[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20160912152200.565384804@linuxfoundation.org>
Date: Mon, 12 Sep 2016 18:59:08 +0200
From: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
To: linux-kernel@...r.kernel.org
Cc: Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
stable@...r.kernel.org, Dexuan Cui <decui@...rosoft.com>,
"K. Y. Srinivasan" <kys@...rosoft.com>,
Sasha Levin <alexander.levin@...izon.com>
Subject: [PATCH 4.4 039/192] [PATCH 040/135] Drivers: hv: vmbus: serialize process_chn_event() and vmbus_close_internal()
4.4-stable review patch. If anyone has any objections, please let me know.
------------------
[ Upstream commit 63d55b2aeb5e4faa170316fee73c3c47ea9268c7 ]
process_chn_event(), running in the tasklet, can race with
vmbus_close_internal() in the case of SMP guest, e.g., when the former is
accessing channel->inbound.ring_buffer, the latter could be freeing the
ring_buffer pages.
To resolve the race, we can serialize them by disabling the tasklet when
the latter is running here.
Signed-off-by: Dexuan Cui <decui@...rosoft.com>
Signed-off-by: K. Y. Srinivasan <kys@...rosoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Signed-off-by: Sasha Levin <alexander.levin@...izon.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
---
drivers/hv/channel.c | 21 +++++++++++++++++++--
1 file changed, 19 insertions(+), 2 deletions(-)
--- a/drivers/hv/channel.c
+++ b/drivers/hv/channel.c
@@ -28,6 +28,7 @@
#include <linux/module.h>
#include <linux/hyperv.h>
#include <linux/uio.h>
+#include <linux/interrupt.h>
#include "hyperv_vmbus.h"
@@ -496,8 +497,21 @@ static void reset_channel_cb(void *arg)
static int vmbus_close_internal(struct vmbus_channel *channel)
{
struct vmbus_channel_close_channel *msg;
+ struct tasklet_struct *tasklet;
int ret;
+ /*
+ * process_chn_event(), running in the tasklet, can race
+ * with vmbus_close_internal() in the case of SMP guest, e.g., when
+ * the former is accessing channel->inbound.ring_buffer, the latter
+ * could be freeing the ring_buffer pages.
+ *
+ * To resolve the race, we can serialize them by disabling the
+ * tasklet when the latter is running here.
+ */
+ tasklet = hv_context.event_dpc[channel->target_cpu];
+ tasklet_disable(tasklet);
+
channel->state = CHANNEL_OPEN_STATE;
channel->sc_creation_callback = NULL;
/* Stop callback and cancel the timer asap */
@@ -525,7 +539,7 @@ static int vmbus_close_internal(struct v
* If we failed to post the close msg,
* it is perhaps better to leak memory.
*/
- return ret;
+ goto out;
}
/* Tear down the gpadl for the channel's ring buffer */
@@ -538,7 +552,7 @@ static int vmbus_close_internal(struct v
* If we failed to teardown gpadl,
* it is perhaps better to leak memory.
*/
- return ret;
+ goto out;
}
}
@@ -555,6 +569,9 @@ static int vmbus_close_internal(struct v
if (channel->rescind)
hv_process_channel_removal(channel,
channel->offermsg.child_relid);
+out:
+ tasklet_enable(tasklet);
+
return ret;
}
Powered by blists - more mailing lists