Design question backlight controller powered by bridge

[Fuzzey, Martin]

Hi all,

I have a design problem I'm not sure how to solve cleanly.

I have a board using a PTN3460  eDP => LVDS bridge and a LP8556
backlight controller.
Both of these are already supported by the kernel.

The PTN3460 has an electrical output signal (PVCGEN) that is intended
to control power to the panel and goes high once valid video data is
received on the eDP link.
However on this board this signal actually switches the power to the LP8556.

The consequences of this are
* Until the video link is up it is not possible to communicate with
the LP8556 over I2C (indeed the driver fails to probe due to this)
* If the video link goes down the LP8556 loses all its state

The LP8556 can either control the backlight via I2C or by an external
PWM input signal (which is connected to the PTN3460)
So I could *mostly* solve the problem of getting a backlight control
by not using I2C communication with the LP8556 and controlling the
backlight using the PWM.
Currently the PTN3460 driver does not implement PWM control but it
would be fairly easy to add and there is already a precedent in the
parade-ps8622 driver that implements a backlight control.

However that does not solve the I2C communications issue, just avoids
the need for it for setting the backlight brightness.
So if I need to do other things with the LP8556 (like tuning the drive
strengths etc it's back to the same problem).

What seems to be needed is a way of expressing that the PTN3460
supplies power to the LP8556 but the LP8556 does not control it.
A normal regulator relationship doesn't work because there it is
expected that the consumer (LP8556) controls the regulator.

Any ideas appreciated.

Martin