RspConnection. The constructor, provided in two flavors. The first takes a single integer argument, the port number to listen on. The second takes an optional string argument, the name of a TCP/IP service to listen on (which defaults to "or1ksim-rsp").

Both use the private function, rspInit to save the port number (0 if the service name is to be used) and service name and set the file descriptor for the connection (private variable clientFd) to -1, indicating no connection is present.

~RspConnection. The destructor, which closes any connection which has been established.

rspConnect (). This function listens for client GDB connections over TCP/IP using the port or service specified in the constructor. The connection is established with a keep-alive flag (for long debug sessions) and with Nagel's algorithm disabled. This ensures that the characters of a packet are transmitted as a soon as they are available, rather than being grouped for more efficient transmission. This is appropriate for an interactive interface such as a debugger.

Once a connection is established, its file descriptor (which cannot be negative) is stored in the private variable, clientFd.

The function returns false if a catastrophic failure occurs in setting up the connection. This will cause the GDB server to terminate. A result of true indicates either a successful connection, or that the connection attempt may be retried. The two can be distinguished by use of the isConnected function (see below).

rspClose. This closes the connection if it is open and sets the file descriptor (clientFd) to -1. It may be safely called without checking if the connection is still open, since it will check the value of clientFd before closing it.

isConnected. This returns true if a connection is established (i.e. clientFd is not negative) and false otherwise.

getPkt. This takes a pointer to a packet and populates it with the data from the next packet read from the TCP/IP stream. It will continue to retry until a packet is successfully read (i.e. is complete, fits into the packet data structure and has a correct checksum) or the connection fails. The packet is acknowledged back to the client.

The RSP protocol requires certain characters ('$', '#' and '}') in the data field of incoming packets to be escaped. However rather than inspecting every packet for escaped characters, it is the raw data which is stored. This is because there is only one packet type (X) which can have these characters, and it is more efficient to unescape the characters only in the handler for that packet.

The private function getRspChar is used to read individual characters from the socket. It will automatically retry in the event of an interrupt.

The function returns true if a packet is successfully read and false if the connection fails.

putPkt. This is the complement to getPkt, sending its argument packet back to the client. It will continue to retry until receipt is successfully acknowledged or the connection fails.

Unlike getPkt, putPkt escapes its data field. Characters '$', '#', '*' and '}' are escaped by preceding them with '}' and then the original character XORed with 0x20. It makes sense to centralize escaping here, since several commands may generate results using these characters.

The private function putRspChar is used to write individual characters to the socket. It will automatically retry in the event of an interrupt.

The function returns true if a packet is successfully sent and false if the connection fails.