Skip to content

[LLDB][NativePDB] Find functions by basename #152295

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 10 commits into from
Aug 12, 2025
Merged

[LLDB][NativePDB] Find functions by basename #152295

Show file tree
Hide file tree
Changes from 4 commits
Commits
Show all changes
10 commits
Select commit Hold shift + click to select a range
b6886d6
[LLDB][NativePDB] Find functions by basename
Nerixyz Aug 6, 2025
f342ea1
fix: use return value of `insert()`
Nerixyz Aug 7, 2025
24808f9
fix: check for nullptr
Nerixyz Aug 7, 2025
b88192f
fix: comments for class fields
Nerixyz Aug 7, 2025
096b814
Update lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp
Nerixyz Aug 8, 2025
0ee5016
Update lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp
Nerixyz Aug 8, 2025
8be4fc6
Update lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp
Nerixyz Aug 8, 2025
75179a8
Update lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp
Nerixyz Aug 8, 2025
64b2f5d
Update lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp
Nerixyz Aug 8, 2025
1cc0ccc
tests: add methods/functions with existing basename
Nerixyz Aug 8, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
153 changes: 134 additions & 19 deletions lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -39,6 +39,7 @@
#include "llvm/DebugInfo/PDB/Native/ModuleDebugStream.h"
#include "llvm/DebugInfo/PDB/Native/NativeSession.h"
#include "llvm/DebugInfo/PDB/Native/PDBFile.h"
#include "llvm/DebugInfo/PDB/Native/PublicsStream.h"
#include "llvm/DebugInfo/PDB/Native/SymbolStream.h"
#include "llvm/DebugInfo/PDB/Native/TpiStream.h"
#include "llvm/DebugInfo/PDB/PDB.h"
Expand Down Expand Up @@ -1641,6 +1642,94 @@ void SymbolFileNativePDB::DumpClangAST(Stream &s, llvm::StringRef filter) {
clang->GetNativePDBParser()->Dump(s, filter);
}

void SymbolFileNativePDB::CacheFunctionNames() {
if (!m_func_full_names.IsEmpty())
return;

// (segment, code offset) -> gid
std::map<std::pair<uint16_t, uint32_t>, uint32_t> addr_ids;

// First, find all function references in the globals table.
for (const uint32_t gid : m_index->globals().getGlobalsTable()) {
CVSymbol ref_sym = m_index->symrecords().readRecord(gid);
auto kind = ref_sym.kind();
if (kind != S_PROCREF && kind != S_LPROCREF)
continue;

ProcRefSym ref =
llvm::cantFail(SymbolDeserializer::deserializeAs<ProcRefSym>(ref_sym));
if (ref.Name.empty())
continue;

// Find the function this is referencing
CompilandIndexItem &cci =
m_index->compilands().GetOrCreateCompiland(ref.modi());
auto iter = cci.m_debug_stream.getSymbolArray().at(ref.SymOffset);
if (iter == cci.m_debug_stream.getSymbolArray().end())
continue;
kind = iter->kind();
if (kind != S_GPROC32 && kind != S_LPROC32)
continue;

ProcSym proc =
llvm::cantFail(SymbolDeserializer::deserializeAs<ProcSym>(*iter));
if ((proc.Flags & ProcSymFlags::IsUnreachable) != ProcSymFlags::None)
continue;
if (proc.Name.empty())
continue;

// The function/procedure symbol only contains the demangled name.
// The mangled names are in the publics table. Save the address of this
// function to lookup the mangled name later.
addr_ids.emplace(std::make_pair(proc.Segment, proc.CodeOffset), gid);

llvm::StringRef basename = MSVCUndecoratedNameParser::DropScope(proc.Name);
if (basename.empty())
basename = proc.Name;

m_func_base_names.Append(ConstString(basename), gid);
m_func_full_names.Append(ConstString(proc.Name), gid);

// To see if this is a member function, check the type
auto type = m_index->tpi().getType(proc.FunctionType);
if (type.kind() == LF_MFUNCTION) {
MemberFunctionRecord mfr;
llvm::cantFail(
TypeDeserializer::deserializeAs<MemberFunctionRecord>(type, mfr));
if (!mfr.getThisType().isNoneType())
m_func_method_names.Append(ConstString(basename), gid);
}
}

