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]
Date:	Tue, 24 Sep 2013 18:29:15 -0700
From:	"Paul E. McKenney" <paulmck@...ux.vnet.ibm.com>
To:	linux-kernel@...r.kernel.org
Cc:	mingo@...nel.org, laijs@...fujitsu.com, dipankar@...ibm.com,
	akpm@...ux-foundation.org, mathieu.desnoyers@...icios.com,
	josh@...htriplett.org, niv@...ibm.com, tglx@...utronix.de,
	peterz@...radead.org, rostedt@...dmis.org, dhowells@...hat.com,
	edumazet@...gle.com, darren@...art.com, fweisbec@...il.com,
	sbw@....edu, "Paul E. McKenney" <paulmck@...ux.vnet.ibm.com>
Subject: [PATCH tip/core/rcu 05/11] rcu: Make list_splice_init_rcu() account for RCU readers

From: "Paul E. McKenney" <paulmck@...ux.vnet.ibm.com>

The list_splice_init_rcu() function allows a list visible to RCU readers
to be spliced into another list visible to RCU readers.  This is OK,
except for the use of INIT_LIST_HEAD(), which does pointer updates
without doing anything to make those updates safe for concurrent readers.

Of course, most of the time INIT_LIST_HEAD() is being used in reader-free
contexts, such as initialization or cleanup, so it is OK for it to update
pointers in an unsafe-for-RCU-readers manner.  This commit therefore
creates an INIT_LIST_HEAD_RCU() that uses ACCESS_ONCE() to make the updates
reader-safe.  The reason that we can use ACCESS_ONCE() instead of the more
typical rcu_assign_pointer() is that list_splice_init_rcu() is updating the
pointers to reference something that is already visible to readers, so
that there is no problem with pre-initialized values.

Signed-off-by: Paul E. McKenney <paulmck@...ux.vnet.ibm.com>
---
 include/linux/rculist.h | 23 +++++++++++++++++++++--
 1 file changed, 21 insertions(+), 2 deletions(-)

diff --git a/include/linux/rculist.h b/include/linux/rculist.h
index 4106721..45a0a9e 100644
--- a/include/linux/rculist.h
+++ b/include/linux/rculist.h
@@ -19,6 +19,21 @@
  */
 
 /*
+ * INIT_LIST_HEAD_RCU - Initialize a list_head visible to RCU readers
+ * @list: list to be initialized
+ *
+ * You should instead use INIT_LIST_HEAD() for normal initialization and
+ * cleanup tasks, when readers have no access to the list being initialized.
+ * However, if the list being initialized is visible to readers, you
+ * need to keep the compiler from being too mischievous.
+ */
+static inline void INIT_LIST_HEAD_RCU(struct list_head *list)
+{
+	ACCESS_ONCE(list->next) = list;
+	ACCESS_ONCE(list->prev) = list;
+}
+
+/*
  * return the ->next pointer of a list_head in an rcu safe
  * way, we must not access it directly
  */
@@ -191,9 +206,13 @@ static inline void list_splice_init_rcu(struct list_head *list,
 	if (list_empty(list))
 		return;
 
-	/* "first" and "last" tracking list, so initialize it. */
+	/*
+	 * "first" and "last" tracking list, so initialize it.  RCU readers
+	 * have access to this list, so we must use INIT_LIST_HEAD_RCU()
+	 * instead of INIT_LIST_HEAD().
+	 */
 
-	INIT_LIST_HEAD(list);
+	INIT_LIST_HEAD_RCU(list);
 
 	/*
 	 * At this point, the list body still points to the source list.
-- 
1.8.1.5

--
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/

Powered by blists - more mailing lists