On Sun, 22 Feb 2009, Hans Verkuil wrote:
> Also the v4l1 ioctls have been removed and instead zoran relies on the v4l1
> compat layer.
I tried testing v4l1 with mplayer and it doesn't seem to work correctly.
[pid 29030] ioctl(3, VIDIOCSYNC, 0x884d790) = -1 EBUSY (Device or resource busy)
[pid 29030] gettimeofday({1235920259, 869629}, NULL) = 0
[pid 29030] ioctl(3, VIDIOCMCAPTURE, 0x884d790) = -1 EBUSY (Device or resource busy)
[pid 29030] write(2, "\nioctl mcapture failed: Device o"..., 48
VIDIOCSYNC and VIDIOCMCAPTURE always return EBUSY. Though, it does appear
that mplayer and the driver actually do work and capture video.
It looks like this is because v4l1-compat calls VIDIOC_STREAMON for each
call to VIDIOCMCAPTURE or VIDIOCSYNC. The zoran driver returns EBUSY if
you try to do that while streaming is already on.
Maybe the zoran driver should detect stream on from a file handle that
already has streaming on as a no-op? Or maybe v4l1-compat should ignore
EBUSY errors from VIDIOC_STREAMON in those compat functions?
Fixing this in correctly in v4l1-compat is hard because that module doesn't
have any device state so it can't keep track if it needs to call
VIDIOC_STREAMON or not. Lots of v4l1 apps expect that the driver will
start streaming automatically when they call VIDIOCMCAPTURE.
--
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
