linux 2.4.x should have SINGLE_LISTEN_UNSERIALIZED_ACCEPT

[dean gaudet <de...@arctic.org>]

someone want to make a patch to 1.3 which compiles in
SINGLE_LISTEN_UNSERIALIZED_ACCEPT when compiled on 2.4.x or later?  (i
don't see a need to conditionalised this at run-time...)

i was gonna patch it in, 'cept we don't yet check the linux subversion...

thanks
-dean

---------- Forwarded message ----------
Date: Mon, 30 Oct 2000 07:29:51 +0100
From: Andi Kleen <ak...@suse.de>
To: dean gaudet <de...@arctic.org>
Cc: Alan Cox <al...@lxorguk.ukuu.org.uk>, Andrew Morton <an...@uow.edu.au>,
     kumon@flab.fujitsu.co.jp, Andi Kleen <ak...@suse.de>,
     Alexander Viro <vi...@math.psu.edu>,
     Jeff V. Merkey <jm...@timpanogas.org>,
     Rik van Riel <ri...@conectiva.com.br>, linux-kernel@vger.kernel.org,
     Olaf Kirch <ok...@monad.swb.de>
Subject: Re: [PATCH] Re: Negative scalability by removal of
    lock_kernel()?(Was:

On Sun, Oct 29, 2000 at 11:45:49AM -0800, dean gaudet wrote:
> On Sat, 28 Oct 2000, Alan Cox wrote:
> 
> > > The big question is: why is Apache using file locking so
> > > much?  Is this normal behaviour for Apache?
> > 
> > Apache uses file locking to serialize accept on hosts where accept either has
> > bad thundering heard problems or was simply broken with multiple acceptors
> 
> if apache 1.3 is compiled with -DSINGLE_LISTEN_UNSERIALIZED_ACCEPT it'll
> avoid the fcntl() serialisation when there is only one listening port.  
> (it still uses it for multiple listeners... you can read all about my
> logic for that at <http://www.apache.org/docs/misc/perf-tuning.html>.)
> 
> is it appropriate for this to be defined for newer linux kernels?  i
> haven't kept track, sorry.  tell me what versions to conditionalize it on.

It should not be needed anymore for 2.4, because the accept() wakeup has been
fixed.


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