[Lldb-commits] [lldb] 44fefe7 - [LLDB][NativePDB] Estimate symbol sizes (#165727)
via lldb-commits
lldb-commits at lists.llvm.org
Fri Oct 31 02:33:41 PDT 2025
Author: nerix
Date: 2025-10-31T10:33:37+01:00
New Revision: 44fefe70e41a3b8e68545f45798740f74f4231cd
URL: https://github.com/llvm/llvm-project/commit/44fefe70e41a3b8e68545f45798740f74f4231cd
DIFF: https://github.com/llvm/llvm-project/commit/44fefe70e41a3b8e68545f45798740f74f4231cd.diff
LOG: [LLDB][NativePDB] Estimate symbol sizes (#165727)
In #165604, a test was skipped on Windows, because the native PDB plugin
didn't set sizes on symbols. While the test isn't compiled with debug
info, it's linked with `-gdwarf`, causing a PDB to be created on
Windows. This PDB will only contain the public symbols (written by the
linker) and section information. The symbols themselves don't have a
size, however the DIA SDK sets a size for them.
It seems like, for these data symbols, the size given from DIA is the
distance to the next symbol (or the section end).
This PR implements the naive approach for the native plugin. The main
difference is in function/code symbols. There, DIA searches for a
corresponding `S_GPROC32` which have a "code size" that is sometimes
slightly smaller than the difference to the next symbol.
Added:
Modified:
lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp
lldb/test/API/functionalities/multiple-slides/TestMultipleSlides.py
lldb/test/Shell/SymbolFile/NativePDB/symtab.cpp
Removed:
################################################################################
diff --git a/lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp b/lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp
index e76b7a3cf274a..aaec1600dacff 100644
--- a/lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp
+++ b/lldb/source/Plugins/SymbolFile/NativePDB/SymbolFileNativePDB.cpp
@@ -1130,7 +1130,35 @@ void SymbolFileNativePDB::AddSymbols(Symtab &symtab) {
if (!section_list)
return;
- for (auto pid : m_index->publics().getPublicsTable()) {
+ PublicSym32 last_sym;
+ size_t last_sym_idx = 0;
+ lldb::SectionSP section_sp;
+
+ // To estimate the size of a symbol, we use the
diff erence to the next symbol.
+ // If there's no next symbol or the section/segment changed, the symbol will
+ // take the remaining space. The estimate can be too high in case there's
+ // padding between symbols. This similar to the algorithm used by the DIA
+ // SDK.
+ auto finish_last_symbol = [&](const PublicSym32 *next) {
+ if (!section_sp)
+ return;
+ Symbol *last = symtab.SymbolAtIndex(last_sym_idx);
+ if (!last)
+ return;
+
+ if (next && last_sym.Segment == next->Segment) {
+ assert(last_sym.Offset <= next->Offset);
+ last->SetByteSize(next->Offset - last_sym.Offset);
+ } else {
+ // the last symbol was the last in its section
+ assert(section_sp->GetByteSize() >= last_sym.Offset);
+ assert(!next || next->Segment > last_sym.Segment);
+ last->SetByteSize(section_sp->GetByteSize() - last_sym.Offset);
+ }
+ };
+
+ // The address map is sorted by the address of a symbol.
