[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[ns] FullTcp/Sack bug?
In ns-2.1b7a/tcp-full.cc, it appears that:
void
SackFullTcpAgent::send_holes(int force, int maxburst)
{
int npack = 0;
int save_tseq = t_seqno_;
t_seqno_ = sack_nxt_;
if (sq_.empty()) {
// no holes, just return
return;
}
int nxtblk[2]; // left, right of 1 sack block
...
}
should actually be:
void
SackFullTcpAgent::send_holes(int force, int maxburst)
{
if (sq_.empty()) {
// no holes, just return
// ** don't screw up t_seqno_ before just returning **
return;
}
int npack = 0;
int save_tseq = t_seqno_;
t_seqno_ = sack_nxt_;
int nxtblk[2]; // left, right of 1 sack block
...
}
Is this correct? Also, could someone let me know the validation
status of FullTcp/Sack? Is this planned for the next release?
Regards,
Rajesh
P.S.:
Wow, since when did ns-users get so unfriendly? I am referring to
the annoying bounce+FAQ that I got when I posted this message.
========================================================================
Rajesh Krishnan
Network Scientist Email: [email protected]
Internetwork Research Department Phone: +1 617 873 4684
BBN Technologies, Verizon Fax: +1 617 873 6091
10 Moulton Street, Mailstop 6/5c GPS: N42�23.36' W71�08.86'
Cambridge, MA 02138, USA Web: http://www.ir.bbn.com
========================================================================