Hi, I opened a tcp connection and after a few bytes were sent I got the tcp_info structure using the getsockopt call. The contents of the structure returned were: 1, 0, 0, 0, 0, 7, 119, 201000, 40000, 1448, 536, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1500, 5792, 1000, 750, 2147483647, 4, 1448, 3, 0, 5792, 0 The interpretation of some of the values according to the linux/tcp.h file is tcpi_rto; 21000 tcpi_ato; 40000 tcpi_snd_mss; 1448 tcpi_rcv_mss; 536 tcpi_pmtu; 1500 tcpi_rcv_ssthresh; 5792 tcpi_rtt; 1000 tcpi_rttvar; 750 tcpi_snd_ssthresh; 2147483647 tcpi_snd_cwnd; 4 tcpi_advmss; 1448 119 is 0111 0111 in binary, hence tcpi_snd_wscale : 4, 0111 => 7 tcpi_rcv_wscale : 4; 0111 => 7 I have the following doubts. 1. Why is the tcpi_snd_ssthresh so large? I noticed in function tcp_v4_init_sock [http://lxr.linux.no/linux+v2.6.32/net/ipv4/tcp_ipv4.c#L1789], the snd_ssthresh is set to a value TCP_INFINITE_SSTHRESH which is according to draft-stevens-tcpca-spec-01 however, RFC 2001 (page 2) [http://tools.ietf.org/html/rfc2001] that the snd_ssthresh be set to 65535. 3. What does the tcpi_rcv_ssthresh stand for? 4. What is the unit for rto? Is it in seconds/milliseconds/microseconds? Further I did a getsockopt before a send of 1380 bytes and just after a send. The tcp_info before the send was: 1, 0, 0, 0, 0, 7, 119, 203000, 40000, 1448, 536, 0, 0, 0, 0, 0, 248, 0, 244, 244, 1500, 5792, 3250, 3000, 2147483647, 4, 1448, 3, 0, 5792, 0 and after the send was 1, 0, 0, 0, 0, 7, 119, 203000, 40000, 1448, 536, 1, 0, 0, 0, 0, 1, 0, 245, 245, 1500, 5792, 3250, 3000, 2147483647, 3, 1448, 3, 0, 5792, 0. This value after the send shows that one packet in unacked (field after value 536) however the window size is now reduced to 3. I am not able to figure out why this happens. Regards, Ashwin -- To unsubscribe from this list: send an email with "unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx Please read the FAQ at http://kernelnewbies.org/FAQ
