Kernel support for graphics cards
Carsten Haitzler (The Rasterman)
raster at rasterman.com
Sun Jan 8 09:11:06 PST 2006
On Sun, 8 Jan 2006 14:42:56 +0200 Dimitar Toshev <mo6eeeb at gmail.com> babbled:
> On Sunday 08 January 2006 13:42, Carsten Haitzler wrote:
> > On Sun, 08 Jan 2006 21:23:23 +1100 Bojan Smojver <bojan at rexursive.com>
> > > I read Keith's excellent paper "The (Re)Architecture of the X Window
> > > System"
> > > (http://keithp.com/~keithp/talks/xarch_ols2004/xarch-ols2004-html/) and
> > > was encouraged by the direction that X is finally taking - the one of
> > > rapid and modern development.
> > >
> > > I do have a few questions in relation to kernel support for graphics
> > > cards though. Is hardware support for graphics cards in X going to be
> > > marked deprecated in the near future (i.e. in favour of complete
> > > Linux/*BSD/Solaris/<insert_flavour_here> kernel support)? If yes, will
> > > it be accompanied by the dual release of X (i.e. one source tree with
> > > deprecated hardware drivers, one with "hands-off" hardware), like we've
> > > seen in the case of modularisation? And finally, is there a page on
> > > freedesktop.org wiki that deals with the transition (specs for kernel
> > > folks, status of kernel drivers, roadmap etc.)?
> > >
> > > PS. What prompted me to ask this is the following sentence from the
> > > paper: "All of these systems directly manipulate hardware registers
> > > without any coordination among them." That really scares me - it almost
> > > sounds like the mess that DOS used to be...
> > quick answer: no. kernels are not going to have graphics drivers any time
> > soon - if ever. :)
> Why not? Kernels have drivers for every other piece for hardware, so what's
> stopping them from having gfx drivers?
i'd argue "why not" as well. but the kernel developers disagree. and i woudl
say that doing a gfx driver in userspace definitely makes for easier
development & debugging. :) also it means for better performance
characteristics of an OS if the gfx subsystem is doing some heavy software
fallbacks - the driver can be pre-empted trivially... like any process...
because... it is a process! :) seriously though - all that is really needed is
enough support hooks in the kernel to allow userspace gfx drivers - that's it.
and thats mostly there already.
------------- Codito, ergo sum - "I code, therefore I am" --------------
The Rasterman (Carsten Haitzler) raster at rasterman.com
Tokyo, Japan (東京 日本)
More information about the xorg