[Sema] Avoid crash on tag-type mismatch (Fixes PR24610)

[vsk]

Summary

We currently assert-fail on the following input:

enum Color { R, G, B };
typedef struct Color C;

This patch makes setTagNameForLinkagePurposes return early if the declaration is invalid.

https://llvm.org/bugs/show_bug.cgi?id=24610

[majnemer]

I think it'd be nice if we keep the assert but make setTagNameForLinkagePurposes early-return if TagFromDeclSpec->isInvalidDecl() returns true. How does that sound to you?

[vsk]

• Keep the assert, return early if the declaration is invalid.

[majnemer]

LGTM

cfe/trunk/lib/Sema/SemaDecl.cpp

if (MangleNumberingContext *MCtx = getCurrentMangleNumberContext(
     Context.setManglingNumber(
         Tag, MCtx->getManglingNumber(
                  Tag, getMSManglingNumber(getLangOpts(), TagScope)));
   }
 }
 
 void Sema::setTagNameForLinkagePurposes(TagDecl *TagFromDeclSpec,
                                         TypedefNameDecl *NewTD) {
+  if (TagFromDeclSpec->isInvalidDecl())
+    return;
+
   // Do nothing if the tag already has a name for linkage purposes.
   if (TagFromDeclSpec->hasNameForLinkage())
     return;
 
   // A well-formed anonymous tag must always be a TUK_Definition.
   assert(TagFromDeclSpec->isThisDeclarationADefinition());
 
   // The type must match the tag exactly;  no qualifiers allowed.

cfe/trunk/test/Sema/enum.c

 void crash(enum E* e) // expected-warning {{declaration of 'enum E' will not be visible outside of this function}} \
                       // expected-warning {{ISO C forbids forward references to 'enum' types}}
 {
         PR8694(e); // expected-warning {{incompatible pointer types passing 'enum E *' to parameter of type 'int *'}}
 }
 
 typedef enum { NegativeShort = (short)-1 } NegativeShortEnum;
 int NegativeShortTest[NegativeShort == -1 ? 1 : -1];
+
+// PR24610
+enum Color { Red, Green, Blue }; // expected-note{{previous use is here}}
+typedef struct Color NewColor; // expected-error {{use of 'Color' with tag type that does not match previous declaration}}