lkml.org 
[lkml]   [1999]   [Jun]   [4]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: zero-copy TCP fileserving
Alan Cox wrote:

> > I've seen mbufs mentioned before - BSD.
> > I've seen mentioned that our skbs are faster than mbufs, without justification.
> > What's the difference between skbs and mbufs?
>
> Fundamentally. skbuffs are single physically linear blocks, BSD mbufs are
> chains of small blocks. The BSD code nowdays makes heavy use of "mbuf
> clusters" to effectively get linear buffers most of the time.
>
> Having chains means you keep having to say 'is the rest of this structure
> in this buffer' and 'copy this around a bit to make the structure in one
> buffer'.

Why do several prominent firewall and networking companies shun or speak foul of our
networking, saying we drop so many packets under load that linux simply isn't an
acceptable platform for their product? This is most recently in reference to
certain other lists conversations about watching a packet stream with tcpdump and
noticing missing packets. Is there any truth in this or is it FUD? Some people say
skbuffs are faster, some say they are overloaded with extraneous information. What
is more correct?

-d



-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:52    [from the cache]
©2003-2011 Jasper Spaans