Re: Review: Be smarter about handling ENOSPC during writeback

[Nathan Scott <nscott@xxxxxxxxxx>]

On Fri, 2007-06-08 at 17:33 +1000, David Chinner wrote:
> On Fri, Jun 08, 2007 at 03:28:14PM +1000, Timothy Shimmin wrote:

> > Will we get questions from people about reduced space from df? :)
> 
> If we do, I think you just volunteered to write the FAQ entry ;)

It would be more correct of XFS to start doing the right thing by
reporting different values for b_free and b_avail in statfs(2) -
this code in xfs_mount.c::xfs_statvfs() ...

    statp->f_bfree = statp->f_bavail =
                            sbp->sb_fdblocks - XFS_ALLOC_SET_ASIDE(mp);

I know this wasn't done for the original per-mount space reserving
ioctls (that was for one user though - dmapi, so I can see why there
may have been a shortcut done there) ... but if it affects everyone
now, there will be questions asked, and there is a standard interface
for reporting this space discrepency that tools like df(1) already
use.

cheers.