linux-kernel - Re: [PATCH v7 10/10] lib/dlock-list: Fix use-after-unlock problem in dlist_for_each_entry

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: <20171030141147.leqcsaxebwiq6dq6@linux-n805>
Date:   Mon, 30 Oct 2017 07:11:48 -0700
From:   Davidlohr Bueso <dave@...olabs.net>
To:     Waiman Long <longman@...hat.com>
Cc:     Alexander Viro <viro@...iv.linux.org.uk>, Jan Kara <jack@...e.com>,
        Jeff Layton <jlayton@...chiereds.net>,
        "J. Bruce Fields" <bfields@...ldses.org>,
        Tejun Heo <tj@...nel.org>,
        Christoph Lameter <cl@...ux-foundation.org>,
        linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org,
        Ingo Molnar <mingo@...hat.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Andi Kleen <andi@...stfloor.org>,
        Dave Chinner <dchinner@...hat.com>,
        Boqun Feng <boqun.feng@...il.com>
Subject: Re: [PATCH v7 10/10] lib/dlock-list: Fix use-after-unlock problem in
 dlist_for_each_entry_safe()

On Fri, 27 Oct 2017, Waiman Long wrote:

>The dlist_for_each_entry_safe() macro in include/linux/dlock-list has
>a use-after-unlock problem where racing condition can happen because
>of a lack of spinlock protection.  Fortunately, this macro is not
>currently being used in the kernel.
>
>This patch changes the dlist_for_each_entry_safe() macro so that the
>call to __dlock_list_next_list() is deferred until the next entry is
>being used. That should eliminate the use-after-unlock problem.
>
>Reported-by: Boqun Feng <boqun.feng@...il.com>
>Signed-off-by: Waiman Long <longman@...hat.com>

Reviewed-by: Davidlohr Bueso <dbueso@...e.de>

But would it not be better to merge this patch (among others) into 1/N?
Specifically the newer patches 7-10 should be in the original dlock
implementation instead of adding fixes to incorrect code in the original
commit. Also less of a pita for backporting.

Thanks,
Davidlohr


>---
> include/linux/dlock-list.h | 28 +++++++++++++++++-----------
> 1 file changed, 17 insertions(+), 11 deletions(-)
>
>diff --git a/include/linux/dlock-list.h b/include/linux/dlock-list.h
>index 02c5f4d..f4b7657 100644
>--- a/include/linux/dlock-list.h
>+++ b/include/linux/dlock-list.h
>@@ -191,17 +191,17 @@ extern void dlock_list_add(struct dlock_list_node *node,
> }
>
> /**
>- * dlock_list_first_entry - get the first element from a list
>+ * dlock_list_next_list_entry - get first element from next list in iterator
>  * @iter  : The dlock list iterator.
>- * @type  : The type of the struct this is embedded in.
>+ * @pos   : A variable of the struct that is embedded in.
>  * @member: The name of the dlock_list_node within the struct.
>- * Return : Pointer to the next entry or NULL if all the entries are iterated.
>+ * Return : Pointer to first entry or NULL if all the lists are iterated.
>  */
>-#define dlock_list_first_entry(iter, type, member)			\
>+#define dlock_list_next_list_entry(iter, pos, member)			\
> 	({								\
> 		struct dlock_list_node *_n;				\
> 		_n = __dlock_list_next_entry(NULL, iter);		\
>-		_n ? list_entry(_n, type, member) : NULL;		\
>+		_n ? list_entry(_n, typeof(*pos), member) : NULL;	\
> 	})
>
> /**
>@@ -231,7 +231,7 @@ extern void dlock_list_add(struct dlock_list_node *node,
>  * This iteration function is designed to be used in a while loop.
>  */
> #define dlist_for_each_entry(pos, iter, member)				\
>-	for (pos = dlock_list_first_entry(iter, typeof(*(pos)), member);\
>+	for (pos = dlock_list_next_list_entry(iter, pos, member);	\
> 	     pos != NULL;						\
> 	     pos = dlock_list_next_entry(pos, iter, member))
>
>@@ -245,14 +245,20 @@ extern void dlock_list_add(struct dlock_list_node *node,
>  * This iteration macro is safe with respect to list entry removal.
>  * However, it cannot correctly iterate newly added entries right after the
>  * current one.
>+ *
>+ * The call to __dlock_list_next_list() is deferred until the next entry
>+ * is being iterated to avoid use-after-unlock problem.
>  */
> #define dlist_for_each_entry_safe(pos, n, iter, member)			\
>-	for (pos = dlock_list_first_entry(iter, typeof(*(pos)), member);\
>+	for (pos = NULL;						\
> 	    ({								\
>-		bool _b = (pos != NULL);				\
>-		if (_b)							\
>-			n = dlock_list_next_entry(pos, iter, member);	\
>-		_b;							\
>+		if (!pos ||						\
>+		   (&(pos)->member.list == &(iter)->entry->list))	\
>+			pos = dlock_list_next_list_entry(iter, pos,	\
>+							 member);	\
>+		if (pos)						\
>+			n = list_next_entry(pos, member.list);		\
>+		pos;							\
> 	    });								\
> 	    pos = n)
>
>-- 
>1.8.3.1
>