Subject: kern/13966: ktrace on local FFS filesystem causes a kernel fault
To: None <gnats-bugs@gnats.netbsd.org>
From: None <nathanw@mit.edu>
List: netbsd-bugs
Date: 09/15/2001 21:59:52
>Number:         13966
>Category:       kern
>Synopsis:       ktrace on local FFS filesystem causes a kernel fault
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sat Sep 15 19:00:00 PDT 2001
>Closed-Date:
>Last-Modified:
>Originator:     Nathan J. Williams
>Release:        NetBSD-current 2001-09-15
>Organization:
	Massachvsetts Institvte of Technology
>Environment:
System: NetBSD speedy-gonzales.putnam 1.5Y NetBSD 1.5Y (SPEEDY-OLDPMAP) #0: Sat Sep 15 21:11:37 EDT 2001 nathanw@speedy-gonzales.putnam:/u1/build/SPEEDY-OLDPMAP macppc
Architecture: powerpc
Machine: macppc
>Description:
	On a macppc system with today's sources (1.5Y with the
"ubcperf" code), I observed config(8) behaving oddly - reporting a
syntax error at line 1 of a file that hadn't changed
recently. Suspecting something odd, I ktrace'd config, and got a
useless ktrace:
   443 config   CALL  exit(0x1)

I noticed that I was in a NFS-mounted directory, wondered if that had
anything to do with it, and ran "ktrace -f /var/tmp/ktrace.out config
...". The machine wedged.

I rebooted, didn't start X, and did:

cd /var/tmp
ktrace -f foo ls

and was rewarded with "panic: trap". The traceback in DDB was as
follows:

panic + 158
trap + 8a4
kernel MCHK trap by __lshrdi3 + 0
ffs_gop_size (forgot to copy the offset)
0x615b07c
ufs_balloc_range + 188
ffs_write + 4ec
VOP_WRITE + 40
vn_write + 150
ktrwrite + fc
ktrnamei + 58
namei + 130
check_exec + 54
sys_execve + e0
trap + 5c4

(My typing fingers are getting tired of this lack of crash dumps on
powerpc).

>How-To-Repeat:
	See above.
>Fix:
	Unknown.

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