[AArch64][3/3] Migrate aarch64_expand_prologue/epilogue to aarch64_add_constant
gcc/
* config/aarch64/aarch64.c (aarch64_add_constant): New parameter
"frame_related_p". Generate CFA annotation when it's necessary.
(aarch64_expand_prologue): Use aarch64_add_constant.
(aarch64_expand_epilogue): Likewise.
(aarch64_output_mi_thunk): Pass "false" when calling
aarch64_add_constant.
[AArch64][1/3] Migrate aarch64_add_constant to new interface & kill aarch64_build_constant
gcc/
* config/aarch64/aarch64.c (aarch64_add_constant): New parameter "mode".
Use aarch64_internal_mov_immediate instead of aarch64_build_constant.
(aarch64_output_mi_thunk): Pass Pmode when calling aarch64_add_constant.
(aarch64_build_constant): Delete.
Call get_ops just for SSA_NAMEs (PR tree-optimization/71987)
PR tree-optimization/71987
* tree-ssa-reassoc.c (maybe_optimize_range_tests): Call get_ops
just for SSA_NAMEs. Fix GNU coding style.
* gcc.dg/torture/pr71987.c: New test.
Use std::mt19937, std::thread and std::atomic to simplify tests
* testsuite/20_util/shared_ptr/thread/default_weaktoshared.cc: Use
std::mt19937, std::thread and std::atomic to simplify test.
* testsuite/20_util/shared_ptr/thread/mutex_weaktoshared.cc: Likewise.
The 2015-11-23 sanitizer merge from upstream lost the changes from f6528435 to sanitizer_common/sanitizer_common_interceptors.inc, which
made use of _OBSTACK_SIZE_T. So the configury changes to define
_OBSTACK_SIZE_T don't do anything. This wasn't such a bad thing
anyway.. The configure test wrongly adds -I${srcdir}/../include,
effectively resulting in a test of libiberty/obstack rather than libc
obstack support, and it's the latter that asan and tsan need to work
with.
So, remove the useless configure test. Upstream santizer project has
been made aware of the problem if glibc obstack support is ever
updated. Bootsrapped etc. x86_64-linux and committed as obvious.
PR middle-end/66726
* tree-ssa-reassoc.c (optimize_vec_cond_expr): Handle tcc_compare stmt
whose result is used in PHI.
(final_range_test_p): Likewise.
(maybe_optimize_range_tests): Likewise.
PR fortran/71935
* check.c (is_c_interoperable): Simplify right expression.
2016-07-22 Steven G. Kargl <kargl@gcc.gnu.org>
PR fortran/71935
* gfortran.dg/c_sizeof_1.f90: Move invalid code to ...
* gfortran.dg/c_sizeof_6.f90: here. Test for error.
* gfortran.dg/pr71935.f90: New test.
PR c++/71675 - __atomic_compare_exchange_n returns wrong type for typed enum
gcc/c-family/ChangeLog:
PR c++/71675
* c-common.c (resolve_overloaded_builtin): Avoid converting
__atomic_compare_exchange_n return type to that of what its
first argument points to.
gcc/testsuite/ChangeLog:
PR c++/71675
* g++.dg/ext/atomic-3.C: New test.
* gcc.dg/atomic/pr71675.c: New test.
* config/rs6000/rs6000.c (rs6000_option_override_internal): Add
comments to explain why certain error messages make mention of
undocumented options.
(rs6000_invalid_builtin): Change error messages to replace mention
of undocumented options with mention of the -mcpu=power9 option
that enables those undocumented options.
* config/rs6000/rs6000.h (MASK_FLOAT128): New macro.
(RS6000_BTM_FLOAT128): Use the new MASK_FLOAT128 macro in the
definition of this macro to correct an existing error.
* config/rs6000/rs6000.opt: Add the Undocumented qualifier to the
mpower9-fusion, mpower9-vector, mpower9-dform, and mmodulo entries.
* doc/extend.texi (PowerPC AltiVec Built-in Functions): Modify
descriptions of built-in functions so that they depend on
-mcpu=power9 instead of on the corresponding undocumented flags.
