On Wed, 2010-06-30 at 23:11 +0200, Johannes Berg wrote: > I'm pretty sure this was a regression in .34, but haven't checked right > now, can bisect when I find time but wanted to inquire first if somebody > had ideas. All I get is: > > [57372.078968] uvcvideo: Failed to query (130) UVC control 5 (unit 3) : > -32 (exp. 1).
Didn't have to bisect it. The problem is caused by commit 59529081e092506edb81a42d914e2d0522f65ca7 Author: Laurent Pinchart <laurent.pinch...@ideasonboard.com> Date: Sat Jan 23 06:30:20 2010 -0300 V4L/DVB: uvcvideo: Cache control min, max, res and def query results which reverts cleanly, but then the driver doesn't compile and one also needs to revert commit cf7a50eeb6f462a0b7d1619fcb27a727a2981769 Author: Laurent Pinchart <laurent.pinch...@ideasonboard.com> Date: Sun Apr 25 16:27:14 2010 -0300 V4L/DVB: uvcvideo: Prevent division by 0 when control step value is 0 and commit e54532e591cd5b9ce77dbc8d9786ae9f600f101a Author: Laurent Pinchart <laurent.pinch...@ideasonboard.com> Date: Sat Jan 23 07:07:53 2010 -0300 V4L/DVB: uvcvideo: Clamp control values to the minimum and maximum values in that order. These are commits introduced in the .34/.35 cycles. Please fix. Can we revert things that are regressions pre-dating the current kernel version? johannes -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html