Fix zpool on zvol deadlock

[zfs.git] / module / zfs / zvol.c

diff --git a/module/zfs/zvol.c b/module/zfs/zvol.c
index 7a448f1..c6cfac2 100644 (file)
--- a/module/zfs/zvol.c
+++ b/module/zfs/zvol.c
@@ -891,25 +891,49 @@ zvol_first_open(zvol_state_t *zv)
 {
        objset_t *os;
        uint64_t volsize;
+       int locked = 0;
        int error;
        uint64_t ro;
 
+       /*
+        * In all other cases the spa_namespace_lock is taken before the
+        * bdev->bd_mutex lock.  But in this case the Linux __blkdev_get()
+        * function calls fops->open() with the bdev->bd_mutex lock held.
+        *
+        * To avoid a potential lock inversion deadlock we preemptively
+        * try to take the spa_namespace_lock().  Normally it will not
+        * be contended and this is safe because spa_open_common() handles
+        * the case where the caller already holds the spa_namespace_lock.
+        *
+        * When it is contended we risk a lock inversion if we were to
+        * block waiting for the lock.  Luckily, the __blkdev_get()
+        * function allows us to return -ERESTARTSYS which will result in
+        * bdev->bd_mutex being dropped, reacquired, and fops->open() being
+        * called again.  This process can be repeated safely until both
+        * locks are acquired.
+        */
+       if (!mutex_owned(&spa_namespace_lock)) {
+               locked = mutex_tryenter(&spa_namespace_lock);
+               if (!locked)
+                       return (-ERESTARTSYS);
+       }
+
        /* lie and say we're read-only */
        error = dmu_objset_own(zv->zv_name, DMU_OST_ZVOL, 1, zvol_tag, &os);
        if (error)
-               return (-error);
+               goto out_mutex;
 
        error = zap_lookup(os, ZVOL_ZAP_OBJ, "size", 8, 1, &volsize);
        if (error) {
-               dmu_objset_disown(os, zvol_tag);
-               return (-error);
+               dmu_objset_disown(os, zvol_tag);
+               goto out_mutex;
        }
 
        zv->zv_objset = os;
        error = dmu_bonus_hold(os, ZVOL_OBJ, zvol_tag, &zv->zv_dbuf);
        if (error) {
-               dmu_objset_disown(os, zvol_tag);
-               return (-error);
+               dmu_objset_disown(os, zvol_tag);
+               goto out_mutex;
        }
 
        set_capacity(zv->zv_disk, volsize >> 9);
@@ -918,13 +942,17 @@ zvol_first_open(zvol_state_t *zv)
 
        VERIFY(dsl_prop_get_integer(zv->zv_name, "readonly", &ro, NULL) == 0);
        if (ro || dmu_objset_is_snapshot(os)) {
-                set_disk_ro(zv->zv_disk, 1);
-               zv->zv_flags |= ZVOL_RDONLY;
+               set_disk_ro(zv->zv_disk, 1);
+               zv->zv_flags |= ZVOL_RDONLY;
        } else {
-                set_disk_ro(zv->zv_disk, 0);
-               zv->zv_flags &= ~ZVOL_RDONLY;
+               set_disk_ro(zv->zv_disk, 0);
+               zv->zv_flags &= ~ZVOL_RDONLY;
        }
 
+out_mutex:
+       if (locked)
+               mutex_exit(&spa_namespace_lock);
+
        return (-error);
 }