Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752117AbbEHAh5 (ORCPT ); Thu, 7 May 2015 20:37:57 -0400 Received: from relay4-d.mail.gandi.net ([217.70.183.196]:52202 "EHLO relay4-d.mail.gandi.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752075AbbEHAhy (ORCPT ); Thu, 7 May 2015 20:37:54 -0400 Date: Thu, 7 May 2015 17:37:48 -0700 From: josh@joshtriplett.org To: Andrew Morton Cc: Linus Torvalds , Fengguang Wu , Iulia Manda , "Paul E. McKenney" , Fabian Frederick , Linux Memory Management List , linux-kernel@vger.kernel.org Subject: Re: [PATCH] devpts: If initialization failed, don't crash when opening /dev/ptmx Message-ID: <20150508003748.GA1033@cloud> References: <20150507003547.GA6862@jtriplet-mobl1> <20150507155919.16ab7177e4956d8f47803750@linux-foundation.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20150507155919.16ab7177e4956d8f47803750@linux-foundation.org> User-Agent: Mutt/1.5.20 (2009-06-14) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 2002 Lines: 45 On Thu, May 07, 2015 at 03:59:19PM -0700, Andrew Morton wrote: > On Wed, 6 May 2015 17:35:47 -0700 Josh Triplett wrote: > > > If devpts failed to initialize, it would store an ERR_PTR in the global > > devpts_mnt. A subsequent open of /dev/ptmx would call devpts_new_index, > > which would dereference devpts_mnt and crash. > > > > Avoid storing invalid values in devpts_mnt; leave it NULL instead. > > Make both devpts_new_index and devpts_pty_new fail gracefully with > > ENODEV in that case, which then becomes the return value to the > > userspace open call on /dev/ptmx. > > It looks like the system is pretty crippled if init_devptr_fs() fails. > Can the user actually get access to consoles and do useful things in > this situation? Maybe it would be better to just give up and panic? Mounting devpts doesn't work without it, but you don't *need* to do that to run a viable system. A full-featured terminal might be unhappy. init=/bin/sh works, and a console login doesn't strictly require /dev/pts. A substantial initramfs or rescue system should work without /dev/pts mounted. I think this falls under Linus's comments elsewhere about BUG versus WARN. The system can continue and will function to some degree. panic() is more suitable for "if I even return from this function, horrible things will start happening". With this patch, all the functions provided by devpts gracefully fail if devpts did, so I don't see a good reason to panic(). > > @@ -676,12 +689,15 @@ static int __init init_devpts_fs(void) > > struct ctl_table_header *table; > > > > if (!err) { > > + static struct vfsmount *mnt; > > static is weird. I assume this was a braino? Copy/paste issue, yes. Fixed in v2. - Josh Triplett -- 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/