Jason Merrill [Mon, 1 Apr 2019 20:43:13 +0000 (16:43 -0400)]
PR c++/86946 - ICE with function call in template argument.
DR 1321 clarified that two dependent names are equivalent if the names are
the same, even if the result of name lookup is different. We need to
implement that in hashing like we already do in comparison and mangling.
* pt.c (iterative_hash_template_arg) [CALL_EXPR]: Use
dependent_name.
Use single-visitation in variant assignment and swap and relops.
Also use indices instead of types when checking whether
variants hold the same thing.
* include/std/variant (__do_visit): Add a template parameter
for index visitation, invoke with indices if index visitation
is used.
(__variant_idx_cookie): New.
(__visit_with_index): Likewise.
(_Copy_assign_base::operator=): Do single-visitation with
an index visitor.
(_Move_assign_base::operator=): Likewise.
(_Extra_visit_slot_needed): Adjust.
(__visit_invoke): Call with indices if it's an index visitor.
(relops): Do single-visitation with an index visitor.
(swap): Likewise.
(__visitor_result_type): New.
Martin Liska [Mon, 1 Apr 2019 07:17:38 +0000 (09:17 +0200)]
Enhance option suggestion for options expected an argument (PR driver/89861).
2019-04-01 Martin Liska <mliska@suse.cz>
PR driver/89861
* opt-suggestions.c (option_proposer::build_option_suggestions):
Add variant without any argument in order to provide better
hints.
2019-04-01 Martin Liska <mliska@suse.cz>
PR driver/89861
* gcc.dg/spellcheck-options-18.c: New test.
* gcc.dg/spellcheck-options-19.c: New test.
* gcc.dg/spellcheck-options-20.c: New test.
* gcc.dg/spellcheck-options-13.c: Adjust expected output.
* gcc.dg/completion-2.c: Add one variant with no argument.
The D2 testsuite script has been updated to handle EXTRA_SOURCES and
EXTRA_FILES settings being split across multiple lines, which is how
they appear in upstream.
* gdc.test/gdc-test.exp (gdc-copy-extra): Append copied files to
cleanup_extra_files.
(dmd2dg): Copy additional files after test is processed.
(gdc-do-test): Remove all copied files after test.
Paul Thomas [Sat, 30 Mar 2019 15:39:00 +0000 (15:39 +0000)]
re PR fortran/89841 (improper descriptor information passed to C)
2019-03-30 Paul Thomas <pault@gcc.gnu.org>
PR fortran/89841
* trans-expr.c (gfc_conv_gfc_desc_to_cfi_desc): Use the formal
argument attributes rather than those of the actual argument.
PR fortran/89842
* trans-expr.c (gfc_conv_gfc_desc_to_cfi_desc): Call
'set_dtype_for_unallocated' for any type of arrayspec.
2019-03-30 Paul Thomas <pault@gcc.gnu.org>
PR fortran/89841
* gfortran.dg/ISO_Fortran_binding_1.f90: Change the interfaces
for c_deallocate, c_allocate and c_assumed_size so that the
attributes of the array arguments are correct and are typed.
* gfortran.dg/ISO_Fortran_binding_7.f90: New test.
* gfortran.dg/ISO_Fortran_binding_7.c: Additional source.
PR fortran/89842
* gfortran.dg/ISO_Fortran_binding_8.f90: New test.
* gfortran.dg/ISO_Fortran_binding_8.c: Additional source.
Jason Merrill [Sat, 30 Mar 2019 15:23:37 +0000 (11:23 -0400)]
PR c++/89744 - ICE with specialization of member class template.
My fix five years ago for PR 60241 was incomplete: when we reassign implicit
instances of a partial instantiation of a member template to the explicit
specialization of that partial instantiation, we also need to adjust the
CLASSTYPE_TI_ARGS to match what we'd get when looking up that instance after
the explicit specialization. We also need to do this when we later look up
the instance in a way that only finds the explicit specialization halfway
through lookup_template_class_1.
* pt.c (lookup_template_class_1): If the partial instantiation is
explicitly specialized, adjust.
