Date: Fri, 31 Aug 2001 19:11:48 +0300
From: "Eli Zaretskii" <eliz AT is DOT elta DOT co DOT il>
Sender: halo1 AT zahav DOT net DOT il
To: "Dykstra Sean" <Sean_Dykstra AT maxtor DOT com>
Message-Id: <1190-Fri31Aug2001191148+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 AT delorie DOT com"
	<68C4CF842BD2D411AC1600902740B6DA02CDC54E AT mcoexc02 DOT mlm DOT maxtor DOT com>
	(Sean_Dykstra AT maxtor DOT com)
Subject: Re: Help! - INT x22 problem.
References:  <68C4CF842BD2D411AC1600902740B6DA02CDC54E AT mcoexc02 DOT mlm DOT maxtor DOT com>
Reply-To: djgpp AT delorie DOT com

> From: "Dykstra, Sean" <Sean_Dykstra AT maxtor DOT com>
> Date: Fri, 31 Aug 2001 09:06:50 -0600
> 
> I am only hooking the protected mode interrupt.  There is some old lock code
> below, but I planning on removing it as I now lock the entire app into
> memory  (I have already removed the lock code for locking the vars...But I
> wanted to send you the file unedited from what I am running now).

One thing your code does not do is lock the stack.  However, each
interrupt handler has a local automatic variable Count which is on the
stack.  So your handlers access unlocked data.

Solution: either lock everything, or make those variable static and
lock them.

> I have
> found it to lock up on several machines so far, I will attempt to
> characterize them for BIOS, manufacturer, etc..  The interrupt frequency is
> about 1 interrupt every 40us - 70us.

This is a very high frequency.  You might be hitting problems with
wrapper code being non-reentrant.

> One of the other reasons for
> using C code is that I call back into the app (another reason I suspect the
> stack might be corrupted).

This is yet another reason that might cause a failure: the callback
code and data is probably not locked, and could potentially run for a
long time.

> Can I still do this if I write the handler assembly?

Sure, assembly functions can call C functions.