linux-kernel - Re: [linux-pm] Is it supposed to be ok to call del

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <AANLkTikH7DvizJ9340SiksoIWU5KU2n-OCqWc4f_-z9v@mail.gmail.com>
Date:	Tue, 11 May 2010 16:55:38 -0700
From:	Matt Reimer <mattjreimer@...il.com>
To:	"Rafael J. Wysocki" <rjw@...k.pl>
Cc:	linux-kernel <linux-kernel@...r.kernel.org>,
	Alan Stern <stern@...land.harvard.edu>,
	Jens Axboe <jens.axboe@...cle.com>,
	Maxim Levitsky <maximlevitsky@...il.com>,
	linux-pm <linux-pm@...ts.linux-foundation.org>,
	Andrew Morton <akpm@...ux-foundation.org>,
	Pavel Machek <pavel@....cz>
Subject: Re: [linux-pm] Is it supposed to be ok to call del_gendisk while 
	userspace is frozen?

On Thu, Apr 22, 2010 at 10:17 PM, Rafael J. Wysocki <rjw@...k.pl> wrote:
> On Friday 23 April 2010, Matt Reimer wrote:
>> On Thu, Mar 4, 2010 at 1:15 PM, Pavel Machek <pavel@....cz> wrote:
>> > Hi!
>> >
>> >> > My thought exactly.  This is the only approach that also solves the
>> >> > following race:
>> >> >
>> >> >     A driver is unloaded at the same time as a suspend starts.
>> >> >
>> >> >     The writeback thread gets frozen.
>> >> >
>> >> >     Then before the rmmod thread is frozen, it calls del_gendisk.
>> >> >
>> >> > Delaying things by means of a workqueue (or the equivalent) might also
>> >> > work, but it doesn't seem as safe.  For example, some important
>> >> > writebacks might end up getting delayed until too late.
>> >
>> > Delaying writebacks during sleep should be ok... That's why we do
>> > sync() after userspace is frozen -- nothing really important should be
>> > waiting for writeback after that point.
>>
>> Has this been fixed,
>
> No, it hasn't.
>
>> or has a consensus about how to fix this been
>> achieved? I'm hitting the same problem and have some time to work on a
>> fix.
>
> Generally, it looks like del_gendisk should thaw writeback threads, but not
> during suspend, only during resume.

Thawing the writeback thread only during resume does fix the case
Maxim originally presented:

0. build kernel with CONFIG_MMC_UNSAFE_RESUME
1. insert SD card
2. suspend
3. remove SD card while suspended
4. resume from suspend hangs

But if CONFIG_MMC_UNSAFE_RESUME is not set, the kernel oopses during
suspend because the MMC device suspend times out:

mmc0: card e624 removed
**** DPM device timeout: pxa2xx-mci.0 (pxa2xx-mci)
kernel BUG at /home/mreimer/sdg/android/android-2.1/kernel/drivers/base/power/main.c:453!
Unable to handle kernel NULL pointer dereference at virtual address 00000000
pgd = c0004000
[00000000] *pgd=00000000
Internal error: Oops: 817 [#1] PREEMPT

If I thaw the writeback thread unconditionally in del_gendisk() then
suspend and resume work as expected for both CONFIG_MMC_UNSAFE_RESUME
set/not set, even when the card is removed while suspended.

So what is the proper fix?

Matt
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/