* doc/invoke.texi (Option Summary): Remove all mention of newly
undocumented flags.
(IBM RS/6000 and PowerPC Options): Likewise.
* doc/md.texi (Constraints for Particuliar Machines): Remove all
mention of newly undocumented flags.
gcc/
* ipa-cp.c (determine_versionability): Do not create constprop clones,
when target_clones attribute is set.
gcc/testsuite/
* gcc.target/i386/mvc8.c: New test.
* tree-ssa-loop-niter.h (number_of_iterations_exit_assumptions): New
Parameter.
* tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions): New
Parameter.
(number_of_iterations_exit): Warn missed loop optimization for
possible infinite loops.
gcc/testsuite
* gcc.dg/tree-ssa/pr19210-1.c: Refine test strings.
* gcc.dg/tree-ssa/pr19210-2.c: Delete.
Subject: [PATCH] rs6000: Fix logic for when to emit .machine (PR71216)
The current logic determining whether to use .machine in the generated
asm code puts it there if the compiler is not configured with a default
target cpu, _or_ no -mcpu= was given on the command line. It should
be "and" instead.
PR target/71216
* config/rs6000/rs6000.c (rs6000_file_start): Fix condition for
when to emit a ".machine" pseudo-op.
PR gcov-profile/69028
PR gcov-profile/62047
* g++.dg/cilk-plus/pr69028.C: New test.
PR gcov-profile/69028
PR gcov-profile/62047
* cilk.c (create_cilk_helper_decl): Set location of a new decl
to the current_function_decl.
PR gcov-profile/69028
PR gcov-profile/62047
* coverage.c (coverage_compute_lineno_checksum): Do not
calculate checksum for fns w/o xloc.file.
(coverage_compute_profile_id): Likewise.
* caf/libcaf.h: Add parameter stat to caf_get() and
caf_send()'s function prototypes.
* caf/single.c (_gfortran_caf_get): Implement reporting
error using stat instead of abort().
(_gfortran_caf_send): Same.
(_gfortran_caf_sendget): Use NULL for stat when calling
caf_send().
jason [Fri, 22 Jul 2016 03:45:30 +0000 (03:45 +0000)]
* call.c (build_over_call): Check unsafe_copy_elision_p even for
trivial constructors.
* method.c (do_build_copy_constructor): Don't copy tail padding
even in a trivial constructor.
ian [Fri, 22 Jul 2016 00:21:51 +0000 (00:21 +0000)]
compiler: fix check for duplicate declaration
The compiler check that issued a duplicate declaration error for
a, a, a := 1, 2, 3
was incorrectly issuing an error for
a, a, a = 1, 2, 3
While this is not particularly useful, it is valid Go.
Fix out-of-bound array accesses in libstdc++ testsuite
* testsuite/21_strings/basic_string/modifiers/append/char/1.cc: Fix
reads past the end of strings.
* testsuite/21_strings/basic_string/operations/compare/char/1.cc:
Likewise.
* testsuite/21_strings/char_traits/requirements/short/1.cc: Fix
invalid array accesses.
* testsuite/experimental/string_view/operations/compare/char/1.cc:
Fix read past the end of string.
* testsuite/util/testsuite_character.cc: Fix out-of-bounds write.
PR middle-end/71876
* builtin-attrs.def (ATTR_RT_NOTHROW_LEAF_LIST): New return twice
attribute.
* builtins.def (BUILT_IN_SETJMP): Use ATTR_RT_NOTHROW_LEAF_LIST here.
* calls.c (special_function_p): Remove the special handling of the
"__builtin_" prefix.
PR middle-end/71876
* calls.c (gimple_maybe_alloca_call_p): New function. Return true
if STMT may be an alloca call.
(gimple_alloca_call_p, alloca_call_p): Return only true for the
builtin alloca call.
* calls.h (gimple_maybe_alloca_call_p): New function.
