kfifo | 内核的kfifo

The Wikipedia page you mentioned states that a circular buffer behaves as if the buffer was connected end to end. In practice, circular buffer is simply an array of a certain fixed length, with two index pointers (usually called head and tail, or in and out) representing the "boundaries" of the written data. To avoid writing outside the buffer bounds, all arithmetic operations on these indices are done modulo length of the buffer.

Usually, the meaning of the pointers is:

• head or in index, indicates the next available slot for writing, and
• tail or out index, indicates the last read ("removed") slot.

There are also two boundary states:

• if tail is equal to head, then the buffer is empty.
• if incrementing the tail modulo buffer length would make tail and head equal, then the buffer is full.

Most implementations will keep the indices within the buffer bounds using one of these approaches: