From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S267475AbTGHPcL (ORCPT ); Tue, 8 Jul 2003 11:32:11 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S267470AbTGHPcL (ORCPT ); Tue, 8 Jul 2003 11:32:11 -0400 Received: from ev2.cpe.orbis.net ([209.173.192.122]:20874 "EHLO srv.foo21.com") by vger.kernel.org with ESMTP id S267473AbTGHPcG (ORCPT ); Tue, 8 Jul 2003 11:32:06 -0400 Date: Tue, 8 Jul 2003 10:46:36 -0500 From: Eric Varsanyi To: Davide Libenzi Cc: Linux Kernel Mailing List Subject: Re: epoll vs stdin/stdout Message-ID: <20030708154636.GM9328@srv.foo21.com> References: <20030707154823.GA8696@srv.foo21.com> <20030707194736.GF9328@srv.foo21.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.4i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Mon, Jul 07, 2003 at 04:26:21PM -0700, Davide Libenzi wrote: > On Mon, 7 Jul 2003, Davide Libenzi wrote: > Try out this one, either over 2.5.74 or over an existing epoll-patched > 2.4.{20,21} ... This appears to be working as advertised, thanks! IMO it doesn't seem that evil to deliver events per-fd rather than per-file, this is similar to the semantic you get from select() on fd's sharing an object. To be surprised someone would have to have coded to the (previous) sharing visible nature of epoll and be expecting the EEXIST back when sharing was in play. -Eric Varsanyi