On 2009-04-28, Daniel Ouellet <dan...@presscom.net> wrote: > Henning Brauer wrote: >> * Daniel Ouellet <dan...@presscom.net> [2009-04-28 02:49]: >>> shut up! All are real and I even learn from Henning about the lost of >>> Queue here as well, witch I haven't thought of then. So, loose of queue, >>> mean also lost of AltQ too. >> >> no, this is not related to altq at all. > > Thanks for the correction here Henning. I was wrong. > > I assume AltQ was working with the queue, so, no queue would mean > loosing altq capability. Hmmm. Looks like something I miss understood > and I will go back looking at it. > > Thanks for the tip.
this is the other queue; sysctl net.inet.ip.ifq I thought PF would use it in pf_check_congestion() as a hint, but I can't work out how this happens for ethernet interfaces, only these.. ./net/if_ppp.c: if_congestion(inq); ./net/if_sl.c: if_congestion(&ipintrq); ./net/if_spppsubr.c.orig: if_congestion(inq); ./net/if_spppsubr.c: if_congestion(inq); ./net/if_strip.c: if_congestion(&ipintrq); ./net/if_tun.c: if_congestion(ifq); from Henning's post; > i told you before it is not an OpenBSD problem. > it is implemented the way it is because you kind of have to do it this > way, or similiar. not to mention at least 4 OS are using substantially the same code. at least 5 if you count miros.
