From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Sat, 2 Nov 2002 16:31:33 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Sat, 2 Nov 2002 16:31:33 -0500 Received: from outpost.ds9a.nl ([213.244.168.210]:61898 "EHLO outpost.ds9a.nl") by vger.kernel.org with ESMTP id ; Sat, 2 Nov 2002 16:31:29 -0500 Date: Sat, 2 Nov 2002 22:37:59 +0100 From: bert hubert To: Davide Libenzi Cc: Linux Kernel Mailing List Subject: Re: [patch] total-epoll r2 ... Message-ID: <20021102213759.GA9440@outpost.ds9a.nl> Mail-Followup-To: bert hubert , Davide Libenzi , Linux Kernel Mailing List References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.3.28i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Sat, Nov 02, 2002 at 11:40:55AM -0800, Davide Libenzi wrote: > http://www.xmailserver.org/linux-patches/epoll.txt This page still contains: Furthermore, the function do_use_fd() immediately tries to read(2) from the fd. If it does not do so, there is a race condition with the risk of losing an event before interest in it was registered. This read may well return EAGAIN, which should be a cause for calling epoll_wait(). Am I correct in understanding that this is no longer true because epoll_ctl inserts an event if the poll condition is met? Furthermore, I don't think the epoll(2) page is that helpful as even an application developer that follows lkml (ie, me) has any use of the 'waitqueue' analogy. Regards, bert -- http://www.PowerDNS.com Versatile DNS Software & Services http://lartc.org Linux Advanced Routing & Traffic Control HOWTO