🚀 go-pugleaf

stumble@utopia.hacktic.nl (Yeats) writes:

> I recently discovered what may be a rare hardware bug -
>the following code will freeze my old 6-8-10mhz 286 AT :
> 
>   MOV AX,[DI]
>   INC DI
>   MOV AX,[DI]
> 
> Other index registers cause the same hangups but the [DI] never
>fails to hang at once.
> Surprisingly, a keyboard reset still worked - as did the LED's.
>Otherwise my system works fine (WIN3 crashes, so what).
> On accessing memory using REP MOVSW and then again in a loop with
>STOSB in two SEPARATE but close pieces of code, there was a more
>subtle messing up of certain bytes as it run over the 64K segment
>wrap-around - this bug disappeared when the REP MOVSW was replaced
>by REP MOVSB with twice the count.
> 
> Does anyone know anything about this bug ?
> 

The above piece of code is supposed to generate a generel protection
exception if (and only if) DI is 0FFFFh before or "after". That should
certainly crash the system: the GP exception corresponds to Int 13h -- 
disk io!  Try loading a zero into DI before.

Then again you might to having some other kind of trouble.

Morten Welinder
terra@diku.dk
-- 
-------------------------------------------------------------------------
Visit the lyrics archive at ftp.uwp.edu (mirrored to nic.funet.fi, a site
in Finland). All kinds of lyrics available -- upload "yours" and join.
-------------------------------------------Morten Welinder, terra@diku.dk