On 8/19/13 3:19 PM, Mark Tinguely wrote:
<an attachment that doesn't show up on reply, moving d_type support to v4
superblocks ;) >
Has you been able to test this at all?
I do still owe a promised xfstest - but for that, we'll need at least mkfs
& xfs_repair support.
Did you patch up mkfs to do basic tests of your kernelspace patch?
Talking to Dave, he reminds me of a few things this needs, if it's
going to be complete & compatible on V4:
* mkfs.xfs commandline option support & manpage updates
* xfs_db support (including version friendly-text output)
* XFS_IOC_FSGEOM support so that xfs_info can report the difference
* xfs_repair needs to know that it's a valid feature on V4
Some of that may overlap w/ things still needed on V5, but some is unique
to allowing it on V4.
Mark, do you plan to do some of those unique-to-V4 parts, too?
As an aside: I would support getting this feature onto V4 superblocks,
after we have confidence that all the bits are done, tested, and robust.
I still would really like to see it go forward in parallel on V5, and
not be blocked by the extra work needed for proper V4 integration.