Hi,
When my networking card died yesterday (sundance based d-link card) i
used dhclient to update the route. Unfortunately it caused it to change
ip, thus today i had some NFS problems...
fstab entry:
blue:/mnt/large /mnt/large nfs
user,async,soft,rsize=32768,wsize=32768,tcp 0 0
Anyways, i did a bunch of umount -f's and so on and so forth as i have
done a million times before, but when i tried remounting it, nothing
happened... A quick ps aux |grep revealed:
root 26793 0.0 0.0 1740 696 ? D 14:23 0:00
/bin/umount /mnt/large
root 26800 0.0 0.0 1868 828 ? D 14:23 0:00
/bin/mount /mnt/large
root 26805 0.0 0.0 1868 828 ? D 14:23 0:00
/bin/mount /mnt/large
root 26807 0.0 0.0 1868 828 ? D 14:23 0:00
/bin/mount /mnt/large
root 26809 0.0 0.0 1868 828 ? D 14:23 0:00
/bin/mount /mnt/large
And to further my agony a dmesg revealed:
------------[ cut here ]------------
kernel BUG at fs/nfs/inode.c:152!
invalid operand: 0000 [#1]
Modules linked in: nls_cp437 vfat fat floppy nvidia twofish serpent blowfish sha256 crypto_null ip6table_filter ip6_tables forcedethCPU: 0
EIP: 0060:[<c01a8bb0>] Tainted: P VLI
EFLAGS: 00210286 (2.6.9-rc1-bk7)
EIP is at nfs_clear_inode+0x50/0x70
eax: ffffffff ebx: f70595a8 ecx: edc36c60 edx: f710cf60
esi: f7059490 edi: 000000ce ebp: c042792c esp: c05cbf04
ds: 007b es: 007b ss: 0068
Process umount (pid: 26773, threadinfo=c05ca000 task=df308bb0)
Stack: 00000002 f70595a8 c05cbf2c c016058d f70595a8 c01605dc 00000000 c1bff64c
c1bff600 c0160736 c05cbf2c c05cbf2c c1bff600 c1bff64c c042a460 c05ca000
c01506f5 ef95c000 00000015 c042a5a0 08050278 c0150dd9 f7e8cc00 c01ab2dc
Call Trace:
[<c016058d>] clear_inode+0x8d/0xa0
[<c01605dc>] dispose_list+0x3c/0x70
[<c0160736>] invalidate_inodes+0x66/0x80
[<c01506f5>] generic_shutdown_super+0x55/0x100
[<c0150dd9>] kill_anon_super+0x9/0x20
[<c01ab2dc>] nfs_kill_super+0xc/0x70
[<c0150623>] deactivate_super+0x43/0x60
[<c0162e4b>] sys_umount+0x3b/0x90
[<c03a693d>] schedule+0x27d/0x450
[<c0103db9>] sysenter_past_esp+0x52/0x71
Code: d4 39 43 d4 75 2f 8b 86 b4 00 00 00 85 c0 74 05 e8 f6 f8 1e 00 8b 86 ac 00 00 00 85 c0 75 0c 8b 5c 24 04 8b 74 24 08 83 c4 0c c3 <0f> 0b 98 00 7a 28 3c c0 eb ea 0f 0b 94 00 7a 28 3c c0 eb c7 8d
This is, as stated before, just FYI... Anyone feeling like making nvidia
comments can read this and just skip mailing "/me no care".
CC, Since i'm not subbed.
--
Ian Kumlien <pomac () vapor ! com> -- http://pomac.netswarm.net
Hello Ian,
This looks very much like the problem in iget() I described today in
LKML. Hope I'll post a patch here soon.
Kirill
> When my networking card died yesterday (sundance based d-link card) i
> used dhclient to update the route. Unfortunately it caused it to change
> ip, thus today i had some NFS problems...
>
> fstab entry:
> blue:/mnt/large /mnt/large nfs
> user,async,soft,rsize=32768,wsize=32768,tcp 0 0
>
> Anyways, i did a bunch of umount -f's and so on and so forth as i have
> done a million times before, but when i tried remounting it, nothing
> happened... A quick ps aux |grep revealed:
>
> root 26793 0.0 0.0 1740 696 ? D 14:23 0:00
> /bin/umount /mnt/large
> root 26800 0.0 0.0 1868 828 ? D 14:23 0:00
> /bin/mount /mnt/large
> root 26805 0.0 0.0 1868 828 ? D 14:23 0:00
> /bin/mount /mnt/large
> root 26807 0.0 0.0 1868 828 ? D 14:23 0:00
> /bin/mount /mnt/large
> root 26809 0.0 0.0 1868 828 ? D 14:23 0:00
> /bin/mount /mnt/large
>
> And to further my agony a dmesg revealed:
>
> ------------[ cut here ]------------
> kernel BUG at fs/nfs/inode.c:152!
> invalid operand: 0000 [#1]
> Modules linked in: nls_cp437 vfat fat floppy nvidia twofish serpent blowfish sha256 crypto_null ip6table_filter ip6_tables forcedethCPU: 0
> EIP: 0060:[<c01a8bb0>] Tainted: P VLI
> EFLAGS: 00210286 (2.6.9-rc1-bk7)
> EIP is at nfs_clear_inode+0x50/0x70
> eax: ffffffff ebx: f70595a8 ecx: edc36c60 edx: f710cf60
> esi: f7059490 edi: 000000ce ebp: c042792c esp: c05cbf04
> ds: 007b es: 007b ss: 0068
> Process umount (pid: 26773, threadinfo=c05ca000 task=df308bb0)
> Stack: 00000002 f70595a8 c05cbf2c c016058d f70595a8 c01605dc 00000000 c1bff64c
> c1bff600 c0160736 c05cbf2c c05cbf2c c1bff600 c1bff64c c042a460 c05ca000
> c01506f5 ef95c000 00000015 c042a5a0 08050278 c0150dd9 f7e8cc00 c01ab2dc
> Call Trace:
> [<c016058d>] clear_inode+0x8d/0xa0
> [<c01605dc>] dispose_list+0x3c/0x70
> [<c0160736>] invalidate_inodes+0x66/0x80
> [<c01506f5>] generic_shutdown_super+0x55/0x100
> [<c0150dd9>] kill_anon_super+0x9/0x20
> [<c01ab2dc>] nfs_kill_super+0xc/0x70
> [<c0150623>] deactivate_super+0x43/0x60
> [<c0162e4b>] sys_umount+0x3b/0x90
> [<c03a693d>] schedule+0x27d/0x450
> [<c0103db9>] sysenter_past_esp+0x52/0x71
> Code: d4 39 43 d4 75 2f 8b 86 b4 00 00 00 85 c0 74 05 e8 f6 f8 1e 00 8b 86 ac 00 00 00 85 c0 75 0c 8b 5c 24 04 8b 74 24 08 83 c4 0c c3 <0f> 0b 98 00 7a 28 3c c0 eb ea 0f 0b 94 00 7a 28 3c c0 eb c7 8d
>
> This is, as stated before, just FYI... Anyone feeling like making nvidia
> comments can read this and just skip mailing "/me no care".
>
> CC, Since i'm not subbed.
On Mon, 2004-09-06 at 16:27, Kirill Korotaev wrote:
> Hello Ian,
>
> This looks very much like the problem in iget() I described today in
> LKML. Hope I'll post a patch here soon.
ahhhh, ok, Must have missed it etc, though i only read lkml trough
marc.theaimsgroup.com =P
I'm looking forward to seeing a patch included =)
--
Ian Kumlien <pomac () vapor ! com> -- http://pomac.netswarm.net