// The publics stream contains all mangled function names and their address.
for (auto pid : m_index->publics().getPublicsTable()) {
PdbGlobalSymId global{pid, true};
CVSymbol sym = m_index->ReadSymbolRecord(global);
auto kind = sym.kind();
if (kind != S_PUB32)
continue;
PublicSym32 pub =
llvm::cantFail(SymbolDeserializer::deserializeAs<PublicSym32>(sym));
// We only care about mangled names - if the name isn't mangled, it's
// already in the full name map
if (!Mangled::IsMangledName(pub.Name))
continue;

// Check if this symbol is for one of our functions
auto it = addr_ids.find({pub.Segment, pub.Offset});
if (it != addr_ids.end())
m_func_full_names.Append(ConstString(pub.Name), it->second);
}

// Sort them before value searching is working properly
m_func_full_names.Sort();
m_func_full_names.SizeToFit();
m_func_method_names.Sort();
m_func_method_names.SizeToFit();
m_func_base_names.Sort();
m_func_base_names.SizeToFit();
}

void SymbolFileNativePDB::FindGlobalVariables(
ConstString name, const CompilerDeclContext &parent_decl_ctx,
uint32_t max_matches, VariableList &variables) {
Expand Down Expand Up @@ -1677,34 +1766,60 @@ void SymbolFileNativePDB::FindFunctions(
if (name_type_mask & eFunctionNameTypeFull)
name = lookup_info.GetName();

// For now we only support lookup by method name or full name.
if (!(name_type_mask & eFunctionNameTypeFull ||
name_type_mask & eFunctionNameTypeBase ||
name_type_mask & eFunctionNameTypeMethod))
return;
CacheFunctionNames();

using SymbolAndOffset = std::pair<uint32_t, llvm::codeview::CVSymbol>;
std::set<uint32_t> resolved_ids; // avoid duplicate lookups
auto resolve_from = [&](UniqueCStringMap<uint32_t> &Names) {
std::vector<uint32_t> ids;
if (!Names.GetValues(name, ids))
return;

std::vector<SymbolAndOffset> matches = m_index->globals().findRecordsByName(
name.GetStringRef(), m_index->symrecords());
for (const SymbolAndOffset &match : matches) {
if (match.second.kind() != S_PROCREF && match.second.kind() != S_LPROCREF)
continue;
ProcRefSym proc(match.second.kind());
cantFail(SymbolDeserializer::deserializeAs<ProcRefSym>(match.second, proc));
for (uint32_t id : ids) {
if (!resolved_ids.insert(id).second)
continue;

if (!IsValidRecord(proc))
continue;
PdbGlobalSymId global{id, false};
if (parent_decl_ctx.IsValid() &&
GetDeclContextContainingUID(toOpaqueUid(global)) != parent_decl_ctx)
continue;

CompilandIndexItem &cci =
m_index->compilands().GetOrCreateCompiland(proc.modi());
SymbolContext sc;
CVSymbol sym = m_index->ReadSymbolRecord(global);
auto kind = sym.kind();
lldbassert(kind == S_PROCREF || kind == S_LPROCREF);

sc.comp_unit = GetOrCreateCompileUnit(cci).get();
PdbCompilandSymId func_id(proc.modi(), proc.SymOffset);
sc.function = GetOrCreateFunction(func_id, *sc.comp_unit).get();
ProcRefSym proc =
cantFail(SymbolDeserializer::deserializeAs<ProcRefSym>(sym));

sc_list.Append(sc);
}
if (!IsValidRecord(proc))
continue;

CompilandIndexItem &cci =
m_index->compilands().GetOrCreateCompiland(proc.modi());
SymbolContext sc;

sc.comp_unit = GetOrCreateCompileUnit(cci).get();
if (!sc.comp_unit)
continue;

PdbCompilandSymId func_id(proc.modi(), proc.SymOffset);
sc.function = GetOrCreateFunction(func_id, *sc.comp_unit).get();
if (!sc.function)
continue;

sc_list.Append(sc);
}
};

if (name_type_mask & eFunctionNameTypeFull)
resolve_from(m_func_full_names);
if (name_type_mask & eFunctionNameTypeBase)
resolve_from(m_func_base_names);
if (name_type_mask & eFunctionNameTypeMethod)
resolve_from(m_func_method_names);
}

void SymbolFileNativePDB::FindFunctions(const RegularExpression &regex,
Expand Down
9 changes: 9 additions & 0 deletions lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -260,6 +260,8 @@ class SymbolFileNativePDB : public SymbolFileCommon {

std::vector<CompilerContext> GetContextForType(llvm::codeview::TypeIndex ti);

void CacheFunctionNames();

llvm::BumpPtrAllocator m_allocator;

lldb::addr_t m_obj_load_address = 0;
Expand All @@ -282,6 +284,13 @@ class SymbolFileNativePDB : public SymbolFileCommon {
m_parent_types;

lldb_private::UniqueCStringMap<uint32_t> m_type_base_names;

