Index: [Article Count Order] [Thread]

Date: Tue, 1 Apr 2008 16:41:08 +0900
From: Mitsuru Chinen <mitch@linux.vnet.ibm.com>
Subject: [dhcptest:00268] Re: RT calculation of #42, rfc3315 client test
To: dhcptest@tahi.org
Cc: Hiroshi MIYATA <miyata@tahi.org>
Message-Id: <20080401164108.8217a266.mitch@linux.vnet.ibm.com>
In-Reply-To: <18C28442-08CF-41A4-BB82-7ED223C2044E@tahi.org>
References: <20080229173011.1d19524b.mitch@linux.vnet.ibm.com>	<E26EB179-F90F-430B-B393-EDE89EB9FCD0@tahi.org>	<18C28442-08CF-41A4-BB82-7ED223C2044E@tahi.org>
X-Mail-Count: 00268

On Tue, 1 Apr 2008 16:02:12 +0900
Hiroshi MIYATA <miyata@tahi.org> wrote:

> Do you mean this is also talking about 1.0.8?

Yes, but I merged this patch into MRD fix patch as I mentioned in
separate mail.
Now I have 3 patches I would like you to review:

o Tests for Maximum Retransmission Time
  http://www.tahi.org/dhcpv6/mail-list/200803.month/259.html

o [Client Test] Timeout value is short at #28
  http://www.tahi.org/dhcpv6/mail-list/200803.month/261.html

o [Client Test] Wrong calculation at MRD tests
  http://www.tahi.org/dhcpv6/mail-list/200804.month/263.html

The patch you mentioned is included in the last patch.
All 3 patches are for 1.0.8.

Best Regards,
----
Mitsuru Chinen <mitch@linux.vnet.ibm.com>


> ...miyata
> 
> On 2008/04/01, at 11:22, Hiroshi MIYATA wrote:
> 
> > Chinen-san,
> >
> > Sorry for my delay.
> > Let me confirm, which version of test tool are you talking about?
> >
> > DHCPv6 Self Test (version 1.0.8) (updated at 2007/11/26)
> >
> > is latest one.
> >
> > Are you using this one?
> >
> > Thanks,
> >
> > ...miyata
> >
> > On 2008/02/29, at 17:30, Mitsuru Chinen wrote:
> >
> >> Hi there,
> >>
> >> I found a bug in one of the client tests:
> >> #42 Part D : Maximum Retransmission Duration of Confirm message
> >>
> >> Section 14 in RFC3315 says:
> >>
> >>>  RT for each subsequent message transmission is based on the  
> >>> previous
> >>>  value of RT:
> >>>
> >>>     RT = 2*RTprev + RAND*RTprev
> >>>
> >>>  MRT specifies an upper bound on the value of RT (disregarding the
> >>>  randomization added by the use of RAND).  If MRT has a value of 0,
> >>>  there is no upper limit on the value of RT.  Otherwise:
> >>>
> >>>     if (RT > MRT)
> >>>        RT = MRT + RAND*MRT
> >>
> >> However, the test handles latter case only.
> >>
> >> Best Regards,
> >> ----
> >> Mitsuru Chinen <mitch@linux.vnet.ibm.com>
> >>
> >> --- DHCPv6_Self_Test/rfc3315/C_RFC3315_18_1_2_RetransCnf.seq	 
> >> 2007-10-30 21:56:45.000000000 +0900
> >> +++ DHCPv6_Self_Test_client/rfc3315/C_RFC3315_18_1_2_RetransCnf.seq	 
> >> 2008-02-25 22:03:52.000000000 +0900
> >> @@ -67,6 +67,7 @@ my $timeout = 30;
> >> my $remain = 0;
> >> my $lastflag = 0;
> >> my $mrdflag = 0;
> >> +my $min_rt = 0;
> >>
> >> my $IRT = $CNF_TIMEOUT;
> >> my $MRT = $CNF_MAX_RT;
> >> @@ -253,7 +254,11 @@ while(1){
> >> 		vLogHTML('<FONT COLOR="#FF0000">MRD expired! </FONT><BR>');
> >> 		dhcpExitFail;
> >> 	}
> >> -	if($rd + $interval*(1-$RAND) > $MRD){
> >> +	$min_rt = (2 - $RAND) * $interval;
> >> +	if ($min_rt > $MRT) {
> >> +		$min_rt = (1 - $RAND) * $MRT;
> >> +	}
> >> +	if($rd + $min_rt > $MRD){
> >> 		vLogHTML('<FONT COLOR="#FF0000">Next message must not be  
> >> observed</FONT><BR>');
> >> 		$mrdflag += 1;
> >> 	}
> >>
> >
> >
> 
>