From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jan Kara Subject: Re: kernel oops in jbd2_journal_get_descriptor_buffer Date: Thu, 11 Oct 2012 17:28:28 +0200 Message-ID: <20121011152828.GA8001@quack.suse.cz> References: <50768765.8040406@ti.com> <20121011095932.GA16624@quack.suse.cz> <5076E449.5000004@ti.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Jan Kara , tytso@mit.edu, adilger.kernel@dilger.ca, linux-ext4@vger.kernel.org To: Clement Gallin-Douathe Return-path: Received: from cantor2.suse.de ([195.135.220.15]:34431 "EHLO mx2.suse.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932303Ab2JKP2b (ORCPT ); Thu, 11 Oct 2012 11:28:31 -0400 Content-Disposition: inline In-Reply-To: <5076E449.5000004@ti.com> Sender: linux-ext4-owner@vger.kernel.org List-ID: On Thu 11-10-12 17:22:49, Clement Gallin-Douathe wrote: > Jan, > > Thanks for your answer. > I have checked and I have no tricks with mapping flags (mount ext4 > options : rw,nosuid,nodev,noatime,errors=continue,data=ordered) > I have added high mem verification on grow_dev_page function just > after the page creation with find_or_create_page function. > PageHighMem() is not null. > So you think, it's a good idea to look into mm/page_alloc.c and > verify gfp flags? Yep. Check what gfp flags are passed to find_or_create_page() and if there if no __GFP_HIGH set, drill down into find_or_create_page() and further to mm/page_alloc.c to find out where we get the page from. Honza > On 10/11/2012 11:59 AM, Jan Kara wrote: > >!__GFP_HIGH in particular. So does the device ext4 is mounted on play > >some tricks with mapping fl > > -- Jan Kara SUSE Labs, CR