On Tue, 23 Sep 2003, Nathan Scott wrote:
> On Mon, Sep 22, 2003 at 07:54:38PM -0400, Cory Visi wrote:
> > Hi, I just want to reiterate that this problem is reliably reproducable. I
> > can cut power to the machine as many times as I want and XFS never
> > recovers. I can even do a "sync" beforehand and it still does not recover.
> > Furthermore, if I boot with a boot CD and attempt a mount, XFS recovers
> > fine (on all partitions).
> >
> > For some reason my kernel is not able to recover from an improper
> > shutdown.
> >
> > Attached is a log print generated by xfs_logprint -t off a boot CD as
> > requested.
> >
> > Here's an xfs_info printout of the partition as well:
> > meta-data=/mnt/test isize=256 agcount=3, agsize=436820 blks
> > = sectsz=512
> > data = bsize=4096 blocks=1310460, imaxpct=25
> > = sunit=0 swidth=0 blks, unwritten=0
> > naming =version 2 bsize=4096
> > log =internal bsize=4096 blocks=16384, version=1
> > = sectsz=512 sunit=0 blks
> > realtime =none extsz=65536 blocks=0, rtextents=0
> >
>
> What compiler version are you using Cory? There was a change that
> went into xfs_log.h a little while back in an area that has been
> sensitive to compiler versions in the past (see xfs_log.h) and I
> wonder if that could be the root cause here.
Nathan, I took a look at xfs_log.h from CVS and I'm not sure that's the
issue. I'm using this version of gcc:
gcc (GCC) 3.2.3 20030422 (Gentoo Linux 1.4 3.2.3-r1, propolice)
The lines in xfs_log.h seem to only affect 2.96. Could this still be an
issue? Should I try the 2.96 code?
Thanks,
Cory Visi
|