Hi,
Further to my last post, we have resolved the problem with the driver ids. Quite simply the device nodes for the root file system are generated from a device_table file and thus were hardcoded to expected values which changed when we removed the H3A driver components. Once we matched these ids back up, the fix was complete.
Now we have another problem...
When we start the capture thread we are finding that quite often we dequeue buffers which have not been filled with image data. The capture thread will either start successfully or we will receive unfilled buffers immediately, this problem never happens partway through a capture.
Is there a known issue in the 2.6.32 kernel or v4l drivers which may cause problems with writing the image data to the buffers?
Regards,
John.