On Thu, May 21, 2015 at 12:10 PM, Chris Wilson chris@chris-wilson.co.uk wrote:
On Thu, May 21, 2015 at 06:07:30PM +0200, Daniel Vetter wrote:
On Thu, May 21, 2015 at 11:58:30AM -0400, Rob Clark wrote:
For actual sharing of buffers with other drivers (ie. actual hardware) we'll need to pimp things out a bit better to deal w/ caching, multiple memory domains, etc. See thread:
http://lists.freedesktop.org/archives/dri-devel/2015-May/083160.html
But for the llvmpipe use-case this isn't a problem. Nor do we really need prime/dri3 (dri2 is sufficient). So until the other issues are sorted lets remove DRIVER_PRIME.
v2: also drop the dead code
Signed-off-by: Rob Clark robdclark@gmail.com
Yeah I'm ok with this too, but it pretty much means vgem is a thing for llvmpipe only. And then abusing dumb buffers is imo ok too, since I don't care too much what kind of horrors drivers pull of internally.
Ugh. But llvmpipe + dumb buffer is much inferior to SHM.
it is using shmem for it's dumb buffers.. although I noticed it was mapping writecombine which is probably not what is wanted..
BR, -R
-Chris
-- Chris Wilson, Intel Open Source Technology Centre