* tree-inline.c (inline_forbidden_p_stmt): Use
gimple_maybe_alloca_call_p here.
gcc/ChangeLog:
* spellcheck-tree.c (best_macro_match::best_macro_match):
Explictly specify the template arguments when invoking the base
class constructor, to help older C++ compilers.
[Patch 2/2 ifcvt costs] Introduce a new cost model for ifcvt.
gcc/
* ifcvt.c (noce_if_info): New fields: speed_p, original_cost,
max_seq_cost. Removed fields: then_cost, else_cost, branch_cost.
(noce_conversion_profitable_p): New.
(noce_try_store_flag_constants): Use it.
(noce_try_addcc): Likewise.
(noce_try_store_flag_mask): Likewise.
(noce_try_cmove): Likewise.
(noce_try_cmove_arith): Likewise.
(bb_valid_for_noce_process_p): Add to the cost parameter rather than
overwriting it.
(noce_convert_multiple_sets): Move cost model to here, from...
(bb_ok_for_noce_convert_multiple_sets) ...here.
(noce_process_if_block): Update calls for above changes.
(noce_find_if_block): Record new noce_if_info parameters.
gcc/testsuite/
* gcc.dg/ifcvt-2.c: Use parameter to guide if-conversion heuristics.
* gcc.dg/ifcvt-3.c: Use parameter to guide if-conversion heuristics.
* gcc.dg/pr68435.c: Use parameter to guide if-conversion heuristics.
* gcc.dg/ifcvt-4.c: Use parameter to guide if-conversion heuristics.
* gcc.dg/ifcvt-5.c: Use parameter to guide if-conversion heuristics.
Some patterns for moving one bit around by BST + BLD.
* config/avr/avr.md (any_extract, any_shiftrt): New code iterators.
(*insv.extract, *insv.shiftrt, *insv.not-bit.0, *insv.not-bit.7)
(*insv.xor-extract, *insv.xor1-bit.0): New insns.
(adjust_len) [insv_notbit, insv_notbit_0, insv_notbit_7]: New
values for insn attribute.
* config/avr/avr.c (avr_out_insert_notbit): New function.
(avr_adjust_insn_length): Handle ADJUST_LEN_INSV_NOTBIT,
ADJUST_LEN_INSV_NOTBIT_0/_7.
* config/avr/avr-protos.h (avr_out_insert_notbit): New proto.
On ARM soft-float, the float to double conversion doesn't convert a sNaN
to qNaN as the IEEE Std 754 standard mandates:
"Under default exception handling, any operation signaling an invalid
operation exception and for which a floating-point result is to be
delivered shall deliver a quiet NaN."
Given the soft float ARM code ignores exceptions and always provides a
result, a float to double conversion of a signaling NaN should return a
quiet NaN. Fix this in extendsfdf2.
gcc/ChangeLog:
PR target/59833
* config/arm/ieee754-df.S (extendsfdf2): Convert sNaN to qNaN.
jason [Thu, 21 Jul 2016 06:05:39 +0000 (06:05 +0000)]
PR c++/65168 - -Waddress in unevaluated context.
gcc/c-family/
* c-common.c (c_common_truthvalue_conversion): Check
c_inhibit_evaluation_warnings for warning about address of
reference.
gcc/cp/
* typeck.c (cp_truthvalue_conversion): Compare pointers to nullptr.
Don't set c_inhibit_evaluation_warnings.
jason [Thu, 21 Jul 2016 06:05:24 +0000 (06:05 +0000)]
Improving concepts performance and diagnostics.
PR c++/67565
PR c++/67579
PR c++/71843
gcc/
* timevar.def (TV_CONSTRAINT_SAT, TV_CONSTRAINT_SUB): New time vars
for constraint satisfaction and subsumption.
* timevar.h (auto_timevar): New constructor that matches the push/pop
pattern of usage in pt.c.
gcc/cp/
* cp-tree.def (CHECK_CONSTR): New.
* cp-tree.h (CHECK_CONSTR_CONCEPT): New.
(CHECK_CONSTR_ARGS): New.
* constraint.cc (make_predicate_constraint): Remove in favor of
normalize_expression.
