Hi Alan, all,
I added an extra check to the videodev.c's video_ioctl, to check vfl is
valid before dereferencing. (See attached patch).
The sequence to trigger the oops appears to be:
1. Open philips usb webcam device (any usb video device with ioctls ?).
2. Wait for (fairly rare) USB EMI issue to be flagged by hub.c.
3. usb.c appears to force a disconnect immediately after #2.
4. pwc module warns about a disconnect while open.
5. Next call to video_ioctl with handle from #1 causes oops.
I added a check to video_icotl to ensure video_device[] isn't NULL
before trying to call the ioctl method, and it returns EINVAL in that
case.
I'm not very familiar with the locking of usb/usb-uhci in this EMI case,
but both host drivers oops in the same way. It would seem the open file
handle needs to be disassociated too somehow - or at least 'defered'
until the video camera is re-enabled after the EMI event.
Best regards,
Mark
PS. I've recently changed the physical layout of the machine, and it
seems the new cabling layout causes these occasional EMI issues.
-- Mark Cooke <mpc@star.sr.bham.ac.uk>
--- linux-2.4.21/drivers/media/video/videodev.c.orig 2003-06-13 15:51:34.000000000 +0100 +++ linux-2.4.21/drivers/media/video/videodev.c 2003-07-10 23:00:49.000000000 +0100 @@ -209,7 +209,12 @@ unsigned int cmd, unsigned long arg) { struct video_device *vfl = video_devdata(file); - int err=vfl->ioctl(vfl, cmd, (void *)arg); + int err; + + if (!vfl) + return -EINVAL; + + err = vfl->ioctl(vfl, cmd, (void *)arg); if(err!=-ENOIOCTLCMD) return err;
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Tue Jul 15 2003 - 22:00:44 EST