/// mangled name/full function name -> Global ID(s)
lldb_private::UniqueCStringMap<uint32_t> m_func_full_names;
/// basename -> Global ID(s)
lldb_private::UniqueCStringMap<uint32_t> m_func_base_names;
/// method basename -> Global ID(s)
lldb_private::UniqueCStringMap<uint32_t> m_func_method_names;
};

} // namespace npdb
Expand Down
34 changes: 34 additions & 0 deletions lldb/test/Shell/SymbolFile/NativePDB/find-functions.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,24 +6,52 @@

// RUN: lldb-test symbols --find=function --name=main --function-flags=full %t.exe \
// RUN: | FileCheck %s --check-prefix=FIND-MAIN
// RUN: lldb-test symbols --find=function --name=main --function-flags=method %t.exe \
// RUN: | FileCheck %s --check-prefix=FIND-NO-FUNCTION
// RUN: lldb-test symbols --find=function --name=main --function-flags=base %t.exe \
// RUN: | FileCheck %s --check-prefix=FIND-MAIN

// RUN: lldb-test symbols --find=function --name=static_fn --function-flags=full %t.exe \
// RUN: | FileCheck %s --check-prefix=FIND-STATIC
// RUN: lldb-test symbols --find=function --name=static_fn --function-flags=method %t.exe \
// RUN: | FileCheck %s --check-prefix=FIND-NO-FUNCTION
// RUN: lldb-test symbols --find=function --name=static_fn --function-flags=base %t.exe \
// RUN: | FileCheck %s --check-prefix=FIND-STATIC

// RUN: lldb-test symbols --find=function --name=varargs_fn --function-flags=full %t.exe \
// RUN: | FileCheck %s --check-prefix=FIND-VAR
// RUN: lldb-test symbols --find=function --name=varargs_fn --function-flags=method %t.exe \
// RUN: | FileCheck %s --check-prefix=FIND-NO-FUNCTION
// RUN: lldb-test symbols --find=function --name=varargs_fn --function-flags=base %t.exe \
// RUN: | FileCheck %s --check-prefix=FIND-VAR

// RUN: lldb-test symbols --find=function --name=Struct::simple_method --function-flags=full %t.exe \
// RUN: | FileCheck %s --check-prefix=FIND-SIMPLE
// RUN: lldb-test symbols --find=function --name=Struct::simple_method --function-flags=method %t.exe \
// RUN: | FileCheck %s --check-prefix=FIND-SIMPLE
// RUN: lldb-test symbols --find=function --name=Struct::simple_method --function-flags=base %t.exe \
// RUN: | FileCheck %s --check-prefix=FIND-SIMPLE

// RUN: lldb-test symbols --find=function --name=Struct::virtual_method --function-flags=full %t.exe \
// RUN: | FileCheck %s --check-prefix=FIND-VIRTUAL
// RUN: lldb-test symbols --find=function --name=Struct::virtual_method --function-flags=method %t.exe \
// RUN: | FileCheck %s --check-prefix=FIND-VIRTUAL
// RUN: lldb-test symbols --find=function --name=Struct::virtual_method --function-flags=base %t.exe \
// RUN: | FileCheck %s --check-prefix=FIND-VIRTUAL

// RUN: lldb-test symbols --find=function --name=Struct::static_method --function-flags=full %t.exe \
// RUN: | FileCheck %s --check-prefix=FIND-STATIC-METHOD
// RUN: lldb-test symbols --find=function --name=Struct::static_method --function-flags=method %t.exe \
// RUN: | FileCheck %s --check-prefix=FIND-NO-FUNCTION
// RUN: lldb-test symbols --find=function --name=Struct::static_method --function-flags=base %t.exe \
// RUN: | FileCheck %s --check-prefix=FIND-STATIC-METHOD

// RUN: lldb-test symbols --find=function --name=Struct::overloaded_method --function-flags=full %t.exe \
// RUN: | FileCheck %s --check-prefix=FIND-OVERLOAD
// RUN: lldb-test symbols --find=function --name=Struct::overloaded_method --function-flags=method %t.exe \
// RUN: | FileCheck %s --check-prefix=FIND-OVERLOAD-METHOD
// RUN: lldb-test symbols --find=function --name=Struct::overloaded_method --function-flags=base %t.exe \
// RUN: | FileCheck %s --check-prefix=FIND-OVERLOAD
Copy link
Contributor

@ZequanWu ZequanWu Aug 8, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can you add tests on 2 different classes/namespaces with the same function basename?

