[LLDB][NativePDB] Create functions with mangled name #149701
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
@llvm/pr-subscribers-lldb Author: nerix (Nerixyz) ChangesBefore, functions created using the NativePDB plugin would not know about their mangled name. This showed when printing a stacktrace. There, only the function name was shown. For #143149, the mangled function name is required to separate different parts. This PR adds that name if available. It reimplements functionality from Microsoft's PDB reference to find a symbol given an address. The Clang AST nodes also take in a mangled name, which was previously unset. I don't think this unblocks anything further, because Clang can mangle the function anyway. Full diff: https://github.com/llvm/llvm-project/pull/149701.diff 9 Files Affected:
diff --git a/lldb/source/Plugins/SymbolFile/NativePDB/PdbAstBuilder.cpp b/lldb/source/Plugins/SymbolFile/NativePDB/PdbAstBuilder.cpp
index 702ec5e5c9ea9..79e7de8a75de7 100644
--- a/lldb/source/Plugins/SymbolFile/NativePDB/PdbAstBuilder.cpp
+++ b/lldb/source/Plugins/SymbolFile/NativePDB/PdbAstBuilder.cpp
@@ -38,16 +38,18 @@ struct CreateMethodDecl : public TypeVisitorCallbacks {
TypeIndex func_type_index,
clang::FunctionDecl *&function_decl,
lldb::opaque_compiler_type_t parent_ty,
- llvm::StringRef proc_name, CompilerType func_ct)
+ llvm::StringRef proc_name, ConstString mangled_name,
+ CompilerType func_ct)
: m_index(m_index), m_clang(m_clang), func_type_index(func_type_index),
function_decl(function_decl), parent_ty(parent_ty),
- proc_name(proc_name), func_ct(func_ct) {}
+ proc_name(proc_name), mangled_name(mangled_name), func_ct(func_ct) {}
PdbIndex &m_index;
TypeSystemClang &m_clang;
TypeIndex func_type_index;
clang::FunctionDecl *&function_decl;
lldb::opaque_compiler_type_t parent_ty;
llvm::StringRef proc_name;
+ ConstString mangled_name;
CompilerType func_ct;
llvm::Error visitKnownMember(CVMemberRecord &cvr,
@@ -87,8 +89,8 @@ struct CreateMethodDecl : public TypeVisitorCallbacks {
bool is_artificial = (options & MethodOptions::CompilerGenerated) ==
MethodOptions::CompilerGenerated;
function_decl = m_clang.AddMethodToCXXRecordType(
- parent_ty, proc_name,
- /*mangled_name=*/nullptr, func_ct, /*access=*/access_type,
+ parent_ty, proc_name, mangled_name.GetCString(), func_ct,
+ /*access=*/access_type,
/*is_virtual=*/is_virtual, /*is_static=*/is_static,
/*is_inline=*/false, /*is_explicit=*/false,
/*is_attr_used=*/false, /*is_artificial=*/is_artificial);
@@ -888,6 +890,11 @@ PdbAstBuilder::CreateFunctionDecl(PdbCompilandSymId func_id,
tag_record = CVTagRecord::create(index.tpi().getType(*eti)).asTag();
}
}
+
+ ConstString mangled_name;
+ if (auto mangled_name_opt = pdb->FindMangledFunctionName(func_id))
+ mangled_name = ConstString(*mangled_name_opt);
+
if (!tag_record.FieldList.isSimple()) {
CVType field_list_cvt = index.tpi().getType(tag_record.FieldList);
FieldListRecord field_list;
@@ -895,15 +902,15 @@ PdbAstBuilder::CreateFunctionDecl(PdbCompilandSymId func_id,
field_list_cvt, field_list))
llvm::consumeError(std::move(error));
CreateMethodDecl process(index, m_clang, func_ti, function_decl,
- parent_opaque_ty, func_name, func_ct);
+ parent_opaque_ty, func_name, mangled_name,
+ func_ct);
if (llvm::Error err = visitMemberRecordStream(field_list.Data, process))
