Regarding Buff-Streaming VS read()/write()... I have a philosophical question regarding the use of the v4l2 buffer streaming interface. In particular I am struggling to understand why you would use it. I my particular case, I'm pondering a codec v4l2 driver that might accept data, convert it, and then produce output. In this scenario, an application could push data into the driver at whatever rate the driver will accept it as controlled by the write() (e.g blocking or select() use). And likewise a application could be simultaneously retrieve available data via the read() interface at a rate controlled by the driver. So, with this said, why would anyone want to implement this example using the buffer streaming interface with the extra overhead/complication of setting up the input and output buffers, mmap, etc. I think I must be missing something. Can someone please clarify? Thanks in advance, Steve Kranz skranz@xxxxxxxxxxxx