Hi Alan,
your slab poisoning seems to trigger some bugs in various places.
Attached is a patch for linux/drivers/scsi/scsi.c. It's against 2.2.10-ac8
(with devfs and some other patches, but it should apply cleanly anyway.)
The bug was triggered on removing st, sg, sd, sr modules.
Regards,
--
Kurt Garloff <[EMAIL PROTECTED]> SuSE GmbH, N�rnberg, FRG
Linux kernel development; SCSI driver: DC390 (tmscsim/AM53C974)
--- linux/drivers/scsi/scsi.c.ORIG Mon Jul 5 12:08:43 1999
+++ linux/drivers/scsi/scsi.c Mon Jul 5 11:56:30 1999
@@ -3246,7 +3246,7 @@
static int scsi_unregister_device(struct Scsi_Device_Template * tpnt)
{
Scsi_Device * SDpnt;
- Scsi_Cmnd * SCpnt;
+ Scsi_Cmnd * SCpnt, * SCnext;
struct Scsi_Host * shpnt;
struct Scsi_Device_Template * spnt;
struct Scsi_Device_Template * prev_spnt;
@@ -3274,11 +3274,13 @@
* Nobody is using this device any more. Free all of the
* command structures.
*/
- for(SCpnt = SDpnt->device_queue; SCpnt;
- SCpnt = SCpnt->next)
+ SCpnt = SDpnt->device_queue;
+ if (SCpnt) SCnext = SCpnt->next;
+ for(; SCpnt; SCpnt = SCnext)
{
if(SCpnt == SDpnt->device_queue)
SDpnt->device_queue = SCpnt->next;
+ SCnext = SCpnt->next;
scsi_init_free((char *) SCpnt, sizeof(*SCpnt));
}
SDpnt->has_cmdblocks = 0;
PGP signature
