Re: [BUG 3.0.0-rc1] ksm: NULL pointer dereference in ksm_do_scan()

From: Andrea Righi <andrea@betterlinux.com>
To: Chris Wright <chrisw@sous-sol.org>
Cc: CAI Qian <caiqian@redhat.com>, Hugh Dickins <hughd@google.com>,
	Andrea Arcangeli <aarcange@redhat.com>,
	Rik van Riel <riel@redhat.com>, Mel Gorman <mel@csn.ul.ie>,
	Izik Eidus <ieidus@redhat.com>,
	KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>,
	linux-mm@kvack.org, linux-kernel@vger.kernel.org,
	Andrew Morton <akpm@linux-foundation.org>
Subject: Re: [BUG 3.0.0-rc1] ksm: NULL pointer dereference in ksm_do_scan()
Date: Thu, 2 Jun 2011 22:10:48 +0200	[thread overview]
Message-ID: <20110602201048.GA4114@thinkpad> (raw)
In-Reply-To: <20110602164841.GK23047@sequoia.sous-sol.org>

On Thu, Jun 02, 2011 at 09:48:41AM -0700, Chris Wright wrote:
> * Andrea Righi (andrea@betterlinux.com) wrote:
> > mmh.. I can reproduce the bug also with the standard ubuntu (11.04)
> > kernel. Could you post your .config?
> 
> Andrea (Righi), can you tell me if this WARN fires?  This looks
> like a pure race between removing from list and checking list, i.e.
> insufficient locking.

Yes, it does. With this patch:

[   50.968896] WARNING: at mm/ksm.c:1305 ksm_scan_thread+0x9e3/0xe50()

> 
> ksm_scan.mm_slot == the only registered mm
> 
> CPU 1 (bug program)		CPU 2 (ksmd)
> 				list_empty() is false
> lock
> ksm_scan.mm_slot
> list_del
> unlock
> 				slot == &ksm_mm_head (but list is now empty_)

It seems to be the exact problem.

-Andrea

> 
> 
> diff --git a/mm/ksm.c b/mm/ksm.c
> index 942dfc7..ab79a92 100644
> --- a/mm/ksm.c
> +++ b/mm/ksm.c
> @@ -1301,6 +1301,7 @@ static struct rmap_item *scan_get_next_rmap_item(struct page **page)
>  		slot = list_entry(slot->mm_list.next, struct mm_slot, mm_list);
>  		ksm_scan.mm_slot = slot;
>  		spin_unlock(&ksm_mmlist_lock);
> +		WARN_ON(slot == &ksm_mm_head);
>  next_mm:
>  		ksm_scan.address = 0;
>  		ksm_scan.rmap_list = &slot->rmap_list;