d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

try to fix lor between z_teardown_lock and spa_namespace_lock

Browse Source 
The lock order reversal and a resulting deadlock were introduced
in r285021 / D2865.  The problem is that zfs_register_callbacks() calls
dsl_prop_get_integer() that has to acquire spa_namespace_lock.
At the same time, spa_config_sync() is called with spa_namespace_lock
held and then it performs ZFS vnode operations that acquire
z_teardown_lock in the reader mode.

So, fix the problem by using dsl_prop_get_int_ds() instead of
dsl_prop_get_integer().  The former does not need to look up
the pool and the dataset by name.

Reported by:	many
Reviewed by:	delphij
Tested by:	delphij, Jens Schweikhardt <schweikh@schweikhardt.net>
MFC after:	5 days
X-MFC with:	r285021
This commit is contained in:
Andriy Gapon 2015-08-21 08:17:44 +00:00
parent 23d1046817
commit 243f5e3085
1 changed files with 16 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

26
sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vfsops.c
View File

@ -471,6 +471,19 @@ zfs_register_callbacks(vfs_t *vfsp)
do_atime = B_TRUE;
}
/*
* We need to enter pool configuration here, so that we can use
* dsl_prop_get_int_ds() to handle the special nbmand property below.
* dsl_prop_get_integer() can not be used, because it has to acquire
* spa_namespace_lock and we can not do that because we already hold
* z_teardown_lock. The problem is that spa_config_sync() is called
* with spa_namespace_lock held and the function calls ZFS vnode
* operations to write the cache file and thus z_teardown_lock is
* acquired after spa_namespace_lock.
*/
ds = dmu_objset_ds(os);
dsl_pool_config_enter(dmu_objset_pool(os), FTAG);
/*
* nbmand is a special property. It can only be changed at
* mount time.
@ -482,14 +495,9 @@ zfs_register_callbacks(vfs_t *vfsp)
nbmand = B_FALSE;
} else if (vfs_optionisset(vfsp, MNTOPT_NBMAND, NULL)) {
nbmand = B_TRUE;
} else {
char osname[MAXNAMELEN];
dmu_objset_name(os, osname);
if (error = dsl_prop_get_integer(osname, "nbmand", &nbmand,
NULL)) {
return (error);
}
} else if (error = dsl_prop_get_int_ds(ds, "nbmand", &nbmand) != 0) {
dsl_pool_config_exit(dmu_objset_pool(os), FTAG);
return (error);
}
/*
@ -499,8 +507,6 @@ zfs_register_callbacks(vfs_t *vfsp)
* the first prop_register(), but I guess I like to go
* overboard...
*/
ds = dmu_objset_ds(os);
dsl_pool_config_enter(dmu_objset_pool(os), FTAG);
error = dsl_prop_register(ds,
zfs_prop_to_name(ZFS_PROP_ATIME), atime_changed_cb, zfsvfs);
error = error ? error : dsl_prop_register(ds,