Go to most recent revision | Blame | Compare with Previous | Last modification | View Log | RSS feed
The Tftp protocol implementation under Shark
Initialization:
1. call the tftp_init() routine;
2. call the tftp_net_start(...) to initialize the network card; the third parameter specify if you wantto initialize
the IP address of the card (1) or if you simply pass an already initialized one (0).
3. call tftp_open(...) to get a free socket;
4. call tftp_upload(...) or tftp_download(...) (the last one is not yet implemented) to start the tftp deamons.
The tftp library allocate a new socket for each upload/download connection.
The base port is set to 2000 and the i-th connection uses the [base]+i port address.
To stop uploading use tftp_close(connection, hardness)
hardness could be TFTP_STOP_NOW or TFTP_FLUSH_BUFFER
with TFTP_STOP_NOW you:
- sends immediately a packet with 0 bytes of data (means stop transfer!);
- all data in the buffer are lost;
- deamon stops immediately.
with TFTP_FLUSH_BUFFER you:
- couldn't write no more data in the buffer;
- deamons stay alive to send all data in the buffer to the server;
- then they shut down the service.
To Do:
- Implement the download routines
- Improve the error checking (at now the ACK deamon don't check for the block number)
- Use the [Big-Little]-endian routines provided with Shark instead of the ad-hoc ones.
- Improving the QUEUE library (I'm developing it) and use it in the TFTP library (and provide it to all Shark users).