Re: Question about tcpTick_

["Amr A. Awadallah" <aaa@stanford.edu>]

> It controls both.  The test suite that tests both tcpTick_ and
> retransmit timers is run by "./test-all-tcp" in tcl/test,
> and is described in tcl/test/test-suite-tcp.txt.  The current

  I could not find any of the above files !

> retransmit timeout value, before any retransmit timer backoffs are
> added in, is as follows in the code:
> 
>         //
>         // Current retransmit value is
>         //    (unscaled) smoothed round trip estimate
>         //    plus 4 times (unscaled) rttvar.
>         //
>         t_rtxcur_ = (((t_rttvar_ << (2 + (T_SRTT_BITS - T_RTTVAR_BITS))) + 
>                        t_srtt_)  >> T_SRTT_BITS ) * tcp_tick_;

  I just looked up the equivalent piece of code in ns2 and here is what I 
found (from tcp.cc):

  double timeout = ((t_srtt_ >> 3) + t_rttvar_ + 1) * tcp_tick_; 

  Now, I have a couple of questions:

  (1) Where does the +1 above come from ?

  (2) Since t_rttvar_ and t_srtt_ are based on t_rtt_ which is already
normalized by tcp_tick_ as follows:

	//
	// tao is the measured RTT sample
	// t_rtt_ is tao expressed in tcp_tick_ units and rounded
	//
	t_rtt_ = int((tao / tcp_tick_) + 0.5);

  and since t_rtxcur_ is based on the unscaled values of t_rttvar_ and
t_srtt_, does not this mean that if t_srtt_ and t_rttvar_ have higher
granualities then this means that the timeout has a higher granuality
(since t_srtt_ and t_rttvar_ are integers), but the units of the timeout
is always in seconds irrespective of tcp ticks ?

Thanks,

-- Amr