+ for (auto pid : m_index->publics().getAddressMap()) {
PdbGlobalSymId global{pid, true};
CVSymbol sym = m_index->ReadSymbolRecord(global);
auto kind = sym.kind();
@@ -1138,8 +1166,11 @@ void SymbolFileNativePDB::AddSymbols(Symtab &symtab) {
continue;
PublicSym32 pub =
llvm::cantFail(SymbolDeserializer::deserializeAs<PublicSym32>(sym));
+ finish_last_symbol(&pub);
+
+ if (!section_sp || last_sym.Segment != pub.Segment)
+ section_sp = section_list->FindSectionByID(pub.Segment);
- auto section_sp = section_list->FindSectionByID(pub.Segment);
if (!section_sp)
continue;
@@ -1148,20 +1179,24 @@ void SymbolFileNativePDB::AddSymbols(Symtab &symtab) {
(pub.Flags & PublicSymFlags::Code) != PublicSymFlags::None)
type = eSymbolTypeCode;
- symtab.AddSymbol(Symbol(/*symID=*/pid,
- /*name=*/pub.Name,
- /*type=*/type,
- /*external=*/true,
- /*is_debug=*/true,
- /*is_trampoline=*/false,
- /*is_artificial=*/false,
- /*section_sp=*/section_sp,
- /*value=*/pub.Offset,
- /*size=*/0,
- /*size_is_valid=*/false,
- /*contains_linker_annotations=*/false,
- /*flags=*/0));
- }
+ last_sym_idx =
+ symtab.AddSymbol(Symbol(/*symID=*/pid,
+ /*name=*/pub.Name,
+ /*type=*/type,
+ /*external=*/true,
+ /*is_debug=*/true,
+ /*is_trampoline=*/false,
+ /*is_artificial=*/false,
+ /*section_sp=*/section_sp,
+ /*value=*/pub.Offset,
+ /*size=*/0,
+ /*size_is_valid=*/false,
+ /*contains_linker_annotations=*/false,
+ /*flags=*/0));
+ last_sym = pub;
+ }
+
+ finish_last_symbol(nullptr);
}
size_t SymbolFileNativePDB::ParseFunctions(CompileUnit &comp_unit) {
diff --git a/lldb/test/API/functionalities/multiple-slides/TestMultipleSlides.py b/lldb/test/API/functionalities/multiple-slides/TestMultipleSlides.py
index 7fd2ff4229004..5fd2b767a6237 100644
--- a/lldb/test/API/functionalities/multiple-slides/TestMultipleSlides.py
+++ b/lldb/test/API/functionalities/multiple-slides/TestMultipleSlides.py
@@ -12,10 +12,6 @@
class MultipleSlidesTestCase(TestBase):
NO_DEBUG_INFO_TESTCASE = True
- # The intermediate object main.o is compiled without debug info, but
- # a.out is linked with `-gdwarf` on Windows. This creates a PDB.
- # However, in the native PDB plugin, the symbols don't have a size.
- @expectedFailureWindows
def test_mulitple_slides(self):
"""Test that a binary can be slid multiple times correctly."""
self.build()
@@ -33,10 +29,13 @@ def test_mulitple_slides(self):
first_sym.GetEndAddress().GetOffset()
- first_sym.GetStartAddress().GetOffset()
)
+ int_size = target.FindFirstType("int").GetByteSize()
+ self.assertGreaterEqual(first_size, 2048 * int_size)
second_size = (
second_sym.GetEndAddress().GetOffset()
- second_sym.GetStartAddress().GetOffset()
)
+ self.assertGreaterEqual(second_size, 2048 * int_size)
