From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from psmtp.com (na3sys010amx176.postini.com [74.125.245.176]) by kanga.kvack.org (Postfix) with SMTP id C7A376B006E for ; Thu, 7 Jun 2012 21:04:34 -0400 (EDT) Date: Thu, 7 Jun 2012 18:05:15 -0700 From: Andrew Morton Subject: Re: a whole bunch of crashes since todays -mm merge. Message-Id: <20120607180515.4afffc89.akpm@linux-foundation.org> In-Reply-To: References: <20120608002451.GA821@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-linux-mm@kvack.org List-ID: To: Linus Torvalds Cc: Dave Jones , Linux Kernel , linux-mm@kvack.org, Konstantin Khlebnikov , Oleg Nesterov On Thu, 7 Jun 2012 17:52:52 -0700 Linus Torvalds wrote: > On Thu, Jun 7, 2012 at 5:24 PM, Dave Jones wrote: > > I just started seeing crashes while doing simple things, like logging on a console.. > > I'm looking at it right now, and the sync_mm_rss() patch is pure > garbage. In many ways. > > You can't do sync_mm_rss() from mmdrop(), because there's no reason to > believe that the task that does mmdrop() does it on its own active_mm. > And even if you *could* do it there, it's still horribly wrong, > because it does it at the end *after* it already freed the mm! > > Does it go away if you revert that (commit 40af1bbdca47). I wish I > hadn't merged it, or that I had noticed how horrible it was before I > pushed out. > It appears this is due to me fat-fingering conflict resolution last week. That hunk is supposed to be in mm_release(), not mmput(). It's probably best to throw the patch away for now - we'll try again. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: email@kvack.org