+Cc: dri-devel@xxxxxxxxxxxxxxxxxxxxx
On 30/11/24 - 13:49, Ananta Srikar Puranam wrote:
> Dear Louis Chauvet,
Hi Srikar!
> I was able to successfully compile Linux with VKMS enabled and have set up
> igt-gpu-tests in a QEMU machine. I can now run some tests successfully.
Congratulation!
> I have a surface-level understanding of the DRM and KMS subsystems
> (framebuffers, planes, CRTCs, encoders, and connectors) and am eager to dive
> deeper into the VKMS module and contribute.
I am very happy to see some interest in VKMS development, do you have a
specific feature that you want to test?
> I understand VKMS helps test the DRM subsystem, but I’m struggling to fully
> grasp how the absence of visible display output aligns with its testing
> goals. Any guidance on this would be highly appreciated.
That the whole point of "V"KMS, everything is virtual, even the display
output :)
There are two way to get some information about the output content:
- CRC compuation [1]
- Writeback connector [2]
The CRC will allows you to write a test like this:
create_reference_image();
ref_crc = grab_crc();
test_procedure();
test_crc = grab_crc();
assert(ref_crc, test_crc);
This is used in the igt test "kms_plane" to test color conversion, they
compute a reference CRC and then test the conversion between multiple
formats [3].
The writeback connector is useful to grab a "screenshot" of the output,
but it is a bit less efficient. It can be used when a simple "reference
CRC" is not possible.
To grab a picture from the writeback connector, you have to do something
like this:
setup_pipeline();
wb_fb = create_fb();
while (true) {
send_fb_to_wb_connector(wb_fb);
wait_for_picture();
do_something_with_result(wb_fb);
}
There is a writeback example in igt [4], but I don't know how it works.
[1]:https://elixir.bootlin.com/linux/v6.12.1/source/drivers/gpu/drm/vkms/vkms_composer.c#L209
[2]:https://elixir.bootlin.com/linux/v6.12.1/source/drivers/gpu/drm/vkms/vkms_writeback.c#L171
[3]:https://gitlab.freedesktop.org/drm/igt-gpu-tools/-/blob/master/tests/kms_plane.c#L1036
[4]:https://gitlab.freedesktop.org/drm/igt-gpu-tools/-/blob/master/tests/kms_writeback.c
> Additionally, I noticed the TODO item to add the plane feature (background
> color KMS property). It seems like a good starting point, and I would be
> grateful for direction on how to approach it.
I think this is the good direction, yes!
All your patches must be on top of drm-misc-next [5]. If you are not
familiar with kernel contribution, I invite you to look at the kernel
documentation [6] and at the amazing b4 tool [7].
For the implementation itself, I think you have to work in the function
blend, probably around the fill_background function [8]. You also need to
attach a BACKGROUND property on the CRTC [9]. I think you have to do it
somewhere in crtc initialization [10].
[5]:https://gitlab.freedesktop.org/drm/misc/kernel/-/tree/drm-misc-next
[6]:https://docs.kernel.org/process/contribution-maturity-model.html
[7]:https://b4.docs.kernel.org/
[8]:https://gitlab.freedesktop.org/drm/misc/kernel/-/blob/drm-misc-next/drivers/gpu/drm/vkms/vkms_composer.c#L389
[9]:https://drmdb.emersion.fr/properties/3435973836/BACKGROUND
[10]:https://gitlab.freedesktop.org/drm/misc/kernel/-/blob/drm-misc-next/drivers/gpu/drm/vkms/vkms_crtc.c?ref_type=heads#L273
> Thank you for your time, and I look forward to your guidance.
>
> Best regards,
> Srikar
--
Louis Chauvet, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