(resolve_constraint_check): Actually return error_mark_node when
resolution fails.
(resolve_variable_concept_check): Perform coercion as if processing
a template. Also return errors on resolution failure.
(lift_*): Remove all of these functions. Don't unnecessarily inline
concepts.
(learn_*): Add facilities to memoize implications for subsumption
during normalization.
(expanding_concept): New.
(expand_concept): New. Return the inlined and normalized definition
of a concept when needed.
(transform_*, xform_*): Rename to normalize_* to better reflect the
responsibility of those functions.
(normalize_template_id_expression): Check for non-boolean operands
when possible. Generate check constraints instead of normal variable
references.
(normalize_call_expression): Report errors when resolution fails.
(check_for_logical_overloads): Rewrite this check to more accurately
report the error.
(normalize_atom): Check for overloaded calls and invalid types before
determining if the expression refers to a concept.
(build_constraints): Don't cache normalized constraints or decmposed
assumptions.
(finish_shorthand_constraint): Return a normalized expression instead
of a predicate constraint.
(finish_template_introduction): Same.
(placeholder_extract_concept_and_args): Rewrite this since we only
ever get check constraints here.
(equivalent_placeholder_constraints): Rewrite in terms of check
constraints, and handle error_mark_nodes correctly.
(tsubst_check_constraint, tsubst_expr_constr, tsubst_type_constr)
(tsubst_implicit_conversion_constr)
(tsubst_argument_deduction_constr, tsubst_exception_constr)
(tsubst_parameterized_constraint, tsubst_constraint): New.
(tsbust_conjunection): Replace with tsubst_logical_operator and
actually generate the right kind of constraint.
(tsubst_requirement_body): Reverse the order of substituted arguments
so that they appear in the order written (helps diagnostics).
(satisfy_check_constraint): New.
(satisfy_conjunction): Simplify.
(satisfy_disjunction): Same.
(satisfy_constraint_1): Handle check constraints.
(eval_constr): New (private) global state.
(evaluating_constraints_sentinel): New. Manages eval_constr.
(satisfy_constraint): Add timing variables.
(satisfy_associated_constraints): Add hooks for memoization.
(evaluate_function_concept): Build a check constraint instead of
normalizing its definition.
(evaluate_variable_concept): Same.
(evaluate_constraint_expression): Normalize, but in the current
declaration processing context.
(evaluating_constraints_p): New.
(elide_constraint_failure_p): Actually emit constraint_thresh errors.
(diagnose_*): Remove artificial indentation. Add a new parameter to
each that tracks the current (complete) constraint prior to any
substitutions.
(diagnose_expression): Removed.
(diagnose_call_expression): Same.
(diagnose_template_id): Same.
(diagnose_template_id): New.
(diagnose_logical_constraint): New.
(diagnose_expression_constraint): Show the original expression.
(diagnose_type_constraint): Show the original type.
(diagnose_implicit_conversion_constraint): Be specific about
failures, don't re-diagnose a known-to-be-failed substitutions,
and manage elisions properly.
(diagnose_argument_deduction_constraint): Same.
(diagnose_exception_constraint): Same.
(diagnose_parameterized_constraint): Same.
(constraint_p): Allow EXPR_PACK_EXPANSION.
* logic.cc (next_by_distance): Removed. No longer used.
(any_p): Renamed from any_of.
(term_entry, term_hasher): New.
(term_list): Rewrite to include a hash table for quick lookup.
Also, make less stateful.
(proof_state): Extend to allow goals to be discharged once
satisfied.
(non_atomic_constraint_p): New.
(any_non_atomic_constraints_p): New.
(...rest...): Previous implementation completely replaced with an
iterative algorithm that opportunistically prunes the search space
before committing to using more memory.
* parser.c: (cp_parser_type_parameter): Normalize constraints.
(cp_parser_explicit_template_declaration): Same.
* pt.c: (finish_template_variable): Be less redundant with this error
message.
(template_args_equal): No longer static.
