From owner-freebsd-questions  Tue Apr  8 11:34:39 1997
Return-Path: <owner-questions>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.5/8.8.5) id LAA22626
          for questions-outgoing; Tue, 8 Apr 1997 11:34:39 -0700 (PDT)
Received: from localhost (PM-205.Muskegon.novagate.net [205.138.137.69])
          by freefall.freebsd.org (8.8.5/8.8.5) with SMTP id LAA22618
          for <questions@freebsd.org>; Tue, 8 Apr 1997 11:34:34 -0700 (PDT)
Received: (from eyager@localhost) by localhost (8.6.12/8.6.12) id LAA00615; Tue, 8 Apr 1997 11:30:55 GMT
Date: Tue, 8 Apr 1997 11:30:54 +0000 ()
From: Eric Yagerlener <eyager@novagate.com>
X-Sender: eyager@localhost
To: questions@freebsd.org
Subject: vm_fault bug in 2.1.0?
Message-ID: <Pine.BSF.3.91.970408104952.199A-100000@localhost>
X-Secure: never
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Sender: owner-questions@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

I have a problem with FreeBSD 2.1.0 that might be a well known problem, 
or as a result of me pushing the system beyond capacity.

I have 32M of ram and and 27M of swap.  The error is easily reproducable.

I logged in as root and did the following command on a directory that 
contained 35 files that were 11.6M long total and were compressed with 
'compress'.

zcat * |less

Things work well and I'm able to scroll through the files.  I decide to go
searching for keywords through all these files using less's search
command.  It works fine at first, but as you get closer to the end of the
files, things get gradually slower and slower.  Finally it reaches the
point to where the system will suddenly do nothing but disk swapping, and
all other processes are completely dead in the water and eventually die
because of the lack of memory.  However, less won't die and keeps on
thrashing away despite all the vm_faults it's generating.  My guess is
that less is spawning a child process that will thrash the system until a
vm_fault occurs kills that child process, less will simply ignore it and
create another one that does the exact same thing an infinite amount of 
times.

You can't kill less because you don't have enough memory to execute ps or 
the kill command.  The only way of shutting down is to hit 
control-alt-delete.

Here's are the kernel messages.

Apr  8 10:22:47 wolfden /kernel: Process 63 killed by vm_fault -- out of swap
Apr  8 10:22:47 wolfden /kernel: swap_pager: out of space
Apr  8 10:22:47 wolfden last message repeated 4 times
Apr  8 10:22:47 wolfden /kernel: Process 648 killed by vm_fault -- out of swap
Apr  8 10:22:47 wolfden /kernel: Process 649 killed by vm_fault -- out of swap
Apr  8 10:22:47 wolfden /kernel: Process 650 killed by vm_fault -- out of swap
Apr  8 10:22:47 wolfden /kernel: Process 651 killed by vm_fault -- out of swap
Apr  8 10:22:47 wolfden /kernel: Process 652 killed by vm_fault -- out of swap
Apr  8 10:22:47 wolfden /kernel: Process 653 killed by vm_fault -- out of swap
Apr  8 10:22:47 wolfden /kernel: Process 654 killed by vm_fault -- out of swap
Apr  8 10:22:47 wolfden /kernel: Process 96 killed by vm_fault -- out of swap
Apr  8 10:22:47 wolfden /kernel: Process 655 killed by vm_fault -- out of swap
Apr  8 10:22:47 wolfden /kernel: Process 656 killed by vm_fault -- out of swap

Is this the vm_fault bug that I heard mention of?

eyager@novagate.com