Copy link
Contributor Author

@Nerixyz Nerixyz Aug 8, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I added them. --function-flags=method will only match the basename of methods (the invocation might look a bit confusing). This is consistent with DWARF:

NativePDB 
> &'f:\dev\llvm-project\build\bin\lldb-test.exe' symbols --find=function --name=Struct::overloaded_method --function-flags=method 'F:\Dev\llvm-project\build\tools\lldb\test\Shell\SymbolFile\NativePDB\Output\find-functions.cpp.tmp.exe'
Module: F:\Dev\llvm-project\build\tools\lldb\test\Shell\SymbolFile\NativePDB\Output\find-functions.cpp.tmp.exe
Found 4 functions:
0x0000001f9e98e988: SymbolContextList
  CompileUnit: id = {0x00000000}, file = "F:\Dev\llvm-project\lldb\test\Shell\SymbolFile\NativePDB\find-functions.cpp", language = "c++"
     Function: id = {0x5d000001}, name = "Class::overloaded_method", decl_context = {}, range = find-functions.cpp.tmp.exe[0x0000000140001200-0x000000014000121b)
     FuncType: id = {0x000101b4}, byte-size = 0, compiler_type = "_Bool (int)"
  CompileUnit: id = {0x00000000}, file = "F:\Dev\llvm-project\lldb\test\Shell\SymbolFile\NativePDB\find-functions.cpp", language = "c++"
     Function: id = {0x67000001}, name = "Class::overloaded_method", decl_context = {}, range = find-functions.cpp.tmp.exe[0x0000000140001220-0x000000014000122b)
     FuncType: id = {0x000101a4}, byte-size = 0, compiler_type = "_Bool (void)"
  CompileUnit: id = {0x00000000}, file = "F:\Dev\llvm-project\lldb\test\Shell\SymbolFile\NativePDB\find-functions.cpp", language = "c++"
     Function: id = {0x4c800001}, name = "Struct::overloaded_method", decl_context = {}, range = find-functions.cpp.tmp.exe[0x00000001400011a0-0x00000001400011d9)
     FuncType: id = {0x000100e4}, byte-size = 0, compiler_type = "int (void)"
  CompileUnit: id = {0x00000000}, file = "F:\Dev\llvm-project\lldb\test\Shell\SymbolFile\NativePDB\find-functions.cpp", language = "c++"
     Function: id = {0x6f400001}, name = "Struct::overloaded_method", decl_context = {}, range = find-functions.cpp.tmp.exe[0x0000000140001230-0x0000000140001246)
     FuncType: id = {0x00010114}, byte-size = 0, compiler_type = "int (char)"
DWARF 
> &'f:\dev\llvm-project\build\bin\lldb-test.exe' symbols --find=function --name=Struct::overloaded_method --function-flags=method a.exe
Module: a.exe
Found 6 functions:
0x0000003203f8e5d8: SymbolContextList
       Module: file = "a.exe", arch = "x86_64"
  CompileUnit: id = {0x00000000}, file = "F:\Dev\llvm-project\lldb\test\Shell\SymbolFile\NativePDB\find-functions.cpp", language = "c++14"
     Function: id = {0x0000028f}, name = "int Struct::overloaded_method(void)", mangled = "?overloaded_method@Struct@@QEAAHXZ", decl_context = {ClassOrStruct(Struct)}, range = a.exe[0x0000000140001150-0x0000000140001189)
     FuncType: id = {0x0000028f}, byte-size = 0, decl = F:\Dev\llvm-project\lldb\test\Shell\SymbolFile\NativePDB\find-functions.cpp:69, compiler_type = "int (void)"
       Module: file = "a.exe", arch = "x86_64"
  CompileUnit: id = {0x00000000}, file = "F:\Dev\llvm-project\lldb\test\Shell\SymbolFile\NativePDB\find-functions.cpp", language = "c++14"
     Function: id = {0x000002b3}, name = "int Class::overloaded_method(bool)", mangled = "?overloaded_method@Class@@SAH_N@Z", decl_context = {ClassOrStruct(Class)}, range = a.exe[0x0000000140001190-0x00000001400011ae)
     FuncType: id = {0x000002b3}, byte-size = 0, decl = F:\Dev\llvm-project\lldb\test\Shell\SymbolFile\NativePDB\find-functions.cpp:90, compiler_type = "int (_Bool)"
       Module: file = "a.exe", arch = "x86_64"
  CompileUnit: id = {0x00000000}, file = "F:\Dev\llvm-project\lldb\test\Shell\SymbolFile\NativePDB\find-functions.cpp", language = "c++14"
     Function: id = {0x000002d5}, name = "bool Class::overloaded_method(int)", mangled = "?overloaded_method@Class@@QEAA_NH@Z", decl_context = {ClassOrStruct(Class)}, range = a.exe[0x00000001400011b0-0x00000001400011cb)
     FuncType: id = {0x000002d5}, byte-size = 0, decl = F:\Dev\llvm-project\lldb\test\Shell\SymbolFile\NativePDB\find-functions.cpp:87, compiler_type = "_Bool (int)"
       Module: file = "a.exe", arch = "x86_64"
  CompileUnit: id = {0x00000000}, file = "F:\Dev\llvm-project\lldb\test\Shell\SymbolFile\NativePDB\find-functions.cpp", language = "c++14"
     Function: id = {0x00000307}, name = "bool Class::overloaded_method(void)", mangled = "?overloaded_method@Class@@QEAA_NXZ", decl_context = {ClassOrStruct(Class)}, range = a.exe[0x00000001400011d0-0x00000001400011db)
     FuncType: id = {0x00000307}, byte-size = 0, decl = F:\Dev\llvm-project\lldb\test\Shell\SymbolFile\NativePDB\find-functions.cpp:84, compiler_type = "_Bool (void)"
       Module: file = "a.exe", arch = "x86_64"
  CompileUnit: id = {0x00000000}, file = "F:\Dev\llvm-project\lldb\test\Shell\SymbolFile\NativePDB\find-functions.cpp", language = "c++14"
     Function: id = {0x0000032b}, name = "int Struct::overloaded_method(char)", mangled = "?overloaded_method@Struct@@MEAAHD@Z", decl_context = {ClassOrStruct(Struct)}, range = a.exe[0x00000001400011e0-0x00000001400011f6)
     FuncType: id = {0x0000032b}, byte-size = 0, decl = F:\Dev\llvm-project\lldb\test\Shell\SymbolFile\NativePDB\find-functions.cpp:73, compiler_type = "int (char)"
       Module: file = "a.exe", arch = "x86_64"
  CompileUnit: id = {0x00000000}, file = "F:\Dev\llvm-project\lldb\test\Shell\SymbolFile\NativePDB\find-functions.cpp", language = "c++14"
     Function: id = {0x0000035d}, name = "int Struct::overloaded_method(char, int, ...)", mangled = "?overloaded_method@Struct@@CAHDHZZ", decl_context = {ClassOrStruct(Struct)}, range = a.exe[0x0000000140001200-0x0000000140001210)
     FuncType: id = {0x0000035d}, byte-size = 0, decl = F:\Dev\llvm-project\lldb\test\Shell\SymbolFile\NativePDB\find-functions.cpp:77, compiler_type = "int (char, int, ...)"

