Subject: kern/4364: "panic: vrele: ref cnt"
To: None <gnats-bugs@gnats.netbsd.org>
From: None <nocturne@cygnus.com>
List: netbsd-bugs
Date: 10/27/1997 12:49:59
>Number:         4364
>Category:       kern
>Synopsis:       "panic: vrele: ref cnt" occurs unpredictably
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people (Kernel Bug People)
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Mon Oct 27 09:50:01 1997
>Last-Modified:
>Originator:     Eric Mumpower
>Organization:
Cygnus Solutions
	
>Release:        Oct  6 23:55 EDT 1997
>Environment:
	
System: NetBSD dr-teeth.cygnus.com 1.2G NetBSD 1.2G (TEETH) #6: Tue Oct 7 00:28:22 EDT 1997 nocturne@dr-teeth.cygnus.com:/u2/nocturne/netbsd/usr/src/sys/arch/i386/compile/TEETH i386

>Description:

kernel prints "panic: vrele: ref cnt"; stack trace transcribed here:

_panic (f813b214, f813b1ff f88ee600 fccfaf28) at _panic+0x46
_vrele (f88ee600) at _vrele+0x46
_rename_files (f7bfd8a7, f7bfd8b2, f89e7000, 0, fccfafa8) at _rename_files+0x386
_sys_rename (f89e7000, fccfaf88, fccfaf80, 0, f7bfd8b2) at _sys_rename+0x15
_syscall() at _syscall+0x248
--- syscall (number 128) ---

The kernel I was running before updating my netbsd-current snapshot on
October 6 was from sometime near September 17 1997. That (older)
kernel suffered panics of exactly this sort 3-10 times a week.
However, I was unable to debug the panics because (1) I only
encountered them while my X server was in use, (2) the PC-console
debugger doesn't work when the console is in VGA mode, and (3) the
serial debugger I had configured wasn't accepting input once the
debugger was invoked.

Since building and booting a new kernel after updating my
netbsd-current snapshot on October 6, *these panics had not recurred
until today* (i.e. my machine didn't panic for a period of almost 3
weeks), at which point I discovered (to my great surprise and
happiness) that the serial-console debugger had been fixed. If this
panic occurs again, I'm prepared to generate a crash dump which will
allow a more detailed bug report.

>How-To-Repeat:
Tends to occur in the first few minutes after initiating a login
session via an xdm running on the local display. Cause of correlation
unknown.
>Fix:
No clue.
>Audit-Trail:
>Unformatted: