Re: [PATCH v4 3/3] drm/tiny: Add MIPI DBI compatible SPI driver

[Maxime Ripard <maxime@cerno.tech>]

[-- Attachment #1: Type: text/plain, Size: 3568 bytes --]

On Sun, Feb 20, 2022 at 10:32:25PM +0100, Sam Ravnborg wrote:
> Hi Noralf,
> 
> On Sun, Feb 20, 2022 at 04:59:34PM +0100, Noralf Trønnes wrote:
> > 
> > 
> > Den 19.02.2022 23.10, skrev Sam Ravnborg:
> > > Hi Noralf,
> > > On Fri, Feb 18, 2022 at 04:11:10PM +0100, Noralf Trønnes wrote:
> > >> Add a driver that will work with most MIPI DBI compatible SPI panels.
> > >> This avoids adding a driver for every new MIPI DBI compatible controller
> > >> that is to be used by Linux. The 'compatible' Device Tree property with
> > >> a '.bin' suffix will be used to load a firmware file that contains the
> > >> controller configuration.
> > > A solution where we have the command sequences as part of the DT-overlay
> > > is IMO a much better way to solve this:
> > > - The users needs to deal only with a single file, so there is less that
> > >   goes wrong
> > > - The user need not reading special instructions how to handle a .bin
> > >   file, if the overlay is present everything is fine
> > > - The file that contains the command sequences can be properly annotated
> > >   with comments
> > > - The people that creates the command sequences has no need for a special
> > >   script to create the file for a display - this is all readable ascii.
> > > - Using a DT-overlay the parsing of the DT-overlay can be done by
> > >   well-tested functions, no need to invent something new to parse the
> > >   file
> > > 
> > > 
> > > The idea with a common mipi DBI SPI driver is super, it is the detail
> > > with the .bin file that I am against.
> > > 
> > 
> > The fbtft drivers has an init= DT property that contains the command
> > sequence. Example for the MZ61581 display:
> > 
> > 				init = <0x10000b0 00
> > 					0x1000011
> > 					0x20000ff
> > 					0x10000b3 0x02 0x00 0x00 0x00
> > 					0x10000c0 0x13 0x3b 0x00 0x02 0x00 0x01 0x00 0x43
> > 					0x10000c1 0x08 0x16 0x08 0x08
> > 					0x10000c4 0x11 0x07 0x03 0x03
> > 					0x10000c6 0x00
> > 					0x10000c8 0x03 0x03 0x13 0x5c 0x03 0x07 0x14 0x08 0x00 0x21 0x08
> > 0x14 0x07 0x53 0x0c 0x13 0x03 0x03 0x21 0x00
> > 					0x1000035 0x00
> > 					0x1000036 0xa0
> > 					0x100003a 0x55
> > 					0x1000044 0x00 0x01
> > 					0x10000d0 0x07 0x07 0x1d 0x03
> > 					0x10000d1 0x03 0x30 0x10
> > 					0x10000d2 0x03 0x14 0x04
> > 					0x1000029
> > 					0x100002c>;
> > 
> > Parsed here:
> > https://elixir.bootlin.com/linux/latest/C/ident/fbtft_init_display_from_property
> > 
> > Before fbdev was closed for new drivers I looked at fixing up fbtft for
> > proper inclusion and asked on the Device Tree mailinglist about the init
> > property and how to handle the controller configuration generically.
> > I was politely told that this should be done in the driver, so when I
> > made my first DRM driver I made a driver specifically for a panel
> > (mi0283qt.c).
> > 
> > Later I found another post that in clear words stated that setting
> > random registers from DT was not acceptable.
>
> Understood and thanks for the explanation.
> It is a shame that the users loose here :-(

From a user point-of-view, nothing prevents the overlays and the
firmware to be in the same package, provided by whatever distro or
build-system they would use.

The only case where it's a bit less efficient is for a panel that isn't
supported already, but it's just a documentation and tooling issue, and
Noralf has an awesome track record for both.

And the DT syntax throw so much people off that I'm not sure it's such a
benefit :)

Maxime

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]