[PATCH] xenbus: do not hold transaction_mutex when returning to userspace

[Ian Campbell-10]

  ================================================
  [ BUG: lock held when returning to user space! ]
  ------------------------------------------------
  xenstore-list/3522 is leaving the kernel with locks still held!
  1 lock held by xenstore-list/3522:
   #0:  (&xs_state.transaction_mutex){......}, at: [<c026dc6f>] xenbus_dev_request_and_reply+0x8f/0xa0

The canonical fix for this type of issue appears to be to maintain a
count manually rather than using an rwsem so do that here.

Signed-off-by: Ian Campbell <ian.campbell@...>
---
 drivers/xen/xenbus/xenbus_xs.c |   57 +++++++++++++++++++++++++++++++++-------
 1 files changed, 47 insertions(+), 10 deletions(-)

diff --git a/drivers/xen/xenbus/xenbus_xs.c b/drivers/xen/xenbus/xenbus_xs.c
index eab33f1..6f91e8c 100644
--- a/drivers/xen/xenbus/xenbus_xs.c
+++ b/drivers/xen/xenbus/xenbus_xs.c
@@ -76,6 +76,14 @@ struct xs_handle {
  /*
  * Mutex ordering: transaction_mutex -> watch_mutex -> request_mutex.
  * response_mutex is never taken simultaneously with the other three.
+ *
+ * transaction_mutex must be held before incrementing
+ * transaction_count. The mutex is held when a suspend is in
+ * progress to prevent new transactions starting.
+ *
+ * When decrementing transaction_count to zero the wait queue
+ * should be woken up, the suspend code waits for count to
+ * reach zero.
  */
 
  /* One request at a time. */
@@ -85,7 +93,9 @@ struct xs_handle {
  struct mutex response_mutex;
 
  /* Protect transactions against save/restore. */
- struct rw_semaphore transaction_mutex;
+ struct mutex transaction_mutex;
+ atomic_t transaction_count;
+ wait_queue_head_t transaction_wq;
 
  /* Protect watch (de)register against save/restore. */
  struct rw_semaphore watch_mutex;
@@ -157,6 +167,31 @@ static void *read_reply(enum xsd_sockmsg_type *type, unsigned int *len)
  return body;
 }
 
+static void transaction_start(void)
+{
+ mutex_lock(&xs_state.transaction_mutex);
+ atomic_inc(&xs_state.transaction_count);
+ mutex_unlock(&xs_state.transaction_mutex);
+}
+
+static void transaction_end(void)
+{
+ if (atomic_dec_and_test(&xs_state.transaction_count))
+ wake_up(&xs_state.transaction_wq);
+}
+
+static void transaction_suspend(void)
+{
+ mutex_lock(&xs_state.transaction_mutex);
+ wait_event(xs_state.transaction_wq,
+   atomic_read(&xs_state.transaction_count) == 0);
+}
+
+static void transaction_resume(void)
+{
+ mutex_unlock(&xs_state.transaction_mutex);
+}
+
 void *xenbus_dev_request_and_reply(struct xsd_sockmsg *msg)
 {
  void *ret;
@@ -164,7 +199,7 @@ void *xenbus_dev_request_and_reply(struct xsd_sockmsg *msg)
  int err;
 
  if (req_msg.type == XS_TRANSACTION_START)
- down_read(&xs_state.transaction_mutex);
+ transaction_start();
 
  mutex_lock(&xs_state.request_mutex);
 
@@ -180,7 +215,7 @@ void *xenbus_dev_request_and_reply(struct xsd_sockmsg *msg)
  if ((msg->type == XS_TRANSACTION_END) ||
     ((req_msg.type == XS_TRANSACTION_START) &&
      (msg->type == XS_ERROR)))
- up_read(&xs_state.transaction_mutex);
+ transaction_end();
 
  return ret;
 }
@@ -432,11 +467,11 @@ int xenbus_transaction_start(struct xenbus_transaction *t)
 {
  char *id_str;
 
- down_read(&xs_state.transaction_mutex);
+ transaction_start();
 
  id_str = xs_single(XBT_NIL, XS_TRANSACTION_START, "", NULL);
  if (IS_ERR(id_str)) {
- up_read(&xs_state.transaction_mutex);
+ transaction_end();
  return PTR_ERR(id_str);
  }
 
@@ -461,7 +496,7 @@ int xenbus_transaction_end(struct xenbus_transaction t, int abort)
 
  err = xs_error(xs_single(t, XS_TRANSACTION_END, abortstr, NULL));
 
- up_read(&xs_state.transaction_mutex);
+ transaction_end();
 
  return err;
 }
@@ -662,7 +697,7 @@ EXPORT_SYMBOL_GPL(unregister_xenbus_watch);
 
 void xs_suspend(void)
 {
- down_write(&xs_state.transaction_mutex);
+ transaction_suspend();
  down_write(&xs_state.watch_mutex);
  mutex_lock(&xs_state.request_mutex);
  mutex_lock(&xs_state.response_mutex);
@@ -677,7 +712,7 @@ void xs_resume(void)
 
  mutex_unlock(&xs_state.response_mutex);
  mutex_unlock(&xs_state.request_mutex);
- up_write(&xs_state.transaction_mutex);
+ transaction_resume();
 
  /* No need for watches_lock: the watch_mutex is sufficient. */
  list_for_each_entry(watch, &watches, list) {
@@ -693,7 +728,7 @@ void xs_suspend_cancel(void)
  mutex_unlock(&xs_state.response_mutex);
  mutex_unlock(&xs_state.request_mutex);
  up_write(&xs_state.watch_mutex);
- up_write(&xs_state.transaction_mutex);
+ mutex_unlock(&xs_state.transaction_mutex);
 }
 
 static int xenwatch_thread(void *unused)
@@ -843,8 +878,10 @@ int xs_init(void)
 
  mutex_init(&xs_state.request_mutex);
  mutex_init(&xs_state.response_mutex);
- init_rwsem(&xs_state.transaction_mutex);
+ mutex_init(&xs_state.transaction_mutex);
  init_rwsem(&xs_state.watch_mutex);
+ atomic_set(&xs_state.transaction_count, 0);
+ init_waitqueue_head(&xs_state.transaction_wq);
 
  /* Initialize the shared memory rings to talk to xenstored */
  err = xb_init_comms();
--
1.5.6.5


_______________________________________________
Xen-devel mailing list
Xen-devel@...
http://lists.xensource.com/xen-devel