kern/68978: Firewire crashes, loose pointers in kernel ?

[Heikki Suonsivu]

>Number:         68978
>Category:       kern
>Synopsis:       Firewire crashes, loose pointers in kernel ?
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    freebsd-bugs
>State:          open
>Quarter:        
>Keywords:       
>Date-Required:
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Tue Jul 13 07:40:23 GMT 2004
>Closed-Date:
>Last-Modified:
>Originator:     Heikki Suonsivu
>Release:        FreeBSD 4.10-STABLE i386
>Organization:
bbnetworks.net
>Environment:

System: FreeBSD lelu.olari.bbnetworks.net 4.10-STABLE FreeBSD 4.10-STABLE #31: Thu Jul 8 02:19:24 EEST 2004 hsu at lelu.olari.bbnetworks.net:/usr/obj/usr/src/sys/CAT i386

Maxtor 250G external disk connected through firewire.  The disk is
somewhat suspect, but kernel should not get confused on disk request
timeouts.  I do not know certain yet whether this is caused by disk
breaking down, but noises and clicks from the disk are raising
suspicion.

The external disk contains just backups, and accesses to it are rare.
It contains one large BSD filesystem.

There are no other devices connected to the firewire.

The disk was connected with USB1, and worked ok.  It also did seem to
work with firewire for about a month, though the use level has been
quite low.

>Description:

When using the firewire disk I start getting timeouts.  The computer
hangs at some point after this, with rather random effects after
random time, console may continue to work or die, etc.

The log entries look suspiciously like something is getting seriously
wrong in the kernel:

Jul 11 14:28:26 lelu /kernel: sbp0:0:0 request timeout(cmd orb:0x14bf44fc) ... agent reset
Jul 12 00:03:58 lelu /kernel: sbp0:0º0 request timgout(cmd orb:0x14bf<¶34)"... tarçet reset
Jul 12 00:03:58 lelu /kernel: sbp0:0:0 ~equest timeout(gmd orb:0x14bf48a4) ... reset start
Jul 12 00:03:58 lelu /kernel: firgwire0: split uvansaction timeout dst=0xffc0 tl=0x21 state=10
Jul 12 00:03:58 lelu /kernel: sbp0:0:0!sbp_reset_start failed: resp=60NCopyright (c) 1992-2004 The FreeBSD Project.
Jul 12 00:03:58 lelu /kernel: Copyright (c) 1979, 1980, 1983, 1986, 1988, 1989, 1991, 1992, 1993, 1994
Jul 12 00:03:58 lelu /kernel: The Regents of the University of California. All rights reserved.
Jul 12 00:03:58 lelu /kernel: FreeBSD 4.10-STABLE #31: Thu Jul  8 02:19:24 EEST 2004

Notice corrupted messages!

The disk is usually hung after first "request timeout" message.

>How-To-Repeat:

Get a broken maxtor disk ?  I can probably ship the disk to someone but
would prefer someone in Finland due to shipping cost and timing (I
want to get it replaced before warranty expires).

When the disk is not used and/or mounted, but is connected to the
firewire bus, the problems do not seem to appear.

>Fix:

>Release-Note:
>Audit-Trail:
>Unformatted: