Submitter : nathans Status : open
Assigned Engineer : mostek Priority : 1
*Modified Date : 05/23/00 *Modified User : jtk
*Modified User Domain : sgi.com *Description :
I've seen this panic in the XFS code a few times now, in both
big-endian and little-endian builds (just before Daniel's
BE-only checkin). Only happens under very high load (test is
using sendmail to queue-remote/save-local mail messages to an
xfs fs), and only happens sporadically (increased load seems
to increase the chance of it happening).
kernel is built with debug on, pagebuf meta on.
.....
==========================
ADDITIONAL INFORMATION (ADD)
From: ted kline <jtk@xxxxxxx>
Date: May 23 2000 05:50:04AM
[pvnews version: 1.71]
==========================
> I've seen this panic in the XFS code a few times now, in both
> big-endian and little-endian builds (just before Daniel's
> BE-only checkin). Only happens under very high load (test is
> using sendmail to queue-remote/save-local mail messages to an
> xfs fs), and only happens sporadically (increased load seems
> to increase the chance of it happening).
>
> kernel is built with debug on, pagebuf meta on.
>
>
> EBP EIP Function(args)
> 0xc2c6be10 0xc01c5c73 xfs_inobp_check+0x33( 0xc4a9d000, 0xc551dd40,
> 0xc37727e4, 0xc551d480, 0x1560 )
> 0xc2c6be78 0xc01c9395 xfs_iunlink_remove+0x5c9( 0xc37727e4, 0xc22413c0,
> 0xc4a9d000, 0xc22413c0, 0xc251f3b4 )
I have a fix for the inobp_check problem, will check it in shortly.
-Ted
|