From owner-cvs-sys  Tue Feb 25 11:09:14 1997
Return-Path: <owner-cvs-sys>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.5/8.8.5) id LAA09110
          for cvs-sys-outgoing; Tue, 25 Feb 1997 11:09:14 -0800 (PST)
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19])
          by freefall.freebsd.org (8.8.5/8.8.5) with ESMTP id LAA09077;
          Tue, 25 Feb 1997 11:08:53 -0800 (PST)
Received: (from bde@localhost) by godzilla.zeta.org.au (8.8.3/8.6.9) id GAA17199; Wed, 26 Feb 1997 06:03:58 +1100
Date: Wed, 26 Feb 1997 06:03:58 +1100
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199702251903.GAA17199@godzilla.zeta.org.au>
To: ache@nagual.ru, bde@zeta.org.au
Subject: Re: cvs commit:  src/sys/kern sys_generic.c
Cc: bde@freefall.freebsd.org, cvs-all@freefall.freebsd.org,
        CVS-committers@freefall.freebsd.org, cvs-sys@freefall.freebsd.org,
        michaelh@cet.co.jp
Sender: owner-cvs-sys@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

>> >> What prevents you from deciding
>> >> now?
>> 
>> No real-world experience.  I don't run any processes that need even 256
>> descriptors.
>
>I think any test like
>
>	for (;;)
>		select(4096, ...
>
>will reveal the picture. We testing allocation strategy, not select
>functionality.

You also have to open an fd near 4096.  Otherwise the malloced buffer
is not used.

This is not a real world test.  Applications often sleep in select().
If there are N applications that don't sleep in select(), then reusing
the same buffer N times is probably better than using N different buffers.

Bruce