I'm not sure if it's correct that DWARF returns static functions here.

Copy link
Contributor

@ZequanWu ZequanWu Aug 11, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm also not sure if dwarf plugin is correct to return static functions (need someone to verify this). Other than that, it looks good to me.


struct Struct {
int simple_method() {
Expand Down Expand Up @@ -69,6 +97,8 @@ int main(int argc, char **argv) {
// FIND-MAIN: Function: id = {{.*}}, name = "main"
// FIND-MAIN-NEXT: FuncType: id = {{.*}}, compiler_type = "int (int, char **)"

// FIND-NO-FUNCTION: Found 0 functions

// FIND-STATIC: Function: id = {{.*}}, name = "{{.*}}static_fn{{.*}}"
// FIND-STATIC-NEXT: FuncType: id = {{.*}}, compiler_type = "int (void)"

Expand All @@ -88,3 +118,7 @@ int main(int argc, char **argv) {
// FIND-OVERLOAD: FuncType: id = {{.*}}, compiler_type = "int (void)"
// FIND-OVERLOAD: FuncType: id = {{.*}}, compiler_type = "int (char)"
// FIND-OVERLOAD: FuncType: id = {{.*}}, compiler_type = "int (char, int, ...)"

// FIND-OVERLOAD-METHOD: Function: id = {{.*}}, name = "{{.*}}Struct::overloaded_method{{.*}}"
// FIND-OVERLOAD-METHOD: FuncType: id = {{.*}}, compiler_type = "int (void)"
// FIND-OVERLOAD-METHOD: FuncType: id = {{.*}}, compiler_type = "int (char)"
Loading