updating modesetting driver

Dave Airlie airlied at gmail.com
Thu Sep 29 06:52:12 PDT 2011

Hi guys,

I've pushed a 'restart' branch to the xf86-video-modesetting driver,

Its a basic attempt to write a simple generic unaccelerated, fallback
driver for KMS. I need this to do some hotplug testing and work.

argb cursor support
randr 1.2
dirty tracking ioctl support.

Differences from modesetting driver:
drops all dri2/exa knowledge.
uses kernel dumb interface directly for bo create/map
based on mature radeon modesetting code.
probing based on fbdev driver (can be improved).

While working on it I noticed we don't have enough information from
the kernel to do certain things in a generic way:

we don't know the max cursor dimensions or whether the hw supports
argb/monochrome cursors
we don't know the allow set of bpp/depth combinations for
autoconfiguration of every device to work.

Now I could expose these from the kernel, but I get the feeling I'd be
walking into a world of pain with people
requesting an info ioctl exposes all kinds of arcane or insane
depth/bpp combos. I'm thinking of maybe
exposing a preferred depth/bpp and 2 sets of cursor limits (one for
argb, one for mono), in order to get autoconfig
working and userspace can override and try insane things with xorg.conf.


More information about the dri-devel mailing list