Computer  Science 826
                         Quiz 4  -     Name ________________________

 1. Which of the following "offenses" is most likely to be encountered
    by a user of UDP

    a. Delivery of duplicate copies       b. Loss of a complete packet.
       of a single packet.

    c. Loss of part of a packet.          d. Delivery of a packet with
                                             several bits inverted.

 2. The maximum size of a UDP datagram that can be transmitted with
    successfully with sendto() is:

    a. is always 64K bytes.               b. Varies from system to system
                                             but is generally larger than
                                             8K bytes

    c. is always 65507 bytes.             d. Is 576 bytes.

 3. If a user sends a UDP packet which is dropped by an intermediate
    router because of congestion the user is:

    a. informed by getting a               b. informed by getting an
       negative return code from              ICMP error message on
       the sendto() function                  his next recvfrom() on
                                              the sending socket
    c. not informed that the packet
       was lost.

 4. The ARP pathology example in the book shows that a UDP datagram
    transmission from a host on an ethernet will fail:

   a. Any time the packet to be sent   b. Whenever the next hop hardware
      is larger than Ethernets MTU        address is not in the ARP cache.

   c. Either a. or b. is true.         d. both a. and b. are true.

 5. Assuming that DF is not set. Fragmentation works the following
    way in IPV4:

   a. A packet must be broken down      b. A packet will be fragmented
      into MTU size frags by the source    by the router owning the link
      host and will be reassembled at      whose MTU is too small and
      the destination.                     will be reassembled by the
                                           next router that can handle the
                                           full packet.

   c. A packet will be fragmented       d. A packet must be broken down
      by the router owning the link        into MTU size units by the
      whose MTU is too small but not       originating host but will
      reassembled until it reaches the     be reassembled as soon as it
      destination.                         passes the bottleneck router.

 6. The "Path MTU Discovery" mechanism is designed to permit source hosts
    to avoid IP fragmentation altogether.  Describe how it is supposed to
    work (We discussed an example in the text in which it did NOT work as
    advertised.. but what I want here is how it is SUPPOSED to work.)







 7. Suppose a path contains a SLIP link with a very small MTU. Use of
    Path MTU discovery is MORE advantageous if the SLIP link is

    a. near the source                   b. near the destination

    c. makes no difference

 8. When a single fragment of a fragmented packet is discarded in transit

    a. IP rules require that all         b. A special ICMP packet bearing
       other fragments be discarded         the ID of the discarded packet
       at the same time                     is sent to the destination to
                                            inform it.

    c. The destination will buffer       d. The destination will buffer the
       the remaining fragments until        remaining fragments until they
       the discarded fragment is            time out and then discard them
       retransmitted.

 9. The UDP checksum covers:

    a. The UDP data only                 b. The UDP header and UDP data

    c. The IP and UDP header and         d. An IP pseudo-header, the UDP
       UDP data                             header and the UDP data

10. The struct sockaddr_in contains 3 fields that <<must>> be filled
    in prior to its use in the "sendto()" system call. What are they:

    a -

    b -

    c -

11. If I am writing a UDP application and I wish it to receive data
    ONLY from host jmw2-lane I should:

    a. bind the socket to a              b. connect the socket to a
       sockaddr_in structure                sockaddr_in structure
       specfying the IP address of          specfying the IP address of
       jmw2-lane                            jmw2-lane

    c. I must do both a. and b.

12. Suppose I am writing a UDP server to run on host jmw.
    Explain the implications of binding the socket to a
    struct sockaddr_in specifying the IP address returned
    by gethostbyname("jmw") versus using an IP address of 0.