Re: [Freeipmi-users] rmcpping and simulated pong replies

[Al Chu]

Hey Thomas,

> In the file ping-tool-common.c at line 522 (freeipmi v0.8.6), only 
> the message sequence number equality is tested.
> 
> My question is rather simple : what is exactly tested by rmcpping to 
> identify a correct pong reply ?

In rmcpping.c it checks that the packet is of the right message type (a
pong) and the message sequence number is valid too.

If I had to bet, you may be failing in this part:

  if ((ret = unassemble_rmcp_pkt (buf,
                                  buflen,
                                  obj_rmcp_hdr,
                                  obj_rmcp_cmd)) < 0)
    ipmi_ping_err_exit ("unassemble_rmcp_pkt: %s", strerror (errno));
    
  if (!ret)
    {
      rv = 0;
      goto cleanup;
    }

If the packet is ill-formed, unassemble_rmcp_pkt will return 0.
Ill-formed usually means a packet is not a correct length.

Are you adding the 6 bytes of "reserved" data at the end of the packet?
I could see how one might think it's optional, but looking at the spec,
it appears to be required.

Al

On Thu, 2010-06-03 at 06:59 -0700, Thomas Favre-Bulle wrote:
> Hi everyone,
> 
> I'm trying to simulate a BMC and I'm currently working on reproducing 
> its networking behaviour (i.e. reply to RMCP-ASF ping, IPMI auth 
> challenge, etc.).
> According to ASF v2.0 specs (DSP0136), a pong reply should have the same 
> message tag as the ping message, a sequence number of 0xff and message 
> type of 0x40.  It also carries a data payload (which is pretty simple 
> for a standard message and should be equal to this -> 
> 00:00:11:be:00:00:00:00:81:00:00:00:00:00:00:00h).
> I'm using rmcpping to run tests, but so far I'm unable to get a working 
> pong reply.
> Analyzing packets with wireshark as shown no incorrect data (both in 
> ping and pong messages).
> In the file ping-tool-common.c at line 522 (freeipmi v0.8.6), only the 
> message sequence number equality is tested.
> 
> My question is rather simple : what is exactly tested by rmcpping to 
> identify a correct pong reply ?
> 
> I've been searching through the mailing list's archives but I couldn't 
> find a similar problem (I've done some googling as well).
> 
> Thanks,
> 
-- 
Albert Chu
address@hidden
Computer Scientist
High Performance Systems Division
Lawrence Livermore National Laboratory