Graphics Driver Frameworks and Security

[Michel Dänzer]

On Thu, 2006-05-25 at 21:30 +0200, olafBuddenhagen at gmx.net wrote:
> 
> On Thu, May 25, 2006 at 02:06:46PM +0200, Michel Dänzer wrote:
> > On Wed, 2006-05-17 at 12:26 +0200, olafBuddenhagen at gmx.net wrote: 
> 
> > > Also, as I already pointed out, there are people willing to do this
> > > work, or at least there used to be -- most are scattered now, due to
> > > the continued missing acceptance. The former KGI attempts did *not*
> > > fail because of lack of people to do the work, but because of lack
> > > of interest in the work they did.
> > 
> > I'm not buying that. What stopped them from making a proof-of-concept
> > implementation?
> 
> Nothing! That's exactly my point: A proof-of-concept implementation of
> KGI *does* exist -- only it's beeing ignored totally.
> 
> The existing implementation is pretty obsolete by now, and quite a pain
> to get it working. (The FreeBSD variant is presently better than the
> Linux one in most regards, though.) Nevertheless it does exist, and I
> have a machine running XGGI on top of KGI. (There is at least one person
> on this list who can confirm having seen it. In fact, if *any* interest
> had existed, I would have presented it at FOSDEM.)
> 
> I readily admit that the existing implementation is not very useful by
> today's standards without some major improvements. But that's not the
> point.

But of course, that's the whole point! When I tried Xggi a couple of
years ago, it didn't seem to provide any advantage over what I was
dealing with, so I quickly lost interest.

Maybe I should clarify what I mean by proof of concept: a working
implementation with top-to-bottom acceleration and otherwise current
features on at least one platform. We already have plenty of cool meta X
servers that run on top of another X server. :)


> > > Unless we can expect at least some support among X developers, there
> > > is no point in even starting to write more code. And the nearly zero
> > > resonance my previous posting(s) on that topic received, isn't
> > > exactly encouraging :-(
> > 
> > I don't understand what kind of support you're looking for. Don't pull
> > a Jon Smirl and expect everybody to suddenly drop what they're doing
> > and focus exclusively on something radically new.
> 
> As I already pointed out, I'm not expecting that. All I wish for would
> be to hear as much as "sounds interesting, show us the code", or
> alternatively "we see the following problems with this approach: ..."

I can only speak for myself, but I'd definitely be interested in seeing
a proof of concept as described above. As for problems, I know from
experience that it's hard to define a kernel<->userspace interface that
is both fast and secure.

> Anything giving hope that work done on GGI/KGI (or some similar system)
> won't go to waste.

I think that's up to you guys, not anyone else. So long as it's useful
to you, does it really matter that much whether it is to everybody else?


-- 
Earthling Michel Dänzer           |          http://tungstengraphics.com
Libre software enthusiast         |          Debian, X and DRI developer