Date: Mon, 14 Jun 1999 11:00:51 +0300 (WET)
From: Andris Pavenis <pavenis AT ieva01 DOT lanet DOT lv>
To: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
cc: djgpp-workers AT delorie DOT com
Subject: Re: save_npx and FSDB/Edebug32
In-Reply-To: <Pine.SUN.3.91.990614103337.21962O-100000@is>
Message-ID: <Pine.A41.4.05.9906141055490.71812-100000@ieva01.lanet.lv>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Reply-To: djgpp-workers AT delorie DOT com
X-Mailing-List: djgpp-workers AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk



On Mon, 14 Jun 1999, Eli Zaretskii wrote:

> 
> On Mon, 14 Jun 1999, Andris Pavenis wrote:
> 
> > Most my own applications use floating point. At least my own experience
> > with debuggers shows that FP supports works there.
> 
> The problem should only show up if the debuggee changes the FPU state by 
> fiddling with the control word, or gets an FP exception.  Otherwise, 
> FNINIT is a no-op.
> 
> > However I think that
> > having 2 FPU states (one for debugger, second for debugee) which are 
> > swapped before and after running child (as suggested in one of earlier
> > posts to this mailing list) would be better.  
> 
> I agree.  But this should probably wait for a later DJGPP version.
> 

I think it would be very small patch (adding one more structure for saving
FPU state, one frstor at end of save_fpu and one fnsave at begin of
load_npx).

I'll test it slightly later when I'll reboot Wincrap95 (I'm currently
running Linux)

Andris