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

[George Neacsu <geocausa@gmail.com>]

Hi Benjamin, Sergey,

On 20/10/2026 15:26, Sergey Lebedev wrote:
> So I agree the register itself
> should not matter there; something else in v1 probably does, and George
> can narrow it down on CAMSS better than I can.

You were both right, and I have to correct my earlier report.

I looked at the receiver side with a CSID change that only reads
registers at stream start. RDI0 is set up for data type 0x2b (RAW10),
VC 0, with no ECC or CRC errors, so CAMSS filters on the data type and
the ISL lines (DT 0x12) should not reach the buffer.

With that in place I reran the exact v1 module that stalled this morning:
same build, 403.2 MHz in the DT, and the ISL_ENABLE=0 write still in. It
streamed 5/5 frames, 120/120 at about 60.2 fps, and 10/10 on an
immediate restart. Builds that wrote 0 and builds that left the default
behaved the same, and read back 0x00 and 0x01 respectively.

To rule out the diagnostic reads changing the timing, I then booted
again with the unmodified CAMSS module and the same exact v1 module:
three rounds of 5, 120 and 10 frames, all complete, at about 60.2 fps,
with no CAMSS errors.

So clearing 0x0329 does not cause the stall on this machine. My earlier
A/B was a correlation, not the cause; something else in those runs made
the difference, and I have not pinned it down yet.

I tried to count the DT 0x12 packets directly with the CSID's
long-packet capture, but enabling that capture changed the behaviour
itself (one frame only), so it could not serve as a passive observer.

Sergey, the write can stay or go as you prefer; the Snapdragon model
does not need it dropped. The 403.2 MHz link frequency still stands.
I will test v2 on the X1E80100 and send a Tested-by if it passes.

Sorry for the noise.

Thanks,
George