v5.15-rcX regression: video devices on i.MX6 are not created

v5.15-rcX regression: video devices on i.MX6 are not created

[Krzysztof Hałasa]

Hi Saravana et al.,

FYI the patch: 6b2117ad65f1bca9ece6d4b1ee784b42701a2d86

of: property: fw_devlink: Add support for "resets" and "pwms"

Allows better tracking of dependencies between devices.

 drivers/of/property.c | 4 ++++

breaks v4l2 video devices on an i.MX6 CPU. The /dev/video* nodes are not
created, and the drivers (async subdevicess) are "waiting":

# cat /sys/kernel/debug/v4l2-async/pending_async_subdevices

imx-media:
 [fwnode] dev=nil, node=/soc/ipu@2400000/port@0
 [fwnode] dev=nil, node=/soc/ipu@2400000/port@1
 [fwnode] dev=nil, node=/soc/ipu@2800000/port@0
 [fwnode] dev=nil, node=/soc/ipu@2800000/port@1
imx6-mipi-csi2:
 [fwnode] dev=0-0036, node=/soc/bus@2100000/i2c@21a0000/camera@36
ipu2_csi1_mux:
ipu1_csi0_mux:
 [fwnode] dev=21dc000.mipi, node=/soc/bus@2100000/mipi@21dc000

Reverting the patch (or commenting the "resets" out) fixes the problem:

# cat /sys/kernel/debug/v4l2-async/pending_async_subdevices
ipu2_csi1:
ipu2_csi0:
ipu1_csi1:
ipu1_csi0:
imx-media:
imx6-mipi-csi2:
ipu2_csi1_mux:
ipu1_csi0_mux:


What now?
--
Krzysztof "Chris" Hałasa

Sieć Badawcza Łukasiewicz
Przemysłowy Instytut Automatyki i Pomiarów PIAP
Al. Jerozolimskie 202, 02-486 Warszawa

Re: v5.15-rcX regression: video devices on i.MX6 are not created

[Saravana Kannan]

On Mon, Oct 4, 2021 at 5:19 AM Krzysztof Hałasa <khalasa@piap.pl> wrote:
>
> Hi Saravana et al.,
>
> FYI the patch: 6b2117ad65f1bca9ece6d4b1ee784b42701a2d86
>
> of: property: fw_devlink: Add support for "resets" and "pwms"
>
> Allows better tracking of dependencies between devices.
>
>  drivers/of/property.c | 4 ++++
>
> breaks v4l2 video devices on an i.MX6 CPU. The /dev/video* nodes are not
> created, and the drivers (async subdevicess) are "waiting":
>
> # cat /sys/kernel/debug/v4l2-async/pending_async_subdevices
>
> imx-media:
>  [fwnode] dev=nil, node=/soc/ipu@2400000/port@0
>  [fwnode] dev=nil, node=/soc/ipu@2400000/port@1
>  [fwnode] dev=nil, node=/soc/ipu@2800000/port@0
>  [fwnode] dev=nil, node=/soc/ipu@2800000/port@1
> imx6-mipi-csi2:
>  [fwnode] dev=0-0036, node=/soc/bus@2100000/i2c@21a0000/camera@36
> ipu2_csi1_mux:
> ipu1_csi0_mux:
>  [fwnode] dev=21dc000.mipi, node=/soc/bus@2100000/mipi@21dc000
>
> Reverting the patch (or commenting the "resets" out) fixes the problem:
>
> # cat /sys/kernel/debug/v4l2-async/pending_async_subdevices
> ipu2_csi1:
> ipu2_csi0:
> ipu1_csi1:
> ipu1_csi0:
> imx-media:
> imx6-mipi-csi2:
> ipu2_csi1_mux:
> ipu1_csi0_mux:
>
>
> What now?

Hi Krzysztof,

Can you give me the following details?

1. The dts file that corresponds to the device you are testing this on.
2. Output of <debugfs>/devices_deferred base on at least 5.15-rc4 (I
added more info there recently).

