From root@jc.f1.ru  Fri May 10 07:33:23 1996
Received: from relay.f1.ru (jc.f1.ru [194.87.86.1])
          by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id HAA22217
          for <FreeBSD-gnats-submit@freebsd.org>; Fri, 10 May 1996 07:33:07 -0700 (PDT)
Received: by relay.f1.ru id AA00717
  (5.65/IDA-simtel  for FreeBSD-gnats-submit@freebsd.org); Fri, 10 May 1996 14:31:17 GMT
Message-Id: <199605101431.AA00717@relay.f1.ru>
Date: Fri, 10 May 1996 14:31:17 GMT
From: Super user <root@jc.f1.ru>
Reply-To: root@jc.f1.ru
To: FreeBSD-gnats-submit@freebsd.org
Subject: MFS doesn't mark memory free when it's filespace cleaned
X-Send-Pr-Version: 3.2

>Number:         1186
>Category:       kern
>Synopsis:       MFS doesn't mark memory free when it's filespace cleaned
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    freebsd-bugs
>State:          closed
>Quarter:
>Keywords:
>Date-Required:
>Class:          change-request
>Submitter-Id:   current-users
>Arrival-Date:   Fri May 10 07:40:02 PDT 1996
>Closed-Date:    Sun Jun 9 10:26:05 PDT 1996
>Last-Modified:  Sun Jun  9 10:26:34 PDT 1996
>Originator:     Super user
>Release:        FreeBSD 2.1-STABLE i386
>Organization:
F1 Communications
>Environment:

	Any versions before and including 2.1-RELEASE I think..

>Description:

	Memory, occupied by mount_mfs becames dirty and lives in
	swap during work..

>How-To-Repeat:

	1) mount_mfs /dev/wd0s1b /mnt (or whatever device appropriate)
	2) swapinfo (near your normal swap usage)
	3) dd if=/dev/rwd0 of=/tmp/bigfile bs=1024 count=30000
	   (or what you prefer to make long file in /mnt)
	4) swapinfo (swap usage will increase if you have no enough
	             real memory)
        5) rm -f /tmp/bigfile
        6) swapinfo (swap usage will not decrease to your normal values)

	Congratulations - you just converted at least part of your
	swap partitions into /mnt filespace. I think it's not MFS
	was coded for :(

>Fix:
	
	Don't know.. May be mark some pages `clean' (if such
	technology exist?) when file removed..

	Or dynamically allocate memory `on demand'..

>Release-Note:
>Audit-Trail:

From: Poul-Henning Kamp <phk@critter.tfs.com>
To: root@jc.f1.ru
Cc: FreeBSD-gnats-submit@freebsd.org
Subject: Re: kern/1186: MFS doesn't mark memory free when it's filespace cleaned 
Date: Fri, 10 May 1996 17:45:25 +0000

 This is expected behaviour.
 
 (see the paper in /share/doc/papers on the subject)
 
 Real solution:
 MFS is a hack, and a real "VMFS" should be made instead for "/tmp" use.
 Kirk said that this would be "really easy" so now I hope he will assign
 it as homework for Jordan :-)
 
 
 --
 Poul-Henning Kamp           | phk@FreeBSD.ORG       FreeBSD Core-team.
 http://www.freebsd.org/~phk | phk@login.dknet.dk    Private mailbox.
 whois: [PHK]                | phk@ref.tfs.com       TRW Financial Systems, Inc.
 Future will arrive by its own means, progress not so.

From: Garrett Wollman <wollman@lcs.mit.edu>
To: freebsd-gnats-submit@freebsd.org
Cc:  Subject: Re: kern/1186: MFS doesn't mark memory free when it's filespace cleaned 
Date: Fri, 10 May 1996 14:35:03 -0400

 <<On Fri, 10 May 1996 11:10:01 -0700 (PDT), Poul-Henning Kamp <phk@critter.tfs.com> said:
 
 >  MFS is a hack, and a real "VMFS" should be made instead for "/tmp" use.
 >  Kirk said that this would be "really easy" so now I hope he will assign
 >  it as homework for Jordan :-)
  
 This is also documented in the New Daemon Book (just got my copy
 from Quantum yesterday).  0-201-54979-4, $47.00.  The book does a
 pretty good job of explaining the evolution of MFS and why it is
 structured the way it is.
 
 -GAWollman
 
 --
 Garrett A. Wollman   | Shashish is simple, it's discreet, it's brief. ... 
 wollman@lcs.mit.edu  | Shashish is the bonding of hearts in spite of distance.
 Opinions not those of| It is a bond more powerful than absence.  We like people
 MIT, LCS, ANA, or NSA| who like Shashish.  - Claude McKenzie + Florent Vollant
State-Changed-From-To: open->closed 
State-Changed-By: phk 
State-Changed-When: Sun Jun 9 10:26:05 PDT 1996 
State-Changed-Why:  
working as designed. 
A true VMFS on the otherhand could be fun :-) 
>Unformatted:
