Date:	Tue, 2 May 2000 13:34:15 +0300 (IDT)
From:	Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
X-Sender:	eliz AT is
To:	muller AT cerbere DOT u-strasbg DOT fr
cc:	djgpp-workers AT delorie DOT com
Subject:	Re: interrupt 0x75 limitation in Win95 ?
In-Reply-To:	<3.0.6.32.20000429014940.00aa43e0@ics.u-strasbg.fr>
Message-ID:	<Pine.SUN.3.91.1000502133356.21668M-100000@is>
MIME-Version:	1.0
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

On Sat, 29 Apr 2000 muller AT cerbere DOT u-strasbg DOT fr wrote:

>   Even a win32 executable raising a FPU exception does
> not get it if it is run from a Dos executable
> that sets the 0x75 interrupt vector!
> The interrupt called is allways that of the first Dos
> program loaded in the calling chain.

This might or might not be relevant: the FP exception handling in DOS
programs is very different (for historical reasons) from the
``normal'' way FP exceptions are raised and processed by Intel CPUs in
protected mode.  The normal way is to generate exception 0 or 16, not
to trigger Int 75h.

So it might be that the juggling done by Windows to translate the FP
exception to the DOS-style interrupt is somehow involved in this mess,
especially when there's a mix of DOS and native Win32 programs.

Also note that the Win32 program runs in a different virtual machine,
which could further complicate the issue of passing the exception to
the correct program.

- Raw text -