lkml.org 
[lkml]   [2006]   [Jun]   [2]   [last100]   RSS Feed
Views: [more markup]  [less markup]  [headers]  [forward] 
 
Messages in this thread
    /
    DateFri, 2 Jun 2006 11:01:49 +0400
    FromEvgeniy Polyakov <>
    SubjectRe: Question about tcp hash function tcp_hashfn()
    On Thu, Jun 01, 2006 at 12:40:10PM -0600, Brian F. G. Bidulock (bidulock@openss7.org) wrote:
    > So what are your thoughts about my sequence number approach (for
    > connected sockets)?
    
    Depending on how you are going to use it.
    Generic socket code does not have TCP sequence numbers since it must
    work with all supported protocols.
    Netchannels also do not know about internals of the packet by design,
    since all protocol processing is performed at the end peer.
    
    Sequence number can be wrapped in minutes in current networks and even
    faster tomorrow, that is why PAWS was created.
    
    Your idea about reinserting the socket does not scale in 1Gbit
    environment, and definitely will not in 10Gbit.
    
    Probably it is possible to create second hash table for TCP sockets only
    and use that table first in protocol handler, but it requires some
    research to prove the idea.
    
    -- 
    	Evgeniy Polyakov
    -
    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to majordomo@vger.kernel.org
    More majordomo info at  http://vger.kernel.org/majordomo-info.html
    Please read the FAQ at  http://www.tux.org/lkml/
    
    
    \
     
     \ /
      Last update: 2006-06-02 09:04    [from the cache]
    ©2003-2008