Le Mon, Oct 25, 2021 at 05:13:04PM +0200, Hans Verkuil a écrit :

On 25/10/2021 16:21, LABBE Corentin wrote:

quoted

Le Mon, Oct 25, 2021 at 02:45:02PM +0200, Hans Verkuil a écrit :

quoted

Hi Corentin,

On 13/10/2021 20:58, Corentin Labbe wrote:

quoted

Hello

The main change of this serie is to fusion all zoran related modules in
one.
This fixes the load order problem when everything is built-in.

Regards

Changes since v1:
- add missing debugfs cleaning
- clean some remaining module_get/put functions which made impossible to
  remove the zoran module
- added the two latest patchs

Something weird is wrong with this series. I have a DC30, but loading this with:

modprobe zr36067 card=3

results in this error message in the kernel log:

[   58.645557] zr36067: module is from the staging directory, the quality is unknown, you have been warned.
[   58.646658] zr36067 0000:03:00.0: Zoran MJPEG board driver version 0.10.1
[   58.646793] zr36067 0000:03:00.0: Zoran ZR36057 (rev 1), irq: 18, memory: 0xf4000000
[   58.648821] zr36067 0000:03:00.0: Initializing i2c bus...
[   58.662420] vpx3220 22-0047: vpx3216b found @ 0x8e (DC30[0])
[   58.737445] zr36067 0000:03:00.0: Fail to get encoder

This works before, so why this is now failing is not clear to me.

It does work with 'card=0', but I really have a DC30.

If I test with 'card=0' then the rmmod issue is now solved.

Everything normal, since card 0 does not have encoder.
Could you check that adv7175 is compiled ?

Yes, and it loaded as well (I see it with lsmod).

However, there is no adv7175 on my board, instead it appears to have an ITT MSE3000.
There is no driver for this one (and I don't even think it is an i2c device), so
I suspect that before the driver just continued without encoder support, whereas now
it fails when it can't load the encoder.

Could that be the reason? In the absence of an encoder, I think it should just
continue, esp. since the driver doesn't use the encoder anyway.

So probably the card list is wrong against DC30.
I checked high resolution photo of DC30 on internet, and it confirms the fact that DC30 does not have adv7175.

Since DC30 and DC30+ are identical in the card list, perhaps it is a very old copy/paste error.

So I will add a patch removing adv7175 from DC30.

Thanks for the report
Regards