Mail Archives: djgpp/2009/05/03/20:31:13
Hi,
On May 3, 1:35=A0pm, "Charles Sandmann" <cwsd DOT DOT DOT AT earthlink DOT net> wrote:
> >"Eli Zaretskii" <e DOT DOT DOT AT gnu DOT org> wrote in message
> >news:83ws8zeb05 DOT fsf AT gnu DOT org...
> >> From: "Charles Sandmann" <cwsd DOT DOT DOT AT earthlink DOT net>
> >> >"Eli Zaretskii" <e DOT DOT DOT AT gnu DOT org> wrote in message
> >> >Yes, for the same reason: it thinks DJGPP supports SSE registers.
> >> What is required to make that work?
> > There are 2 possible fixes: the easy one and the more complicated one.
>
> > The easy one is a two-liner: tell GDB that we don't support SSE. =A0I
> > already did that, and the patch is in the GDB CVS repository since
> > about 2 days ago.
>
> OK, that seems to be reasonable. =A0I suspect that any SSE capable
> software will probably be written on other platforms and rarely
> debugged at the register level with DJGPP (mostly ported).
For the record, 386SWAT supports SSE1, OpenWatcom's WD supports SSE2
(and SSE3??). So at least DOS isn't completely bereft although I'm
skeptical OW's WD can handle DJGPP .EXEs properly (debug info).
Japheth claims to love WD, but even he says it has issues with
FreeDOS. (As for debugging on other hosts, at least on XP you need the
VDMSERV for DOS apps, and gfx don't work. In other words, Windows
ain't for DOS no more, oh noes ....)
> > The more complex solution is to extend save_npx and load_npx to use
> > FXSAVE and FXRSTOR instructions, when they are available, instead of
> > FNSAVE and FRSTOR that we use now.
>
> Rugxulo did convince me to update CWSDPMI to support SSE, which
> was a good thing.
So good that DJ never updated his server with it. ;-) Or maybe
you told him to wait until you made some tweaks?? The only thing worse
than not updating something is updating and having nobody know about
it!! (Hi, Raine dude!)
> But needing to debug is another issue, agreed. =A0There
> are the complications of determining support, etc, making sure you don't
> break old platforms. =A0Not worth the hassle.
> (For someone wanting to work on this, the referenced code is in dbgcom.c)=
.
Lots of inline AT&T in that file. Very confusing although I
(unsurprisingly) admit to knowing nothing about the FPU and DPMI
exceptions. Maybe somebody can find Pierre Muller and bug him to work
on it (doubtful).
- Raw text -