Mucho thanks to the multitudes of people (and old friends -- Spiros,
did you go to PSU?) who juggled my memory and told me about Purify from
Pure Software. That was indeed the software I was thinking of. Since
there seem to be many different phone numbers, I will only post the e-mail
address info@pure.com as a general source of information.
Two other interesting software packages were also suggested. The messages
are below.
Date: Tue, 6 Oct 1992 15:41:43 -0400
>From: Scott Schwartz <schwartz@groucho.cs.psu.edu>
Subject: Re: Memory leak detection software???
| I'm looking for software to detect memory leaks. I saw some companies
| advertising such stuff at Uniforum, but cannot locate the info.
You are probably thinking of purify, from Pure Software. It's pretty
spiffy. Hans Boehm from Xerox has a garbage collecting memory
allocator, which, as a side effect, complains about leaks. That's less
spiffy, but it is free. (ftp from parcftp.xerox.com.)
Date: Tue, 6 Oct 92 16:52:59 -0700
>From: David Fetrow <fetrow@biostat.washington.edu>
Subject: electric fence
Haven't tried this or the others I've run accross but:
>From news.u.washington.edu!uw-beaver!micro-heart-of-gold.mit.edu!news.bbn.com!olivea!spool.mu.edu!uunet!psinntp!pixar!bruce Wed Mar 25 02:00:03 PST 1992
Article: 1529 of comp.sys.sun.misc
Path: news.u.washington.edu!uw-beaver!micro-heart-of-gold.mit.edu!news.bbn.com!olivea!spool.mu.edu!uunet!psinntp!pixar!bruce
From: bruce@pixar.com (Bruce Perens)
Newsgroups: comp.sys.sun.misc
Subject: Electric Fence (a malloc buffer-overrun debugger) posted
Message-ID: <1992Mar24.203856.2680@pixar.com>
Date: 24 Mar 92 20:38:56 GMT
Sender: news@pixar.com (Usenet Newsmaster)
Distribution: na
Organization: Pixar -- Point Richmond, California
Lines: 12
Originator: bruce@janis
Nntp-Posting-Host: janis
I have posted my Electric Fence malloc buffer-overrun debugger to
alt.sources and comp.sources.sun .
This is different from the malloc debugger that comes with your Sun.
Electric Fence uses the virtual-memory system to position an
inaccessable memory page immediately after the end of each buffer
returned by malloc(). When your program attempts to write or even
_read_ beyond the end of a malloc-ed buffer, it hits this dead page and
gets an immediate segmentation fault. Finding the errant code with a
debugger then becomes simple.
Bruce Perens
--
Angela Thomas Hoynowski Internet: angela@tandem.com
Tandem Computers, Inc. UUCP: {...}!tandem!angela
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:51 CDT