[RFC PATCH] dt-bindings: display: ti,tfp410.txt: convert to yaml

Laurent Pinchart laurent.pinchart at ideasonboard.com
Wed May 13 14:08:32 UTC 2020

Hi Ricardo,

On Wed, May 13, 2020 at 01:09:57PM +0200, Ricardo Cañuelo wrote:
> On mié 06-05-2020 18:53:20, Laurent Pinchart wrote:
> > I didn't if I remember correctly, I just mapped it to the hardware
> > features. The hardware register indeed takes a value between 0 and 7,
> > and that is mapped to [-4,3] x t(STEP). I don't mind either option.
> I was taking a look at the ti-tfp410.c driver to see if it needs any
> changes to support the updated deskew property ranges [0-7], but I don't
> fully understand what this does (line 276):
> 	/* Get the setup and hold time from vendor-specific properties. */
> 	of_property_read_u32(dvi->dev->of_node, "ti,deskew", (u32 *)&deskew);
> 	if (deskew < -4 || deskew > 3)
> 		return -EINVAL;
> 	timings->setup_time_ps = min(0, 1200 - 350 * deskew);
> 	timings->hold_time_ps = min(0, 1300 + 350 * deskew);
> It looks like that the driver doesn't really apply the deskew settings
> to the device and that this has not been really tested, so it's probably
> not a big deal.

The driver doesn't apply any setting to the device :-) The ti,deskew
property is meant to report the deskew settings selected by the chip's
configuration pins, not to set a value to be programmed to the device.

> I guess what you wanted to do was to adjust the setup and hold times
> around 1200 and 1300 ps respectively in increments/decrements of 350 ps
> depending on the deskew value, as the datasheet describes. But this code
> would set timings->setup_time_ps to 0 regardless of the deskew value,
> and timings->hold_time_ps would be either 0 or a very big integer value
> if deskew is -4 (both setup_time_ps and hold_time_ps are u32).
> Am I missing something? Was this intentional?

Oops. That's embarassing... It should clearly be a max(), not a min().
And only for hold_time_ps is this required.

Would you like to send a patch, or should I do so ?


Laurent Pinchart

More information about the dri-devel mailing list