mbox series

[RFC,V3,0/2] Attach and Set vrr_enabled property

Message ID 20220517072636.3516381-1-bhanuprakash.modem@intel.com (mailing list archive)
Headers show
Series Attach and Set vrr_enabled property | expand

Message

Modem, Bhanuprakash May 17, 2022, 7:26 a.m. UTC
This series will add a support to set the vrr_enabled property for
crtc based on the platform support and the request from userspace.
And userspace can also query to get the status of "vrr_enabled".

Test-with: 20220422075223.2792586-2-bhanuprakash.modem@intel.com

Bhanuprakash Modem (2):
  drm/vrr: Attach vrr_enabled property to the drm crtc
  drm/i915/vrr: Set drm crtc vrr_enabled property

 drivers/gpu/drm/drm_crtc.c               | 26 ++++++++++++++++++++++++
 drivers/gpu/drm/drm_mode_config.c        |  2 +-
 drivers/gpu/drm/i915/display/intel_vrr.c |  8 ++++++++
 include/drm/drm_crtc.h                   |  3 +++
 4 files changed, 38 insertions(+), 1 deletion(-)

--
2.35.1

Comments

Daniel Vetter May 31, 2022, 1:56 p.m. UTC | #1
On Tue, May 17, 2022 at 12:56:34PM +0530, Bhanuprakash Modem wrote:
> This series will add a support to set the vrr_enabled property for
> crtc based on the platform support and the request from userspace.
> And userspace can also query to get the status of "vrr_enabled".
> 
> Test-with: 20220422075223.2792586-2-bhanuprakash.modem@intel.com
> 
> Bhanuprakash Modem (2):
>   drm/vrr: Attach vrr_enabled property to the drm crtc
>   drm/i915/vrr: Set drm crtc vrr_enabled property

I'm rather confused by this patch set:

- This seems to move the property from connector to crtc without any
  justification. For uapi that we want to have standardized (anything
  around kms really) that's no good, unless there's really a mandatory
  semantic reason pls stick to existing uapi.

- If the driver interface doesn't fit (maybe the helper should be on the
  crtc and adjust the property for all connector) pls roll that change out
  to all drivers.

- This is uapi, so needs igt tests and userspace. For igts we should make
  sure they're generic so that they apply across all drivers which already
  support this property, and not just create new intel-only testcases.

- Finally the property is set up, but not wired through. Or at least I'm
  not seeing how this can even work.

So no idea what exactly you're aiming for here and what kind of comments
you want, but this doesn't look like it's on the right path at all.

Cheers, Daniel


> 
>  drivers/gpu/drm/drm_crtc.c               | 26 ++++++++++++++++++++++++
>  drivers/gpu/drm/drm_mode_config.c        |  2 +-
>  drivers/gpu/drm/i915/display/intel_vrr.c |  8 ++++++++
>  include/drm/drm_crtc.h                   |  3 +++
>  4 files changed, 38 insertions(+), 1 deletion(-)
> 
> --
> 2.35.1
>
Modem, Bhanuprakash June 1, 2022, 10:31 a.m. UTC | #2
On Tue-31-05-2022 07:26 pm, Daniel Vetter wrote:
> On Tue, May 17, 2022 at 12:56:34PM +0530, Bhanuprakash Modem wrote:
>> This series will add a support to set the vrr_enabled property for
>> crtc based on the platform support and the request from userspace.
>> And userspace can also query to get the status of "vrr_enabled".
>>
>> Test-with: 20220422075223.2792586-2-bhanuprakash.modem@intel.com
>>
>> Bhanuprakash Modem (2):
>>    drm/vrr: Attach vrr_enabled property to the drm crtc
>>    drm/i915/vrr: Set drm crtc vrr_enabled property
> 
> I'm rather confused by this patch set >
> - This seems to move the property from connector to crtc without any
>    justification. For uapi that we want to have standardized (anything
>    around kms really) that's no good, unless there's really a mandatory
>    semantic reason pls stick to existing uapi.

Thanks for the reply.

No, we are not moving any property from connector to crtc.
"vrr_capable" belongs to connector & "vrr_enabled" belongs to crtc.

I am trying to manipulate this "vrr_enabled" property.

> 
> - If the driver interface doesn't fit (maybe the helper should be on the
>    crtc and adjust the property for all connector) pls roll that change out
>    to all drivers.
> 
> - This is uapi, so needs igt tests and userspace. For igts we should make
>    sure they're generic so that they apply across all drivers which already
>    support this property, and not just create new intel-only testcases.

Yeah, IGT patches are already in ML:
https://patchwork.freedesktop.org/series/100539/

> 
> - Finally the property is set up, but not wired through. Or at least I'm
>    not seeing how this can even work.
> 
> So no idea what exactly you're aiming for here and what kind of comments
> you want, but this doesn't look like it's on the right path at all.

This "vrr_enabled" prop was userspace write only & driver read only prop 
which would be used by the userspace to request to enable the VRR on 
that CRTC.

Now I would like to modify this prop to be used as a bidirectional 
property. So, that userspace can request to enable the VRR and also to 
get the status of VRR on that CRTC.

Manasi is already replied to the Patch [1/2] in this series.

- Bhanu

> 
> Cheers, Daniel
> 
> 
>>
>>   drivers/gpu/drm/drm_crtc.c               | 26 ++++++++++++++++++++++++
>>   drivers/gpu/drm/drm_mode_config.c        |  2 +-
>>   drivers/gpu/drm/i915/display/intel_vrr.c |  8 ++++++++
>>   include/drm/drm_crtc.h                   |  3 +++
>>   4 files changed, 38 insertions(+), 1 deletion(-)
>>
>> --
>> 2.35.1
>>
>