# View the first element of `first` and `second` while
# they have no load address set.
diff --git a/lldb/test/Shell/SymbolFile/NativePDB/symtab.cpp b/lldb/test/Shell/SymbolFile/NativePDB/symtab.cpp
index beb5ae2f90256..75c59c560fad9 100644
--- a/lldb/test/Shell/SymbolFile/NativePDB/symtab.cpp
+++ b/lldb/test/Shell/SymbolFile/NativePDB/symtab.cpp
@@ -42,18 +42,18 @@ int main(int argc, char **argv) {
return ns::a_function() + b.b_func();
}
-// CHECK-DAG: Code {{.*}} main
-// CHECK-DAG: Code {{.*}} ?b_func@?$B at F@ns@@QEBAHXZ
-// CHECK-DAG: Code {{.*}} ?something at A@@QEAAXXZ
-// CHECK-DAG: Code {{.*}} ??_GDyn at ns@@UEAAPEAXI at Z
-// CHECK-DAG: Code {{.*}} ??2 at YAPEAX_K@Z
-// CHECK-DAG: Code {{.*}} ??3 at YAXPEAX_K@Z
-// CHECK-DAG: Code {{.*}} ?static_fn at C@?$B at H@ns@@SAHXZ
-// CHECK-DAG: Code {{.*}} ?a_function at ns@@YAHXZ
-// CHECK-DAG: Code {{.*}} ?static_fn at C@?$B at _N@ns@@SAHXZ
-// CHECK-DAG: Code {{.*}} ??1Dyn at ns@@UEAA at XZ
-// CHECK-DAG: Code {{.*}} ??0Dyn at ns@@QEAA at XZ
-// CHECK-DAG: Data {{.*}} ?global_int@@3HA
-// CHECK-DAG: Data {{.*}} ??_7Dyn at ns@@6B@
-// CHECK-DAG: Data {{.*}} ?global_a@@3UA@@A
-// CHECK-DAG: Data {{.*}} ?global_c@@3UC@?$B at _J@ns@@A
+// CHECK-DAG: Code 0x{{[0-9a-f]+}} 0x{{0*[1-9a-f][0-9a-f]*}} 0x00000000 main
+// CHECK-DAG: Code 0x{{[0-9a-f]+}} 0x{{0*[1-9a-f][0-9a-f]*}} 0x00000000 ?b_func@?$B at F@ns@@QEBAHXZ
+// CHECK-DAG: Code 0x{{[0-9a-f]+}} 0x{{0*[1-9a-f][0-9a-f]*}} 0x00000000 ?something at A@@QEAAXXZ
+// CHECK-DAG: Code 0x{{[0-9a-f]+}} 0x{{0*[1-9a-f][0-9a-f]*}} 0x00000000 ??_GDyn at ns@@UEAAPEAXI at Z
+// CHECK-DAG: Code 0x{{[0-9a-f]+}} 0x{{0*[1-9a-f][0-9a-f]*}} 0x00000000 ??2 at YAPEAX_K@Z
+// CHECK-DAG: Code 0x{{[0-9a-f]+}} 0x{{0*[1-9a-f][0-9a-f]*}} 0x00000000 ??3 at YAXPEAX_K@Z
+// CHECK-DAG: Code 0x{{[0-9a-f]+}} 0x{{0*[1-9a-f][0-9a-f]*}} 0x00000000 ?static_fn at C@?$B at H@ns@@SAHXZ
+// CHECK-DAG: Code 0x{{[0-9a-f]+}} 0x{{0*[1-9a-f][0-9a-f]*}} 0x00000000 ?a_function at ns@@YAHXZ
+// CHECK-DAG: Code 0x{{[0-9a-f]+}} 0x{{0*[1-9a-f][0-9a-f]*}} 0x00000000 ?static_fn at C@?$B at _N@ns@@SAHXZ
+// CHECK-DAG: Code 0x{{[0-9a-f]+}} 0x{{0*[1-9a-f][0-9a-f]*}} 0x00000000 ??1Dyn at ns@@UEAA at XZ
+// CHECK-DAG: Code 0x{{[0-9a-f]+}} 0x{{0*[1-9a-f][0-9a-f]*}} 0x00000000 ??0Dyn at ns@@QEAA at XZ
+// CHECK-DAG: Data 0x{{[0-9a-f]+}} 0x{{0*[1-9a-f][0-9a-f]*}} 0x00000000 ?global_int@@3HA
+// CHECK-DAG: Data 0x{{[0-9a-f]+}} 0x{{0*[1-9a-f][0-9a-f]*}} 0x00000000 ??_7Dyn at ns@@6B@
+// CHECK-DAG: Data 0x{{[0-9a-f]+}} 0x{{0*[1-9a-f][0-9a-f]*}} 0x00000000 ?global_a@@3UA@@A
+// CHECK-DAG: Data 0x{{[0-9a-f]+}} 0x{{0*[1-9a-f][0-9a-f]*}} 0x00000000 ?global_c@@3UC@?$B at _J@ns@@A
More information about the lldb-commits
mailing list