Cherry-pick fix for zpl_mount deadlock

Signed-off-by: Stoiko Ivanov <s.ivanov@proxmox.com>
This commit is contained in:
Stoiko Ivanov 2018-07-16 11:29:24 +02:00 committed by Thomas Lamprecht
parent 50ad4ffe78
commit f3341ff73f
2 changed files with 93 additions and 0 deletions

View File

@ -0,0 +1,92 @@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Brian Behlendorf <behlendorf1@llnl.gov>
Date: Wed, 11 Jul 2018 15:49:10 -0700
Subject: [PATCH] Fix zpl_mount() deadlock
Commit 93b43af10 inadvertently introduced the following scenario which
can result in a deadlock. This issue was most easily reproduced by
LXD containers using a ZFS storage backend but should be reproducible
under any workload which is frequently mounting and unmounting.
-- THREAD A --
spa_sync()
spa_sync_upgrades()
rrw_enter(&dp->dp_config_rwlock, RW_WRITER, FTAG); <- Waiting on B
-- THREAD B --
mount_fs()
zpl_mount()
zpl_mount_impl()
dmu_objset_hold()
dmu_objset_hold_flags()
dsl_pool_hold()
dsl_pool_config_enter()
rrw_enter(&dp->dp_config_rwlock, RW_READER, tag);
sget()
sget_userns()
grab_super()
down_write(&s->s_umount); <- Waiting on C
-- THREAD C --
cleanup_mnt()
deactivate_super()
down_write(&s->s_umount);
deactivate_locked_super()
zpl_kill_sb()
kill_anon_super()
generic_shutdown_super()
sync_filesystem()
zpl_sync_fs()
zfs_sync()
zil_commit()
txg_wait_synced() <- Waiting on A
Reviewed by: Alek Pinchuk <apinchuk@datto.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #7598
Closes #7659
Closes #7691
Closes #7693
(Cherry-picked from ac09630d8b0bf6c92084a30fdaefd03fd0adbdc1)
Signed-off-by: Stoiko Ivanov <s.ivanov@proxmox.com>
---
include/sys/zfs_vfsops.h | 1 +
module/zfs/zpl_super.c | 11 ++++++++++-
2 files changed, 11 insertions(+), 1 deletion(-)
diff --git a/include/sys/zfs_vfsops.h b/include/sys/zfs_vfsops.h
index 2326da42..927153b2 100644
--- a/include/sys/zfs_vfsops.h
+++ b/include/sys/zfs_vfsops.h
@@ -32,6 +32,7 @@
#include <sys/zil.h>
#include <sys/sa.h>
#include <sys/rrwlock.h>
+#include <sys/dsl_dataset.h>
#include <sys/zfs_ioctl.h>
#ifdef __cplusplus
diff --git a/module/zfs/zpl_super.c b/module/zfs/zpl_super.c
index fc10271b..5c426b0a 100644
--- a/module/zfs/zpl_super.c
+++ b/module/zfs/zpl_super.c
@@ -271,8 +271,17 @@ zpl_mount_impl(struct file_system_type *fs_type, int flags, zfs_mnt_t *zm)
if (err)
return (ERR_PTR(-err));
+ /*
+ * The dsl pool lock must be released prior to calling sget().
+ * It is possible sget() may block on the lock in grab_super()
+ * while deactivate_super() holds that same lock and waits for
+ * a txg sync. If the dsl_pool lock is held over over sget()
+ * this can prevent the pool sync and cause a deadlock.
+ */
+ dsl_pool_rele(dmu_objset_pool(os), FTAG);
s = zpl_sget(fs_type, zpl_test_super, set_anon_super, flags, os);
- dmu_objset_rele(os, FTAG);
+ dsl_dataset_rele(dmu_objset_ds(os), FTAG);
+
if (IS_ERR(s))
return (ERR_CAST(s));

View File

@ -1,3 +1,4 @@
0001-remove-DKMS-modules-and-dracut-build.patch
0002-import-with-d-dev-disk-by-id-in-scan-service.patch
0003-always-load-ZFS-module-on-boot.patch
0004-Fix-zpl_mount-deadlock.patch