On Mon, Oct 28, 2002 at 03:25:17PM -0600, Steve Lord wrote:
> On Sun, 2002-10-27 at 22:05, Eric Sandeen wrote:
> > Hi, I'll have to look a bit more closely... but your patch does look
> > correct.
> > Back in 2000 (v. 1.279), it was as you suggest, but Steve changed it.
> > :) Steve?
>
> I no longer remember the why of this change, I think we can just revert
> to this code.
Whem I look at the diff at
http://oss.sgi.com/cgi-bin/cvsweb.cgi/linux-2.4-xfs/linux/fs/xfs/xfs_vfsops.c.diff?r1=1.279&r2=1.280
I don't know either what your rationale for this was, but it seems we need
to revert the ip = ip->i_mnext assignment before the continue statements then.
Still trying to understand that commit, especially as the commit message
isn't exactly helpfull..
>
> Steve
>
> >
> > (The Irix code also has your "new" test.)
> >
> > Thanks,
> >
> > -Eric
> >
> > On Mon, 28 Oct 2002, ASANO Masahiro wrote:
> >
> > > Hi,
> > >
> > > I have a question.
> > > Why don't we sync the last xfs_inode in xfs_syncsub()?
> > > I am wondering if the condition check is wrong... (see below)
> > >
> > > --
> > > Masano
> > >
> > > --- linux/fs/xfs/xfs_vfsops.c Thu Oct 24 07:46:17 2002
> > > +++ linux/fs/xfs/xfs_vfsops.c.new Mon Oct 28 11:29:21 2002
> > > @@ -1433,7 +1433,7 @@
> > > ASSERT(ipointer_in == B_FALSE);
> > > ip = ip->i_mnext;
> > >
> > > - } while (ip->i_mnext != mp->m_inodes);
> > > + } while (ip != mp->m_inodes);
> > >
> > > XFS_MOUNT_IUNLOCK(mp);
> > >
> > >
> > >
> >
> --
>
> Steve Lord voice: +1-651-683-3511
> Principal Engineer, Filesystem Software email: lord@xxxxxxx
>
>
---end quoted text---
|