Subject: kern/25339: VM subsystem broken.
To: None <gnats-bugs@gnats.NetBSD.org>
From: Denis Lagno <dlagno@mail.nnov.ru>
List: netbsd-bugs
Date: 04/27/2004 01:18:41
>Number:         25339
>Category:       kern
>Synopsis:       starting jobs that allocates big chunks of swappable memory can hang up system
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Mon Apr 26 21:19:00 UTC 2004
>Closed-Date:
>Last-Modified:
>Originator:     Denis Lagno
>Release:        NetBSD 2.0E (Apr 25 snapshot)
>Organization:
>Environment:
System: NetBSD flamante.local.domain 2.0E NetBSD 2.0E (FLAMANTE) #0: Mon Apr 26 12:34:19 MSD 2004 dina@flamante.local.domain:/var/worksrc/netbsd-current/src/sys/arch/i386/compile/FLAMANTE i386
Architecture: i386
Machine: i386
>Description:
Earlier I filed PR kern/25315 "reproducible hangups when creating virtual screens in screen(1)"
I discovered that real reason of hangups was somewhere in swap subsystem.
Hangups occur always when I use swap (I use swap on cgd but swap on normal device
hangs as well) and have jobs allocating big chunks of memory.
Recently I bumped size of scrollback buffer in screen from 10000 to 100000 lines
so screen was such job in my case.
>How-To-Repeat:
have some swap attached.
start some memory consuming activity.
Observe hangup.
>Fix:
>Release-Note:
>Audit-Trail:
>Unformatted: