[Ns-developers] 64 bit port of ns

[Tom Henderson]

Mathieu Lacage wrote:
> On Tue, 2005-08-30 at 08:55 +0200, Mathieu Lacage wrote:
> 
> 
>>been regtested against the regression suite on x86. Furthermore, I have
>>made sure that the test suite runs on x86-64 even though it does
>>generate a few different results. I don't think I will really have time
> 
> 
> I was sort of annoyed by this so I started looking into it. The first
> problem is this one:
> 
> --- test-output-sctp/sctp-AMR-Exceeded  2005-08-30 08:53:24.000000000 +0200
> +++ test-output-sctp/sctp-AMR-Exceeded.test     2005-08-30 08:53:24.000000000 +0200
> @@ -145,29 +145,29 @@
>  d 6.17008 0 1 sctp 1500 -------D 0 0.0 1.0 1 13 62 0 12
>  d 10.17008 0 1 sctp 1500 -------D 0 0.0 1.0 1 13 63 0 12
>  d 18.17008 0 1 sctp 1500 -------D 0 0.0 1.0 1 13 64 0 12
> -+ 19.102368 1 0 sctp 56 -------H 0 1.0 0.0 1 -1 65 65535 65535
> -- 19.102368 1 0 sctp 56 -------H 0 1.0 0.0 1 -1 65 65535 65535
> -r 19.303264 1 0 sctp 56 -------H 0 1.0 0.0 1 -1 65 65535 65535
> -d 19.303264 0 1 sctp 56 -------B 0 0.0 1.0 1 -1 66 65535 65535
> ++ 19.102368 1 0 sctp 64 -------H 0 1.0 0.0 1 -1 65 65535 65535
> +- 19.102368 1 0 sctp 64 -------H 0 1.0 0.0 1 -1 65 65535 65535
> +r 19.303392 1 0 sctp 64 -------H 0 1.0 0.0 1 -1 65 65535 65535
> +d 19.303392 0 1 sctp 64 -------B 0 0.0 1.0 1 -1 66 65535 65535
>  d 34.17008 0 1 sctp 1500 -------D 0 0.0 1.0 1 13 67 0 12
> 
> On x86-64, one of the transmitted packets is 64 bytes big and it is 56
> bytes big on x86. The specific size field printed here is HDR_CMN ()-
> 
>>size () since this packet goes through trace.cc L325.
> 
> 

<snip>

This problem has just been checked in for SCTP without changing 
regression output.  Basically:
-  SendPacket((u_char *) &sHeartbeatChunk, sizeof(SctpHeartbeatChunk_S), 
spDest);
+   SendPacket((u_char *) &sHeartbeatChunk, SCTP_CHUNK_HEARTBEAT_LENGTH, 
spDest)

Are there other occurrences that broke regression output for 64-bit?

Tom