Re: [PATCH v2 0/5] drm/panel-simple: Patches for N116BCA-EA1

[Doug Anderson]

Hi,

On Wed, Mar 10, 2021 at 3:25 PM Linus Walleij <linus.walleij@xxxxxxxxxx> wrote:
>
> On Fri, Jan 15, 2021 at 11:44 PM Douglas Anderson <dianders@xxxxxxxxxxxx> wrote:
>
> > - ("drm/panel-simple: Don't wait longer for HPD...") new for v2.
> > - ("drm/panel-simple: Retry if we timeout waiting for HPD") new for v2.
>
> I couldn't find these patches in my inbox

Doh! Sorry about that. I think get_maintainer tagged you only on the
patches that had the explicit "fixes" in them on something you were
involved in. I tend to rely on get_maintainer heavily unless I think
someone is particularly interested in the whole series. I will make
sure to CC you on the whole series if I post it again. In the meantime
the whole series is on lore:

https://lore.kernel.org/lkml/20210115224420.1635017-1-dianders@xxxxxxxxxxxx/

> but my concern would
> be that at some point panel-simple will turn from simple into
> panel-rube-goldberg-machine.

Yes, it's definitely something to watch out for. I guess you're
commenting on the general number of changes I've made to simple-panel
in the last year or two? I guess my comment on those:

* Many of the changes I made were around HPD handling and supporting a
GPIO (and also supporting the "hpd absent delay"). The fact that we
use a GPIO for HPD is actually an attribute of our board design,
though, so if I had forked panel drivers for each of the panels that
needed it then it would have required copying the code lots of places
(or implementing some code sharing). I was specifically asked to do
the HPD handling code at the panel layer.

* The other big change I made recently was around allowing for
relative rather than absolute timings (instead of a fixed delay at a
given stage adding a constraint that enough time had passed since a
previous event). When I proposed that the feedback I got back was that
it was a good improvement for all panels and something that had been
on a TODO list for a while.

...so while it feels like I keep piling crap onto simple-panel, I
_think_ they've been good general improvements that many people will
be able to benefit from and I don't think they've uglified things
lots.

> Given that the talk with the manufacturer may result
> in even more quirks... maybe this should just be a separate
> panel driver?

I don't _think_ this will end up with more quirks. At least I sure
hope not. So far it seems like pure luck that I even noticed it
because only one other device in the whole batch produced had similar
problems. That leads me to believe that there could be other panels
with a similar need.

> (I expect pushback because I see how handy it is, but
> I am the guy writing new panel drivers all the time rather than
> using simple.)

I guess what I'd say in summary is:

* If you object to the retries in simple panel, I still hope the rest
of the series can land.

* If somehow this panel gets out into real users hands and we find
that the retries are necessary and people still don't want the retries
in simple panel, I can fork a special panel driver just for it then.

-Doug