| Subject: Re: Bug report: discovery fails if non-expected packets hit interface |
Author:
David F. Skoll
|
[
Next Thread |
Previous Thread |
Next Message |
Previous Message
]
Date Posted: 17:47:06 2006-07-28 Fri
Author Host/IP: 209.217.122.117
In reply to:
Mike Ireton
's message, "Bug report: discovery fails if non-expected packets hit interface" on 15:39:45 2006-07-28 Fri
>The problem however is that if a packet is received,
>and it's not destined for or or not a pado, the loop
>is looped. And the tv structure is re-initialized
>again, giving it anohter 5,10 or or 40 seconds with
>which to sit and listen for more packets. So one
>packet every x number of seconds (where x < timeout
>value) is enough to keep the loop going indefinately.
Well, it won't go indefinitely. As soon as a PADO is
received, it will break out of the loop. Of course, if
the PADO is lost *and* you keep getting more packets, it
will sit indefinitely, but that would only happen if a
malicious person were sending PPPoE discovery packets to
you.
>In my test version I have moved the initialization of
>tv to outside of the do loop and that seems to me to
>be the correct solution as select decrements tv and
>eventually it'll reach zero (or we get the packet
>we're waiting for).
That's not portable; select() is not guaranteed to modify
tv in the (useful) way you describe. A proper fix would
call gettimeofday() to see how much time we should
initialize tv to each time through the loop.
I don't think this is a serious enough problem in the
real world to be worth a fix, but if enough people
disagree, I will fix it.
Regards,
David.
[
Next Thread |
Previous Thread |
Next Message |
Previous Message
]
| |
