Re: Backporting DFly patches to FreeBSD?

[Tom Hummel <tom@xxxxxxxxxxxxx>]

> I don't think I have to "prove" it.

Nobody forces you, but it has some advantages to do so.

>  I'm telling you what happens.

Please don't forget to make sure, i believe it.

> I can't prove that the Sun is bright either. It

however... you can prove what's less brighter than the sun, by measuring 
two things in cl and tell me how. That's what i expect. If i already did 
it myself, i could compare our ways and learn...
What if the sun is 'BSD and another lightsource is Linux...

> just is. You shouldn't get RX_OVERRUNS with a constant
> traffic stream running at 60% usage, but you do. And if 
> you do something in user-land, like a database lookup, it
> gets worse, which also shouldn't happen. On BSD the lookup
> will take a real long time, but you won't drop packets on 
> the network interface. It can probably be tuned, 
> but only Linus and Allan know how to do it most likely.

I'm really interested in some exact measures, because i didn't have the 
"luck" to encounter such a situation and could measure it myself.