From owner-freebsd-bugs  Sun May 28 06:10:02 1995
Return-Path: bugs-owner
Received: (from majordom@localhost)
          by freefall.cdrom.com (8.6.10/8.6.6) id GAA09205
          for bugs-outgoing; Sun, 28 May 1995 06:10:02 -0700
Received: (from gnats@localhost)
          by freefall.cdrom.com (8.6.10/8.6.6) id GAA09196
          ; Sun, 28 May 1995 06:10:01 -0700
Date: Sun, 28 May 1995 06:10:01 -0700
Message-Id: <199505281310.GAA09196@freefall.cdrom.com>
From: Pierre Beyssac <pb@fasterix.freenix.fr>
Reply-To: Pierre Beyssac <pb@fasterix.freenix.fr>
To: freebsd-bugs
Subject: kern/452: page fault in pagedaemon on vnode swapping
In-Reply-To: Your message of Sun, 28 May 1995 14:56:52 +0200
	<199505281256.OAA17468@fasterix.frmug.fr.net>
Sender: bugs-owner@FreeBSD.org
Precedence: bulk


>Number:         452
>Category:       kern
>Synopsis:       vnode swapping panics
>Confidential:   no
>Severity:       critical
>Priority:       high
>Responsible:    freebsd-bugs (FreeBSD bugs mailing list)
>State:          open
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Sun May 28 06:10:01 1995
>Originator:     Pierre Beyssac
>Organization:
Pierre Beyssac
>Release:        FreeBSD 2.0-BUILT-19950528 i386
>Environment:

	Diskless 486 PC.
	RAM : 4Mb
	Swap : 8Mb vnconfig'd in /swp

>Description:

	Swapping works for awhile (swapinfo indicates 10-15% used).
	After some short time (apparently when a big amount of pages
	is swapped out at the same time to make room, e.g. to start
	"vi") the same panic occurs.

Hand-copied from DDB :

trap 12
page fault while in kernel mode
supervisor read, page not present
process 2 (pagedaemon)

Stack trace:

bcopy		(apparently in macro nfsm_fhtom() in nfs_vnops.c:912)
nfs_writerpc
nfs_doio
nfs_strategy
vnstrategy
spec_strategy
swstrategy
spec_strategy
swap_pager_output
swap_pager_putmulti
vm_pager_put_pages
vm_pageout_clean
vm_pageout_scan
vm_pageout
getpbuf

>How-To-Repeat:

	Make /swp a 8Mb file.

	Boot the diskless (4Mb) PC.

	# vnconfig -e /dev/vn0c /swp swap
	# swapinfo
	# vi something

>Fix:
	
	None known.
	Direct NFS-swap doesn't help : many processes get a SIGSEGV
	and the kernel quickly panics too.
	Probably local swap works, I'll try that next.
>Audit-Trail:
>Unformatted: