From: Gordon Ross <gwr@nexenta.com>
Date: Tue, 26 Jul 2011 18:37:06 +0000 (-0700)
Subject: Illumos #764: panic in zfs:dbuf_sync_list
X-Git-Tag: debian/0.7.9-2~588^2~3^2~8
X-Git-Url: https://git.proxmox.com/?a=commitdiff_plain;h=ef3c1dea7024b07b4ace6115de9f22a99c1394d8;p=mirror_zfs-debian.git

Illumos #764: panic in zfs:dbuf_sync_list

Hypothesis about what's going on here.

At some time in the past, something, i.e. dnode_reallocate()
calls one of:
dbuf_rm_spill(dn, tx);

These will do:
dbuf_rm_spill(dnode_t *dn, dmu_tx_t *tx)
dbuf_free_range(dn, DMU_SPILL_BLKID, DMU_SPILL_BLKID, tx)
dbuf_undirty(db, tx)

Currently dbuf_undirty can leave a spill block in dn_dirty_records[],
(it having been put there previously by dbuf_dirty) and free it.
Sometime later, dbuf_sync_list trips over this reference to free'd
(and typically reused) memory.

Also, dbuf_undirty can call dnode_clear_range with a bogus
block ID. It needs to test for DMU_SPILL_BLKID, similar to
how dnode_clear_range is called in dbuf_dirty().

References to Illumos issue and patch:
- https://www.illumos.org/issues/764
- https://github.com/illumos/illumos-gate/commit/3f2366c2bb

Reviewed by: George Wilson <gwilson@zfsmail.com>
Reviewed by: Mark.Maybe@oracle.com
Reviewed by: Albert Lee <trisk@nexenta.com
Approved by: Garrett D'Amore <garrett@nexenta.com>

Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #340
---

diff --git a/module/zfs/dbuf.c b/module/zfs/dbuf.c
index e166c81d..34ce2f62 100644
--- a/module/zfs/dbuf.c
+++ b/module/zfs/dbuf.c
@@ -20,6 +20,7 @@
  */
 /*
  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Copyright 2011 Nexenta Systems, Inc.  All rights reserved.
  */
 
 #include <sys/zfs_context.h>
@@ -1347,13 +1348,17 @@ dbuf_undirty(dmu_buf_impl_t *db, dmu_tx_t *tx)
 	 * it, since one of the current holders may be in the
 	 * middle of an update.  Note that users of dbuf_undirty()
 	 * should not place a hold on the dbuf before the call.
+	 * Also note: we can get here with a spill block, so
+	 * test for that similar to how dbuf_dirty does.
 	 */
 	if (refcount_count(&db->db_holds) > db->db_dirtycnt) {
 		mutex_exit(&db->db_mtx);
 		/* Make sure we don't toss this buffer at sync phase */
-		mutex_enter(&dn->dn_mtx);
-		dnode_clear_range(dn, db->db_blkid, 1, tx);
-		mutex_exit(&dn->dn_mtx);
+		if (db->db_blkid != DMU_SPILL_BLKID) {
+			mutex_enter(&dn->dn_mtx);
+			dnode_clear_range(dn, db->db_blkid, 1, tx);
+			mutex_exit(&dn->dn_mtx);
+		}
 		DB_DNODE_EXIT(db);
 		return (0);
 	}
@@ -1366,11 +1371,18 @@ dbuf_undirty(dmu_buf_impl_t *db, dmu_tx_t *tx)
 
 	*drp = dr->dr_next;
 
+	/*
+	 * Note that there are three places in dbuf_dirty()
+	 * where this dirty record may be put on a list.
+	 * Make sure to do a list_remove corresponding to
+	 * every one of those list_insert calls.
+	 */
 	if (dr->dr_parent) {
 		mutex_enter(&dr->dr_parent->dt.di.dr_mtx);
 		list_remove(&dr->dr_parent->dt.di.dr_children, dr);
 		mutex_exit(&dr->dr_parent->dt.di.dr_mtx);
-	} else if (db->db_level+1 == dn->dn_nlevels) {
+	} else if (db->db_blkid == DMU_SPILL_BLKID ||
+	    db->db_level+1 == dn->dn_nlevels) {
 		ASSERT(db->db_blkptr == NULL || db->db_parent == dn->dn_dbuf);
 		mutex_enter(&dn->dn_mtx);
 		list_remove(&dn->dn_dirty_records[txg & TXG_MASK], dr);