Re: expiring cached routes on in_pcb entries.

On Tue, Nov 20, 2001 at 08:45:31PM -0600, Mike Silbersack wrote: > > On Wed, 21 Nov 2001, Ian West wrote: > > > Hi, I have been looking at the code in ip_ouput in relation to a problem > > I have had with named using the wrong route for talking to forwarders > > after

expiring cached routes on in_pcb entries.

Hi, I have been looking at the code in ip_ouput in relation to a problem I have had with named using the wrong route for talking to forwarders after route changes occur. There is a check for cached routes, and a check for validity, but not as far as I can see a check for expiry. The situation is

Netgraph/fxp/pppoe causing system panic in 4.3 stable.

I have a reproducible kernel panic in 4.3 stable caused (sort of) by netgraph, or maybe by fxp0 in conjunction with pppoe. It is caused by setting up and attempting to open a pppoe connection over an fxp interface, due to fxp_start being called before fxp device has been opened. (This only occurs