From: locke AT mcs DOT net (Peter Johnson)
Newsgroups: comp.os.msdos.djgpp
Subject: Re: Win 2000 and "forcedos"
Message-ID: <MPG.144f0ef3948b7ef09897e2@news-proxy.cso.uiuc.edu>
References: <Pine DOT SUN DOT 3 DOT 91 DOT 1001011161155 DOT 26304E-100000 AT is> <200010111709 DOT NAA12265 AT envy DOT delorie DOT com> <8011-Wed11Oct2000210042+0300-eliz AT is DOT elta DOT co DOT il>
X-Newsreader: MicroPlanet Gravity v2.12
X-Proxy-Client: pljohnsn AT uiuc DOT edu from isr5498.urh.uiuc.edu
Lines: 43
Date: Thu, 12 Oct 2000 00:58:13 -0500
NNTP-Posting-Host: 128.174.5.27
X-Complaints-To: abuse AT uiuc DOT edu
X-Trace: vixen.cso.uiuc.edu 971330242 128.174.5.27 (Thu, 12 Oct 2000 00:57:22 CDT)
NNTP-Posting-Date: Thu, 12 Oct 2000 00:57:22 CDT
Organization: University of Illinois at Urbana-Champaign
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com

In article <8011-Wed11Oct2000210042+0300-eliz AT is DOT elta DOT co DOT il>, 
eliz AT is DOT elta DOT co DOT il says...
> > Date: Wed, 11 Oct 2000 13:09:18 -0400
> > From: DJ Delorie <dj AT delorie DOT com>
> > 
> > It's in the NT 4 help file and MSDN.  FORCEDOS disables the OS/2
> > subsystem (etc), forcing NT to run a program in a VDM even if NT
> > wouldn't normally have recognized the program as an MS-DOS program.
> 
> Thanks for the info.
> 
> > I'm not sure why this would solve the W2K nesting problem, though.
> 
> I'm not sure; from your description it looks like it shouldn't.
> However, we won't lose anything if we try...
> 

I just tried and it doesn't appear that it works (eg, "forcedos make" 
still exhibits the crash problem).

On a side note, I found another thing that Win2K breaks: hardware 
breakpoints.  This isn't particularly a problem for gdb, as it defaults 
to software (int3) breakpoints, but edebug fails to single-step (and 
reports back error messages saying it can't get a DPMI breakpoint), and 
fsdb just dies horribly (lock-up essentially).. and gdb fails when you 
try to use hbreak, so it's not just the edebug-based code.

Another fsdb-related item: the method it uses to get the LDT causes FSDB 
to crash under Win2K.  I've managed to work around this by rewriting 
that code to use the DPMI get LDT entry call, but it doesn't let you 
read nearly so much :(.  I also managed to fix the hardware breakpoint 
problem by making the original code think that all the hardware 
breakpoints have been already used as startup, at which point fsdb 
defaults to using software breakpoints.

I can send patches for a "win2k-compat" version of fsdb to whoever the 
current maintainer is, but unfortunately I doubt it would be good to 
integrate them into the main source, as I don't believe there's a 
reliable method to find out if the program is running under Win2K versus 
NT4 (if I'm wrong on this, please let me know!).

-- 
Peter Johnson             locke AT mcs DOT net