On 02/08/18 16:13, Philipp Zabel wrote: > On Thu, 2018-02-08 at 13:01 +0100, Hans Verkuil wrote: >>> These are likely to be filled correctly already. I've just added a commit >>> to v4l2-compliance to make it easier to see what function is used: >>> >>> v4l2-compliance -m0 -v >> >> Actually, can you run this using the latest v4l-utils version for the imx >> and post the output? > > I have tried with v4l-utils-1.14.0-111-g542af94c on a platform with a > Toshiba TC358743 connected via MIPI CSI-2. Apart from a crash , I get > a few: > - missing G_INPUT on the capture devices - is that really a bug?
Depends who you ask :-) It's an open issue, we had a patch for this that was never picked up. Hopefully this will resurrect that patch. > - cap->timeperframe.numerator == 0 || cap->timeperframe.denominator == 0, > where there is nothing connected that could provide timing information > - missing enum_mbus_code > - check_0(reserved) errors on subdev ioctls Should be fixed by https://patchwork.linuxtv.org/patch/46955/ and https://patchwork.linuxtv.org/patch/46950/ > - node->enum_frame_interval_pad != (int)pad > - subscribe event failures > - g_ext_ctrls does not support count == 0 (which no subdev implements) Was this patch applied? https://patchwork.linuxtv.org/patch/46958/ That will probably fix this fail. > >  https://patchwork.linuxtv.org/patch/46979/ > > The CSIs are currently marked as pixel formatters instead of IF bridges, > > the vdics are marked as pixel formatters instead of deinterlacers, and > the ic_prp is marked as scaler instead of video splitter. The other > entity functions are initialized correctly. Thanks for testing this! Hans > Total: 1307, Succeeded: 1089, Failed: 218, Warnings: 4 Total: 1307 Cool! Hans