TinyDRM ILI9486 overlay, and minor bcm2835-codec logging tweak

[6by9]

https://forums.raspberrypi.com/viewtopic.php?t=355484 for the TinyDRM overlay.

bcm2835-codec change for #5582

[pelwell]

Are you sure this works?:

$ dtmerge arch/arm/boot/dts/bcm2711-rpi-4-b.dtb base.dtb -
$ dtmerge base.dtb merged.dtb arch/arm/boot/dts/overlays/piscreen.dtbo drm
DTOVERLAY[error]: invalid override value 'GPIO_ACTIVE_HIGH' - ignored

[6by9]

You're right - my update script to push the change to the Pi obviously didn't for some reason (probably PEBKAC).

[6by9]

Actually there's a bigger problem. In the non-drm case (compatible ilitek,ili9486), both drivers get loaded

pi@pi:~ $ modinfo fb_ili9486
filename:       /lib/modules/6.1.47-v8+/kernel/drivers/staging/fbtft/fb_ili9486.ko.xz
license:        GPL
author:         Noralf Tronnes
description:    FB driver for the ILI9486 LCD Controller
alias:          platform:ili9486
alias:          spi:ili9486
alias:          platform:fb_ili9486
alias:          spi:fb_ili9486
srcversion:     A87A9FEB7E5C3553E7CDE5A
alias:          of:N*T*Cilitek,ili9486C*
alias:          of:N*T*Cilitek,ili9486
depends:        fbtft
staging:        Y
intree:         Y
name:           fb_ili9486
vermagic:       6.1.47-v8+ SMP preempt mod_unload modversions aarch64

pi@pi:~ $ modinfo ili9486
filename:       /lib/modules/6.1.47-v8+/kernel/drivers/gpu/drm/tiny/ili9486.ko.xz
license:        GPL
author:         Kamlesh Gurudasani <[email protected]>
description:    Ilitek ILI9486 DRM driver
srcversion:     6FBFCC0E15DDF43CA36470A
alias:          of:N*T*Cozzmaker,piscreenC*
alias:          of:N*T*Cozzmaker,piscreen
alias:          of:N*T*Cwaveshare,rpi-lcd-35C*
alias:          of:N*T*Cwaveshare,rpi-lcd-35
alias:          spi:ili9486
depends:        drm_kms_helper,drm,drm_mipi_dbi,backlight,drm_dma_helper
intree:         Y
name:           ili9486
vermagic:       6.1.47-v8+ SMP preempt mod_unload modversions aarch64

Presumably it's of:N*T*Cilitek,ili9486 vs spi:ili9486 causing the issue.

[pelwell]

Presumably it's of:NTCilitek,ili9486 vs spi:ili9486 causing the issue.

Sorry, missed this pseudo-question - yes, that is probably the cauase.

Can we just disable the fbtft version?

[6by9]

Can we just disable the fbtft version?

Not really unless we change the overlay to only supporting the DRM one.

Both are mainline drivers, so probably worth reporting to mainline to see what they want to do. The fact that SPI has this secondary matching (and complaining if the SPI alias is missing) is really a bit ugly.

[pelwell]

In the same way that the fbtft driver uses spi:fb_ili9486, we could patch the upstream DRM driver so it also uses a prefix for the SPI alias, e.g. spi:drm_ili9486.

[6by9]

I'm not totally sure on how the spi_id table is used, except that the SPI core whinges if a compatible string device name isn't present in the spi_id table as well. So it complains on ili9486 as rpi-lcd-35 and piscreen aren't present.
(It also complains that ili9486 is missing for fb_ili9486).

Easiest fix is

@ -187,7 +187,8 @@ static const struct of_device_id ili9486_of_match[] = {
 MODULE_DEVICE_TABLE(of, ili9486_of_match);
 
 static const struct spi_device_id ili9486_id[] = {
-       { "ili9486", 0 },
+       { "rpi-lcd-35", 0 },
+       { "piscreen", 0 },
        { }
 };
 MODULE_DEVICE_TABLE(spi, ili9486_id);

thus silencing the SPI core code on the mismatch, and removing the clash.

Whilst at it I'll add the missing spi_device_id entries in ads7846 (the touchscreen driver) so at least with the DRM option the SPI core is happy.

[pelwell]

This also works, if you prefer:

<&piscreen>,"reset-gpios:8=",<GPIO_ACTIVE_HIGH>;

but I'm happy to merge as is.

[6by9]

This also works, if you prefer:

Give me two mins to update to that - it's much nicer.