llvm::consumeError(std::move(err));
}
if (!function_decl) {
function_decl = m_clang.AddMethodToCXXRecordType(
- parent_opaque_ty, func_name,
- /*mangled_name=*/nullptr, func_ct,
+ parent_opaque_ty, func_name, mangled_name.GetCString(), func_ct,
/*access=*/lldb::AccessType::eAccessPublic,
/*is_virtual=*/false, /*is_static=*/false,
/*is_inline=*/false, /*is_explicit=*/false,
diff --git a/lldb/source/Plugins/SymbolFile/NativePDB/PdbUtil.h b/lldb/source/Plugins/SymbolFile/NativePDB/PdbUtil.h
index 1f888f4de1fed..d910b6e8ad28e 100644
--- a/lldb/source/Plugins/SymbolFile/NativePDB/PdbUtil.h
+++ b/lldb/source/Plugins/SymbolFile/NativePDB/PdbUtil.h
@@ -89,6 +89,16 @@ struct SegmentOffset {
SegmentOffset(uint16_t s, uint32_t o) : segment(s), offset(o) {}
uint16_t segment = 0;
uint32_t offset = 0;
+
+ bool operator==(SegmentOffset rhs) const {
+ return segment == rhs.segment && offset == rhs.offset;
+ }
+
+ bool operator<(SegmentOffset rhs) const {
+ if (segment == rhs.segment)
+ return offset < rhs.offset;
+ return segment < rhs.segment;
+ }
};
struct SegmentOffsetLength {
diff --git a/lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp b/lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp
index 20d8c1acf9c42..ca4718b7e0261 100644
--- a/lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp
+++ b/lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp
@@ -38,6 +38,7 @@
#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"
@@ -496,7 +497,9 @@ lldb::FunctionSP SymbolFileNativePDB::CreateFunction(PdbCompilandSymId func_id,
return nullptr;
PdbTypeSymId sig_id(proc.FunctionType, false);
- Mangled mangled(proc.Name);
+ auto mangled_opt =
+ FindMangledSymbol(SegmentOffset(proc.Segment, proc.CodeOffset));
+ Mangled mangled(mangled_opt.value_or(proc.Name));
FunctionSP func_sp = std::make_shared<Function>(
&comp_unit, toOpaqueUid(func_id), toOpaqueUid(sig_id), mangled,
func_type.get(), func_addr,
@@ -2353,3 +2356,68 @@ SymbolFileNativePDB::GetParentType(llvm::codeview::TypeIndex ti) {
return std::nullopt;
return parent_iter->second;
}
+
+std::optional<llvm::StringRef>
+SymbolFileNativePDB::FindMangledFunctionName(PdbCompilandSymId func_id) {
+ const CompilandIndexItem *cci =
+ m_index->compilands().GetCompiland(func_id.modi);
+ if (!cci)
+ return std::nullopt;
+
+ CVSymbol sym_record = cci->m_debug_stream.readSymbolAtOffset(func_id.offset);
+ if (sym_record.kind() != S_LPROC32 && sym_record.kind() != S_GPROC32)
+ return std::nullopt;
+
+ ProcSym proc(static_cast<SymbolRecordKind>(sym_record.kind()));
+ cantFail(SymbolDeserializer::deserializeAs<ProcSym>(sym_record, proc));
+ return FindMangledSymbol(SegmentOffset(proc.Segment, proc.CodeOffset));
+}
+
+/// Find the mangled name of a function at \a so.
+///
+/// This is similar to the NearestSym function from Microsoft's PDB reference:
+/// https://github.com/microsoft/microsoft-pdb/blob/805655a28bd8198004be2ac27e6e0290121a5e89/PDB/dbi/gsi.cpp#L1492-L1581
+/// The main difference is that we search for the exact symbol.
+///
+/// \param so[in] The address of the function given by its segment and code
+/// offset.
+/// \return The mangled function name if found. Otherwise an empty optional.
+std::optional<llvm::StringRef>
+SymbolFileNativePDB::FindMangledSymbol(SegmentOffset so) {
+ // The address map is sorted by address, so we do binary search.
