From owner-freebsd-current  Sat Sep 19 14:05:04 1998
Return-Path: <owner-freebsd-current@FreeBSD.ORG>
Received: (from majordom@localhost)
          by hub.freebsd.org (8.8.8/8.8.8) id OAA25681
          for freebsd-current-outgoing; Sat, 19 Sep 1998 14:05:04 -0700 (PDT)
          (envelope-from owner-freebsd-current@FreeBSD.ORG)
Received: from nash.pr.mcs.net (nash.pr.mcs.net [204.95.47.72])
          by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id OAA25659
          for <current@FreeBSD.ORG>; Sat, 19 Sep 1998 14:04:58 -0700 (PDT)
          (envelope-from alex@nash.pr.mcs.net)
Received: (from alex@localhost)
	by nash.pr.mcs.net (8.8.8/8.8.7) id QAA02198;
	Sat, 19 Sep 1998 16:04:12 -0500 (CDT)
	(envelope-from alex)
Message-ID: <19980919160412.G16974@pr.mcs.net>
Date: Sat, 19 Sep 1998 16:04:12 -0500
From: Alex Nash <nash@mcs.net>
To: chaos@ultra.net.au, current@FreeBSD.ORG
Subject: Re: free() isn't threads safe!
References: <Pine.BSF.4.02.9809200113170.9082-100000@chaotic.oz.org>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
X-Mailer: Mutt 0.93.2i
In-Reply-To: <Pine.BSF.4.02.9809200113170.9082-100000@chaotic.oz.org>; from Simon Coggins on Sun, Sep 20, 1998 at 01:18:00AM +1000
X-Operating-System: FreeBSD 3.0-CURRENT i386
Sender: owner-freebsd-current@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG

On Sun, Sep 20, 1998 at 01:18:00AM +1000, Simon Coggins wrote:
> This is the code from lib/libc/stdlib/malloc.c
> 
> void
> free(void *ptr)
> {
>     malloc_func = " in free():"; <- isn't there a slim chance this can be
>                                     changed before the thread is locked?

Right you are.  In fact malloc() and realloc() had the same problem.

>     THREAD_LOCK();
>     if (malloc_active++) {
>         wrtwarning("recursive call.\n");
>         malloc_active--;
>         return; <--------- Shouldn't the thread be unlocked before returning?

Yes.

> Wouldn't something like:
> 
> void
> free(void *ptr)
> {
>     THREAD_LOCK();
>     malloc_func = " in free():";
>     if (malloc_active++) {
>         wrtwarning("recursive call.\n");
>         malloc_active--;
> 	THREAD_UNLOCK();
>         return;
>     }
[...]
> Be more sensable?

I committed a slightly different fix to -current (I'll merge this back
into -stable later this weekend) using a single exit point from free().
Cue Terry to explain the advantages of single entry/single exit
functions...

Alex

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message