Date: Wed, 20 Jun 2001 10:05:06 +0300
From: "Eli Zaretskii" <eliz AT is DOT elta DOT co DOT il>
Sender: halo1 AT zahav DOT net DOT il
To: "Mark E." <snowball3 AT softhome DOT net>
Message-Id: <7704-Wed20Jun2001100505+0300-eliz@is.elta.co.il>
X-Mailer: Emacs 20.6 (via feedmail 8.3.emacs20_6 I) and Blat ver 1.8.9
CC: djgpp-workers AT delorie DOT com
In-reply-to: <3B2FEDD5.12239.459453@localhost> (snowball3@softhome.net)
Subject: Re: libtool's max cmd line crash
References:  <3B2FEDD5 DOT 12239 DOT 459453 AT localhost>
Reply-To: djgpp-workers AT delorie DOT com
Errors-To: nobody AT delorie DOT com
X-Mailing-List: djgpp-workers AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk

> From: "Mark E." <snowball3 AT softhome DOT net>
> Date: Wed, 20 Jun 2001 00:27:01 -0400
> 
> Once the response file is processed, expand_wildcards is called which then 
> calls glob to expand any wildcards, etc. glob allocates a 2000 byte buffer 
> for an argument, but the mega argument is over 250K.

It sounds like we need to malloc that buffer in glob, and then add
code to all of its subroutines to check if the buffer's size is about
to be exceeded, and if so, realloc it.

This would fix the crashes, but the result of the test performed by
libtool will be meaningless anyway: it will probably be the amount of
available VM on the machine where the test runs.  So it still makes
sense IMHO to introduce a special change somewhere to return something
reasonable, like 12K or some such.  That's what I did for GDB in the
current CVS.