+ // Each element is an offset into the symbols for a public symbol.
+ auto lo = m_index->publics().getAddressMap().begin();
+ auto hi = m_index->publics().getAddressMap().end();
+ hi -= 1;
+
+ while (lo < hi) {
+ auto tgt = lo + ((hi - lo + 1) / 2);
+ auto val = tgt->value();
+ auto sym = m_index->symrecords().readRecord(val);
+ if (sym.kind() != S_PUB32)
+ return std::nullopt; // this is most likely corrupted debug info
+
+ PublicSym32 psym =
+ llvm::cantFail(SymbolDeserializer::deserializeAs<PublicSym32>(sym));
+ SegmentOffset cur(psym.Segment, psym.Offset);
+ if (so < cur) {
+ tgt -= 1;
+ hi = tgt;
+ } else if (so == cur)
+ return psym.Name;
+ else
+ lo = tgt;
+ }
+
+ // We might've found something, check if it's the symbol we're searching for
+ auto val = lo->value();
+ auto sym = m_index->symrecords().readRecord(val);
+ if (sym.kind() != S_PUB32)
+ return std::nullopt;
+ PublicSym32 psym =
+ llvm::cantFail(SymbolDeserializer::deserializeAs<PublicSym32>(sym));
+ if (psym.Segment != so.segment || psym.Offset != so.offset)
+ return std::nullopt;
+ return psym.Name;
+}
+
diff --git a/lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.h b/lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.h
index 9891313f11d0b..abb70600447c1 100644
--- a/lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.h
+++ b/lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.h
@@ -140,6 +140,9 @@ class SymbolFileNativePDB : public SymbolFileCommon {
std::optional<PdbCompilandSymId> FindSymbolScope(PdbCompilandSymId id);
+ std::optional<llvm::StringRef> FindMangledFunctionName(PdbCompilandSymId id);
+ std::optional<llvm::StringRef> FindMangledSymbol(SegmentOffset so);
+
void FindTypes(const lldb_private::TypeQuery &match,
lldb_private::TypeResults &results) override;
diff --git a/lldb/test/Shell/SymbolFile/NativePDB/break-by-function.cpp b/lldb/test/Shell/SymbolFile/NativePDB/break-by-function.cpp
index a580d574a9ca3..d4499373bb860 100644
--- a/lldb/test/Shell/SymbolFile/NativePDB/break-by-function.cpp
+++ b/lldb/test/Shell/SymbolFile/NativePDB/break-by-function.cpp
@@ -50,9 +50,9 @@ int main(int argc, char **argv) {
// CHECK: 1: name = 'main', locations = 1
// CHECK: 1.1: where = break-by-function.cpp.tmp.exe`main + {{[0-9]+}}
// CHECK: 2: name = 'OvlGlobalFn', locations = 3
-// CHECK: 2.1: where = break-by-function.cpp.tmp.exe`OvlGlobalFn + {{[0-9]+}}
-// CHECK: 2.2: where = break-by-function.cpp.tmp.exe`OvlGlobalFn
-// CHECK: 2.3: where = break-by-function.cpp.tmp.exe`OvlGlobalFn + {{[0-9]+}}
+// CHECK: 2.1: where = break-by-function.cpp.tmp.exe`int OvlGlobalFn(int) + {{[0-9]+}}
+// CHECK: 2.2: where = break-by-function.cpp.tmp.exe`int OvlGlobalFn(int, int)
