On 08/22/2014 01:39 PM, Mikhail Ulianov wrote: > Hi Hans, > > Thanks for your comments. > > I have a question on default width, height and sizeimage values > m2m driver should return in case when we try to run G_FMT on just > opened device. Is there any preferable values? v4l2-compliance tool > want it to be non zero, but at that moment we have no information > about output and capture, so i see few options here: > 1) ignore v4l-complince error in such case >>> v4l2-test-formats.cpp(417): !pix.width || !pix.height" > 2) set some default values e.g. 640x480 > Do you have a suggestion? \ Use 2: just set a default value. V4L2 should always be in a sane state, which in this case means that the driver should just pick some default initial format. Regards, Hans > > Thanks, Mikhail. > >> Hi Mikhail, >> >> I did a quick scan over the source code and I noticed a few things >> that aren't right. The easiest for you is probably to run the >> v4l2-compliance tool over your driver and it should tell you what >> needs to be fixed. The things I noticed are: querycap doesn't fill in >> bus_info (should be 'platform:<foo>') and device_caps, the vid_cap >> try_fmt fails on a wrong field setting, instead it should just set it. >> >> I also have some doubts about g_selection, but I need to look at that >> again when I have more time next week. It does look like it is not >> properly separating the capture and output streams. I would expect >> g_selection to return different things for capture and output. Note >> that v4l2-compliance doesn't yet check the selection API, so it won't >> help you there. >> >> Regards, >> >> Hans -- To unsubscribe from this list: send the line "unsubscribe devicetree" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
