[Mesa-dev] [PATCH] wayland/egl: initialize window surface size to window size

[Brendan King]

The IMG patch Eric was referring to was to dEQP. I submitted bug reports 
to both Google and Khronos:

https://issuetracker.google.com/issues/64059452
https://gitlab.khronos.org/Tracker/vk-gl-cts/issues/594

I've attached the patch to this email. The problem is that the dEQP  
*resize.surface_size* tests check the native window dimensions after 
creating the window, without having called eglSwapBuffers, and hence get 
back a native window width and height of zero.

Here is the patch description:

"The Wayland version of deqp-egl calls wl_egl_window_get_attached_size
to get window dimensions. On Mesa at least, this initially returns a
width and height of zero for a new window. The surface resize tests
check the initial surface size, and fail as a result.

Add a new native window capability, CAPABILITY_INITIAL_SURFACE_SIZE_QUERY,
which is only set if the initial window size can be queried after
creation. If native windows don't have the capability, return the
required size from getNativeSurfaceSize, rather than the size obtained
from the window system.

A parameter has been added to getNativeSurfaceSize, to indicate if
the initial size of a surface is being queried".

Brendan.


On 01/06/18 09:36, Juan A. Suarez Romero wrote:
> On Thu, 2018-05-31 at 17:24 +0100, Daniel Stone wrote:
>> Hi Eric,
>>
>> On 31 May 2018 at 17:13, Eric Engestrom <eric.engestrom at intel.com> wrote:
>>> On Thursday, 2018-05-31 16:57:17 +0100, Daniel Stone wrote:
>>>> Not initialising attached_{width,height} should not cause any problems
>>>> with these checks. By definition there cannot have been any buffers
>>>> allocated or attached between creation and the first draw call, so
>>>> there are no old buffers to release.
>>>>
>>>> So with that initialisation removed, this is (assuming dEQP still passes):
>>> IMG has a similar patch locally (Cc'ed Frank), because (IIRC) dEQP
>>> creates a window, and before attaching any surface to it, it reads back
>>> the size and asserts out of most tests because it's 0 or -1 or something
>>> like that, and I think this attached_width/height hack was needed
>>> because of that.
>> That's odd. As I understand it, eglQuerySurface(EGL_{WIDTH,HEIGHT})
>> should be returning the base surface values. attached_{width,height}
>> should only be returned to the user when calling
>> wl_egl_window_get_attached_size(). I'm happy to fix the former for
>> consistency with other platforms, but anyone calling the latter
>> without ever having attached anything deserves whatever they get, to
>> be honest.
> What dEQP-EGL.functional.resize.surface_size.* is doing:
>
> - Initialize:
>    - Create a wayland window with specific size (128x128)
>    - Create the surface with eglQuerySurface()
> - Perform the proper test
>    - Check if current surface size is the same as window size
>    - Resize the window
>    - Swap buffers
>    - Check again if the current surface size is the same as the window size
>
> The first check is the one that it is failing.
>
> I'm not sure if the test should swap buffers before the first check, as I didn't
>   find any specification or documentation stating it. But without such call,
> seems to work in other platforms.
>
>
> 	J.A.
>
>
>
>> Cheers,
>> Daniel
>>
> _______________________________________________
> mesa-dev mailing list
> mesa-dev at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-dev

-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-Fix-the-deqp-egl-resize.surface_size-for-Wayland.patch
Type: text/x-patch
Size: 5249 bytes
Desc: not available
URL: <https://lists.freedesktop.org/archives/mesa-dev/attachments/20180601/8def5cbd/attachment.bin>