linux-kernel - Re: [PATCH] mm/mmu_notifier: restore set_pte_at

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: <52DFCF2B.1010603@mellanox.com>
Date:	Wed, 22 Jan 2014 16:01:15 +0200
From:	Haggai Eran <haggaie@...lanox.com>
To:	Andrea Arcangeli <aarcange@...hat.com>,
	Mike Rapoport <mike.rapoport@...ellosystems.com>
CC:	<linux-mm@...ck.org>, <linux-kernel@...r.kernel.org>,
	Izik Eidus <izik.eidus@...ellosystems.com>,
	Peter Zijlstra <a.p.zijlstra@...llo.nl>,
	Or Gerlitz <ogerlitz@...lanox.com>,
	Sagi Grimberg <sagig@...lanox.com>,
	Shachar Raindel <raindel@...lanox.com>
Subject: Re: [PATCH] mm/mmu_notifier: restore set_pte_at_notify semantics

On 22/01/2014 15:10, Andrea Arcangeli wrote:
> On Wed, Jan 15, 2014 at 11:40:34AM +0200, Mike Rapoport wrote:
>> Commit 6bdb913f0a70a4dfb7f066fb15e2d6f960701d00 (mm: wrap calls to
>> set_pte_at_notify with invalidate_range_start and invalidate_range_end)
>> breaks semantics of set_pte_at_notify. When calls to set_pte_at_notify
>> are wrapped with mmu_notifier_invalidate_range_start and
>> mmu_notifier_invalidate_range_end, KVM zaps pte during
>> mmu_notifier_invalidate_range_start callback and set_pte_at_notify has
>> no spte to update and therefore it's called for nothing.
>>
>> As Andrea suggested (1), the problem is resolved by calling
>> mmu_notifier_invalidate_page after PT lock has been released and only
>> for mmu_notifiers that do not implement change_ptr callback.
>>
>> (1) http://thread.gmane.org/gmane.linux.kernel.mm/111710/focus=111711
>>
>> Reported-by: Izik Eidus <izik.eidus@...ellosystems.com>
>> Signed-off-by: Mike Rapoport <mike.rapoport@...ellosystems.com>
>> Cc: Andrea Arcangeli <aarcange@...hat.com>
>> Cc: Haggai Eran <haggaie@...lanox.com>
>> Cc: Peter Zijlstra <a.p.zijlstra@...llo.nl>
>> ---
>>  include/linux/mmu_notifier.h | 31 ++++++++++++++++++++++++++-----
>>  kernel/events/uprobes.c      | 12 ++++++------
>>  mm/ksm.c                     | 15 +++++----------
>>  mm/memory.c                  | 14 +++++---------
>>  mm/mmu_notifier.c            | 24 ++++++++++++++++++++++--
>>  5 files changed, 64 insertions(+), 32 deletions(-)
> 
> Reviewed-by: Andrea Arcangeli <aarcange@...hat.com>
> 

Hi Andrea, Mike,

Did you get a chance to consider the scenario I wrote about in the other
thread?

I'm worried about the following scenario:

Given a read-only page, suppose one host thread (thread 1) writes to
that page, and performs COW, but before it calls the
mmu_notifier_invalidate_page_if_missing_change_pte function another host
thread (thread 2) writes to the same page (this time without a page
fault). Then we have a valid entry in the secondary page table to a
stale page, and someone (thread 3) may read stale data from there.

Here's a diagram that shows this scenario:

Thread 1                                | Thread 2        | Thread 3
========================================================================
do_wp_page(page 1)                      |                 |
  ...                                   |                 |
  set_pte_at_notify                     |                 |
  ...                                   | write to page 1 |
                                        |                 | stale access
  pte_unmap_unlock                      |                 |
  invalidate_page_if_missing_change_pte |                 |

This is currently prevented by the use of the range start and range end
notifiers.

Do you agree that this scenario is possible with the new patch, or am I
missing something?

Regards,
Haggai

--
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/