Specifications

Transmit

pointer

Receive

pointer

End Address

of the buffer

Starting Address

of the buffer

Memory

addresses

Figure 13 The ring buffer

The buffer works by using two pointers a transmit pointer and a receive pointer. The

receive pointer is passed to the function that reads the TCP data and stores the data from

this pointer forward. When the read is finished, it is determined how many bytes were

put into the buffer and the pointer is moved to this next position, ready to receive more

data. If the pointer goes past the end of the memory space allocated it is reset back to

the start of the memory space, hence the name ring buffer. The transmit pointer is used

by the MP3 function, which was yet to be implemented, to determine from where in the

buffer it was to read data. This pointer wraps back to the beginning of the buffer the

same as the receive pointer.

The buffer management algorithm starts by calculating the amount of data that is in the

buffer by counting how far the receive pointer is in front of the transmit pointer. If the

buffer becomes full, meaning the receive buffer catches up to the transmit buffer, the

program waits until the buffer starts to become empty before reading more TCP data.

This situation is ideal; it indicates when the data is coming in faster than the data is

going out. At this point a test of how the Nut OS regulated the TCP stream from the

server when the buffer became full was done. The hope was that it would simply stop

the stream when its own buffer became full and wait for data to be taken out. It was

tested by using the same TCP file transfer model as above, but this time the Ethernut

took all the data from the server and put it into the buffer as shown in figure 14.