> On Monday 07 June 2004 12:23 pm, Tobias DiPasquale wrote: > > > Hello all, > > > > I was wondering if someone could point me to a discussion of what LPM > > algorithm the Linux routing FIB uses? > > Er, what do LPM and FIB mean please? I think I know netfilter quite > well, and I've not come across these abbreviations so far... Longest Prefix Match and Forwarding Information Base. To answer your question: Linux has routing cache which stores complete 5-tuples. If it is not found in cache, it is looked up in the FIB which is stored as 33 hashes (based on netmask size). Hashes are looked up from the /32 back to /0. Yes, it is extremely inefficient ;) -alex
