First, a bit of background: A bug in the em28xx implementation of s_ctrl() was present where it would always return 1, even in success cases, regardless of what the subdev servicing the request said (in this case the video decoder). It was using v4l2_device_call_all(), and disregarding the return value from any of the subdevs. This prompted me to change the code so that it started using v4l2_device_call_until_err(), figuring that subdevs that did not support it would simply return -ENOIOCTLCMD. However, as Mauro correctly pointed out, subdevs that do implement s_ctrl, but not the desired control will return -EINVAL, which would cause the bridge to stop sending the command to other subdevs and return an error. I looked at various other bridges, and don't see any consistent approach for this case. Some of the bridges always return zero (regardless of what happened during the call). Some of them look at the content of the resulting struct for some value that suggests it was changed. Others feed the call to different classes of subdevice depending on what the actual control being set was. So what's the "right" approach? I'm willing to conform to whatever the recommendation is here, since it will obviously be an improvement over always returning 1 (even always returning zero would be better since at least applications wouldn't treat it as a failure). Hans? Devin -- Devin J. Heitmueller - Kernel Labs http://www.kernellabs.com -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
