By Date: <-- -->
By Thread: <-- -->

V4L2 *STD ioctls with webcam drivers

Hi Eduardo,

> I would like to know what the correct implementation of the *STD ioctls
> in V4L2 webcam drivers is.
>
> Either they return -EINVAL or they get implemented like the attached
> patch in the http://linux-uvc.berlios.de/ USB driver or like the
> http://www.saillard.org/linux/pwc/ driver.
>
> I'm asking because different USB webcam drivers implement different code
> and I think a consensus should be reached if we want all the
> applications working properly with different drivers.

Thanks for raising the issue here, I was planning to send a mail myself but 
I'm unfortunately a bit busy at the moment.

As I already stated, my understanding is that the V4L2 spec doesn't make those 
ioctls mandatory. Returning -EINVAL is the proper way to deal with the *STD 
ioctls when they are not supported by the driver.

Applications which handle -EINVAL as a critical failure should be fixed. Some 
people adviced me to add a dummy implementation of those ioctls to the 
linux-uvc driver to ensure compatibility with broken user-space applications. 
I'd rather not go that way, as adding workarounds for user-space bugs in the 
kernel will not help to fix the bugs. Application developpers testing against 
workaround-enabled drivers will not catch bugs in their software. It would be 
much more usefull to fix the broken applications.

I can of course be wrong, and I'm waiting for comments from all V4L2 
developers.

Laurent Pinchart

--
video4linux-list mailing list
Unsubscribe mailto:video4linux-list-request (at) redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/video4linux-list