[OpenAFS] OpenAFS 1.3.84 on Linux 2.4.21-15.0.3 - "dcache hc" panic in afs_HashOutDCache

Rainer Toebbicke rtb@pclella.cern.ch
Thu, 23 Jun 2005 12:18:56 +0200


I can reproducibly crash a 2.4.21-27 i386 SMP machine running 1.3.84 
with just "STABLE14-linux-change-dentry-cleanup-20050619" on top of it.

typical traceback:

dcache hc<1>Unable to handle kernel paging request at virtual address 
ffffffff
  printing eip:
f8a9dbd0
*pde = 00000000
Oops: 0002
xfs_quota xfs_dmapi xfs openafs tg3 e100 floppy sg microcode keybdev 
mousedev hid input usb-ohci usbcore ext3 jbd raid1 3w-xxxx aic7xxx 
sd_mod scsi_mod
CPU:    0
EIP:    0060:[<f8a9dbd0>]    Tainted: PF
EFLAGS: 00010282

EIP is at osi_Panic [openafs] 0x20 (2.4.21-15.0.3.EL.cernsmp/i686)
eax: 00000009   ebx: f8c5f000   ecx: 00000046   edx: c03a0e94
esi: f8c82488   edi: 00000000   ebp: 00014263   esp: f58efdd4
ds: 0068   es: 0068   ss: 0068
Process afs_cachetrim (pid: 3272, stackpage=f58ef000)
Stack: f8ac48d9 f8c38f08 c03c879c f8c824fc f8c82488 f58ee000 00014263 
f8a5d83f
        f8ac48d9 f8c38f08 c03c879c f8c824fc f8ad8ae0 f8c82488 f58ee000 
f8a5d4fb
        f8c82488 00000000 00000000 00000008 f6eb2c54 f88902df f6eb2c54 
0000000b
Call Trace:   [<f8ac48d9>] .rodata.str1.1 [openafs] 0x111 (0xf58efdd4)
[<f8a5d83f>] afs_HashOutDCache [openafs] 0x7f (0xf58efdf0)
[<f8ac48d9>] .rodata.str1.1 [openafs] 0x111 (0xf58efdf4)
[<f8ad8ae0>] afs_global_lock [openafs] 0x0 (0xf58efe04)
[<f8a5d4fb>] afs_GetDownD [openafs] 0x6ab (0xf58efe10)
[<f88902df>] ext3_do_update_inode [ext3] 0x16f (0xf58efe28)
[<f8a5ca90>] afs_CacheTruncateDaemon [openafs] 0x110 (0xf58eff80)
[<f8ab0b48>] afsd_thread [openafs] 0x378 (0xf58effc0)
[<f8ac5a4f>] .rodata.str1.1 [openafs] 0x1287 (0xf58effc8)
[<f8ab07d0>] afsd_thread [openafs] 0x0 (0xf58effe0)
[<c010957d>] kernel_thread_helper [kernel] 0x5 (0xf58efff0)

Code: c6 05 ff ff ff ff 2a 83 c4 1c c3 90 8d 74 26 00 b8 09 54 ac

Kernel panic: Fatal exception


Entering kdb (current=0xf58ee000, pid 3272) on processor 0 due to 
KDB_ENTER()


The same [stress] test runs flawlessly on a 2.6 amd64 machine - though 
without the "linux-change-dentry-cleanup" patch. As the latter mainly 
affects 2.4 kernels I didn't bother yet trying withdrawing it, should I?

-- 
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Rainer Toebbicke
European Laboratory for Particle Physics(CERN) - Geneva, Switzerland
Phone: +41 22 767 8985       Fax: +41 22 767 7155