Hi Yu, The TCP behavior you see is determined solely by the Linux kernel. M5 is running the actual Linux kernel code. I can't say offhand why you're seeing the behavior you describe, but I'm 99% positive it's a Linux issue and not an M5 issue.
Steve Yu Zhang wrote: > Hi, > > I'm running some simulation on TCP/IP workload. And something unexpected > happened to me. As we know, TCP will play fast retransmit when the > sender receive 3 or more duplicate ACKs asap. However, on my > observation, sometimes, when the sender see the duplicate ACKs, it will > not trigger a fast retransmit, instead, a timeout is set up, which means > it will take much longer time to complete the retransmission. I'm quite > confused on this. I tracked the packets, and I'm sure that the duplicate > ACKs successfully arrived at the sender side. As much as I know, such > behavior of TCP is not reasonable. How is TCP implemented on M5? > > Thanks, > Yu > > > ------------------------------------------------------------------------ > > _______________________________________________ > m5-users mailing list > [email protected] > http://m5sim.org/cgi-bin/mailman/listinfo/m5-users _______________________________________________ m5-users mailing list [email protected] http://m5sim.org/cgi-bin/mailman/listinfo/m5-users
