From:	sandmann AT clio DOT rice DOT edu (Charles Sandmann)
Message-Id:	<9607151316.AA13525@clio.rice.edu>
Subject:	Re: Debuggers under Windows
To:	eliz AT is DOT elta DOT co DOT il (Eli Zaretskii)
Date:	Mon, 15 Jul 1996 08:16:59 -0600 (CDT)
Cc:	djgpp-workers AT delorie DOT com, terra AT diku DOT dk
In-Reply-To:	<Pine.SUN.3.91.960715133313.13725E-100000@is> from "Eli Zaretskii" at Jul 15, 96 01:47:26 pm

> breakpoints, so it seems.  ANY breakpoint that is set by writing the CCh 
> opcode (Int 03) seems to not work under Windows, in either GDB or FSDB.  
> Both the debuggers get Exception 0Dh with an error code of 80000000h 
> (does that error code mean anything?).

Ah, this might explain alot.  Rainer told me a long, long time ago that
he couldn't make GDB work under Windows since breakpoint int 3's wouldn't
work.  I never investigated.  I was under the assumption we were using
debug registers everywhere.

Only the low 16 bits of the error code are defined, so no, this doesn't
mean anything.  However, a full register listing would tell alot :-)

> Another thing that I saw is that the address where Exception 0Dh happens 
> is NOT the address where the breakpoint was set, but somewhat after that.
> 
> If anybody has any clue as to where to look further, please tell me.  
> I've checked all the code around the place where the breakpoint is set 
> and didn't find anything that even remotedly seems suspicious.

It might be a selector problem.  Are CS,DS,ES,SS,FS,GS as expected?
Is the exception really in the child or in the debugger or in the DPMI
provider?  Is the instruction of pukage a memory access, where all the
ones before were register only?

- Raw text -