(tsubst_decl): Don't try to find specializations of variables that
have already been instantiated.
(build_non_dependent_expr): Avoid infinite recursion during concept
expansion.
(make_constrained_auto): Normalize constraints.
(do_auto_deduction): When doing auto deduction from a
partial-concept-id, be sure to include the explicit args checking
the constraints.
(constraint_sat_*): New. Memoize satisfied constraints.
(concept_spec_*): New. Memoize expressions associated with a concept
specialization.
(constraint_memos, concept_memos): New.
(lookup_constraint_satisfaction, memoize_constraint_satisfaction): New.
(lookup_concept_satisfaction, memoize_concept_satisfaction): New.
(get_concept_expansion, save_concept_expansion): New.
(hash_subsumption_args): New.
(comp_subsumption_args): New.
(subsumption_*): New. Memoize parts of the subsumption relation.
(lookup_subsumption_result, save_subsumption_result): New.
(init_constraint_processing): Initialize memo tables.
(get_constraints): Shortcut if !flag_concepts.
* decl.c (grokfndecl): Normalize constraints.
* error.c (dump_simple_decl): Print "concept" when appropriate.
(dump_function_decl): Same.
(dump_template_decl): Don't write requirements when we're not
printing the header.
(dump_expr): Handle fold expressions.
* cxx-pretty-print.c (cxx_pretty_printer::expression): Handle
fold expressions.
(get_fold_operator): New.
(pp_cxx_unary_left_fold_expression): New.
(pp_cxx_unary_right_fold_expression): New.
(pp_cxx_binary_fold_expression): New.
(pp_cxx_check_constraint): New.
(pp_cxx_*_constraint): Rewrite the grammar of internal constraints
to make them easier to read when debugging.
* search.c (accessible_p): Don't shortcut when evaluating constraints.
* tree.c (cp_tree_equal): Handle CHECK_CONSTR.
C++ FE: handle misspelled identifiers and typenames
gcc/cp/ChangeLog:
PR c/70339
PR c/71858
* name-lookup.c: Include gcc-rich-location.h, spellcheck-tree.h,
and parser.h.
(suggest_alternatives_for): If no candidates are found, try
lookup_name_fuzzy and report if if finds a suggestion.
(consider_binding_level): New function.
(lookup_name_fuzzy) New function.
* parser.c: Include gcc-rich-location.h.
(cp_lexer_next_token_is_decl_specifier_keyword): Move most of
logic into...
(cp_keyword_starts_decl_specifier_p): ...this new function.
(cp_parser_diagnose_invalid_type_name): When issuing
"does not name a type" errors, attempt to make a suggestion using
lookup_name_fuzzy.
* parser.h (cp_keyword_starts_decl_specifier_p): New prototype.
* search.c (lookup_field_fuzzy_info::fuzzy_lookup_field): Reject
types that are not CLASS_TYPE_P, rather than rejecting individual
tree codes.
gcc/testsuite/ChangeLog:
PR c/70339
PR c/71858
* g++.dg/spellcheck-identifiers.C: New test case, based on
gcc.dg/spellcheck-identifiers.c.
* g++.dg/spellcheck-identifiers-2.C: New test case, based on
gcc.dg/spellcheck-identifiers-2.c.
* g++.dg/spellcheck-typenames.C: New test case, based on
gcc.dg/spellcheck-typenames.c
S/390: Remove mode size check in encode_section_info.
With the last change the not-aligned symbol ref markers are always set
for modes with size zero. This is wrong since for larl the size of
the access does not matter. This patch removes that check entirely
from s390_encode_section_info. Modes with a size of 0 get rejected in
s390_check_symref_alignment which is used for the load/store relative
instructions to check for natural alignment.
Bootstrapped and regression tested on s390 and s390x with
--with-arch=z900 and --with-arch=z13.
gcc/ChangeLog:
2016-07-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
The attached patch rewrites the pr67443.c testcase in a different way
so that the test still works with the changed allocation of globals
pinned to registers. The test ist hopefully more robust now.
projects / thirdparty / gcc.git / log
