[PATCH] D109703: [DebugInfo] Fix scope for local static variables
Kristina Bessonova via Phabricator via cfe-commits
cfe-commits at lists.llvm.org
Mon Sep 13 13:02:48 PDT 2021
krisb added a comment.
In D109703#2998155 <https://reviews.llvm.org/D109703#2998155>, @ellis wrote:
> I've added this to the added test case.
>
> !18 = !DIImportedEntity(tag: DW_TAG_imported_declaration, scope: !4, entity: !19, line: 122)
> !19 = !DIGlobalVariable(name: "imported_static_var", scope: !6, file: !5, line: 3, type: !10, isLocal: false, isDefinition: true)
Thank you for pointing to this! Just in case, do you have a C++ example, that reproduces the issue with imported static local declaration? I'm asking, cause in your example `DW_TAG_imported_declaration` has CU as a scope, while `DIGlobalVariable` is function-scoped. This looks a bit strange. I mean, `constructImportedEntityDIE()` called from `DwarfDebug::beginModule()` skips all local-scoped entities, and unless I'm missing something static locals shouldn't go here. Another call for `constructImportedEntityDIE()` is from `DwarfCompileUnit::createScopeChildrenDIE` where I'd not expect any issue with parents.
But it seems imported declarations are broken not just for static locals, but for all inlined entities, for example
namespace ns {
inline __attribute__((always_inline))
int foo() {
int a = 42;
return a;
}
}
int main() {
using ns::foo;
return foo();
}
produces (with or w/o this patch) imported declaration for `foo()` that refers to an empty subprogram:
x0000002a: DW_TAG_namespace
DW_AT_name ("ns")
0x0000002f: DW_TAG_subprogram
DW_AT_name ("foo")
DW_AT_inline (DW_INL_inlined)
0x0000003f: DW_TAG_variable
DW_AT_name ("a")
0x0000004a: NULL
0x0000004b: DW_TAG_subprogram
0x0000004c: NULL
0x00000054: DW_TAG_subprogram
DW_AT_name ("main")
0x0000006d: DW_TAG_imported_declaration
DW_AT_import (0x0000004b)
while it should point to `0x0000002f`.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D109703/new/
https://reviews.llvm.org/D109703
More information about the cfe-commits
mailing list