Prevent a crash when a global variable has debug metadata

[gbMattN]

This patch fixes a crash I found when trying to compile some codebases with -fsanitize=type and -g

[llvmbot]

@llvm/pr-subscribers-debuginfo

@llvm/pr-subscribers-llvm-ir

Author: None (gbMattN)

Changes

This patch fixes a crash I found when trying to compile some codebases with -fsanitize=type and -g

---

Full diff: https://github.com/llvm/llvm-project/pull/145918.diff

2 Files Affected:

• (modified) llvm/lib/IR/DebugInfo.cpp (+2-2)
• (modified) llvm/unittests/IR/DebugInfoTest.cpp (+38)

diff --git a/llvm/lib/IR/DebugInfo.cpp b/llvm/lib/IR/DebugInfo.cpp
index 5c645ffe3f3f7..c0ca896247ff3 100644
--- a/llvm/lib/IR/DebugInfo.cpp
+++ b/llvm/lib/IR/DebugInfo.cpp
@@ -50,7 +50,7 @@ TinyPtrVector<DbgDeclareInst *> llvm::findDbgDeclares(Value *V) {
   // DenseMap lookup. This check is a bitfield datamember lookup.
   if (!V->isUsedByMetadata())
     return {};
-  auto *L = LocalAsMetadata::getIfExists(V);
+  auto *L = ValueAsMetadata::getIfExists(V);
   if (!L)
     return {};
   auto *MDV = MetadataAsValue::getIfExists(V->getContext(), L);
@@ -69,7 +69,7 @@ TinyPtrVector<DbgVariableRecord *> llvm::findDVRDeclares(Value *V) {
   // DenseMap lookup. This check is a bitfield datamember lookup.
   if (!V->isUsedByMetadata())
     return {};
-  auto *L = LocalAsMetadata::getIfExists(V);
+  auto *L = ValueAsMetadata::getIfExists(V);
   if (!L)
     return {};
 
diff --git a/llvm/unittests/IR/DebugInfoTest.cpp b/llvm/unittests/IR/DebugInfoTest.cpp
index 35bdbf8cc8321..1f88f249789f8 100644
--- a/llvm/unittests/IR/DebugInfoTest.cpp
+++ b/llvm/unittests/IR/DebugInfoTest.cpp
@@ -195,6 +195,44 @@ TEST(MetadataTest, DeleteInstUsedByDbgRecord) {
   EXPECT_TRUE(isa<UndefValue>(DVRs[0]->getValue(0)));
 }
 
+TEST(MetadataTest, GlobalConstantMetadataUsedByDbgRecord) {
+    LLVMContext C;
+  std::unique_ptr<Module> M = parseIR(C, R"(
+    @x = dso_local global i32 0, align 4
+    define i16 @f(i16 %a) !dbg !6 {
+      %b = add i16 %a, 1, !dbg !11
+      call void @llvm.dbg.declare(metadata ptr @x, metadata !9, metadata !DIExpression()), !dbg !11
+      ret i16 0, !dbg !11
+    }
+    declare void @llvm.dbg.value(metadata, metadata, metadata) #0
+    attributes #0 = { nounwind readnone speculatable willreturn }
+
+    !llvm.dbg.cu = !{!0}
+    !llvm.module.flags = !{!5}
+
+    !0 = distinct !DICompileUnit(language: DW_LANG_C, file: !1, producer: "debugify", isOptimized: true, runtimeVersion: 0, emissionKind: FullDebug, enums: !2)
+    !1 = !DIFile(filename: "t.ll", directory: "/")
+    !2 = !{}
+    !5 = !{i32 2, !"Debug Info Version", i32 3}
+    !6 = distinct !DISubprogram(name: "foo", linkageName: "foo", scope: null, file: !1, line: 1, type: !7, scopeLine: 1, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: !0, retainedNodes: !8)
+    !7 = !DISubroutineType(types: !2)
+    !8 = !{!9}
+    !9 = !DILocalVariable(name: "1", scope: !6, file: !1, line: 1, type: !10)
+    !10 = !DIBasicType(name: "ty16", size: 16, encoding: DW_ATE_unsigned)
+    !11 = !DILocation(line: 1, column: 1, scope: !6)
+)");
+
+  // Find the global @x
+  Value* V = M->getNamedValue("x");
+
+  // Find the dbg.value
+  auto DVIs = findDbgDeclares(V);
+  auto DVRs = findDVRDeclares(V);
+
+  EXPECT_EQ(DVRs[0]->getNumVariableLocationOps(), 1u);
+  EXPECT_FALSE(isa<UndefValue>(DVRs[0]->getValue(0)));
+}
+
 TEST(DbgVariableIntrinsic, EmptyMDIsKillLocation) {
   LLVMContext Ctx;
   std::unique_ptr<Module> M = parseIR(Ctx, R"(

[github-actions]

✅ With the latest revision this PR passed the C/C++ code formatter.

[jmorse]

Yikes, that's a pain -- so 'C' s being looked up from somewhere else and is wrong/bad? Any idea where the 'C' variable is coming from here? (Possibly we should do something else to ensure this isn't an error latent in other unit tests).

[jmorse]

Hmmmph, got my wires crossed on reviews and patches there, lemmie try again...

[jmorse]

Urgh. Discussed slightly offline, the fix is correct but it indicates that we've never been able to lookup debug-users of global variables in LLVM, ever, because of the VAM/LocalAsMetadata differences you're fixing here. Which is horrible, but I imagine it's not a common access pattern -- usually we look up the debug-users of a local instruction that's being fiddled with.

Would you be able to repeat the fix/unit-test for findDVRValues please, as it looks like that's vulnerable to the same issue.

[gbMattN]

Done so!

[jmorse]

LGTM; ship it!