Re: kernel oops with linux 2.4.17-xfs

[Robert Sander <ml-linux-xfs@xxxxxxxxxxxxxxx>]

On Fri, 18 Jan 2002 12:05:47 +0000 (UTC),
 "Johannes Schritz" <johannes@xxxxxxxxxx> wrote:
> 
> No, I used the CVS source.

And I got the same oops right now:

kernel BUG at /mnt/raid/1/src/linux-2.4.17-xfs-new/include/asm/spinlock.h:86!
invalid operand: 0000
CPU:    0
EIP:    0010:[<f88e04d6>]    Tainted: P
EFLAGS: 00010002
eax: 0000004e   ebx: 00000086   ecx: c02d81c4   edx: 00002fce
esi: 00000001   edi: f7c69414   ebp: f7db9ec0   esp: f7db9e8c
ds: 0018   es: 0018   ss: 0018
Process kupdated (pid: 7, stackpage=f7db9000)
Stack: f89134a0 00000056 f1923e24 f16eecfc f78f5a4c 00000000 00000082 f7db9ec0
       00000000 00000000 00000002 f7c69400 f1923e24 f7db9ef8 f88dff01 f16eecfc
       f7292e10 f7c69520 f1228d7c f766a58c 00000282 00000003 f78f5a48 00000000
Call Trace: [<f89134a0>] [<f88dff01>] [<f88f5e4c>] [<f88f569a>] [<f890cc11>]
   [<f89261e0>] [<c014904f>] [<c014823b>] [<c0148845>] [<c0105787>]

Code: 0f 0b 83 c4 08 90 8d 74 26 00 8b 55 f8 c6 42 14 01 53 9d 8b

Entering kdb (current=0xf7db8000, pid 7) on processor 0 Oops: invalid operand 
due to oops @ 0xf88e04d6
eax = 0x0000004e ebx = 0x00000086 ecx = 0xc02d81c4 edx = 0x00002fce
esi = 0x00000001 edi = 0xf7c69414 esp = 0xf7db9e8c eip = 0xf88e04d6
ebp = 0xf7db9ec0 xss = 0x00000018 xcs = 0x00000010 eflags = 0x00010002
xds = 0x00000018 xes = 0x00000018 origeax = 0xffffffff &regs = 0xf7db9e58

Using kdb I got this backtrace:

[0]kdb> bt
    EBP       EIP         Function(args)
0xf7db9ec0 0xf88e04d6 [xfs]xfs_iflush_int+0x32a (0xf16eecfc, 0xf7292e10, 
0xf7c69520, 0xf1228d7c, 0xf766a58c)
                               xfs .text 0xf889e060 0xf88e01ac 0xf88e0520
0xf7db9ef8 0xf88dff01 [xfs]xfs_iflush+0x229 (0xf1228d7c, 0x1, 0xf7c69520, 
0xf7c69800, 0xf7c69848)
                               xfs .text 0xf889e060 0xf88dfcd8 0xf88e01ac
0xf7db9f78 0xf88f5e4c [xfs]xfs_syncsub+0x7ac (0xf7c69400, 0x31, 0x0, 0x0)
                               xfs .text 0xf889e060 0xf88f56a0 0xf88f627c
0xf7db9f90 0xf88f569a [xfs]xfs_sync+0x16 (0xf7c69400, 0x31, 0xf89261e0)
                               xfs .text 0xf889e060 0xf88f5684 0xf88f56a0
0xf7db9fa4 0xf890cc11 [xfs]linvfs_write_super+0x2d (0xf7c69800, 0xf7db8000, 
0xf7db8000)
                               xfs .text 0xf889e060 0xf890cbe4 0xf890cc18
0xf7db9fb8 0xc014904f sync_supers+0x14f (0x0, 0xf7db8000, 0xc022d517)
                               kernel .text 0xc0100000 0xc0148f00 0xc01490c8
0xf7db9fcc 0xc014823b sync_old_buffers+0x67 (0x10f00, 0xc2113f98, 0xc2112000, 
0xf7db87f4, 0xf7db87e4)
                               kernel .text 0xc0100000 0xc01481d4 0xc014836c
0xf7db9fec 0xc0148845 kupdate+0x25d
                               kernel .text 0xc0100000 0xc01485e8 0xc014884c
           0xc0105787 kernel_thread+0x23
                               kernel .text 0xc0100000 0xc0105764 0xc010579c


I was running dbench on an NFSv3 (read kernel nfsd) exported filesystem
and am able to reproduce the problem. It bugs out in spinlock.h,
spinlock debugging is enabled.

Greetings
-- 
Robert Sander
Computer Scientist                                       Epigenomics AG
Bioinformatics R&D        www.epigenomics.com         Kastanienallee 24
+493024345330                                              10435 Berlin