This happened when doing a bash completion on an NFS mounted directory.
The server was (at least every other machine still thinks so).
nfs: server osdlab not responding, timed out
-----------[ cut here ]------------
kernel BUG at mm/highmem.c:177!
invalid operand: 0000
ide-cd cdrom soundcore mga agpgart autofs nfs lockd sunrpc eepro100 mii
mousede
CPU: 1
EIP: 0060:[<c014df77>] Not tainted
EFLAGS: 00013246
EIP is at kunmap_high+0x47/0xe0
eax: 00000000 ebx: c19bc7b0 ecx: 00000000 edx: c040fa00
esi: f7c05dd0 edi: f5635f18 ebp: c19bc7b0 esp: f5635d48
ds: 0068 es: 0068 ss: 0068
Process bash (pid: 10674, threadinfo=f5634000 task=f6816000)
Stack: c19bc7b0 c19bc7b0 f7c05dd0 f5635f18 fffffffb f8ace19e f7c05dd0
f672bd20
00000000 00000000 c19bc7b0 00001000 00000002 f672bd20 c19bc7b0
00000000
00000000 f7c05e58 c013eb84 f5635f18 c19bc7b0 00000000 f7c05dd0
00000000
Call Trace:
[<f8ace19e>] nfs_readdir_filler+0x13e/0x1a0 [nfs]
[<c013eb84>] __read_cache_page+0x64/0x140
[<c013dd38>] read_cache_page+0x38/0x110
[<f8ace060>] nfs_readdir_filler+0x0/0x1a0 [nfs]
[<f8ad0d40>] find_dirent_page+0x40/0x190 [nfs]
[<f8ace060>] nfs_readdir_filler+0x0/0x1a0 [nfs]
[<f8ad0a58>] readdir_search_pagecache+0x28/0x90 [nfs]
[<f8ace4f8>] nfs_readdir+0x148/0x2b0 [nfs]
[<c0145fad>] __alloc_pages+0x7d/0x280
[<c01399b7>] do_anonymous_page+0x2b7/0x310
[<c0139a5d>] do_no_page+0x4d/0x410
[<c0139ed4>] handle_mm_fault+0xb4/0x1b0
[<f8aded70>] nfs3_decode_dirent+0x0/0x1f0 [nfs]
[<c0169c52>] vfs_readdir+0xb2/0xc0
[<c0169f60>] filldir64+0x0/0x110
[<c016a0cb>] sys_getdents64+0x5b/0x1d9
[<c0169f60>] filldir64+0x0/0x110
[<c012f1ed>] sys_rt_sigprocmask+0x10d/0x270
[<c01099df>] syscall_call+0x7/0xb
Code: 0f 0b b1 00 2f 12 2a c0 05 00 00 00 02 ba 40 e6 40 c0 c1 e8
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs
>>>>> " " == Stephen Hemminger <[email protected]> writes:
> This happened when doing a bash completion on an NFS mounted
> directory. The server was (at least every other machine still
> thinks so).
Hmm. Looks like I missed a 'kunmap()' call when I made the RPC code
copy directly into pages. Does the following fix it?
Cheers,
Trond
--- linux-2.5.41/fs/nfs/dir.c.orig 2002-10-01 19:43:15.000000000 -0400
+++ linux-2.5.41/fs/nfs/dir.c 2002-10-10 22:09:08.000000000 -0400
@@ -130,7 +130,6 @@
return 0;
error:
SetPageError(page);
- kunmap(page);
unlock_page(page);
invalidate_inode_pages(inode->i_mapping);
desc->error = error;
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs