From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753461Ab2ABWVw (ORCPT ); Mon, 2 Jan 2012 17:21:52 -0500 Received: from mail-ee0-f46.google.com ([74.125.83.46]:37414 "EHLO mail-ee0-f46.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753004Ab2ABWVu (ORCPT ); Mon, 2 Jan 2012 17:21:50 -0500 Message-ID: <4F022DFA.1090105@openvz.org> Date: Tue, 03 Jan 2012 02:21:46 +0400 From: Konstantin Khlebnikov User-Agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.1.19) Gecko/20111108 Iceape/2.0.14 MIME-Version: 1.0 To: Greg KH CC: "linux-kernel@vger.kernel.org" Subject: Re: [PATCH 2/2] tty: rework pty count limiting References: <20120102172254.11535.2653.stgit@zurg> <20120102172301.11535.10338.stgit@zurg> <20120102172903.GB19782@suse.de> In-Reply-To: <20120102172903.GB19782@suse.de> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Greg KH wrote: > On Mon, Jan 02, 2012 at 09:23:01PM +0400, Konstantin Khlebnikov wrote: >> After adding devpts multiple-insrances sysctl kernel.pty.max limit pty count for >> each devpts instance independently, while kernel.pty.nr shows total pty count. >> >> This patch restores sysctl kernel.pty.max as global limit (4096 by default), >> adds pty reseve for main devpts (mounted without "newinstance" argument), >> and new sysctl to tune it: kernel.pty.reserve (1024 by default) >> >> Also it adds devpts mount option "max=%d" to limit pty count for each devpts >> instance independently. (by default NR_UNIX98_PTY_MAX == 2^20) >> >> Thus devpts instances in containers cannot eat up all available pty even if we didn't >> set any limits, while with "max" argument we can adjust limits more precisely. >> >> Plus, now open("/dev/ptmx") return -ENOSPC in case lack of pty indexes, >> this is more informative than -EIO. > > That's a userspace api change, why is this going to be allowed? open() can return -ENOSPC in case normal fs, meanwhile error codes for /dev/ptmx wasn't documented at all: there is nothing in manpage pts, pty and ptmx. manpage posix_openpt and getpt refers to manpage open(2). I can return EIO back, if this is critical. > > greg k-h