(maybe_process_partial_specialization): Also adjust
CLASSTYPE_TI_ARGS.
re PR rtl-optimization/89862 (LTO bootstrap fails for ARM)
2019-03-29 Kugan Vivekanandarajah <kuganv@linaro.org>
Eric Botcazou <ebotcazou@adacore.com>
PR rtl-optimization/89862
* rtl.h (word_register_operation_p): Exclude CONST_INT from operations
that operates on the full registers for WORD_REGISTER_OPERATIONS
architectures.
Co-Authored-By: Eric Botcazou <ebotcazou@adacore.com>
From-SVN: r270030
Jim Wilson [Fri, 29 Mar 2019 21:51:58 +0000 (21:51 +0000)]
RISC-V: Fix __riscv_compressed regression.
This was accidentally broken by the patch adding ELF attribute support.
The __riscv_compressed predefined macro is not being defined when the
target supports the C extension. The fix adds testcases to verify that
all of the predefined macros are correct for various option combinations.
gcc/
* common/config/riscv/riscv-common.c (riscv_parse_arch_string):
Clear MASK_RVC and then set if C subset supported.
Jakub Jelinek [Fri, 29 Mar 2019 20:51:15 +0000 (21:51 +0100)]
re PR rtl-optimization/89865 (FAIL: gcc.target/i386/pr49095.c scan-assembler-times \\\\), % 45)
PR rtl-optimization/89865
* gcc.target/i386/pr49095.c: Include in scan-assembler-times patterns
the first argument register, so that occassional spills/fills are
ignored.
Jakub Jelinek [Fri, 29 Mar 2019 20:10:19 +0000 (21:10 +0100)]
re PR sanitizer/89869 (-fsanitize=undefined miscompilation)
PR sanitizer/89869
* typeck.c: Include gimplify.h.
(cp_build_modify_expr) <case COND_EXPR>: Unshare rhs before using it
for second time. Formatting fixes.
Martin Liska [Fri, 29 Mar 2019 11:45:39 +0000 (12:45 +0100)]
Fix multiple values for -fdbg-cnt.
2019-03-29 Martin Liska <mliska@suse.cz>
* dbgcnt.c (dbg_cnt_process_single_pair): Fix GNU coding style.
(dbg_cnt_process_opt): Parse first tokens aas
dbg_cnt_process_single_pair is also using strtok.
Jakub Jelinek [Thu, 28 Mar 2019 22:33:29 +0000 (23:33 +0100)]
re PR middle-end/89621 (ICE with allocatable character and openmp)
PR middle-end/89621
* tree-inline.h (struct copy_body_data): Add
dont_remap_vla_if_no_change flag.
* tree-inline.c (remap_type_3, remap_type_2): New functions.
(remap_type): Don't remap vla types if id->dont_remap_vla_if_no_change
and remap_type_2 returns false.
* omp-low.c (new_omp_context): Set ctx->cb.dont_remap_vla_if_no_change.
Move ctx->cb.adjust_array_error_bounds setting to the outermost ctx
only from where it is copied to nested contexts.
Marek Polacek [Thu, 28 Mar 2019 20:24:48 +0000 (20:24 +0000)]
PR c++/89612 - ICE with member friend template with noexcept.
* pt.c (maybe_instantiate_noexcept): For function templates, use their
template result (function decl). Don't set up local specializations.
Temporarily turn on processing_template_decl. Update the template type
too.
* g++.dg/cpp0x/noexcept38.C: New test.
* g++.dg/cpp0x/noexcept39.C: New test.
* g++.dg/cpp1z/noexcept-type21.C: New test.
Jakub Jelinek [Thu, 28 Mar 2019 14:47:47 +0000 (15:47 +0100)]
re PR c++/89785 (Incorrect "not a constant expression" error with switch statement that returns)
PR c++/89785
* constexpr.c (struct check_for_return_continue_data): New type.
(check_for_return_continue): New function.
(potential_constant_expression_1) <case SWITCH_STMT>: Walk
SWITCH_STMT_BODY to find RETURN_EXPRs or CONTINUE_STMTs not nested
in loop bodies and set *jump_target to that if found.
* g++.dg/cpp1y/constexpr-89785-1.C: New test.
* g++.dg/cpp1y/constexpr-89785-2.C: New test.
optrecord_json_writer::optinfo_to_json can in theory be called from any
optimization pass, but currently uses get_fnname_from_decl, which
is RTL-specific.
In that PR, Jakub suggested using either DECL_ASSEMBLER_NAME or the
"printable name" (via current_function_name).
