Re: [PATCH 2/4] media: i2c: Add ST VD55G0 global shutter sensor driver

[Benjamin Mugnier <benjamin.mugnier@foss.st.com>]

Hi,

Le 20/10/2026 à 10:52, Sergey Lebedev a écrit :
> Hi George,
> 
> On 20/10/2026 07:44, George Neacsu wrote:
>> Would you be comfortable dropping that write in the next revision? I can
>> retest the revision on the Snapdragon model and add a Tested-by once it
>> is clean.
> 
> Yes, I will drop that write in v2 and leave 0x0329 at its default, and
> Cc you on it. Thank you for narrowing it down to one line.
> 
> On the Intel model it makes no difference. In one boot, with the sensor
> held powered and 0x0329 set to 1, then 0, then 1 again, each state
> streamed 150 of 150 frames at 59.14 fps and restarted 10 of 10 times.
> The test-pattern frames showed no shift or extra lines in any state, so
> IPU7 does not put the status lines into the buffer, and its error
> count did not change. The full test run without the write
> matches the run with it, v4l2-compliance 46 of 46 included.
> 
> One thing I cannot explain: on 16/10 your vd55g1 build streamed 2/2
> with 0x0329 = 0 written once per power-on, as v1 does it. So the write
> alone may not be what stops CAMSS; something else in v1 may take part.
> Leaving the register alone is simpler either way.

I'm a bit puzzled on this.

ISL are output on a different data type than regular data. They are
output with 0x12, which translates to MIPI_CSI2_DT_EMBEDDED_8B in V4L2.
Your receiver should drop these frames regardless as it only waits for
data and not ISL. Outputting them doesn't affect sensor timings either.

I guess there could be an issue on the receiver side ? Maybe your
receiver is ignoring data types and completes the frame with ISL data,
effectively filling up the DMA ?

> 
> Sergey

-- 
Regards,
Benjamin