Really quick summary: pppd will not let go of its allocated ppp0 device
when hung up upon (sometimes) or sent a KILL signal. It does work
fine when sent an INT or HUP.
Detailed testing explaination:
I had PPP built as part of the kernel, at first, with the bsd_comp as
a module (not used, to my knowledge). I had established a connection,
it had somehow died, and it had tried to reconnect (as per the persist
option to pppd--2.3.5 for all these cases). It tried to reconnect,
but after running the chat script, I couldn't ping anything. I tried
examining my route table, got tired of waiting for it to resolve, and
ran ifconfig. Wow, there was an extra ppp1 device! And top showed
pppd using 67.99% of one of my P II 266s.
I tried killing off pppd (with a big bad signal 9), it died, but each
attempt to re-establish a connection would result in another ppp device
being created (up to ppp3). I didn't try changing my routing table
so that I could route the net through the latest in the pppX chain, but
I suspect it would have worked.
Now for the exciting part. I rebuilt PPP as a module, rebooted, things
seemed to work. It dialed up, did connect. I then killed pppd with a
signal 9. It died a quick death and the wild pppX device creation
thing continued on each reload.
I rebooted to clear this up, and give me a clean test bed for further
experimentations. I booted, it connected. I then tried sending not
KILL but INT and HUP signals to pppd. pppd behaves great when sent
these signals, but freaks out on a KILL and doesn't let go of the allocated
ppp0 device when it does so. However, it also doesn't let go of ppp0
when it gets hung up on (it seems it only has a problem in this case
when it doesn't log a "NO CARRIER" through syslog).
This sounds like a pppd problem, but I haven't looked into the changes
to ppp in 2.1.109. 2.1.108 works great with pppd. Oh, if you're currently
confused as to why I included both in-kernel and module accounts for
my testing, I did it just to be thorough. It behaved identically for
both cases.
-- Shaw Terwilliger (ten.tenecnavda@giwt)- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html