That should tell me all the info I need to debug this. In the end
it'll essentially boil down to a reset controller driver not using
device driver model to initialize the device and directly parsing the
node in some kind of initcall.

Thanks,
Saravana
> --
> Krzysztof "Chris" Hałasa
>
> Sieć Badawcza Łukasiewicz
> Przemysłowy Instytut Automatyki i Pomiarów PIAP
> Al. Jerozolimskie 202, 02-486 Warszawa

Re: v5.15-rcX regression: video devices on i.MX6 are not created

[Fabio Estevam]

Hi Krzysztof,

On Tue, Oct 5, 2021 at 9:10 PM Saravana Kannan <saravanak@google.com> wrote:
>
> On Mon, Oct 4, 2021 at 5:19 AM Krzysztof Hałasa <khalasa@piap.pl> wrote:
> >
> > Hi Saravana et al.,
> >
> > FYI the patch: 6b2117ad65f1bca9ece6d4b1ee784b42701a2d86
> >
> > of: property: fw_devlink: Add support for "resets" and "pwms"
> >
> > Allows better tracking of dependencies between devices.
> >
> >  drivers/of/property.c | 4 ++++
> >
> > breaks v4l2 video devices on an i.MX6 CPU. The /dev/video* nodes are not
> > created, and the drivers (async subdevicess) are "waiting":

Phillip has sent a fix for this issue:
http://lists.infradead.org/pipermail/linux-arm-kernel/2021-October/687868.html

Re: v5.15-rcX regression: video devices on i.MX6 are not created

[Krzysztof Hałasa]

Hi Fabio, Saravana,

Fabio Estevam <festevam@gmail.com> writes:

>> > FYI the patch: 6b2117ad65f1bca9ece6d4b1ee784b42701a2d86
>> >
>> > of: property: fw_devlink: Add support for "resets" and "pwms"
>> >
>> > Allows better tracking of dependencies between devices.
>> >
>> >  drivers/of/property.c | 4 ++++
>> >
>> > breaks v4l2 video devices on an i.MX6 CPU. The /dev/video* nodes are not
>> > created, and the drivers (async subdevicess) are "waiting":
>
> Phillip has sent a fix for this issue:
> http://lists.infradead.org/pipermail/linux-arm-kernel/2021-October/687868.html

This patch makes the /dev/video* devices to show up on my i.MX6-based
system again, so I consider the problem fixed. Thanks.
-- 
Krzysztof "Chris" Hałasa

Sieć Badawcza Łukasiewicz
Przemysłowy Instytut Automatyki i Pomiarów PIAP
Al. Jerozolimskie 202, 02-486 Warszawa

Re: v5.15-rcX regression: video devices on i.MX6 are not created

[Saravana Kannan]

On Tue, Oct 5, 2021 at 9:56 PM Krzysztof Hałasa <khalasa@piap.pl> wrote:
>
> Hi Fabio, Saravana,
>
> Fabio Estevam <festevam@gmail.com> writes:
>
> >> > FYI the patch: 6b2117ad65f1bca9ece6d4b1ee784b42701a2d86
> >> >
> >> > of: property: fw_devlink: Add support for "resets" and "pwms"
> >> >
> >> > Allows better tracking of dependencies between devices.
> >> >
> >> >  drivers/of/property.c | 4 ++++
> >> >
> >> > breaks v4l2 video devices on an i.MX6 CPU. The /dev/video* nodes are not
> >> > created, and the drivers (async subdevicess) are "waiting":
> >
> > Phillip has sent a fix for this issue:
> > http://lists.infradead.org/pipermail/linux-arm-kernel/2021-October/687868.html
>
> This patch makes the /dev/video* devices to show up on my i.MX6-based
> system again, so I consider the problem fixed. Thanks.

Ah, I helped Phillip come up with that fix. I didn't realize this was
the same issue. Glad that I have one less issue to debug :)

-Saravana