Implementation Guide

You are provided with a fair amount of scaffolding code, some of which you will need to use or modify to implement your router. This page provides an overview of the files that you will need to modify or be familiar with. Most of the documentation on individual functions can be found in the header files themselves.

chirouter.h

This header file defines the main data structures of the router. All these data structures are accessible through the router context struct (chirouter_ctx_t) that is passed to most functions. Our code takes care of creating and populating this struct with all the information relevant to the router:

  • An array of chirouter_interface_t structs, each representing an Ethernet interface in the router.

  • An array of chirouter_rtable_entry_t structs, each representing an entry in the routing table. Each entry in the routing table, in turn, contains the destination network (specified by an IPv4 address and a subnet mask), the gateway for that entry, and the Ethernet interface for that entry (a pointer to a chirouter_interface_t struct).

    Take into account that, in this project, you can ignore the gateway. i.e., you can assume that there is a single router in the network, and that its interfaces are directly connected to their destination networks.

  • An ARP cache, represented as an array of chirouter_arpcache_entry_t structs.

  • A list of pending ARP requests, representing ARP requests that have been sent but for which the router has not received a reply yet. This is explained in more detail in Assignment: Implementing an IP router.

  • A mutex that must be locked any time the ARP cache or the list of pending ARP requests is accessed.

This header file also defines an ethernet_frame_t struct representing an inbound Ethernet frame. This struct contains not just the frame itself, but also a pointer to the chirouter_interface_t struct corresponding to the Ethernet interface on which the frame arrived.

You should not modify this header file in any way.

router.c

Most of your work will take place in this file. In particular, you must implement the chirouter_process_ethernet_frame function. This function will get called every time an Ethernet frame is received by the router. The router uses a single thread to process inbound Ethernet frames, so you can assume that chirouter_process_ethernet_frame is always called sequentially (there will never be concurrent calls to this function).

Your implementation of the chirouter_process_ethernet_frame function must process the frame meeting the requirements described in Assignment: Implementing an IP router. You are allowed, and encouraged, to use helper functions to implement chirouter_process_ethernet_frame. For example, it would make sense to write separate functions to handle ARP messages, ICMP messages directed to the router, and IP datagrams.

arp.c / arp.h

Part of your work will take place in this file. In particular, you must implement the chirouter_arp_process_pending_req function. Besides the router’s main thread (which is in charge of calling chirouter_process_ethernet_frame when an Ethernet frame arrives), the router has an additional thread, the ARP thread, that runs function chirouter_arp_process. This thread will wake up every second to purge stale entries in the ARP cache (entries that are more than 15 seconds old) and to traverse the list of pending ARP requests.

For each pending request in the list, it will call chirouter_arp_process_pending_req, which must either re-send the pending ARP request or cancel the request and send ICMP Host Unreachable messages in reply to all the withheld frames (this is described in more detail in Assignment: Implementing an IP router.).

Because the main thread and the ARP thread may both need to access the ARP cache or the list of pending ARP requests at the same time, you must always lock the lock_arp mutex (in the router context) before accessing either the ARP cache or the list of pending ARP requests (even if you are just reading them).

You must not modify any code in this file other than chirouter_arp_process_pending_req. However, this file does provide several functions to access and/or manipulate the ARP cache and list of pending ARP requests, which you can use in your implementation. Take into account that these functions assume that the lock_arp mutex has already been locked before the functions are called.

The chirouter_pox_send_frame function

When an Ethernet frame arrives and is processed in chirouter_process_ethernet_frame you will, in many cases, have to send an Ethernet frame through one of the router’s interfaces. This is done using the chirouter_pox_send_frame function, defined in the pox.h header file.

ethernet.h, arp.h, icmp.h, and ipv4.h

These header files (located in the src/router/protocols directory) provide structs that allow easy access to the values contained in Ethernet, ARP, ICMP, and IPv4 headers.

Given an inbound Ethernet frame (an ethernet_frame_t struct), you can access the different headers like this:

/* Accessing the Ethernet header */
 ethhdr_t* hdr = (ethhdr_t*) frame->raw;

/* Accessing the IP header */
iphdr_t* ip_hdr = (iphdr_t*) (frame->raw + sizeof(ethhdr_t));

/* Accessing an ARP message */
arp_packet_t* arp = (arp_packet_t*) (frame->raw + sizeof(ethhdr_t));

/* Accessing an ICMP message */
icmp_packet_t* icmp = (icmp_packet_t*) (frame->raw + sizeof(ethhdr_t) + sizeof(iphdr_t));

utils.c / utils.h

This module provides two useful functions: one to compute an IP or ICMP checksum, and one to compare MAC addresses. If you need to add functions in your implementation that need to be shared by router.c and arp.c, you should add them to this module.

The logging functions

chirouter prints out detailed information to standard output using a series of logging functions declared in src/router/log.h. We encourage you to use these logging functions instead of using printf directly. More specifically, you should use the printf-style chilog() function to print messages:

chilog(DEBUG, "Received Ethernet frame with unsupported Ethertype: %i)", ntohs(hdr->type));

And the chilog_ethernet(), chilog_arp(), chilog_ip(), and chilog_icmp() functions to dump the contents of an Ethernet header, ARP message, IP header, or ICMP message. For example:

int reply_len = sizeof(ethhdr_t) + sizeof(iphdr_t) + ICMP_HDR_SIZE + payload_len;
uint8_t reply[reply_len];
memset(reply, 0, reply_len);

ethhdr_t* reply_ether_hdr = (ethhdr_t*) reply;
iphdr_t* reply_ip_hdr = (iphdr_t*) (reply + sizeof(ethhdr_t));
icmp_packet_t* reply_icmp = (icmp_packet_t*) (reply + sizeof(ethhdr_t) + sizeof(iphdr_t));

/* Set values in all the headers */

chilog(DEBUG, "Sending ICMP packet");
chilog_ip(DEBUG, reply_ip_hdr, LOG_OUTBOUND);
chilog_icmp(DEBUG, reply_icmp, LOG_OUTBOUND);

The last parameter of these functions can be LOG_INBOUND or LOG_OUTBOUND to designate a message that is being received or sent, respectively (this affects the formatting of the message in the log). LOG_NO_DIRECTION can also be used to indicate that the message is neither inbound nor outbound.

In all the functions, the first parameter is used to specify the log level:

  • CRITICAL: Used for critical errors for which the only solution is to exit the program.
  • ERROR: Used for non-critical errors, which may allow the program to continue running, but a specific part of it to fail (e.g., an individual socket).
  • WARNING: Used to indicate unexpected situation which, while not technically an error, could cause one.
  • INFO: Used to print general information about the state of the program.
  • DEBUG: Used to print detailed information about the state of the program.
  • TRACE: Used to print low-level information, such as function entry/exit points, dumps of entire data structures, etc.

The level of logging is controlled by the -v argument when running chirouter:

  • No -v argument: Print only CRITICAL and ERROR messages.
  • -v: Also print WARNING and INFO messages.
  • -vv: Also print DEBUG messages.
  • -vvv: Also print TRACE messages.

We recommend running at the -vv level, which will print all the inbound Ethernet frames. The -vvv contains much lower-level information that the instructors may need to debug a specific issue, but which is typically not relevant in most situations when implementing chirouter.