[PATCH] D21489: [codeview] Added support for bitfield type

[Amjad Aboud via llvm-commits]

aaboud added a comment.

In http://reviews.llvm.org/D21489#462178, @majnemer wrote:

> This approach still seems a little problematic, consider:
>
>   #pragma pack(1)
>   struct S {
>     char  : 0;
>     int y : 1;
>   };
>   S s;
>
>
> Your code will produce:
>
>   BitField (0x1001) {
>     TypeLeafKind: LF_BITFIELD (0x1205)
>     Type: int (0x74)
>     BitSize: 1
>     BitOffset: 8
>   }
>   UnknownLeaf (0x1002) {
>     TypeLeafKind: LF_FIELDLIST (0x1203)
>     DataMember {
>       AccessSpecifier: Public (0x3)
>       Type: 0x1001
>       FieldOffset: 0x0
>       Name: y
>     }
>   }
>   
>
> However, I think it should produce:
>
>   BitField (0x1001) {
>     TypeLeafKind: LF_BITFIELD (0x1205)
>     Type: int (0x74)
>     BitSize: 1
>     BitOffset: 0
>   }
>   UnknownLeaf (0x1002) {
>     TypeLeafKind: LF_FIELDLIST (0x1203)
>     DataMember {
>       AccessSpecifier: Public (0x3)
>       Type: 0x1001
>       FieldOffset: 0x1
>       Name: y
>     }
>   }
>   
>
> I'm pretty sure that the only real way to get bitfields right is by encoding the storage offset in the DI metadata.


I think my code will produce:

  BitField (0x1001) {
    TypeLeafKind: LF_BITFIELD (0x1205)
    Type: int (0x74)
    BitSize: 1
    BitOffset: 0
  }
  UnknownLeaf (0x1002) {
    TypeLeafKind: LF_FIELDLIST (0x1203)
    DataMember {
      AccessSpecifier: Public (0x3)
      Type: 0x1001
      FieldOffset: 0x0
      Name: y
    }
  }

Which is the same as MS cl.


http://reviews.llvm.org/D21489