From owner-cvs-all  Thu Jun  4 10:41:58 1998
Return-Path: <owner-cvs-all@FreeBSD.ORG>
Received: (from majordom@localhost)
          by hub.freebsd.org (8.8.8/8.8.8) id KAA21542
          for cvs-all-outgoing; Thu, 4 Jun 1998 10:41:58 -0700 (PDT)
          (envelope-from owner-cvs-all@FreeBSD.ORG)
Received: from burka.rdy.com (dima@burka.rdy.com [205.149.163.30])
          by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id KAA21531
          for <committers@FreeBSD.ORG>; Thu, 4 Jun 1998 10:41:55 -0700 (PDT)
          (envelope-from dima@burka.rdy.com)
Received: (from dima@localhost)
	by burka.rdy.com (8.8.8/RDY&DVV) id KAA08545;
	Thu, 4 Jun 1998 10:41:01 -0700 (PDT)
Message-Id: <199806041741.KAA08545@burka.rdy.com>
Subject: Re: FD_SETSIZE
In-Reply-To: <199806041439.WAA24579@spinner.netplex.com.au> from Peter Wemm at "Jun 4, 98 10:39:21 pm"
To: peter@netplex.com.au (Peter Wemm)
Date: Thu, 4 Jun 1998 10:41:01 -0700 (PDT)
Cc: committers@FreeBSD.ORG
X-Class: Fast
Organization: HackerDome
Reply-To: dima@best.net
From: dima@best.net (Dima Ruban)
X-Mailer: ELM [version 2.4ME+ PL40 (25)]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-cvs-all@FreeBSD.ORG
Precedence: bulk

Peter Wemm writes:
> Dima Ruban wrote:
> > Hey guys!
> > 
> > -current has Subject defined to 1024, -stable - to 256.
> > Any objections from bumping it in -stable, so it would match
> > value that we have in -current?
> 
> If I recall correctly, the -current select() implementation was enhanced to
> make better use of memory and copyin/out with large vector sizes.  This has
> not happened with -stable, so increasing the vector size might not be an
> idea as it could have a [slight?] detrimental effect.  I don't remember 
> the details of the situation well enough but I am pretty sure that there 
> was some reason why it wasn't taken across.

I just went through a commit logs, and here we go:
--
revision 1.19
date: 1996/08/20 07:17:48;  author: smpatel;  state: Exp;  lines: +43 -15
Remove the kernel FD_SETSIZE limit for select().
Make select()'s first argument 'int' not 'u_int'.

Reviewed by:    bde
--
revision 1.22
date: 1997/02/20 11:51:52;  author: bde;  state: Exp;  lines: +51 -36
Improved select():
- avoid malloc() if the number of fds is small.
- pack the bits better so that `small' is quite large.
- don't waste time generating zero bits for null fd_set pointers or
  scanning these bits.

Possibly improved select():
- free malloc()ed storage before returning.  This is simpler and I
  think huge select()s aren't worth optimizing since they are rare,
  relative gain would be small and there would be tiny costs for all
  selects().

Reviewed by:    ache (first version by him too)
--
(that's for sys_generic.c)

It looks like it would be relatively easy to bring this into -stable.

> 
> Cheers,
> -Peter
> --
> Peter Wemm <peter@netplex.com.au>   Netplex Consulting
> 
> 

-- dima

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe cvs-all" in the body of the message