+// CHECK: 2.3: where = break-by-function.cpp.tmp.exe`int OvlGlobalFn(int, int, int) + {{[0-9]+}}
// CHECK: 3: name = 'StaticFn', locations = 1
// CHECK: 3.1: where = break-by-function.cpp.tmp.exe`StaticFn + {{[0-9]+}}
// CHECK: 4: name = 'DoesntExist', locations = 0 (pending)
diff --git a/lldb/test/Shell/SymbolFile/NativePDB/break-by-line.cpp b/lldb/test/Shell/SymbolFile/NativePDB/break-by-line.cpp
index 90ac633b01632..3d7de3275ed65 100644
--- a/lldb/test/Shell/SymbolFile/NativePDB/break-by-line.cpp
+++ b/lldb/test/Shell/SymbolFile/NativePDB/break-by-line.cpp
@@ -24,4 +24,4 @@ int main(int argc, char **argv) {
// CHECK: (lldb) target create "{{.*}}break-by-line.cpp.tmp.exe"
// CHECK: Current executable set to '{{.*}}break-by-line.cpp.tmp.exe'
// CHECK: (lldb) break set -f break-by-line.cpp -l 15
-// CHECK: Breakpoint 1: where = break-by-line.cpp.tmp.exe`NS::NamespaceFn + {{[0-9]+}} at break-by-line.cpp:15
+// CHECK: Breakpoint 1: where = break-by-line.cpp.tmp.exe`int NS::NamespaceFn(int) + {{[0-9]+}} at break-by-line.cpp:15
diff --git a/lldb/test/Shell/SymbolFile/NativePDB/disassembly.cpp b/lldb/test/Shell/SymbolFile/NativePDB/disassembly.cpp
index db3b85fa7e59f..3603db80ba8a7 100644
--- a/lldb/test/Shell/SymbolFile/NativePDB/disassembly.cpp
+++ b/lldb/test/Shell/SymbolFile/NativePDB/disassembly.cpp
@@ -27,7 +27,7 @@ int main(int argc, char **argv) {
// CHECK-NEXT: disassembly.cpp.tmp.exe[{{.*}}] <+12>: mov qword ptr [rsp + 0x28], rdx
// CHECK-NEXT: disassembly.cpp.tmp.exe[{{.*}}] <+17>: mov dword ptr [rsp + 0x24], ecx
// CHECK: ** 15 foo();
-// CHECK: disassembly.cpp.tmp.exe[{{.*}}] <+21>: call {{.*}} ; foo at disassembly.cpp:12
+// CHECK: disassembly.cpp.tmp.exe[{{.*}}] <+21>: call {{.*}} ; int foo(void) at disassembly.cpp:12
// CHECK: ** 16 return 0;
// CHECK-NEXT: 17 }
// CHECK-NEXT: 18
diff --git a/lldb/test/Shell/SymbolFile/NativePDB/local-variables.cpp b/lldb/test/Shell/SymbolFile/NativePDB/local-variables.cpp
index 44a8dc14c6158..f44a5b9dd56e2 100644
--- a/lldb/test/Shell/SymbolFile/NativePDB/local-variables.cpp
+++ b/lldb/test/Shell/SymbolFile/NativePDB/local-variables.cpp
@@ -55,7 +55,7 @@ int main(int argc, char **argv) {
// CHECK-NEXT: (lldb) step
// CHECK-NEXT: Process {{.*}} stopped
// CHECK-NEXT: * thread #1, stop reason = step in
-// CHECK-NEXT: frame #0: {{.*}} local-variables.cpp.tmp.exe`Function(Param1=16, Param2='a') at local-variables.cpp:{{.*}}
+// CHECK-NEXT: frame #0: {{.*}} local-variables.cpp.tmp.exe`int Function(Param1=16, Param2='a') at local-variables.cpp:{{.*}}
// CHECK-NEXT: 6
// CHECK-NEXT: 7
// CHECK-NEXT: 8 int Function(int Param1, char Param2) {
@@ -71,7 +71,7 @@ int main(int argc, char **argv) {
// CHECK-NEXT: (lldb) step
// CHECK-NEXT: Process {{.*}} stopped
// CHECK-NEXT: * thread #1, stop reason = step in
-// CHECK-NEXT: frame #0: {{.*}} local-variables.cpp.tmp.exe`Function(Param1=16, Param2='a') at local-variables.cpp:{{.*}}
+// CHECK-NEXT: frame #0: {{.*}} local-variables.cpp.tmp.exe`int Function(Param1=16, Param2='a') at local-variables.cpp:{{.*}}
// CHECK-NEXT: 7
// CHECK-NEXT: 8 int Function(int Param1, char Param2) {
// CHECK-NEXT: 9 unsigned Local1 = Param1 + 1;
@@ -89,7 +89,7 @@ int main(int argc, char **argv) {
// CHECK-NEXT: (lldb) step
// CHECK-NEXT: Process {{.*}} stopped
// CHECK-NEXT: * thread #1, stop reason = step in
-// CHECK-NEXT: frame #0: {{.*}} local-variables.cpp.tmp.exe`Function(Param1=16, Param2='a') at local-variables.cpp:{{.*}}
+// CHECK-NEXT: frame #0: {{.*}} local-variables.cpp.tmp.exe`int Function(Param1=16, Param2='a') at local-variables.cpp:{{.*}}
// CHECK-NEXT: 8 int Function(int Param1, char Param2) {
// CHECK-NEXT: 9 unsigned Local1 = Param1 + 1;
// CHECK-NEXT: 10 char Local2 = Param2 + 1;
@@ -109,7 +109,7 @@ int main(int argc, char **argv) {
// CHECK-NEXT: (lldb) step
// CHECK-NEXT: Process {{.*}} stopped
// CHECK-NEXT: * thread #1, stop reason = step in
-// CHECK-NEXT: frame #0: {{.*}} local-variables.cpp.tmp.exe`Function(Param1=16, Param2='a') at local-variables.cpp:{{.*}}
+// CHECK-NEXT: frame #0: {{.*}} local-variables.cpp.tmp.exe`int Function(Param1=16, Param2='a') at local-variables.cpp:{{.*}}
// CHECK-NEXT: 9 unsigned Local1 = Param1 + 1;
// CHECK-NEXT: 10 char Local2 = Param2 + 1;
// CHECK-NEXT: 11 ++Local1;
@@ -129,7 +129,7 @@ int main(int argc, char **argv) {
// CHECK-NEXT: (lldb) step
// CHECK-NEXT: Process {{.*}} stopped
// CHECK-NEXT: * thread #1, stop reason = step in
-// CHECK-NEXT: frame #0: {{.*}} local-variables.cpp.tmp.exe`Function(Param1=16, Param2='a') at local-variables.cpp:{{.*}}
+// CHECK-NEXT: frame #0: {{.*}} local-variables.cpp.tmp.exe`int Function(Param1=16, Param2='a') at local-variables.cpp:{{.*}}
// CHECK-NEXT: 10 char Local2 = Param2 + 1;
// CHECK-NEXT: 11 ++Local1;
// CHECK-NEXT: 12 ++Local2;
diff --git a/lldb/test/Shell/SymbolFile/NativePDB/stack_unwinding01.cpp b/lldb/test/Shell/SymbolFile/NativePDB/stack_unwinding01.cpp
index 596a826f4a11b..87eeebe7aa1b6 100644
--- a/lldb/test/Shell/SymbolFile/NativePDB/stack_unwinding01.cpp
+++ b/lldb/test/Shell/SymbolFile/NativePDB/stack_unwinding01.cpp
@@ -24,19 +24,19 @@ int main(int argc, char **argv) {
// CHECK: (lldb) thread backtrace
// CHECK-NEXT: * thread #1, stop reason = breakpoint 1.1
-// CHECK-NEXT: * frame #0: {{.*}} stack_unwinding01.cpp.tmp.exe`Struct::simple_method(this={{.*}}, a=2, b=2) at stack_unwinding01.cpp:12
+// CHECK-NEXT: * frame #0: {{.*}} stack_unwinding01.cpp.tmp.exe`void Struct::simple_method(this={{.*}}, a=2, b=2) at stack_unwinding01.cpp:12
