STM32 DSI controller driver: mode_valid clock tolerance

[Maxime Ripard]

On Thu, Mar 21, 2024 at 09:47:18AM +0100, Sean Nyekjaer wrote:
> >> If HDMI is requiring a tolerance of 50 Hz, would it be better to do
> >> the check in the HDMI bridge driver?
> > 
> > Judging from the code, it would be hard to do because the clock that
> > generates the pixel clock isn't exposed to the HDMI bridge.
> 
> What do you think about Raphael’s proposal to add the tolerance as a
> device tree option?

I don't think it's the right solution, for multiple reasons:

  - There's no way for OEM to come up with a decent value;

  - That value isn't platform specific in the first place;

  - Allowing any value will create troubles somewhere else in the stack,
    for example for applications that rely on accurate vblank;

  - We could come up with something that would make this entirely
    irrelevant, like being able to come up with proper drm_display_mode
    based on display_timings and a given clock rate.

So it's a software problem, it needs to be solved at the software layer,
not in the platform description layer, no matter how convenient it is.

Like I said, VESA DMT and CVT specs use a .5% tolerance and it works
quite well for panels too for other platforms. Just use that and see if
it's still an issue.

Maxime
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 228 bytes
Desc: not available
URL: <https://lists.freedesktop.org/archives/dri-devel/attachments/20240322/67ebf142/attachment.sig>