Re: ppp and LCP ConfReq ignored

In the Usenet newsgroup uk.comp.os.linux, in article
<AqudnTybQs1nymHfRVnyvg@xxxxxxxxxxxxxx>, Keith Matthews wrote:

>Paul Kinsler wrote:
>
>> No solution yet, and their support service is, erm, failing
>> to help. However, from reading around, even M$ servers seem
>> to understand how to exchange LCP frames (I've seen a sensible
>> looking negotiation on some M$ web doco); so I'm now beginning
>> to think that their end is not even starting ppp.
>
>I've forgotten who the ISP is and a large set of the thread so this may
>have already been covered, but this is a fairly standard BT trick ISTR.
>Wvdial has a comment in it's config about it. Seems like something has
>to be sent to trigger of the other end - have you checked that out ?

When all is said and done, there are just three ways to connect. The
original way - see some of the ancient HOWTOs like

33014 Mar 2 2000 ISP-Hookup-HOWTO
183087 Jul 19 2000 PPP-HOWTO

is what most "helper" tools default to. They expect you to log in to a
shell account on the ISP end, and after you get in, starting an application
that runs pppd. If you are running your own dialin server, this is the
default mode with mgetty or mingetty. This is the original mode from 1989
when ppp was initially proposed. This mode went out of fashion when
microsoft invented the Internet, or whatever in 1995. Virtually _NO_ ISP
anywhere in the world is still using this. But this only occurred nine
years ago, and people writing applications for Linux (specifically stuff
like WvDial, or KPPP) have not yet gotten the word of this change and
default to this mode.

The problem with this is that this old mode of operation is the default
mode on many terminal servers - but because windoze users don't use it,
it is often unconfigured - you get a login prompt, followed by a password
prompt, and no matter what you do, you get a "Bad Login" or "Access Denied"
error, and you are screwed until you hang up the phone and do it the right
way. The key here is to NOT send anything - not even a carriage return,
before starting ppp transmissions.

The second mode - used by virtually every ISP is the "AutoPPP" style,
demanded by microsoft. They didn't invent it - it actually dates from
1992, three years before microsoft invented the telephone. If you are
running your own dialin server - the keyword is 'autoppp'. Briefly, the
person who dials in starts sending ppp frames AS SOON AS THE MODEM
CONNECTS, and doesn't send even a hint of ASCII (not even a carriage
return or line feed) which would trigger the unconfigured text mode noted
in the previous paragraph. For users of /usr/sbin/chat, the dial script
looks like this:

ABORT BUSY ABORT 'NO CARRIER' "" AT&F1 OK ATDT2662902 CONNECT \d\c

Note: AT&F1 = USR/3Com AT&F (AT&F0) = Rockwell & Lucent

This script sets two abort conditions (the line is busy, or something
_Other_Than_ a modem answers), then waits for nothing ( "" ) and sends an
init string (check your modem manual). When the modem responds with OK,
it dials the phone number, and waits for the modem to report establishing
a connection. The script then waits one second ( \d ) and exits without
sending the normal newline character ( \c ).

Authentication is handled by pppd, using either PAP (secret is in
/etc/ppp/pap-secrets) or one of two or three types of CHAP (secret
is in /etc/ppp/chap-secret). The two or three types differ in the
algorithm used to encrypt the username and password sent over the wire.
Up to ppp-2.4.1, ppp would use CHAP-MD5 (the standard), and microsoft's
broken MS-CHAP.v1 (a.k.a. CHAP-80). Microsoft decided to improve that,
and brought out MS-CHAP.v2 (a.k.a. CHAP-80) that just happens to be totally
incompatible with MS-CHAP.v1. ppp-2.4.2 has added that capability, though
I don't see it used that often, even by microsoft.

Finally, there is a third mode of connecting - introduced in windoze2000.
This is mid-way between the previous two modes - where you wait for the
modems to connect, and then send the ASCII text "CLIENT" (without the
quotes). The peer may or may not respond with "SERVER" or "CLIENT/SERVER"
but that is irrelevant. You then start sending ppp frames, and the peer
will probably expect you to authenticate using MS-CHAP.v2 which means you
need to have ppp-2.4.2 or later. Thankfully, this last mode is fairly
rare, and is usually set up by the MCSE your company hired to run the
networks - he wouldn't know what ppp means if you asked, and probably
can't spell it either. Luckily, NO ISP has ever used this abortion. But
then, I don't know of any "helper" tool that has ever heard of it either,
so we're even.

Old guy
.

Relevant Pages

  • Re: w32 blaster worm
    ... >> Each computer has a modem and dials up to the Internet as needed. ... > All your ports are exposed when they dial into the ISP, ... > user has the ability to access the other users computer the RCP calls ...
    (alt.computer.security)
  • Problems with dial2net and dhcp on sorcerer
    ... configured it to dial to an isp using a modem on ttyS0. ... network address. ... modem dials the isp. ...
    (comp.os.linux.networking)
  • Problems with dial2net and dhcp on sorcerer
    ... configured it to dial to an isp using a modem on ttyS0. ... network address. ... modem dials the isp. ...
    (comp.os.linux.misc)
  • Dial Up Problem
    ... I use xp home and a 56k modem on a new Dell 2400. ... working okay until my ISP issued a new dial in phone number. ... I have tried these accounts on ... What happens is that I hear the dial tone, the modem dials and shrieks ...
    (microsoft.public.windowsxp.network_web)
  • Dial Up Problem
    ... I use xp home and a 56k modem on a new Dell 2400. ... working okay until my ISP issued a new dial in phone number. ... I have tried these accounts on ... What happens is that I hear the dial tone, the modem dials and shrieks ...
    (microsoft.public.windowsxp.general)