree.c (cmp_attrib_identifiers): Use cmp_attribs.
(private_is_attribute_p): Remove.
(private_lookup_attribute): Likewise.
(private_lookup_attribute_by_prefix): Simplify.
(remove_attribute): Use is_attribute_p.
* tree.h: Remove removed declarations.
2017-08-07 Martin Liska <mliska@suse.cz>
* array-notation-common.c: Add new includes.
* c-format.c( handle_format_attribute): Canonicalize a format
function name.
* c-lex.c (c_common_has_attribute): Canonicalize name of an
attribute.
* c-pretty-print.c: Add new include.
2017-08-07 Martin Liska <mliska@suse.cz>
* parser.c (cp_parser_gnu_attribute_list): Canonicalize name of an
attribute.
(cp_parser_std_attribute): Likewise.
* tree.c: Add new include.
2017-08-07 Martin Liska <mliska@suse.cz>
* c-parser.c (c_parser_attributes): Canonicalize name of an
attribute.
2017-08-07 Martin Liska <mliska@suse.cz>
* go-gcc.cc (Gcc_backend::function): Look up for no_split_stack
and not __no_split_stack__.
2017-08-07 Martin Liska <mliska@suse.cz>
jakub [Mon, 7 Aug 2017 08:29:21 +0000 (08:29 +0000)]
PR middle-end/81698
* stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
instead of computing it in the function. Formatting fix.
(expand_case): Don't rely on default_edge being the first edge,
clear it if removing it, pass default_edge to
emit_case_dispatch_table.
(expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
fix.
vries [Sun, 6 Aug 2017 07:54:29 +0000 (07:54 +0000)]
Fix call arguments mismatch in gcc.dg/torture/pr78218.c
2017-08-06 Tom de Vries <tom@codesourcery.com>
PR testsuite/81731
* gcc.dg/torture/pr78218.c (struct v): New type, factored type out of
variable 'a' definition.
(b): Remove unused variable.
(gp): New variable.
(check): Add parameter corresponding to call in main, and handle.
ebotcazou [Sat, 5 Aug 2017 22:00:41 +0000 (22:00 +0000)]
* c-ada-spec.c (has_static_fields): Look only into variables.
(print_constructor): Add TYPE parameter and use it for the name.
(print_destructor): Likewise.
(print_ada_declaration): Adjust to new constructor/destructor names.
Adjust calls to print_constructor and print_destructor.
(print_ada_struct_decl): Do not test TREE_STATIC on FIELD_DECL.
Look only into variables in the final loop.
paolo [Fri, 4 Aug 2017 22:15:48 +0000 (22:15 +0000)]
/cp
2017-08-04 Paolo Carlini <paolo.carlini@oracle.com>
PR c++/79790
* pt.c (do_class_deduction): Handle the case of no viable implicit
deduction guides; simplify the code generating implicit deduction
guides.
/testsuite
2017-08-04 Paolo Carlini <paolo.carlini@oracle.com>
paolo [Fri, 4 Aug 2017 22:13:46 +0000 (22:13 +0000)]
/cp
2017-08-04 Paolo Carlini <paolo.carlini@oracle.com>
PR c++/79790
* pt.c (do_class_deduction): Handle the case of no viable implicit
deduction guides; simplify the code generating implicit deduction
guides.
/testsuite
2017-08-04 Paolo Carlini <paolo.carlini@oracle.com>
ygribov [Fri, 4 Aug 2017 20:29:12 +0000 (20:29 +0000)]
Remove useless floating point casts in comparisons.
2017-08-04 Yury Gribov <tetra2005@gmail.com>
PR tree-optimization/57371
gcc/
* match.pd: New pattern.
gcc/testsuite/
* c-c++-common/pr57371-1.c: New test.
* c-c++-common/pr57371-2.c: New test.
* c-c++-common/pr57371-3.c: New test.
* c-c++-common/pr57371-4.c: New test.
* gcc.dg/pr57371-5.c: New test.
carll [Fri, 4 Aug 2017 18:39:30 +0000 (18:39 +0000)]
gcc/testsuite/ChangeLog:
2017-08-04 Carl Love <cel@us.ibm.com>
* gcc.target/powerpc/builtins-3.c: Remove ISA 3.0 word variant
builtin test cases for vec_mule, and vec_mulo.
* gcc.target/powerpc/builtins-3-p8.c: Add ISA 3.0 word variant
builtin test cases for vec_mule, and vec_mulo.
visit0r [Fri, 4 Aug 2017 15:50:14 +0000 (15:50 +0000)]
Fix PR 81713
* brigfrontend/brig-basic-inst-handler.cc: replace build_int_cst with
bitsize_int in building BIT_FIELD_REF.
* brigfrontend/brig-code-entry-handler.cc: likewise.
rsandifo [Fri, 4 Aug 2017 10:42:53 +0000 (10:42 +0000)]
Pool alignment information for common bases
This patch is a follow-on to the fix for PR81136. The testcase for that
PR shows that we can (correctly) calculate different base alignments
for two data_references but still tell that their misalignments wrt the
vector size are equal. This is because we calculate the base alignments
for each dr individually, without looking at the other drs, and in
general the alignment we calculate is only guaranteed if the dr's DR_REF
actually occurs.
This is working as designed, but it does expose a missed opportunity.
We know that if a vectorised loop is reached, all statements in that
loop execute at least once, so it should be safe to pool the alignment
information for all the statements we're vectorising. The only catch is
that DR_REFs for masked loads and stores only occur if the mask value is
nonzero. For example, in:
struct s __attribute__((aligned(32))) {
int misaligner;
int array[N];
};
int *ptr;
for (int i = 0; i < n; ++i)
ptr[i] = c[i] ? ((struct s *) (ptr - 1))->array[i] : 0;
we can only guarantee that ptr points to a "struct s" if at least
one c[i] is true.
This patch adds a DR_IS_CONDITIONAL_IN_STMT flag to record whether
the DR_REF is guaranteed to occur every time that the statement
executes to completion. It then pools the alignment information
for references that aren't conditional in this sense.
2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
gcc/
PR tree-optimization/81136
* tree-vectorizer.h: Include tree-hash-traits.h.
(vec_base_alignments): New typedef.
(vec_info): Add a base_alignments field.
(vect_record_base_alignments): Declare.
* tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
field.
(DR_IS_CONDITIONAL_IN_STMT): New macro.
(create_data_ref): Add an is_conditional_in_stmt argument.
* tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
the is_conditional_in_stmt field.
(data_ref_loc): Add an is_conditional_in_stmt field.
(get_references_in_stmt): Set the is_conditional_in_stmt field.
(find_data_references_in_stmt): Update call to create_data_ref.
(graphite_find_data_references_in_stmt): Likewise.
* tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
* tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
(vect_record_base_alignment): New function.
(vect_record_base_alignments): Likewise.
(vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
for nested statements even if we fail to compute a misalignment.
Use pooled base alignments for unconditional references.
(vect_find_same_alignment_drs): Compare base addresses instead
of base objects.
(vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
* tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
rsandifo [Fri, 4 Aug 2017 10:41:12 +0000 (10:41 +0000)]
C++-ify vec_info structures
This patch uses new, delete, constructors and desctructors to manage
vec_info. This includes making ~vec_info free all the data shared
by bb_vec_info and loop_vec_info, whereas previously the code was
duplicated in destroy_bb_vec_info and destroy_loop_vec_info. This
in turn meant changing the order of:
FOR_EACH_VEC_ELT (slp_instances, i, instance)
vect_free_slp_instance (instance);
and:
gimple_set_uid (stmt, -1);
in destroy_bb_vec_info/~_bb_vec_info, so that now vect_free_slp_instance
could see a uid of -1 as well as 0. The patch updates vinfo_for_stmt
so that it returns NULL for a uid of -1.
2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
gcc/
* tree-vectorizer.h (vec_info): Add a constructor and destructor.
Add an explicit name for the enum. Use auto_vec for slp_instances
and grouped_stores.
(_loop_vec_info): Add a constructor and destructor. Use auto_vec
for all vectors.
(_bb_vec_info): Add a constructor and destructor.
(vinfo_for_stmt): Return NULL for uids of -1 as well.
(destroy_loop_vec_info): Delete.
(vect_destroy_datarefs): Likewise.
* tree-vectorizer.c (vect_destroy_datarefs): Delete.
(vec_info::vec_info): New function.
(vec_info::~vec_info): Likewise.
(vectorize_loops): Use delete instead of destroy_loop_vec_info.
* tree-parloops.c (gather_scalar_reductions): Use delete instead of
destroy_loop_vec_info.
* tree-vect-loop.c (new_loop_vec_info): Replace with...
(_loop_vec_info::_loop_vec_info): ...this.
(destroy_loop_vec_info): Replace with...
(_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
the stmt_vec_infos. Leave handling of vec_info information to its
destructor. Remove explicit vector releases.
(vect_analyze_loop_form): Use new instead of new_loop_vec_info.
(vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
* tree-vect-slp.c (new_bb_vec_info): Replace with...
(_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
(destroy_bb_vec_info): Replace with...
(_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
information to its destructor.
(vect_slp_analyze_bb_1): Use new and delete instead of
new_bb_vec_info and destroy_bb_vec_info.
(vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
single delete.
rsandifo [Fri, 4 Aug 2017 10:40:35 +0000 (10:40 +0000)]
Use base inequality for some vector alias checks
This patch checks whether two data references x and y cannot
partially overlap and so are independent whenever &x != &y.
We can then use this in the vectoriser to optimise alias checks.
gcc/
2016-08-04 Richard Sandiford <richard.sandiford@linaro.org>
* hash-traits.h (pair_hash): New struct.
* tree-data-ref.h (data_dependence_relation): Add object_a and
object_b fields.
(DDR_OBJECT_A, DDR_OBJECT_B): New macros.
* tree-data-ref.c (initialize_data_dependence_relation): Initialize
DDR_OBJECT_A and DDR_OBJECT_B.
