ch8 mget example not unicast #7

Activity

1. timClicks commented on Jul 12, 2020

   

   timClicks

   on Jul 12, 2020

   Contributor

   More actions

   I'll need to spend some time addressing these issues. This may involve re-working the examples somewhat.
2. jiayihu commented on Aug 12, 2020

   

   jiayihu

   on Aug 12, 2020

   Last edited by jiayihu

   More actions

   The least significant bit of the first octate should be zero to 0, not 1, for be considered as unicast. I managed to do so in this way:

   octets[0] |= 0b_0000_0010; // Set bit for local address
   octets[0] &= 0b_1111_1110; // Clear bit for unicast
   

   Unfortunately, running the example still results in error:

   Poll error: Unrecognized
   Connecting
   Poll error: Unrecognized
   

   The error is from these lines:

   match iface.poll(&mut sockets, timestamp) {
               Ok(_) => {}
               Err(e) => {
                   eprintln!("Poll error: {:?}", e);
               }
   }

   The error seems to be caused by unknown packets in the socket, but I have no idea what is the cause since I don't know much how to debug it. I tried running the code with cargo run -- http://www.rustinaction.com/ tap-rust.

   In case, full source code of my attempt is available here: https://github.com/jiayihu/lab/blob/master/rust/mget-tcp/src/http.rs but it should be almost identical to the book

   👍React with 👍2Reacted by Tim McNamara and Alexander
3. mgiles commented on Nov 14, 2020

   

   mgiles

   on Nov 14, 2020

   More actions

   I don't think those errors from poll are a problem. From the poll documentation:

   This method will routinely return errors in response to normal network activity as well as certain boundary conditions such as buffer exhaustion. These errors are provided as an aid for troubleshooting, and are meant to be logged and ignored.

   As a special case, Err(Error::Unrecognized) is returned in response to packets containing any unsupported protocol, option, or form, which is a very common occurrence and on a production system it should not even be logged.
4. jiayihu commented on Nov 14, 2020

   

   jiayihu

   on Nov 14, 2020

   More actions

   @mgiles You're right, my mistake
5. ewenmcneill

   added a commit that references this issue on Jan 10, 2023

   ch8/ch8-mac: Clear Unicast/Multicast bit

   ...

   5a9a8ab