Date:	Thu, 8 Aug 1996 10:54:29 +0200 (IST)
From:	Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To:	David M Barrett <dbarrett AT srvr1 DOT engin DOT umich DOT edu>
Cc:	djgpp AT delorie DOT com
Subject:	Re: Enhanced FSDB Questions
In-Reply-To:	<4tobm6$psp@srvr1.engin.umich.edu>
Message-Id:	<Pine.SUN.3.91.960808104832.19168f-100000@is>
Mime-Version:	1.0

On 31 Jul 1996, David M Barrett wrote:

> - Is this program stable for other people?  I'm running it under Win NT 
> 3.51 with 32 megs o' ram on a Dell P90.  Sometimes I'm running perhaps a 
> telnet application in the background, but the system is mainly just a dos 
> window.  I get random errors such as :
>   o When I set a breakpoint and press F9 to run up to the breakpoint, 
>     everything is fine.  But, if I just press F4 on the line I want to run 
>     to, it gets an exception.  

There are similar problems under Windows 3.x (but not Windows 95, I'm 
told), so maybe NT has the same problem as 3.x.  Try setting no more than 
3 breakpoints (including the temporary breakpoint used when you 
single-step) and see if that makes the problems go away.  If so, the 
cause is that the breakpoint opcode causes an interrupt (as opposed to 
exception) on some versions of Windows, and the current version of DJGPP 
debugging support expects to get an exception.  For the first 3 breakpoints 
FSDB sets, it uses debug registers, which don't have these problems, but 
after that it begins using the breakpoint opcode.

There is no work-around currently (except to use 3 breakpoints or less) 
for the above problems.  Somebody should fix the DJGPP debug support for 
this to be solved.

- Raw text -