* tree-vectorizer.h (vec_object_pair): New type.
(_loop_vec_info): Add a check_unequal_addrs field.
(LOOP_VINFO_CHECK_UNEQUAL_ADDRS): New macro.
(LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Return true if there is an
entry in check_unequal_addrs. Check comp_alias_ddrs instead of
may_alias_ddrs.
* tree-vect-loop.c (destroy_loop_vec_info): Release
LOOP_VINFO_CHECK_UNEQUAL_ADDRS.
(vect_analyze_loop_2): Likewise, when restarting.
(vect_estimate_min_profitable_iters): Estimate the cost of
LOOP_VINFO_CHECK_UNEQUAL_ADDRS.
* tree-vect-data-refs.c: Include tree-hash-traits.h.
(vect_prune_runtime_alias_test_list): Try to handle conflicts
using LOOP_VINFO_CHECK_UNEQUAL_ADDRS, if the data dependence allows.
Count such tests in the final summary.
* tree-vect-loop-manip.c (chain_cond_expr): New function.
(vect_create_cond_for_align_checks): Use it.
(vect_create_cond_for_unequal_addrs): New function.
(vect_loop_versioning): Call it.
gcc/testsuite/
* gcc.dg/vect/vect-alias-check-6.c: New test.
rsandifo [Fri, 4 Aug 2017 10:39:44 +0000 (10:39 +0000)]
Handle data dependence relations with different bases
This patch tries to calculate conservatively-correct distance
vectors for two references whose base addresses are not the same.
It sets a new flag DDR_COULD_BE_INDEPENDENT_P if the dependence
isn't guaranteed to occur.
The motivating example is:
struct s { int x[8]; };
void
f (struct s *a, struct s *b)
{
for (int i = 0; i < 8; ++i)
a->x[i] += b->x[i];
}
in which the "a" and "b" accesses are either independent or have a
dependence distance of 0 (assuming -fstrict-aliasing). Neither case
prevents vectorisation, so we can vectorise without an alias check.
I'd originally wanted to do the same thing for arrays as well, e.g.:
void
f (int a[][8], struct b[][8])
{
for (int i = 0; i < 8; ++i)
a[0][i] += b[0][i];
}
I think this is valid because C11 6.7.6.2/6 says:
For two array types to be compatible, both shall have compatible
element types, and if both size specifiers are present, and are
integer constant expressions, then both size specifiers shall have
the same constant value.
So if we access an array through an int (*)[8], it must have type X[8]
or X[], where X is compatible with int. It doesn't seem possible in
either case for "a[0]" and "b[0]" to overlap when "a != b".
However, as the comment above "if (same_base_p)" explains, GCC is more
forgiving: it supports arbitrary overlap of arrays and allows arrays to
be accessed with different dimensionality. There are examples of this
in PR50067. The patch therefore only handles references that end in a
structure field access.
There are two ways of handling these dependences in the vectoriser:
use them to limit VF, or check at runtime as before. I've gone for
the approach of checking at runtime if we can, to avoid limiting VF
unnecessarily, but falling back to a VF cap when runtime checks aren't
allowed.
The patch tests whether we queued an alias check with a dependence
distance of X and then picked a VF <= X, in which case it's safe to
drop the alias check. Since vect_prune_runtime_alias_check_list
can be called twice with different VF for the same loop, it's no
longer safe to clear may_alias_ddrs on exit. Instead we should use
comp_alias_ddrs to check whether versioning is necessary.
2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
gcc/
* tree-data-ref.h (subscript): Add access_fn field.
(data_dependence_relation): Add could_be_independent_p.
(SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
(same_access_functions): Move to tree-data-ref.c.
* tree-data-ref.c (ref_contains_union_access_p): New function.
(access_fn_component_p): Likewise.
(access_fn_components_comparable_p): Likewise.
(dr_analyze_indices): Add a reference to access_fn_component_p.
(dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
DR_ACCESS_FN.
(constant_access_functions): Likewise.
(add_other_self_distances): Likewise.
(same_access_functions): Likewise. (Moved from tree-data-ref.h.)
(initialize_data_dependence_relation): Use XCNEW and remove
explicit zeroing of DDR_REVERSED_P. Look for a subsequence
of access functions that have the same type. Allow the
subsequence to end with different bases in some circumstances.
Record the chosen access functions in SUB_ACCESS_FN.
(build_classic_dist_vector_1): Replace ddr_a and ddr_b with
a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
(subscript_dependence_tester_1): Likewise dra and drb.
(build_classic_dist_vector): Update calls accordingly.
(subscript_dependence_tester): Likewise.
* tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
DDR_COULD_BE_INDEPENDENT_P.
* tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
comp_alias_ddrs instead of may_alias_ddrs.
* tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
New function.
(vect_analyze_data_ref_dependence): Use it if
DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
distance vectors if that fails.
(dependence_distance_ge_vf): New function.
(vect_prune_runtime_alias_test_list): Use it. Don't clear
LOOP_VINFO_MAY_ALIAS_DDRS.
projects / thirdparty / gcc.git / log
