Hi Dan, On Sunday 11 May 2008, Dan McGee wrote: > On Sun, May 11, 2008 at 3:16 AM, Laurent Pinchart > > <[EMAIL PROTECTED]> wrote: > > Hi Dan, > > > > On Friday 09 May 2008, Dan McGee wrote: > > > I'm seeing this same problem on the Eee PC with the integrated webcam. > > > Running an strace on the luvcview process shows it coming to a halt on > > > the following: > > > ioctl(5, VIDIOC_STREAMON, 0x77eede58) = 0 > > > ioctl(5, VIDIOC_DQBUF > > > > > > and stays there. > > > > > > I attempted to track down when this problem started occurring and why, > > > but I couldn't figure out the why. I know I had a setup where > > > everything was working with a near-vanilla 2.6.25 kernel and > > > linux-uvc-svn version 205. The problem started when I compiled > > > 2.6.25.1 and bumped to SVN revision 207. I attempted to drop back down > > > to 205 and the problem was still occurring. > > Doh! I thought I had been dropping down to 205, futher inspection > tells me my build script was always pulling the head revision. Answers > are below. > > > > Looking through the kernel > > > > > > changelog from .25 to .25.1 didn't turn up anything obvious as to what > > > changed causing it to fail. > > > > I'm pretty surprised that the upgrade from .25 to .25.1 alone caused the > > problem. > > > > Could you please retry with r206 and r207 and compare the results ? r207 > > introduces a patch to drop incomplete frames. Check the kernel log to > > make sure the right driver gets loaded (it should print the SVN revision > > number when loaded). > > > > If reverting to r206 helps, I'll ask you to try a patch to get more > > information about the issue. > > The 206 -> 207 transition is in fact what broke it. I'd be happy to > test out any patches you want me to try to help debug this issue. > Commenting out the changes from that revision in > linux-uvc/trunk/uvc_v4l2.c and then compiling the current revision > (209) also solves the problem, as drop_incomplete will never get set. > > For this camera, the only output format supported is yuv if that helps > at all, or is a possible configuration that did not get tested.
Coud you please test the patch attached to https://lists.berlios.de/pipermail/linux-uvc-devel/2008-May/003500.html ? Instructions are available in the mail. Best regards, Laurent Pinchart _______________________________________________ Linux-uvc-devel mailing list [email protected] https://lists.berlios.de/mailman/listinfo/linux-uvc-devel
