Re: help me hang up the phone

Dan Jacobson <jidanni@jidanni.org> wrote:[color=blue]
> As we see, I just can't get the phone hung up right away:[/color]
[color=blue]
> Terminating on signal 15.
> Script /etc/ppp/ip-down started (pid 17072)
> sent [LCP TermReq id=0x2 "User request"]
> Script /etc/ppp/ip-down finished (pid 17072), status = 0x1
> sent [LCP TermReq id=0x3 "User request"]
> sent [LCP TermReq id=0x4 "User request"]
> Hangup (SIGHUP)
> Modem hangup
> Connection terminated.[/color]
[color=blue]
> despite pppd options in effect:[/color]

....
[color=blue]
> lcp-max-terminate 1[/color]

I now think it likely there's an inconsistency in the pppd code
whereby the id field of Terminate-Request is changed with each
retransmission, even though there is no response from the peer.
For a LCP Configure-Request pppd does not change the id field in
retransmissions if there is no response from the peer.

Moreover, the pppd lcp-max-{configure, failure} options appear to
only apply when the id field is unchanged, and any reply before the
maximum is reached resets the counter. If true, it would seem that
the Terminate-Request id field is treated the same way and the id
changes seen above prevented lcp-max-terminate from working.

This is all conjecture, derived from the Terminate-Request behavior
seen above and black-box tests with pppd. I've not been able to
motivate myself to delve deeply into the code to be sure.

--
Clifford Kite Email: "echo [email]xvgr_yvahk-ccc@ri1.arg[/email]|rot13"
PPP-Q&A links, downloads: [url]http://ckite.no-ip.net/[/url]
/* Speak softly and carry a +6 two-handed sword. */

Clifford Kite <kite@see.signature.id> writes:[color=blue]
> I now think it likely there's an inconsistency in the pppd code
> whereby the id field of Terminate-Request is changed with each
> retransmission, even though there is no response from the peer.
> For a LCP Configure-Request pppd does not change the id field in
> retransmissions if there is no response from the peer.[/color]

Maybe. But what does that have to do with the behavior of
lcp-max-terminate?

I tested the current CVS code, and it *does* respect the
lcp-max-terminate parameter correctly.
[color=blue]
> Moreover, the pppd lcp-max-{configure, failure} options appear to
> only apply when the id field is unchanged, and any reply before the
> maximum is reached resets the counter. If true, it would seem that
> the Terminate-Request id field is treated the same way and the id
> changes seen above prevented lcp-max-terminate from working.[/color]

No, I don't think that's the case.

--
James Carlson, IP Systems Group <james.d.carlson@sun.com>
Sun Microsystems / 1 Network Drive 71.234W Vox +1 781 442 2084
MS UBUR02-212 / Burlington MA 01803-2757 42.497N Fax +1 781 442 1677

James Carlson <james.d.carlson@sun.com> wrote:[color=blue]
> Clifford Kite <kite@see.signature.id> writes:[color=green]
>> I now think it likely there's an inconsistency in the pppd code
>> whereby the id field of Terminate-Request is changed with each
>> retransmission, even though there is no response from the peer.
>> For a LCP Configure-Request pppd does not change the id field in
>> retransmissions if there is no response from the peer.[/color][/color]
[color=blue]
> Maybe. But what does that have to do with the behavior of
> lcp-max-terminate?[/color]

Reading it now, it seems to have very little to do with the behavior
of lcp-max-terminate except to say that the behavior in regard to the
Identifier seemed to be the opposite from that of lcp-max-configure
when there are no replies to requests.

It certainly does nothing to answer the question as to why the option
lcp-max-terminate failed to work for the OP, which is what his initial
post appeared to show had happened.
[color=blue]
> I tested the current CVS code, and it *does* respect the
> lcp-max-terminate parameter correctly.[/color]

I'm using the Linux 2.4.2 version, which may not be the current CVS
version. I think the OP was too, and according to his original post
the problem does exist for it.
[color=blue][color=green]
>> Moreover, the pppd lcp-max-{configure, failure} options appear to
>> only apply when the id field is unchanged, and any reply before the
>> maximum is reached resets the counter. If true, it would seem that
>> the Terminate-Request id field is treated the same way and the id
>> changes seen above prevented lcp-max-terminate from working.[/color][/color]
[color=blue]
> No, I don't think that's the case.[/color]

You are probably right. In reviewing what I did, it now seems that
I concluded too much, given too little understanding of the code and
with limited resources available for black-box testing.

--
Clifford Kite Email: "echo [email]xvgr_yvahk-ccc@ri1.arg[/email]|rot13"
PPP-Q&A links, downloads: [url]http://ckite.no-ip.net/[/url]