Fix deadlock in 'zfs rollback'

Currently, the 'zfs rollback' code can end up deadlocked due to
the way the kernel handles unreferenced inodes on a suspended fs.
Essentially, the zfs_resume_fs() code path may cause zfs to spawn
new threads as it reinstantiates the suspended fs's zil. When a
new thread is spawned, the kernel may attempt to free memory for
that thread by freeing some unreferenced inodes. If it happens to
select inodes that are a a part of the suspended fs a deadlock
will occur because freeing inodes requires holding the fs's
z_teardown_inactive_lock which is still held from the suspend.

This patch corrects this issue by adding an additional reference
to all inodes that are still present when a suspend is initiated.
This prevents them from being freed by the kernel for any reason.

Reviewed-by: Alek Pinchuk <[email protected]>
Reviewed-by: Brian Behlendorf <[email protected]>
Signed-off-by: Tom Caputi <[email protected]>
Closes #9203 

3 files changed, 17 insertions, 1 deletions

diff --git a/include/sys/zfs_znode.h b/include/sys/zfs_znode.h
index a0a3dd1ad..acaaf2884 100644
--- a/include/sys/zfs_znode.h
+++ b/include/sys/zfs_znode.h
@@ -200,6 +200,7 @@ typedef struct znode {
 	boolean_t	z_is_mapped;	/* are we mmap'ed */
 	boolean_t	z_is_ctldir;	/* are we .zfs entry */
 	boolean_t	z_is_stale;	/* are we stale due to rollback? */
+	boolean_t	z_suspended;	/* extra ref from a suspend? */
 	uint_t		z_blksz;	/* block size in bytes */
 	uint_t		z_seq;		/* modification sequence number */
 	uint64_t	z_mapcnt;	/* number of pages mapped to file */
diff --git a/module/zfs/zfs_vfsops.c b/module/zfs/zfs_vfsops.c
index af82c7bc4..34f4842d7 100644
--- a/module/zfs/zfs_vfsops.c
+++ b/module/zfs/zfs_vfsops.c
@@ -1737,7 +1737,12 @@ zfsvfs_teardown(zfsvfs_t *zfsvfs, boolean_t unmounting)
 	 * will fail with EIO since we have z_teardown_lock for writer (only
 	 * relevant for forced unmount).
 	 *
-	 * Release all holds on dbufs.
+	 * Release all holds on dbufs. We also grab an extra reference to all
+	 * the remaining inodes so that the kernel does not attempt to free
+	 * any inodes of a suspended fs. This can cause deadlocks since the
+	 * zfs_resume_fs() process may involve starting threads, which might
+	 * attempt to free unreferenced inodes to free up memory for the new
+	 * thread.
 	 */
 	if (!unmounting) {
 		mutex_enter(&zfsvfs->z_znodes_lock);
@@ -1745,6 +1750,9 @@ zfsvfs_teardown(zfsvfs_t *zfsvfs, boolean_t unmounting)
 		    zp = list_next(&zfsvfs->z_all_znodes, zp)) {
 			if (zp->z_sa_hdl)
 				zfs_znode_dmu_fini(zp);
+			if (igrab(ZTOI(zp)) != NULL)
+				zp->z_suspended = B_TRUE;
+
 		}
 		mutex_exit(&zfsvfs->z_znodes_lock);
 	}
@@ -2202,6 +2210,12 @@ zfs_resume_fs(zfsvfs_t *zfsvfs, dsl_dataset_t *ds)
 			remove_inode_hash(ZTOI(zp));
 			zp->z_is_stale = B_TRUE;
 		}
+
+		/* see comment in zfs_suspend_fs() */
+		if (zp->z_suspended) {
+			zfs_iput_async(ZTOI(zp));
+			zp->z_suspended = B_FALSE;
+		}
 	}
 	mutex_exit(&zfsvfs->z_znodes_lock);
 
diff --git a/module/zfs/zfs_znode.c b/module/zfs/zfs_znode.c
index 498547758..8512db9bc 100644
--- a/module/zfs/zfs_znode.c
+++ b/module/zfs/zfs_znode.c
@@ -545,6 +545,7 @@ zfs_znode_alloc(zfsvfs_t *zfsvfs, dmu_buf_t *db, int blksz,
 	zp->z_is_mapped = B_FALSE;
 	zp->z_is_ctldir = B_FALSE;
 	zp->z_is_stale = B_FALSE;
+	zp->z_suspended = B_FALSE;
 	zp->z_sa_hdl = NULL;
 	zp->z_mapcnt = 0;
 	zp->z_id = db->db_object;