On 2/18/20 12:46 PM, PÁSZTOR György wrote:
Hi,
"Alan Coopersmith" <alan.coopersm...@oracle.com> írta 2020-02-05 09:11-kor:
Right - applications expecting DRI/KMS interfaces are going to be sorely
disappointed, and limited to what the underlying illumos kernel provides.
Are you 100% sure about this?
If apps would have direct access to those devices at the same time
X11server does, that could lead to serious deadlock issues, or one writes
over parts in the graphic card's memory what is under the control of the
other one.
Which is why X has the DRI extension to coordinate access with applications
that directly access the graphics hardware (like OpenGL).
https://en.wikipedia.org/wiki/Direct_Rendering_Infrastructure
If you would say, that the Xserver mmap-s or does some other magic to
access the card's memory, than shares the memory area using some X11
extenson and the client (wm/de) could access that like some sysv ipc-way
(eg. shmem), I could understand if that would not work.
Or like fdpassing.
Different generations use mmap, fd passing, or other IPC techniques.
Btw.: In X11 the partys can negotiate who supports which feature.
Thanks, but you really don't need to explain X11 to me, since I've been
working on it for 20 years now. A much more complete explanation of what
you describe is available in the chapters I wrote of
https://www.x.org/wiki/guide/ .
I would think if the Xserver doesn't support some linux-specific drm/kms
interface (which can be easily the point here) than it would just fallback
to some legacy solution.
Some applications will, some won't - it depends on whether the authors
wanted to put in the effort to code a fallback or not.
--
-Alan Coopersmith- alan.coopersm...@oracle.com
Oracle Solaris Engineering - https://blogs.oracle.com/alanc
_______________________________________________
openindiana-discuss mailing list
openindiana-discuss@openindiana.org
https://openindiana.org/mailman/listinfo/openindiana-discuss
