Device loses barrier support

[Timothy Shimmin]

Andi Kleen wrote:
> The write block with barrier bit varies, jbd/gfs2 do it synchronously
> too and xfs does it asynchronously (with io done callbacks), but
> in both cases they handle an EOPNOTSUPP comming out in the final
> io done.
> 
Yes, XFS handles it, however,
it doesn't look like we currently propagate the EOPNOTSUPP
up to where we test it (not set for b_error).
Patch disscussed recently on xfs list to rectify this.

--Tim