Web lists-archives.com

Re: [Spca50x-devs] questions about the internal of the driver

Le dimanche 3 février 2008 11:59, Jean-Francois Moine a écrit :
Jean François,
> When I wrote some code for the v2l version 2, I found the grabbing
> frame state is not clear. I think that an other state should be
> required meaning the frame is treated by the application.
> Then, I was wondering why is there a tasklet do to the conversion.
> In other drivers such as ov51x-jpeg, 
Don't really know ov51x-jpeg sorry :(
> the conversion is done by the 
> application. So, the code may be simpler:
The gspcav1 is v4l1, v4l1 know nothing about mjpeg that is why a tasklet is 
used here.
> - receiving iso packets, move the data to the frame buffer, state
> - when end of frame (start of new frame) is detected, change the
> 	state to READY (and that's all).
> - when the application asks for a new frame (read() or iotcl()
> 	VIDIOCMCAPTURE or VIDIOC_DQBUF), convert the frame to the
> 	desired palette and change the state to DONE.
> - when the application releases the frame (end of read() or ioctl()
> 	VIDIOCSYNC or VIDIOC_QBUF, free the buffer setting its state
> 	to UNUSED.
> Doing so should avoid the problem occuring with 2 frame buffers:
> - one buffer being grabbed,
> - this other one being converted and
> - which one being processed by the application...
My firs idea was to stop developping gspcav1 and start a new branch for v4l2 
 The problem is no one seem interresting to develop this branch, of course it 
is difficult as we need to be sure the > 270 webcam are ok after each small 
change ;-)
I see you change the core driver to add your chipset.I have to double check 
all webcams are working before apply, sorry for the delay
bests regards
Michel Xhaard

This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
Spca50x-devs mailing list