[LLDB][NativePDB] Use undecorated name for types if UniqueName isn't mangled #152114
Conversation
|
@llvm/pr-subscribers-lldb Author: nerix (Nerixyz) ChangesLanguages other than C/C++ don't necessarily emit mangled names in the For example, !266 = !DICompositeType(
tag: DW_TAG_structure_type, name: "tuple$<i32,i32>", file: !9, size: 64, align: 32,
elements: !267, templateParams: !17, identifier: "19122721b0632fe96c0dd37477674472"
)which results in In C++ with Clang and MSVC, a structure similar to this would result in With this PR, if a I'm not sure how to test this. Maybe compiling the LLVM IR that rustc emits? Fixes #152051. Full diff: https://github.com/llvm/llvm-project/pull/152114.diff 2 Files Affected:
diff --git a/lldb/source/Plugins/SymbolFile/NativePDB/PdbAstBuilder.cpp b/lldb/source/Plugins/SymbolFile/NativePDB/PdbAstBuilder.cpp
index 8137622842253..36950c00f9e7f 100644
--- a/lldb/source/Plugins/SymbolFile/NativePDB/PdbAstBuilder.cpp
+++ b/lldb/source/Plugins/SymbolFile/NativePDB/PdbAstBuilder.cpp
@@ -178,7 +178,7 @@ PdbAstBuilder::CreateDeclInfoForType(const TagRecord &record, TypeIndex ti) {
std::string_view sv(record.UniqueName.begin(), record.UniqueName.size());
llvm::ms_demangle::TagTypeNode *ttn = demangler.parseTagUniqueName(sv);
if (demangler.Error)
- return {m_clang.GetTranslationUnitDecl(), std::string(record.UniqueName)};
+ return CreateDeclInfoForUndecoratedName(record.Name);
llvm::ms_demangle::IdentifierNode *idn =
ttn->QualifiedName->getUnqualifiedIdentifier();
diff --git a/lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp b/lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp
index 7af53e16ce9d5..24b80c13a10d8 100644
--- a/lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp
+++ b/lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp
@@ -618,18 +618,14 @@ lldb::TypeSP SymbolFileNativePDB::CreateSimpleType(TypeIndex ti,
}
static std::string GetUnqualifiedTypeName(const TagRecord &record) {
- if (!record.hasUniqueName()) {
- MSVCUndecoratedNameParser parser(record.Name);
- llvm::ArrayRef<MSVCUndecoratedNameSpecifier> specs = parser.GetSpecifiers();
-
- return std::string(specs.back().GetBaseName());
- }
+ if (!record.hasUniqueName())
+ return std::string(MSVCUndecoratedNameParser::DropScope(record.Name));
llvm::ms_demangle::Demangler demangler;
std::string_view sv(record.UniqueName.begin(), record.UniqueName.size());
llvm::ms_demangle::TagTypeNode *ttn = demangler.parseTagUniqueName(sv);
if (demangler.Error)
- return std::string(record.Name);
+ return std::string(MSVCUndecoratedNameParser::DropScope(record.Name));
llvm::ms_demangle::IdentifierNode *idn =
ttn->QualifiedName->getUnqualifiedIdentifier();
|
There was a problem hiding this comment.
Seems reasonable to me. I'd let @ZequanWu or @Walnut356 take a look too
| return std::string(specs.back().GetBaseName()); | ||
| } | ||
| if (!record.hasUniqueName()) | ||
| return std::string(MSVCUndecoratedNameParser::DropScope(record.Name)); |
There was a problem hiding this comment.
Was this change required? Or a drive-by cleanup?
There was a problem hiding this comment.
No, this isn't required. This was basically the implementation of DropScope before. I changed it for clarity - to be identical with the other case.
|
Can you add a test for this if it's not hard to create one? |
Found a way to make the IR a bit smaller by compiling an rlib with a |
Languages other than C/C++ don't necessarily emit mangled names in the
UniqueNamefield of type records. Rust specifically emits a unique ID that doesn't contain the name.For example,
(i32, i32)is emitted aswhich results in
In C++ with Clang and MSVC, a structure similar to this would result in
With this PR, if a
UniqueNameis encountered that couldn't be parsed, it will fall back to using the undecorated (→ do the same as if the unique name is empty/unavailable).I'm not sure how to test this. Maybe compiling the LLVM IR that rustc emits?
Fixes #152051.