From: sandmann AT clio DOT rice DOT edu (Charles Sandmann)
Message-Id: <9804211402.AA12940@clio.rice.edu>
Subject: Re: Temporary files for gcc-2.81
To: Vik DOT Heyndrickx AT rug DOT ac DOT be (Vik Heyndrickx)
Date: Tue, 21 Apr 1998 09:02:57 -0600 (CDT)
Cc: pavenis AT lanet DOT lv, djgpp-workers AT delorie DOT com
In-Reply-To: <353C6A0E.D68@rug.ac.be> from "Vik Heyndrickx" at Apr 21, 98 11:42:38 am
Content-Type: text
Precedence: bulk

> I have a perfectly working and tested PID code available, but I need to
> know whether it is acceptable to put that code in the stub (it really,
> really, really needs to run in real-mode). 

This is really, really, really a bad idea.  The stub is a very stable 
beast which hasn't changed functionality since 2.0 alpha 1 - and has been
cloned in debug loaders, image packers, alternate DPMI stubs and
god knows where else.  This is one of the huge issues in going non-COFF
images.  Since almost no computer runs in real mode, you are at the victim
of the EMM provider to do emulation for anything which needs to run in 
real mode instead of protected mode - which means testing every version
of EMM386 released since Windows 3.0, all versions of OS/2, all patches/
hotfixes of NT, 386sx/486/Pentium etc.

There are also methods for writing a real mode chunk of code to the
transfer buffer and executing it via a simulate real mode far call - this
would be much preferred to messing with the stub.

> Question is: is every
> coff-image started with a stub, i.e. can a coff image been run without
> one? I've been writing my own stub.asm (nearly 1/3 finished) but started
> wondering then.

Yes, coff images can be run without stubs.