Revert r133024, "[format strings] correctly suggest correct type for '%@'
specifiers.  Fixes <rdar://problem/9607158>." because it causes false positives
on some code that uses CF toll free bridging.
 - I'll let Doug or Ted figure out the right fix here, possibly just to accept
   any pointer type.

llvm-svn: 134041
diff --git a/clang/lib/Analysis/FormatString.cpp b/clang/lib/Analysis/FormatString.cpp
index be214e0..c1b5ea8 100644
--- a/clang/lib/Analysis/FormatString.cpp
+++ b/clang/lib/Analysis/FormatString.cpp
@@ -300,8 +300,7 @@
         argTy->isNullPtrType();
 
     case ObjCPointerTy:
-      return argTy->getAs<ObjCObjectPointerType>() != NULL ||
-             argTy->getAs<BlockPointerType>() != NULL;
+      return argTy->getAs<ObjCObjectPointerType>() != NULL;
   }
 
   // FIXME: Should be unreachable, but Clang is currently emitting