// CHECK-NEXT: frame #1: {{.*}} stack_unwinding01.cpp.tmp.exe`main(argc={{.*}}, argv={{.*}}) at stack_unwinding01.cpp:20
// CHECK: (lldb) thread backtrace
// CHECK-NEXT: * thread #1, stop reason = breakpoint 1.1
-// CHECK-NEXT: * frame #0: {{.*}} stack_unwinding01.cpp.tmp.exe`Struct::simple_method(this={{.*}}, a=3, b=2) at stack_unwinding01.cpp:12
-// CHECK-NEXT: frame #1: {{.*}} stack_unwinding01.cpp.tmp.exe`Struct::simple_method(this={{.*}}, a=2, b=2) at stack_unwinding01.cpp:12
+// CHECK-NEXT: * frame #0: {{.*}} stack_unwinding01.cpp.tmp.exe`void Struct::simple_method(this={{.*}}, a=3, b=2) at stack_unwinding01.cpp:12
+// CHECK-NEXT: frame #1: {{.*}} stack_unwinding01.cpp.tmp.exe`void Struct::simple_method(this={{.*}}, a=2, b=2) at stack_unwinding01.cpp:12
// CHECK-NEXT: frame #2: {{.*}} stack_unwinding01.cpp.tmp.exe`main(argc={{.*}}, argv={{.*}}) at stack_unwinding01.cpp:20
// CHECK: (lldb) thread backtrace
// CHECK-NEXT: * thread #1, stop reason = breakpoint 1.1
-// CHECK-NEXT: * frame #0: {{.*}} stack_unwinding01.cpp.tmp.exe`Struct::simple_method(this={{.*}}, a=4, b=2) at stack_unwinding01.cpp:12
-// CHECK-NEXT: frame #1: {{.*}} stack_unwinding01.cpp.tmp.exe`Struct::simple_method(this={{.*}}, a=3, b=2) at stack_unwinding01.cpp:12
-// CHECK-NEXT: frame #2: {{.*}} stack_unwinding01.cpp.tmp.exe`Struct::simple_method(this={{.*}}, a=2, b=2) at stack_unwinding01.cpp:12
+// CHECK-NEXT: * frame #0: {{.*}} stack_unwinding01.cpp.tmp.exe`void Struct::simple_method(this={{.*}}, a=4, b=2) at stack_unwinding01.cpp:12
+// CHECK-NEXT: frame #1: {{.*}} stack_unwinding01.cpp.tmp.exe`void Struct::simple_method(this={{.*}}, a=3, b=2) at stack_unwinding01.cpp:12
+// CHECK-NEXT: frame #2: {{.*}} stack_unwinding01.cpp.tmp.exe`void Struct::simple_method(this={{.*}}, a=2, b=2) at stack_unwinding01.cpp:12
// CHECK-NEXT: frame #3: {{.*}} stack_unwinding01.cpp.tmp.exe`main(argc={{.*}}, argv={{.*}}) at stack_unwinding01.cpp:20
|
|
✅ With the latest revision this PR passed the C/C++ code formatter. |
Nerixyz
force-pushed
the
fix/npdb-use-mangled-name
branch
from
5b48860 to
8ea8a0e
Compare
Michael137
reviewed
Jul 21, 2025
There was a problem hiding this comment.
@rnk any thoughts on this? Could this live somewhere in LLVM's PDB support?
There was a problem hiding this comment.
Ping @rnk - could/should this be a method on the publics stream? Something like lower_bound?
There was a problem hiding this comment.
Sorry for the delay, summits and conferences were involved. I think that's a great idea, presumably the API would return optional<PublicSym32>, and then the caller looks at the flags, take the name string, or does whatever it needs to.
