On 6/27/19 5:56 AM, Philipp Zabel wrote:
Hi Fabio,
On Thu, 2019-06-27 at 09:38 -0300, Fabio Estevam wrote:
Hi Philipp,
On Thu, Jun 27, 2019 at 5:43 AM Philipp Zabel <p.zabel@xxxxxxxxxxxxxx> wrote:
Are there any visual artifacts in the first frame(s) in this case?
I do not observe visual artifacts when running gst-launch-1.0 v4l2src ! kmssink
So in my opinion the next version of this patch should make LP-11
timeout a warning only, but keep the error return on clock lane timeouts.
I agree.
Here is a reworked version of Ezequiel's patch as per the suggestions:
http://code.bulix.org/g5qap5-780475
Does this one look good?
Limiting the change to wait_stopstate is fine, the actual message
makes assumptions that could be misleading. How about:
"Timeout waiting for LP-11 state on all active lanes.
This is most likely caused by a bug in the sensor driver.
Capture might fail or contain visual artifacts."
Yes I agree that is more descriptive, if a bit wordy for a kernel error
message. I think it could be reduced, something like:
"LP-11 wait timeout on all lanes, likely a sensor driver bug, expect
capture failures."
I'd like to keep the phy_state register output though, if only as
dev_dbg(). It contains useful output for debugging, for example if only
some of the lanes are in stop state, which could indicate an issue with
connections or lane configuration.
Agreed!
Steve
