Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Mon, 28 Oct 2002 19:04:59 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Mon, 28 Oct 2002 19:04:59 -0500 Received: from x35.xmailserver.org ([208.129.208.51]:19098 "EHLO x35.xmailserver.org") by vger.kernel.org with ESMTP id ; Mon, 28 Oct 2002 19:04:57 -0500 X-AuthUser: davidel@xmailserver.org Date: Mon, 28 Oct 2002 16:20:42 -0800 (PST) From: Davide Libenzi X-X-Sender: davide@blue1.dev.mcafeelabs.com To: bert hubert cc: Linux Kernel Mailing List , Subject: Re: [PATCH] epoll more scalable than poll In-Reply-To: <20021029000339.GA31212@outpost.ds9a.nl> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 987 Lines: 27 On Tue, 29 Oct 2002, bert hubert wrote: > On Mon, Oct 28, 2002 at 11:44:34PM +0000, Jamie Lokier wrote: > > > :( I was hoping sys_epoll would be scalable without increasing the > > number of system calls per event. > > I see only one call per event? sys_epoll_wait. Perhaps sys_epoll_ctl to > register a new interest. In theory you can register the fd at creation time with the full interest set and you can leave it in there for its whole life without having to call epoll_ctl() every switch between read/write. It's true that you could receive false events, but by studying the frequency of those false events on a "very high loaded" HTTP server, it resulted to be both very little and uneffective on the server performance. - Davide - 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/