Re: [PATCH v4 3/3] drm/panel-fannal-c3004: Add fannal c3004 DSI panel

[Marek Vasut]

On 7/12/23 14:07, Paulo Pavacic wrote:
> Hello all,
>
> sub, 8. srp 2023. u 14:53 Marek Vasut <marex@xxxxxxx> napisao je:
> > On 7/7/23 17:26, Paulo Pavacic wrote:
> > > Hello Marek,
> >
> > Hi,
> >
> > > čet, 6. srp 2023. u 17:26 Marek Vasut <marex@xxxxxxx> napisao je:
> > > > On 7/6/23 17:18, Paulo Pavacic wrote:
> > > > > Hello Linus,
> > > > >
> > > > > čet, 22. lip 2023. u 10:22 Linus Walleij <linus.walleij@xxxxxxxxxx> napisao je:
> > > > > > On Wed, Jun 21, 2023 at 5:09 PM Paulo Pavacic <pavacic.p@xxxxxxxxx> wrote:
> > > > > >
> > > > > > > A lot of modifications to st7701 are required. I believe it would
> > > > > > > result in a driver that doesn't look or work the same. e.g compare
> > > > > > > delays between initialization sequences of panel-fannal-c3004 and
> > > > > > > panel-st7701. I think it would be optimal to create st7701s driver and
> > > > > > > have special handling for st7701s panels. If there was a flag for
> > > > > > > whether panel is st7701 or st7701s it would end up looking like a
> > > > > > > mess.
> > > > > >
> > > > > > What matters is if the original authors of the old st7701 driver are
> > > > > > around and reviewing and testing patches at all. What we need is
> > > > > > active maintainers. (Added Jagan, Marek & Maya).
> > > > > >
> > > > > > I buy the reasoning that the st7701s is perhaps substantially different
> > > > > > from st7701.
> > > > > >
> > > > > > If st7701s is very different then I suppose it needs a separate driver,
> > > > > > then all we need to to name the driver properly, i.e.
> > > > > > panel-sitronix-st7701s.c.
> > > > >
> > > > > I had in person talk with Paul Kocialkowski and I have concluded that
> > > > > this is the best solution.
> > > > > I believe I should rename it to st7701s due to the hardware changes. I
> > > > > would like to create V5 patch with driver renamed to st7701s.
> > > > > Please let me know if you agree / disagree.
> > > >
> > > > If I recall it right, the ST7701 and ST7701S are basically the same
> > > > chip, aren't they ?
> > >
> > > I'm currently exploring all the differences. There aren't a lot of
> > > differences, but there are some.
> > > So far I can see that default register values are different, new
> > > previously unused registers are now used and there has been some
> > > reordering of how info is placed in registers [1] (data bits are in
> > > different order). Moreover, instructions to some commands have been
> > > changed and meaning of what data bits mean [2][3]. Also, new features
> > > have been added [2]; there is now PCLKS 3 for example.
> > >
> > > You can see few differences in following images. Same images were
> > > attached in this mail:
> > > [1] https://ibb.co/NmgbZmy - GAMACTRL_st7701.png
> > > [2] https://ibb.co/G79y235 - PCLKS2.png
> >
> > Ouch. I wonder if this is still something that can be abstracted out
> > with some helper accessor functions like:
> >
> > if (model == ST7701)
> >     write something
> > else
> >     write the other layout
> >
> > Or whether it makes sense to outright have a separate driver. The later
> > would introduce duplication, but maybe that much duplication is OK.
>
> I would like to create new driver because panel-st7701 seems to be
> outdated and is using non-standard macro (ST7701_WRITE()

There is no such macro:

$ git grep ST7701_WRITE drivers/gpu/drm/panel/ | wc -l
0

There never was such a macro used in the driver either, are you sure you 
are not using some hacked up patched downstream fork of the driver ?

$ git log -p next/master -- 
drivers/gpu/drm/panel/panel-sitronix-st7701.c | grep ST7701_WRITE | wc -l
0

> ) and for me
> it is crashing kernel 5.15.

Have you based all the aforementioned discussion and argumentation on 
year and half old Linux 5.15.y code base too ?

If so, you are missing many patches:

$ git log --oneline --no-merges v5.15..next/master -- 
drivers/gpu/drm/panel/panel-sitronix-st7701.c
5a2854e577dc2 drm: panel: Add orientation support for st7701
e89838968ee44 drm: panel: Add Elida KD50T048A to Sitronix ST7701 driver
c62102165dd79 drm/panel/panel-sitronix-st7701: Remove panel on DSI 
attach failure
49ee766b364ed drm/panel/panel-sitronix-st7701: Clean up CMDnBKx selection
c1cdee9b685a1 drm/panel/panel-sitronix-st7701: Fix RTNI calculation
57b2efce45ef5 drm/panel/panel-sitronix-st7701: Add Densitron 
DMT028VGHMCMI-1A TFT
42542c7904cf2 drm/panel/panel-sitronix-st7701: Split GIP and init sequences
83b7a8e7e88e7 drm/panel/panel-sitronix-st7701: Parametrize voltage and 
timing
de2b4917843cd drm/panel/panel-sitronix-st7701: Infer horizontal pixel 
count from TFT mode
82f9cee25598a drm/panel/panel-sitronix-st7701: Adjust porch control 
bitfield name
1ba85119afb5e drm/panel/panel-sitronix-st7701: Infer vertical line count 
from TFT mode
779c84fea3dbd drm/panel/panel-sitronix-st7701: Make gamma correction TFT 
specific
7fa8e07128ed6 drm/panel/panel-sitronix-st7701: Make voltage supplies 
common to ST7701
a6c225be3da7e drm/panel/panel-sitronix-st7701: Enable DSI burst mode, 
LPM, non-continuous clock
6f481afe220d3 drm/panel/panel-sitronix-st7701: Make DSI mode flags 
common to ST7701
79abca2b39900 drm/mipi-dsi: Make remove callback return void

> Does anyone have similar issues with it?

No, I am using it in production.