Re: EGL_MESA_screen_surface proposal

Re: EGL_MESA_screen_surface proposal

[Jon Smirl]

I cc'd this onto the other lists to get more opinions. Maybe we can
turn this into a workable concept.

On Tue, 15 Mar 2005 08:17:13 -0700, Brian Paul
<brian.paul@tungstengraphics.com> wrote:
> Perhaps the eglShowSurface() and eglScreenMode() functions should be
> combined so the new surface and new display mode can be validated
> together.  That way, the undefined state between setting the new
> surface and new mode can be avoided.

In one of the other threads (on x.org&fbdev) people are arguing that
mode setting and surface selection should not be combined. I think
their concern comes from mode setting code is in fbdev and their
allocation code is somewhere else.

The problem is that fbdev doesn't have a memory manager so it has no
concept of where the surfaces are. It relies on some other piece of
software manage the surfaces. Maybe we should teach fbdev about the
concept of surfaces.

Initially the radeonfb dev driver would only know about two surfaces,
one for each head. Name them default. Setting the mode on either head
would have an implied parameter surface=default. surface=default is a
special case since the surface type is mutable into what ever the mode
requires.

fbdev could then be extended to have a plugable API for enumerating
surfaces. DRM would add other surfaces to this list each with a name.
Now you can set the mode and specific surface=my_new_surface. If the
mode and surface are incompatible you get an error.

Another twist would be to set the mode just by specificing the surface
name. That would cause a mode to be picked that can display the
surface.






-- 
Jon Smirl
jonsmirl@gmail.com

Re: EGL_MESA_screen_surface proposal

[Michel Danzer]

On Tue, 2005-03-15 at 11:28 -0500, Jon Smirl wrote:
> 
> On Tue, 15 Mar 2005 08:17:13 -0700, Brian Paul
> <brian.paul@tungstengraphics.com> wrote:
> > Perhaps the eglShowSurface() and eglScreenMode() functions should be
> > combined so the new surface and new display mode can be validated
> > together.  That way, the undefined state between setting the new
> > surface and new mode can be avoided.

Makes sense to me.

> In one of the other threads (on x.org&fbdev) people are arguing that
> mode setting and surface selection should not be combined.

Actually, I was arguing that mode setting and _memory management_
shouldn't be tied together, see my other post to dri-egl in this thread.
The approach Brian is proposing is much more flexible, which I applaud.


-- 
Michel Danzer, Linux Software Engineering       \   Tel: +1 905-882-2600
ATI Technologies Inc., Markham, Ontario, Canada  \       Extension: 3550



-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: EGL_MESA_screen_surface proposal

[Jon Smirl]

On Tue, 15 Mar 2005 11:36:53 -0500, Michel Danzer <mdanzer@ati.com> wrote:
> On Tue, 2005-03-15 at 11:28 -0500, Jon Smirl wrote:
> >
> > On Tue, 15 Mar 2005 08:17:13 -0700, Brian Paul
> > <brian.paul@tungstengraphics.com> wrote:
> > > Perhaps the eglShowSurface() and eglScreenMode() functions should be
> > > combined so the new surface and new display mode can be validated
> > > together.  That way, the undefined state between setting the new
> > > surface and new mode can be avoided.
> 
> Makes sense to me.

What if I just want to show another surface without changing the mode?
I could still use a combined version and since the mode matches the
mode won't actually get changed?

-- 
Jon Smirl
jonsmirl@gmail.com

Re: EGL_MESA_screen_surface proposal

[Vladimir Dergachev]

On Tue, 15 Mar 2005, Jon Smirl wrote:

> On Tue, 15 Mar 2005 11:36:53 -0500, Michel Danzer <mdanzer@ati.com> wrote:
>> On Tue, 2005-03-15 at 11:28 -0500, Jon Smirl wrote:
>>>
>>> On Tue, 15 Mar 2005 08:17:13 -0700, Brian Paul
>>> <brian.paul@tungstengraphics.com> wrote:
>>>> Perhaps the eglShowSurface() and eglScreenMode() functions should be
>>>> combined so the new surface and new display mode can be validated
>>>> together.  That way, the undefined state between setting the new
>>>> surface and new mode can be avoided.
>>
>> Makes sense to me.
>
> What if I just want to show another surface without changing the mode?
> I could still use a combined version and since the mode matches the
> mode won't actually get changed?

Have a flag that describes what needs to be changed ?

                      best

                         Vladimir Dergachev

>
> -- 
> Jon Smirl
> jonsmirl@gmail.com
> _______________________________________________
> xorg mailing list
> xorg@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/xorg
>

Re: EGL_MESA_screen_surface proposal

[Brian Paul]

Jon Smirl wrote:
> On Tue, 15 Mar 2005 11:36:53 -0500, Michel Danzer <mdanzer@ati.com> wrote:
> 
>>On Tue, 2005-03-15 at 11:28 -0500, Jon Smirl wrote:
>>
>>>On Tue, 15 Mar 2005 08:17:13 -0700, Brian Paul
>>><brian.paul@tungstengraphics.com> wrote:
>>>
>>>>Perhaps the eglShowSurface() and eglScreenMode() functions should be
>>>>combined so the new surface and new display mode can be validated
>>>>together.  That way, the undefined state between setting the new
>>>>surface and new mode can be avoided.
>>
>>Makes sense to me.
> 
> 
> What if I just want to show another surface without changing the mode?
> I could still use a combined version and since the mode matches the
> mode won't actually get changed?

Right.  The combined function can just as well be used to change the 
screen's current mode _or_ current surface.  You don't have to change 
both.

-Brian