On Wednesday 06 February 2008, Jean-Christophe Jice Cardot wrote:
> Le dimanche 3 février 2008, vous avez écrit :
> > Hi,
> >
> > On Saturday 26 January 2008, Jice wrote:
> > > Hi Laurent
> > >
> > > I'm using version 173 or the driver now and it works perfectly with
> > > luvcview, and I just managed to have ekiga display the webcam also 5
> > > minutes ago.
> > >
> > > Unfortunately I'm trying to use the webcam to chat with people using
> > > skype (forget it) or msn. Nor Amsn or Kopete are willing to display
> > > anything. Amsn crashes, and Kopete only display garbage :-(
> > > I tried xawtv (which always worked with my old cam and my old tv card)
> > > but it also segfaults (I'm using the 3.95-7mdv version from Mandriva).
> > > Did you hear something about xawtv crashing with uvcvideo? And the 2
> > > above?
> > >
> > > Sorry to bother you but I'm searching for a few weeks and cannot find
> > > anything.
> >
> > Check the kernel log (dmesg) after aMsn/Kopete/xaxtv failures to see if
> > the driver prints any error/warning message. If so, please report them on
> > the linux-uvc-devel mailing list.
> >
> > Best regards,
> >
> > Laurent Pinchart
>
> Hi Laurent
>
> For kopete I already reported the problem on the ml, for xawtv I've
> isolated the crash (thanks gdb ;) and will post another email.
>
> For AMSN crash, nothing in the kernel log, just the following output:
>
> ioctl: VIDIOC_QUERYMENU(id=134217738;index=0;name="60
> Hz";reserved=3081900320): Invalid argument
> ioctl: VIDIOC_QUERYMENU(id=134217738;index=0;name="60
> Hz";reserved=3081900320): Invalid argument
> Corrupt JPEG data: premature end of data segment
> Invalid JPEG file structure: two SOI markers
> Oops: malloc_video_bufs is 1 (expected 0)
> [utils/linux/capture/libng/grab-ng.c:malloc_bufs_check:235]
> Oops: processes is 1 (expected 0)
> [utils/linux/capture/libng/convert.c:process_check:179]
>
> Must be a bug in Amsn (?), or could it be a problem with the webcam (as it
> says that the jpeg stream is corrupted) ?

Data corruption can occur, as isochronous USB transfers offer no guarantee. 
The application shouldn't crash, it's definitely a bug. Unfortunately there 
is no way for the driver to detect data corruption when transfering 
compressed (MJPEG) streams, so I can't drop corrupted compressed frames.

Best regards,

Laurent Pinchart
_______________________________________________
Linux-uvc-devel mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/linux-uvc-devel

Reply via email to