On Thu, Jan 14, 2010 at 1:19 PM, Luca Barbieri <l...@luca-barbieri.com> wrote:
> On Wed, Jan 13, 2010 at 3:55 AM, Chia-I Wu <olva...@gmail.com> wrote:
>> On Wed, Jan 13, 2010 at 2:52 AM, Luca Barbieri <l...@luca-barbieri.com>
>> wrote:
>>> Doesn't this make two DRI2GetBuffers protocol calls, in case of a resize?
>> I expect resizing happens rarely.
> It likely happens every frame while the user is resizing the window,
Still, I expect that happens rarely.
> and it's relatively expensive (at least 7 system calls + scheduling +
> work in the X server and the DRM memory manager).
> Also, having to do this redundant work is a sign that the design can
> be improved.
> It's your call, but I think your:
>> if (not forced) {
>> get_current_sequence_number();
>> if (seq_num has not changed)
>> continue;
>> }
>> validate_and_update_buffers();
> reformulated as
> validate_and_get_current_sequence_number();
> if (not forced and seq_num has not changed)
> continue;
> update_buffers();
> is the best idea, with DRI2GetBuffers done in
> validate_and_get_current_sequence_number(), and the results reused in
> update_buffers().
This works too. It assumes fast texture creation (as they are always asked
for), which is true with your DRI2 texture cache patch. What do you think
settling on this approach?
> Or if you prefer a single function, then my approach puts that logic
> in the validate function by taking the sequence number as in/out.
>
>> Plus, I want to hook winsys->update_buffer. It is called when the user calls
>> glViewport. It will set "force_validate" and avoid the first call.
>
> Shouldn't the glViewport hook simply cause a call to validate which
> otherwise would not have happened? (in the single-buffering case)
> Wouldn't ignoring the sequence number (which is what forced validation
> does) make sense only if the surface or desired attachments change?
In single-buffering case, re-validation happens in egl_g3d_flush_frontbuffer.
Native surface becomes invalid at any time. The best EGL can do is to validate
when a frame is displayed (swap buffers or flush frontbuffer, as can be seen in
the code).
But EGL is not alone here. The application knows when window resize happens,
through display server events. There is no eglValidateSurface, but usually,
the application calls glViewport when it sees a resize event. It can be used
as a hint to EGL that the surface might have become invalid. In my approach,
the hint can be used to avoid a redundant call. With the newly mentioned
approach, since asking for textures becomes so cheap, the hint can be ignored.
-olv
------------------------------------------------------------------------------
Throughout its 18-year history, RSA Conference consistently attracts the
world's best and brightest in the field, creating opportunities for Conference
attendees to learn about information security's most important issues through
interactions with peers, luminaries and emerging and established companies.
http://p.sf.net/sfu/rsaconf-dev2dev
_______________________________________________
Mesa3d-dev mailing list
Mesa3d-dev@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mesa3d-dev
