Re: hires timestamps for netif_rx()

From: kuznet@ms2.inr.ac.ru
Date: Fri Jan 18 2002 - 12:42:58 EST


Hello!

> Unfortunately gettimeofday at least on i386 has a readlock. It would
> add a new cache line bouncing bottleneck for every packet.

Wow! Andi, what cache line? Do you know where to find an SMP 386 or 486? :-)
It is not easy to find even UP one. :-)

get_fast_time() is _identically_ do_getimeofday().

I think Dave is right, get_fast_time() must die.

Actually, it should not be ever born. At least, my poor amber.inr.ac.ru
used to have these bogus function replaced with true stamp anyway.
Couple of IO ops did not affect performance comparing to lots of IO
to ne2000 and right stamp is really critical to get something sensible
from tcpdump.

Alexey
-
To unsubscribe from this list: send the line "unsubscribe linux-net" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html



This archive was generated by hypermail 2b29 : Wed Jan 23 2002 - 21:01:13 EST