Supply missing break in case statement.

[amccarth]

Summary

The FIXME comment is provocative, but I haven't repro'd yet. This changes
the behavior in the surprising case, but that doesn't seem to affect any
tests, and returning a bool type for void doesn't seem any more correct
than letting the default code handle the odd situation.

[zturner]

Aaron, do you remember why you added a check for width == 0 here? Would it ever not be 0?

[asmith]

That's a good question and I'm not sure. If all the tests pass you can make it return unconditionally.

[amccarth]

Per Aaron's suggestion, I ran the tests with an unconditional return (and an assertion). There was no change in the test results, so I'm making this return unconditionally.

[amccarth]

Any remaining concerns?

[asmith]

Lgtm

lldb/trunk/source/Plugins/SymbolFile/PDB/PDBASTParser.cpp

CompilerType GetBuiltinTypeForPDBEncodingAndBitSize(
   if (!ast)
     return CompilerType();
 
   switch (pdb_type->getBuiltinType()) {
   default: break;
   case PDB_BuiltinType::None:
     return CompilerType();
   case PDB_BuiltinType::Void:
-    // FIXME: where is non-zero size of `void` from?
-    if (width == 0)
-      return clang_ast->GetBasicType(eBasicTypeVoid);
   case PDB_BuiltinType::Bool:
     return clang_ast->GetBasicType(eBasicTypeBool);
   case PDB_BuiltinType::Long:
     if (width == ast->getTypeSize(ast->LongTy))
       return CompilerType(ast, ast->LongTy);
     if (width == ast->getTypeSize(ast->LongLongTy))
       return CompilerType(ast, ast->LongLongTy);
     break;