[Linux-cachefs] kernel panic when using fscache to against an nfs server

[mpreston <mark preston gmail com>]

Hi,

RHEL5.1 i386, fscache and cachefilesd. The application is a web server
(lighttpd) serving static content from an nfs mount which is being cached
locally via fscache/cachefilesd. Works great except that machines in the
above configuration are kernel panic'ing after random amounts of time with
the following:

Stack: c1eb4140 00000000 f8aef3e6 f8aef3b5 000000d0 c0471248 c1eb4140
f4a44ef8
       c045c058 f7f6bf78 00000020 00000000 00000000 00000020 00000010
00000001
       00000000 00000020 00000002 00000001 c1e106e0 c1a408a0 c1e16c60
c1f15a00
Call Trace:
 [<f8aef3e6>] nfs_release_page+0x31/0x40 [nfs]
 [<f8aef3b5>] nfs_release_page+0x0/0x40 [nfs]
 [<c0471248>] try_to_release_page+0x34/0x46
 [<c045c058>] shrink_inactive_list+0x477/0x718
 [<c045badd>] shrink_active_list+0x375/0x37d
 [<f8aedc03>] nfs_access_cache_shrinker+0x20/0x160 [nfs]
 [<c04977d2>] mb_cache_shrink_fn+0x1e/0xb7
 [<c045c3a8>] shrink_zone+0xaf/0xd0
 [<c045c815>] kswapd+0x294/0x39d
 [<c0436031>] autoremove_wake_function+0x0/0x2d
 [<c045c581>] kswapd+0x0/0x39d
 [<c0435f65>] kthread+0xc0/0xeb
 [<c0435ea5>] kthread+0x0/0xeb
 [<c0405c3b>] kernel_thread_helper+0x7/0x10
 =======================
Code: 00 b8 00 00 00 00 0f 45 d0 89 d0 c3 56 53 89 c3 8b 40 0c a8 02 74 4d
8b 43
 10 8b 30 81 ee 34 01 00 00 83 be 30 01 00 00 00 75 08 <0f> 0b 0b 01 36 d2
b0 f8
 8b 03 f6 c4 01 74 0c ba 08 00 00 00 89
EIP: [<f8aef379>] nfs_fscache_release_page+0x1f/0x5b [nfs] SS:ESP
0068:f7f6be20
 <0>Kernel panic - not syncing: Fatal exception

do we know what causes the above? and is there a fix for it ?

thanks
mark