From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <linux-kernel-owner@vger.kernel.org>
Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand
	id S1753497Ab0FBMnF (ORCPT <rfc822;w@1wt.eu>);
	Wed, 2 Jun 2010 08:43:05 -0400
Received: from smtpauth.net4india.com ([202.71.129.41]:58223 "EHLO
	smtpauth.net4india.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S1751759Ab0FBMnB (ORCPT
	<rfc822;linux-kernel@vger.kernel.org>);
	Wed, 2 Jun 2010 08:43:01 -0400
Message-ID: <4C0651FF.2080007@xenontk.org>
Date: Wed, 02 Jun 2010 18:13:43 +0530
From: David John <davidjon@xenontk.org>
Reply-To: davidjon@xenontk.org
User-Agent: Mozilla/5.0 (X11; X; Linux x86_64; rv:1.9.1.9) Gecko/20100430 Thunderbird/3.0.4
MIME-Version: 1.0
To: Jens Axboe <jaxboe@fusionio.com>
CC: Divyesh Shah <dpshah@google.com>,
       "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Subject: Re: [2.6.35-rc1 BUG]: Calling sched_clock causes stack trace with
 DEBUG_PREEMPT
References: <4C04AC78.3020702@xenontk.org> <4C04BB3F.2030707@xenontk.org> <20100601102319.GL3564@kernel.dk>
In-Reply-To: <20100601102319.GL3564@kernel.dk>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

On 06/01/2010 03:53 PM, Jens Axboe wrote:
> On Tue, Jun 01 2010, David John wrote:
>> On 06/01/2010 12:15 PM, David John wrote:
>>> Hi,
>>>
>>> I get stack traces like the following:
>>>
>>> BUG: using smp_processor_id() in preemptible [00000000] code: init/1
>>> caller is native_sched_clock+0x37/0x6d
>>> Pid: 1, comm: init Not tainted 2.6.35-rc1 #99
>>> Call Trace:
>>> [<ffffffff811c422b>] debug_smp_processor_id+0xd3/0xec
>>> [<ffffffff81008a1e>] native_sched_clock+0x37/0x6d
>>> [<ffffffff811a7bfd>] blk_rq_init+0x92/0x9e
>>> [<ffffffff811a7f2e>] get_request+0x1c8/0x27e
>>> [<ffffffff811a8011>] get_request_wait+0x2d/0x13c
>>> [<ffffffff810306ee>] ? get_parent_ip+0x11/0x42
>>> [<ffffffff81411ea6>] ? add_preempt_count+0xad/0xb1
>>> [<ffffffff811a83b3>] __make_request+0x293/0x3b8
>>> [<ffffffff81099737>] ? mempool_alloc_slab+0x10/0x12
>>> [<ffffffff811a6bc4>] generic_make_request+0x187/0x1ea
>>> [<ffffffff810306ee>] ? get_parent_ip+0x11/0x42
>>> [<ffffffff811a6ce5>] submit_bio+0xbe/0xc7
>>> [<ffffffff810f1f37>] submit_bh+0xef/0x111
>>> [<ffffffff810f4865>] block_read_full_page+0x1ec/0x20c
>>> [<ffffffff810f6815>] ? blkdev_get_block+0x0/0x5e
>>> [<ffffffff81411de6>] ? sub_preempt_count+0x92/0xa5
>>> [<ffffffff810a0500>] ? __lru_cache_add+0x73/0x93
>>> [<ffffffff810f6e81>] blkdev_readpage+0x13/0x15
>>> [<ffffffff8109f981>] __do_page_cache_readahead+0x161/0x194
>>> [<ffffffff8109f9d0>] ra_submit+0x1c/0x20
>>> [<ffffffff8109fc5f>] ondemand_readahead+0x1bb/0x1ce
>>> [<ffffffff810973c9>] ? rcu_read_unlock+0xe/0x29
>>> [<ffffffff8109fd46>] page_cache_sync_readahead+0x38/0x3a
>>> [<ffffffff810993c2>] generic_file_aio_read+0x24d/0x57f
>>> [<ffffffff810cf973>] ? __mem_cgroup_try_charge+0x78/0x46c
>>> [<ffffffff810d2aa0>] do_sync_read+0xc6/0x103
>>> [<ffffffff810306ee>] ? get_parent_ip+0x11/0x42
>>> [<ffffffff810adca5>] ? handle_mm_fault+0x7cb/0x7e5
>>> [<ffffffff810f69c5>] ? block_ioctl+0x32/0x36
>>> [<ffffffff81189fcc>] ? selinux_file_permission+0x57/0xae
>>> [<ffffffff81184524>] ? security_file_permission+0x11/0x13
>>> [<ffffffff810d34af>] vfs_read+0xa6/0x103
>>> [<ffffffff810d35c2>] sys_read+0x45/0x69
>>> [<ffffffff810029ab>] system_call_fastpath+0x16/0x1b
>>>
>>> Since with commit 9195291e you call 'set_start_time_ns' from
>>> 'blk_rq_init' which in turn calls 'sched_clock' with preempt enabled.
>>> There are other calls to sched_clock in blk-cgroup.c. Maybe they should
>>> be checked as well?
>>>
>>> Regards,
>>> David.
>>>
>>
>> I see this has been reported. Never mind.
> 
> This should fix it for now, if you have time to test, then that would
> not hurt.

Yes, this works.

Thanks,
David.