This patch makes it use DECL_ASSEMBLER_NAME.
gcc/ChangeLog:
PR middle-end/89725
* optinfo-emit-json.cc (optrecord_json_writer::optinfo_to_json):
Use DECL_ASSEMBLER_NAME rather than get_fnname_from_decl.
* regcprop.c (copyprop_hardreg_forward_1): Remove redundant INSN_P
test.
(cprop_hardreg_bb, cprop_hardreg_debug): New functions.
(pass_cprop_hardreg::execute): Use those. Don't repeat bb processing
immediately after first one with df_analyze in between, but rather
process all bbs, queueing ones that need second pass in a worklist,
df_analyze, process queued debug insn changes and if second pass is
needed, process bbs from worklist, df_analyze, process queued debug
insns again.
Jakub Jelinek [Thu, 28 Mar 2019 14:07:40 +0000 (15:07 +0100)]
rtl.h (NONDEBUG_INSN_P): Define as NONJUMP_INSN_P or JUMP_P or CALL_P instead of INSN_P && !DEBUG_INSN_P.
* rtl.h (NONDEBUG_INSN_P): Define as NONJUMP_INSN_P or JUMP_P
or CALL_P instead of INSN_P && !DEBUG_INSN_P.
(INSN_P): Define using NONDEBUG_INSN_P or DEBUG_INSN_P.
Jason Merrill [Wed, 27 Mar 2019 20:39:19 +0000 (16:39 -0400)]
PR c++/89831 - error with qualified-id in const member function.
Since the fix for 15272 we were remembering the wrong function to use at
instantiation time, because the type of the SCOPE_REF didn't reflect the
cv-quals of 'this'. Conveniently, we can fix this by simplifying the code.
* semantics.c (finish_non_static_data_member): Use object cv-quals
in scoped case, too.
Jason Merrill [Wed, 27 Mar 2019 18:39:20 +0000 (14:39 -0400)]
PR c++/89241 - ICE with lambda in template parameter list.
We were getting confused by a lambda in template definition context that
isn't actually in the scope of any templated entity. Fixed by telling
type_dependent_expression_p that such a lambda is type-dependent even if we
can't tell that from its closure type. I've also restored the error for
defining a non-lambda class in a default template argument, and for a lambda
befor C++20.
* parser.c (cp_parser_lambda_expression): Also reject a lambda in a
template parameter list before C++20.
* pt.c (type_dependent_expression_p): True for LAMBDA_EXPR.
* semantics.c (begin_class_definition): Restore error about defining
non-lambda class in template parm list.
Peter Bergner [Wed, 27 Mar 2019 16:59:15 +0000 (16:59 +0000)]
re PR rtl-optimization/89313 (ICE in process_alt_operands, at lra-constraints.c:2962)
gcc/
PR rtl-optimization/89313
* function.c (matching_constraint_num): New static function.
(match_asm_constraints_1): Use it. Fixup white space and comment.
Don't replace inputs with non-matching constraints which conflict
with early clobber outputs.
gcc/testsuite/
PR rtl-optimization/89313
* gcc.dg/pr89313.c: New test.
Jeff Law [Wed, 27 Mar 2019 16:18:06 +0000 (10:18 -0600)]
re PR rtl-optimization/87761 ([MIPS] New FAIL: gcc.target/mips/fix-r4000-10.c -O1 start with r265398)
PR rtl-optimization/87761
PR rtl-optimization/89826
* regcprop.c (copyprop_hardreg_forward_1): Move may_trap_p test
slightly later.
(pass_cprop_hardreg::execute): Call df_analyze after adding the
note problem to get REG_DEAD/REG_UNUSED notes updated.
PR rtl-optimization/87761
PR rtl-optimization/89826
* gcc.c-torture/execute/pr89826.c: New test.
Jason Merrill [Wed, 27 Mar 2019 14:27:00 +0000 (10:27 -0400)]
PR c++/86932 - missed SFINAE with empty pack.
The issue here was that when processing the explicit template args in
fn_type_unification we added an empty argument pack for the parameter pack,
so we never tried to do any deduction for it, and therefore never looked at
its type. We need that empty pack behavior for partial ordering, but we
don't want it here, so let's make it conditional on tf_partial.
* pt.c (coerce_template_parms): Don't add an empty pack if
tf_partial.
