From: Nico Weber <nicolasweber@gmx.de>
Date: Wed, 10 Apr 2019 17:31:34 +0000 (+0000)
Subject: llvm-undname: Fix another crash-on-invalid
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=12041c1bc882b017093a27f44df833ba4f2f0409;p=llvm

llvm-undname: Fix another crash-on-invalid

This fixes a regression from https://reviews.llvm.org/D60354. We used to

  SymbolNode *Symbol = demangleEncodedSymbol(MangledName, QN);
  if (Symbol) {
    Symbol->Name = QN;
  }

but changed that to
  SymbolNode *Symbol = demangleEncodedSymbol(MangledName, QN);
  if (Error)
    return nullptr;
  Symbol->Name = QN;

and one branch somewhere returned a nullptr without setting Error.

Looking at the code changed in r340083 and r340710 that branch looks
like a remnant from an earlier attempt to demangle RTTI descriptors
that has since been rewritten -- so just remove this branch. It
shouldn't change behavior for correctly mangled symbols.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@358112 91177308-0d34-0410-b5e6-96231b3b80d8
---

diff --git a/lib/Demangle/MicrosoftDemangle.cpp b/lib/Demangle/MicrosoftDemangle.cpp
index b0f4d9a8a8f..c649552051f 100644
--- a/lib/Demangle/MicrosoftDemangle.cpp
+++ b/lib/Demangle/MicrosoftDemangle.cpp
@@ -696,8 +696,6 @@ SymbolNode *Demangler::demangleEncodedSymbol(StringView &MangledName,
     StorageClass SC = demangleVariableStorageClass(MangledName);
     return demangleVariableEncoding(MangledName, SC);
   }
-  case '8':
-    return nullptr;
   }
   FunctionSymbolNode *FSN = demangleFunctionEncoding(MangledName);
 
diff --git a/test/Demangle/invalid-manglings.test b/test/Demangle/invalid-manglings.test
index b6021761ed2..40bf92df357 100644
--- a/test/Demangle/invalid-manglings.test
+++ b/test/Demangle/invalid-manglings.test
@@ -89,3 +89,8 @@
 ; CHECK-EMPTY:
 ; CHECK-NEXT: ??__E?Foo@@0HA@@
 ; CHECK-NEXT: error: Invalid mangled name
+
+??8@8
+; CHECK-EMPTY:
+; CHECK-NEXT: ??8@8
+; CHECK-NEXT: error: Invalid mangled name