folks,
In a recent adventure with a PPPOE driver on Linux, Matthew and I stumbled
on what i believe is a PAP state transition that was left out.
- A client sends a PAP auth-request
- The server sends back a auth-ack
- Unfortunately the auth-ack gets lost (actually it was lost
because of a bug in PPPOE).
- The server then issues an IP address to the client. The state machine
says this optional but allows negotiation to proceed quickly should the
auth-ack be lost.
Assuming i=0 is the one executed above:
for (i=1; i<pap-max-authreq;i++) {
-The client sends a PAP auth-request
- the server sends back an auth-ack which gets lost.
- the server issues an IP address to the client
}
LCP timeout as a result
In my opinion, the short-circuit should have been taken to the "opened"
state as soon as the IPCP message (with the IP address) was received.
This would add robustness to pppd and infact conformance to the specs.
I believe that problem is also there in the CHAP code.
cheers
jamal
PS:- is this mailing list archived anywhere?
-
To unsubscribe from this list: send the line "unsubscribe linux-ppp" in
the body of a message to [EMAIL PROTECTED]
