Non blocking buffer transaction APIs
The low level drivers are designed to have a minimal footprint and minimal latency（延迟）. The low level drivers do not maintain read or write buffer queues（队列）. They support only a ping pong buffer scheme that allows the application to have only one or two outstanding（未解决的） transactions at a time.
To use the low level drivers in non-blocking mode and to stream data to or from the driver, an application must allocate two buffers and then proceed to use them in the following ping pong manner.
application pseudo code
adi_xxx_SubmitBuffer // transfer buffer to driver
adi_xxx_SubbmitBuffer // transfer buffer to driver
adi_xxx_GetBuffer // will block until transaction complete
// buffer is owned by application again
// buffer processing
while （） is a mistake ？