ZequanWu
approved these changes
Aug 8, 2025
Nerixyz
force-pushed
the
fix/npdb-use-mangled-name
branch
from
8ea8a0e to
5cb50cc
Compare
Michael137
requested changes
Sep 5, 2025
Nerixyz
added a commit
that referenced
this pull request
Sep 10, 2025
This adds a method on the `PublicsStream` to look up symbols using their address (segment + offset). It's largely a reimplementation of [`NearestSym`](https://github.com/microsoft/microsoft-pdb/blob/805655a28bd8198004be2ac27e6e0290121a5e89/PDB/dbi/gsi.cpp#L1492-L1581) from the reference. However, we don't return the nearest symbol, but the exact symbol. Still, in case of ICF, we return the symbol that's first in the address map. Users can then use the returned offset to read the next records to check if multiple symbols overlap, if desired. From #149701.
|
Ping - this now uses the method from LLVM's PDB support. |
Michael137
approved these changes
Sep 24, 2025
| std::optional<PdbCompilandSymId> FindSymbolScope(PdbCompilandSymId id); | ||
|
|
||
| std::optional<llvm::StringRef> FindMangledFunctionName(PdbCompilandSymId id); | ||
| std::optional<llvm::StringRef> FindMangledSymbol(SegmentOffset so); |
There was a problem hiding this comment.
Could we just make these an overload of each other? Both functions "find a mangled name". It's just the input that differs. Also, can we make FindMangledSymbol private?
There was a problem hiding this comment.
I made FindMangledSymbol private and kept FindMangledFunctionName, because it doesn't work for all PdbCompilandSymIds - it only works for functions. For example, it doesn't work for global variables.
lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp
Outdated
Show resolved
Hide resolved
Michael137
approved these changes
Sep 24, 2025
|
This broke demangling of Itanium names (for mingw) on i686, from PDB files - see https://github.com/mstorsjo/llvm-mingw/actions/runs/17993746384/job/51211649713 for a failed test run. Before this commit, we got backtraces that looked like this (e.g. https://github.com/mstorsjo/llvm-mingw/actions/runs/17963156286/job/51112172736 from yesterday): While now we get this: Somewhat surprisingly, the corresponding case on x86_64 (where symbols don't have a |
Oh are we just not expecting Itanium mangled names on Windows somewhere? |
Not entirely sure (I haven't had time to look into the code change itself yet); it seems like it does demangle the Itanium names fine on x86_64 though, where there isn't any extra |
Ah didn't notice the extra Wonder if we need to adjust LLDB further to accommodate for this. Could be a red herring of course |
|
I won't have time to look at this until tomorrow evening. If this PR causes problems, feel free to revert it. I do wonder why this causes the mangled name to show, because the demangling should detect the scheme automatically. Maybe this also has problems when debugging Rust, because that doesn't use the MSVC mangling scheme either. |
mstorsjo
added a commit
that referenced
this pull request
Sep 25, 2025
This reverts commit e98f34e. This broke demangling of Itanium symbols on i386.
Thanks, appreciated. Yeah I didn't have time to dig into it today either. I'll push a revert then, as it does block my nightly test pipeline. |
mahesh-attarde
pushed a commit
to mahesh-attarde/llvm-project
that referenced
this pull request
Oct 3, 2025
Before, functions created using the NativePDB plugin would not know about their mangled name. This showed when printing a stacktrace. There, only the function name was shown. For llvm#143149, the mangled function name is required to separate different parts. This PR adds that name if available. The Clang AST nodes also take in a mangled name, which was previously unset. I don't think this unblocks anything further, because Clang can mangle the function anyway.
mahesh-attarde
pushed a commit
to mahesh-attarde/llvm-project
that referenced
this pull request
Oct 3, 2025
Nerixyz
added a commit
that referenced
this pull request
Oct 7, 2025
Relands #149701 which was reverted in 185ae5c because it broke demangling of Itanium symbols on i386. The last commit in this PR adds the fix for this (discussed in #160930). On x86 environments, the prefix of `__cdecl` functions will now be removed to match DWARF. I opened #161676 to discuss this for the other calling conventions.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
6 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Before, functions created using the NativePDB plugin would not know about their mangled name. This showed when printing a stacktrace. There, only the function name was shown. For #143149, the mangled function name is required to separate different parts.
This PR adds that name if available. It reimplements functionality from Microsoft's PDB reference to find a symbol given an address.
The Clang AST nodes also take in a mangled name, which was previously unset. I don't think this unblocks anything further, because Clang can mangle the function anyway.