Re: xfsrestore dumps core

[Simon Matter <simon.matter@ch.sauter-bc.com>]

Timothy Shimmin schrieb:
> 
> Hi Simon,
> 
> On Thu, Aug 23, 2001 at 02:39:20PM +0200, Simon Matter wrote:
> >
> > I was just trying to xfsdump and -restore some files on my test system
> > and I get a core dump while restoring.
> > The system is PII/400, 256MB, IDE disk, DLT1 tape drive /DELL PowerVault
> > 110T). RH XFS 1.0.1 release, newest command rpms, including
> > xfsdump-1.1.3-0.
> >
> > BTW /dev/tape is a link to /dev/st0
> >
> > I did the dump with:
> > xfsdump -f /dev/tape -l 0 -s home -o -L session1 -M medium1 /
> >
> > No problems so far, then is tried restore with:
> > [root@looser /root]# xfsrestore -f /dev/tape xfsrestore
> > xfsrestore: drive_scsitape.c:1492: do_next_mark: Assertion
> > `rechdrp->first_mark_offset - rechdrp->file_offset <= ( off64_t )(
> > contextp->dc_recsz )' failed.
> > Aborted (core dumped)
> >
> Ok so this is the assertion bug (pv#831275 in SGI).

I have made a few more tests. It's interesting that it's working
sometimes. When I dump/restore some Mb's, it's okay. With some 100Mb it
dumps core.
If you're interested in the core, just let me know. it's ~2M bzipped.

-Simon

> 
> > I saw some discussion about this some weeks ago on this list but I was
> > wondering it's not yet fixed or I'm doing something completly wrong ?
> I haven't looked into this one yet.
> 
> Someone suggested a fix by not endian converting some value.
> However, I wasn't convinced this was the right thing to do
> and so didn't check it in.
> 
> Ciao,
> Tim.