Re: Examine route.c! -- two bugs fixed -- patch attached!

[Dr. Werner Fink]

> 
> Ok, I have tested a bit more and found that:
>   yes | nc somehost discard &
> does reproduce the problem more reliably even with this patch applied. But
> only when 'somehost' is a linux firewall (that also acts as a router
> here). It also takes some minutes before it start to increase the
> allocation in "route.c:1422", like it has to fill up some buffer in the
> firewall or so. If I make 'somehost' a normal linux machine then it does
> not began to leak.
> 
> Can I supply some useful tcpdump when this starts to happen? Some ideas
> on how to filter the dump, otherwise it will be very large to examine.
> Maybe even tcpdump won't be able to keep up to that amount of data?

ip_rt_put() only frees a route if this is a) not cached and b) if
the reference count is exactly one. This is in the most cases correct.
If Philip found out all possible drops this may be a reason.

            Werner