From owner-freebsd-hackers  Fri Jun 23 22:56:47 1995
Return-Path: hackers-owner
Received: (from majordom@localhost)
          by freefall.cdrom.com (8.6.10/8.6.6) id WAA15064
          for hackers-outgoing; Fri, 23 Jun 1995 22:56:47 -0700
Received: from physics.su.oz.au (dawes@physics.su.OZ.AU [129.78.129.1])
          by freefall.cdrom.com (8.6.10/8.6.6) with SMTP id WAA15048
          for <freebsd-hackers@freebsd.org>; Fri, 23 Jun 1995 22:56:41 -0700
Received: by physics.su.oz.au id AA01618
  (5.67b/IDA-1.4.4 for freebsd-hackers@freebsd.org); Sat, 24 Jun 1995 15:55:30 +1000
From: David Dawes <dawes@physics.usyd.edu.au>
Message-Id: <199506240555.AA01618@physics.su.oz.au>
Subject: Re: Memory leak somewhere?
To: henrich@crh.cl.msu.edu (Charles Henrich)
Date: Sat, 24 Jun 1995 15:55:30 +1000 (EST)
Cc: freebsd-hackers@freebsd.org
In-Reply-To: <199506230630.CAA09279@crh.cl.msu.edu> from "Charles Henrich" at Jun 23, 95 02:30:03 am
X-Mailer: ELM [version 2.4 PL23]
Content-Type: text
Content-Length: 1319      
Sender: hackers-owner@freebsd.org
Precedence: bulk

>I've noticed lately that my S3 X server grows continuously (as does my swap
>utilization) as time goes by, and never shrinks.  Currently my X server process
>looks like so:
>
>root       252  0.0 30.9 11632 9440 ??  I    10:39PM    0:21.43 X :0 (XF86_S3)
>
>and has only been running a few hours.  I've talked with the author of the
>Server and he was astonished when I told him I've seen it as high as 15M of
>ram.  He claims he never see's this, however he's running it under Linux.  He
>suggests perhaps there is a problem with something somewhere in FreeBSD. This
>behaviour seems to be new with the 0412-SNAP, although I dont have any
>proof of this.  This is crazy, I have a 32mb machine, and its performing like
>a dog because of this sort of memory usage (!) :(.  On a 16mb machine, if you
>run any significant apps you go to swaphell because of the memory usage here.
>Could this be a leak in the kernel malloc, or mmap code or some such?

I've noticed it growing fairly large too.  I've tried linking it with
-lgnumalloc (Linux uses GNU malloc by default), and that seems to help
(you can do this using the XFree86 LinkKit).  We (XFree86) are doing
some work on the malloc issue, both regarding memory leaks, and
regarding methods to reduce fragmentation, and allow the system to
regain freed memory.

David