Re: Idea to make a kickass TCP read!

H. Peter Anvin (hpa@transmeta.com)
7 Aug 1998 01:24:32 GMT

Followup to: <199808061818.LAA25038@sun4.apsoft.com>
By author: Perry Harrington <pedward@sun4.apsoft.com>
In newsgroup: linux.dev.kernel
>
> Here's the thought.
>
> What if we created a read call that was identical in every way, except
> you passed it a SHM handle. It would directly write the data to the
> SHM segment rather than doing a NIC -> kernel -> user operation?
>
> This would make a one copy high performance, low-latency TCP read for
> the Cluster nuts.
>

No need for a SHM handle (which is just a pointer anyway); the kernel
could at least theoretically write it directly into user space. It
partially depends on the cleverness of the NIC though (remember, you
need to make sure you're giving it to the right process -- check
destination address and port number -- first of all!)

-hpa

-- 
    PGP: 2047/2A960705 BA 03 D3 2C 14 A8 A8 BD  1E DF FE 69 EE 35 BD 74
    See http://www.zytor.com/~hpa/ for web page and full PGP public key
        I am Bahá'í -- ask me about it or see http://www.bahai.org/
   "To love another person is to see the face of God." -- Les Misérables


-
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.altern.org/andrebalsa/doc/lkml-faq.html