MFV 314243

6676 Race between unique_insert() and unique_remove() causes ZFS fsid change

illumos/illumos-gate@40510e8eba18690b9a9843b26393725eeb0f1dac
https://github.com/illumos/illumos-gate/commit/40510e8eba18690b9a9843b26393725eeb0f1dac

https://www.illumos.org/issues/6676

  The fsid of zfs filesystems might change after reboot or remount. The problem seems to
  be caused by a race between unique_insert() and unique_remove(). The unique_remove()
  is called from dsl_dataset_evict() which is now an asynchronous thread. In a case the
  dsl_dataset_evict() thread is very slow and calls unique_remove() too late we will end
  up with changed fsid on zfs mount.

  This problem is very likely caused by #5056.

  Steps to Reproduce
  Note: I'm able to reproduce this always on a single core (virtual) machine. On multicore
  machines it is not so easy to reproduce.

# uname -a
SunOS openindiana 5.11 illumos-633aa80 i86pc i386 i86pc Solaris
# zfs create rpool/TEST
# FS=$(echo ::fsinfo | mdb -k | grep TEST | awk '{print $1}')
# echo $FS::print vfs_t vfs_fsid | mdb -k
vfs_fsid = {
    vfs_fsid.val = [ 0x54d7028a, 0x70311508 ]
}
# zfs umount rpool/TEST
# zfs mount rpool/TEST
# FS=$(echo ::fsinfo | mdb -k | grep TEST | awk '{print $1}')
# echo $FS::print vfs_t vfs_fsid | mdb -k
vfs_fsid = {
    vfs_fsid.val = [ 0xd9454e49, 0x6b36d08 ]
}
#

  Impact
  The persistent fsid (filesystem id) is essential for proper NFS functionality.
  If the fsid of a filesystem changes on remount (or after reboot) the NFS
  clients might not be able to automatically recover from such event and the
  manual remount of the NFS filesystems on every NFS client might be needed.

Author: Josef 'Jeff' Sipek <josef.sipek@nexenta.com>
Reviewed by: Saso Kiselkov <saso.kiselkov@nexenta.com>
Reviewed by: Sanjay Nadkarni <sanjay.nadkarni@nexenta.com>
Reviewed by: Dan Vatca <dan.vatca@gmail.com>
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: George Wilson <george.wilson@delphix.com>
Reviewed by: Sebastien Roy <sebastien.roy@delphix.com>
Approved by: Robert Mustacchi <rm@joyent.com>

1 files changed, 6 insertions, 5 deletions

diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zap.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zap.c
index 7050d6a341a2..e0bea807d589 100644
--- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zap.c
+++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zap.c
@@ -81,7 +81,8 @@ fzap_upgrade(zap_t *zap, dmu_tx_t *tx, zap_flags_t flags)
 	ASSERT(RW_WRITE_HELD(&zap->zap_rwlock));
 	zap->zap_ismicro = FALSE;
 
-	zap->zap_dbu.dbu_evict_func = zap_evict;
+	zap->zap_dbu.dbu_evict_func_sync = zap_evict_sync;
+	zap->zap_dbu.dbu_evict_func_async = NULL;
 
 	mutex_init(&zap->zap_f.zap_num_entries_mtx, 0, 0, 0);
 	zap->zap_f.zap_block_shift = highbit64(zap->zap_dbuf->db_size) - 1;
@@ -399,7 +400,7 @@ zap_allocate_blocks(zap_t *zap, int nblocks)
 }
 
 static void
-zap_leaf_pageout(void *dbu)
+zap_leaf_evict_sync(void *dbu)
 {
 	zap_leaf_t *l = dbu;
 
@@ -423,7 +424,7 @@ zap_create_leaf(zap_t *zap, dmu_tx_t *tx)
 	VERIFY(0 == dmu_buf_hold(zap->zap_objset, zap->zap_object,
 	    l->l_blkid << FZAP_BLOCK_SHIFT(zap), NULL, &l->l_dbuf,
 	    DMU_READ_NO_PREFETCH));
-	dmu_buf_init_user(&l->l_dbu, zap_leaf_pageout, &l->l_dbuf);
+	dmu_buf_init_user(&l->l_dbu, zap_leaf_evict_sync, NULL, &l->l_dbuf);
 	winner = dmu_buf_set_user(l->l_dbuf, &l->l_dbu);
 	ASSERT(winner == NULL);
 	dmu_buf_will_dirty(l->l_dbuf, tx);
@@ -470,13 +471,13 @@ zap_open_leaf(uint64_t blkid, dmu_buf_t *db)
 	l->l_bs = highbit64(db->db_size) - 1;
 	l->l_dbuf = db;
 
-	dmu_buf_init_user(&l->l_dbu, zap_leaf_pageout, &l->l_dbuf);
+	dmu_buf_init_user(&l->l_dbu, zap_leaf_evict_sync, NULL, &l->l_dbuf);
 	winner = dmu_buf_set_user(db, &l->l_dbu);
 
 	rw_exit(&l->l_rwlock);
 	if (winner != NULL) {
 		/* someone else set it first */
-		zap_leaf_pageout(&l->l_dbu);
+		zap_leaf_evict_sync(&l->l_dbu);
 		l = winner;
 	}