| Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame^] | 1 | How the new TCP output machine [nyi] works. |
| 2 | |
| 3 | |
| 4 | Data is kept on a single queue. The skb->users flag tells us if the frame is |
| 5 | one that has been queued already. To add a frame we throw it on the end. Ack |
| 6 | walks down the list from the start. |
| 7 | |
| 8 | We keep a set of control flags |
| 9 | |
| 10 | |
| 11 | sk->tcp_pend_event |
| 12 | |
| 13 | TCP_PEND_ACK Ack needed |
| 14 | TCP_ACK_NOW Needed now |
| 15 | TCP_WINDOW Window update check |
| 16 | TCP_WINZERO Zero probing |
| 17 | |
| 18 | |
| 19 | sk->transmit_queue The transmission frame begin |
| 20 | sk->transmit_new First new frame pointer |
| 21 | sk->transmit_end Where to add frames |
| 22 | |
| 23 | sk->tcp_last_tx_ack Last ack seen |
| 24 | sk->tcp_dup_ack Dup ack count for fast retransmit |
| 25 | |
| 26 | |
| 27 | Frames are queued for output by tcp_write. We do our best to send the frames |
| 28 | off immediately if possible, but otherwise queue and compute the body |
| 29 | checksum in the copy. |
| 30 | |
| 31 | When a write is done we try to clear any pending events and piggy back them. |
| 32 | If the window is full we queue full sized frames. On the first timeout in |
| 33 | zero window we split this. |
| 34 | |
| 35 | On a timer we walk the retransmit list to send any retransmits, update the |
| 36 | backoff timers etc. A change of route table stamp causes a change of header |
| 37 | and recompute. We add any new tcp level headers and refinish the checksum |
| 38 | before sending. |
| 39 | |