(fn_type_unification): Pass tf_partial to coerce_template_parms.
Paul Thomas [Wed, 27 Mar 2019 12:51:43 +0000 (12:51 +0000)]
re PR fortran/88247 (ICE in get_array_ctor_var_strlen, at fortran/trans-array.c:2068)
2019-03-27 Paul Thomas <pault@gcc.gnu.org>
PR fortran/88247
* expr.c (is_subref_array): Permit substrings to be detected
as subref arrays.
* trans-array.c (get_array_ctor_var_strlen): Obtain the length
of deferred length strings. Handle substrings with a NULL end
expression.
(trans_array_constructor): Remove an unnecessary blank line.
(gfc_conv_scalarized_array_ref): Skip to label 'done' if 'decl'
is a pointer array.
(get_array_charlen): If the expression is an array, convert the
first element of the constructor and use its string length. Get
a new charlen if necessary.
(gfc_conv_expr_descriptor): Call 'get_array_charlen' for array
constructor expressions. If the ss_info string length is
available, use that to set the span of character arrays.
* trans-expr.c (gfc_get_expr_charlen): Handle substrings
* trans-stmt.c (trans_associate_var): Set the pointer array
flag for variable targets and constant array constructors. Take
care not to reset the string length or the span in the case of
expressions that are not converted as direct by reference.
2019-03-27 Paul Thomas <pault@gcc.gnu.org>
PR fortran/88247
* gfortran.dg/associate_47.f90: New test.
Richard Biener [Wed, 27 Mar 2019 09:31:53 +0000 (09:31 +0000)]
re PR debug/89463 (debug information for iterator of an empty loop is gone (at -O3))
2019-03-27 Richard Biener <rguenther@suse.de>
PR tree-optimization/89463
* tree-ssa-dce.c (remove_dead_stmt): Take output vector to
queue edges to remove.
(eliminate_unnecessary_stmts): Remove dead PHIs alongside
dead stmts. Delay edge removal until PHIs are removed to
make debug-stmt creation not confused by seemingly degenerate
PHIs.
Jason Merrill [Tue, 26 Mar 2019 16:02:19 +0000 (12:02 -0400)]
PR c++/86429 - constexpr variable in lambda.
When we refer to a captured variable from a constant-expression context
inside a lambda, the closure (like any function parameter) is not constant
because we aren't in a call, so we don't have an argument. So the capture
is non-constant. But if the captured variable is constant, we might be able
to use it directly in constexpr evaluation.
PR c++/82643
PR c++/87327
* constexpr.c (cxx_eval_constant_expression): In a lambda function,
try evaluating the captured variable directly.
Jonathan Wakely [Tue, 26 Mar 2019 15:28:48 +0000 (15:28 +0000)]
PR libstdc++/85965 delay static assertions until types are complete
The static assertions added for PR libstdc++/48101 were at class scope
and so were evaluated too eagerly, when it might not be possible to
determine whether the function objects are invocable with the key types.
The problematic cases are where the key type is not known to be
convertible to the argument type(s) of the function object until later,
after a type has been completed. Specifically, if the key type is a
pointer to a derived class and the function object's argument type is a
pointer to a base class, then the derived-to-base conversion is only
valid once the derived type is complete.
By moving the static assertions to the destructor they will only be
evaluated when the destructor is instantiated, at which point whether
the key type can be passed to the function object should be knowable.
The ideal place to do the checks would be only when the function objects
are actually invoked, but that would mean adding the checks in numerous
places, so the destructor is used instead.
The tests need to be adjusted because the "required from here" line is
now the location of the destructor, not the point of instantiation in
the test file. For the map and multimap tests which check two
specializations, the dg-error matching the assertion text matches both
cases. Also check the diagnostic output for the template arguments, to
ensure both specializations trigger the assertion.
PR libstdc++/85965
* include/bits/hashtable.h (_Hashtable): Move static assertions to
destructor so they are not evaluated until the _Key type is complete.
* include/bits/stl_tree.h (_Rb_tree): Likewise.
* testsuite/23_containers/set/85965.cc: New test.
* testsuite/23_containers/unordered_set/85965.cc: New test.
* testsuite/23_containers/map/48101_neg.cc: Replace "here" errors
with regexp matching the corresponding _Rb_tree specialization.
* testsuite/23_containers/multimap/48101_neg.cc: Likewise.
* testsuite/23_containers/multiset/48101_neg.cc: Remove "here" error.
* testsuite/23_containers/set/48101_neg.cc: Likewise.
* testsuite/23_containers/unordered_map/48101_neg.cc: Likewise.
* testsuite/23_containers/unordered_multimap/48101_neg.cc: Likewise.
* testsuite/23_containers/unordered_multiset/48101_neg.cc: Likewise.
* testsuite/23_containers/unordered_set/48101_neg.cc: Likewise.
re PR libstdc++/89825 (Jump table for variant visitation could be shortened for never empty variants)
PR libstdc++/89825
Fix based on a suggestion by Antony Polukhin.
* include/std/variant (_Extra_visit_slot_needed): New.
(_Multi_array): Use it.
(_S_apply_all_alts): Likewise.
Jakub Jelinek [Tue, 26 Mar 2019 13:40:43 +0000 (14:40 +0100)]
hash-table.h (hash_table::m_gather_mem_stats): If GATHER_STATISTICS is constant 0...
* hash-table.h (hash_table::m_gather_mem_stats): If GATHER_STATISTICS
is constant 0, turn into static const data member initialized to false.
(hash_table::hash_table): Only initialize m_gather_mem_stats #if
GATHER_STATISTICS. Add ATTRIBUTE_UNUSED to gather_mem_stats param.
re PR libstdc++/89824 (Variant jump table reserves space for __variant_cookie twice)
PR libstdc++/89824
Fix based on a suggestion by Antony Polukhin.
* include/std/variant (__gen_vtable): Don't reserve an
additional table slot, _Multi_array already does that.
re PR libstdc++/89816 (std::variant move construction regressed since GCC 8.3)
PR libstdc++/89816
Fix based on a suggestion by Antony Polukhin.
* include/std/variant (__variant_construct): Capture a pointer
to the storage and visit just one variant.
Bin Cheng [Tue, 26 Mar 2019 11:12:37 +0000 (11:12 +0000)]
re PR tree-optimization/81740 (wrong code at -O3 in both 32-bit and 64-bit modes on x86_64-linux-gnu)
2019-03-26 Bin Cheng <bin.cheng@linux.alibaba.com>
PR tree-optimization/81740
* tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
In case of outer loop vectorization, check for backward dependence
at the inner loop if outer loop dependence is reversed.
* gcc.dg/vect/pr81740-1.c: New testcase.
* gcc.dg/vect/pr81740-2.c: Likewise.
Jeff Law [Mon, 25 Mar 2019 23:33:36 +0000 (17:33 -0600)]
re PR rtl-optimization/88347 (ICE in begin_move_insn, at sched-ebb.c:175)
PR rtl-optimization/88347
PR rtl-optimization/88423
* sched-deps.c (sched_analyze_insn): Take into account that for
tablejumps the barrier appears after a label and a jump_table_data.
* gcc.c-torture/compile/pr88347.c: New test.
* gcc.c-torture/compile/pr88423.c: New test.
Martin Sebor [Mon, 25 Mar 2019 22:56:40 +0000 (22:56 +0000)]
PR c/89812 - incorrect maximum in error: requested alignment '536870912' exceeds maximum 2147483648
gcc/c-family/ChangeLog:
PR c/89812
* c-common.c (check_user_alignment): Rename local. Correct maximum
alignment in diagnostic. Avoid assuming argument fits in SHWI,
convert it to UHWI when it fits.
Johan Karlsson [Mon, 25 Mar 2019 21:19:09 +0000 (21:19 +0000)]
re PR debug/86964 (Too many debug symbols included, especially for extern globals)
PR debug/86964
* dwarf2out.c (premark_used_variables): New function.
(prune_unused_types_walk): Do not mark not premarked external
variables.
(prune_unused_types): Call premark_used_variables.
Vladimir Makarov [Mon, 25 Mar 2019 21:14:40 +0000 (21:14 +0000)]
re PR rtl-optimization/89676 (Redundant moves for long long shift on 32bit x86)
2019-03-25 Vladimir Makarov <vmakarov@redhat.com>
PR rtl-optimization/89676
* lra-constraints.c (curr_insn_transform): Do match reload for
early clobbers when the match was successful only for different
registers.
projects / thirdparty / gcc.git / log
