Date:	Thu, 23 Apr 1998 08:57:38 +0300 (IDT)
From:	Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To:	b52g AT usa DOT net
cc:	djgpp AT delorie DOT com
Subject:	Re: Problems with Real-Mode Transfer Buffer
In-Reply-To:	<6hl5rh$lfc$1@nnrp1.dejanews.com>
Message-ID:	<Pine.SUN.3.91.980423085717.9091B-100000@is>
MIME-Version:	1.0

On Wed, 22 Apr 1998 b52g AT usa DOT net wrote:

> what can I do about it?

You should debug this.  Bugs with RMCBs are one of the nastiest
possible problems in protected-mode programs, since they mean
something is messing with the real-mode interrupt vector table and/or
conventional memory.  You can easily have your machine freeze cold or
wipe out your hard disk partition table this way.  The feature whereby
illegal addresses in RMCBs cause your program to crash is your last
line of defence against these atrocities.

> Is changing the RMTB size masking the problem or fixing it?

IMHO, it masks another, potentially very serious, problem.

> If so, can I set the RMTB size in my program, like the stack size?

That's impossible.  The size of transfer buffer is determined by the
startup code using the stub info alone.

> Alternatively, could I make a new stub with a default of 32k RMTB?

Don't do that!  In addition to leaving a potentially disastrous bug
lurking in your program, this will only get you into subtle bugs,
because too many DJGPP development tools have the stub hard-wired into
them.  You cannot change them all, and you won't always know which
stub is getting used in every particular case.

- Raw text -