[Unionfs] Can't mount NFS-exported union anymore in unionfs-1.4
Wilhelm Meier
wilhelm.meier at fh-kl.de
Wed Nov 29 07:29:42 EST 2006
Am Mittwoch, 29. November 2006 03:04 schrieb hooanon05 at yahoo.co.jp:
> Wilhelm Meier:
> > Here it comes:
> >
> > unionfs_interpose: c126c800, b0, hi425997
> > read_uin: c126c800, b0, i425997 f0x40, cfd10ca4
> > read_uin:543
> > BUG: unable to handle kernel NULL pointer dereference at virtual address
> > 00000008
>
> I could reproduce the problem.
> The bug is related to parsing options.
> - remove all pathces and use plain unionfs-1.4.
> - If you meet a compile error, fix it by referencing my previous
> patches. you don't need entire patch, just to fix the compile error.
O.k., now I have it cleanly compile
> - specify imap option AFTER dirs option.
o.k., but strange
> - you don't need to create and specify two revmaps unless your branches
> are on the same filesystem.
yes, that's what I did.
Now I can cleanly mount the union and use it on the same system. Looks stable.
So far, so good.
I applied the other patches to avoid the ERR_ACCESS (13) and made a new patch
for unionfs-1.4 with all the relevant part (attached).
Now I can cleanly mount the union from another machine. So far, so good.
But, if I try to use it as nfs-root fpr a diskless machine I get on the
diskless client (as I had before using the imap-option):
nfs_stat_to_errno: bad nfs status return value: 45
after the client has mounted the nfs-root and wants to open the initial
console and start init.
I have a tcpdump of the progress and it seems that the nfs-lookup operation is
unsuccessfull. If you want I can send the tcpdump to you (2MB).
--
Wilhelm Meier
email: wilhelm.meier at fh-kl.de
-------------- next part --------------
A non-text attachment was scrubbed...
Name: unionfs-1.4.imap-nfs.patch
Type: text/x-diff
Size: 3599 bytes
Desc: not available
Url : http://www.fsl.cs.sunysb.edu/pipermail/unionfs/attachments/20061129/6274b012/unionfs-1.4.imap-nfs.bin
More information about the unionfs
mailing list