Add support for user/group dnode accounting & quota

This patch tracks dnode usage for each user/group in the
DMU_USER/GROUPUSED_OBJECT ZAPs. ZAP entries dedicated to dnode
accounting have the key prefixed with "obj-" followed by the UID/GID
in string format (as done for the block accounting).
A new SPA feature has been added for dnode accounting as well as
a new ZPL version. The SPA feature must be enabled in the pool
before upgrading the zfs filesystem. During the zfs version upgrade,
a "quotacheck" will be executed by marking all dnode as dirty.

ZoL-bug-id: https://github.com/zfsonlinux/zfs/issues/3500

Signed-off-by: Jinshan Xiong <[email protected]>
Signed-off-by: Johann Lombardi <[email protected]>

9 files changed, 52 insertions, 3 deletions

diff --git a/include/sys/dmu.h b/include/sys/dmu.h
index b67acb52c..ec89e9018 100644
--- a/include/sys/dmu.h
+++ b/include/sys/dmu.h
@@ -257,6 +257,12 @@ void zfs_znode_byteswap(void *buf, size_t size);
 #define	DMU_GROUPUSED_OBJECT	(-2ULL)
 
 /*
+ * Zap prefix for object accounting in DMU_{USER,GROUP}USED_OBJECT.
+ */
+#define	DMU_OBJACCT_PREFIX	"obj-"
+#define	DMU_OBJACCT_PREFIX_LEN	4
+
+/*
  * artificial blkids for bonus buffer and spill blocks
  */
 #define	DMU_BONUS_BLKID		(-1ULL)
diff --git a/include/sys/dmu_objset.h b/include/sys/dmu_objset.h
index 1674897c2..68fb5cffb 100644
--- a/include/sys/dmu_objset.h
+++ b/include/sys/dmu_objset.h
@@ -56,6 +56,7 @@ struct dmu_tx;
 	(arc_buf_size(buf) > OBJSET_OLD_PHYS_SIZE)
 
 #define	OBJSET_FLAG_USERACCOUNTING_COMPLETE	(1ULL<<0)
+#define	OBJSET_FLAG_USEROBJACCOUNTING_COMPLETE	(1ULL<<1)
 
 typedef struct objset_phys {
 	dnode_phys_t os_meta_dnode;
@@ -68,6 +69,8 @@ typedef struct objset_phys {
 	dnode_phys_t os_groupused_dnode;
 } objset_phys_t;
 
+typedef int (*dmu_objset_upgrade_cb_t)(objset_t *);
+
 struct objset {
 	/* Immutable: */
 	struct dsl_dataset *os_dsl_dataset;
@@ -125,6 +128,13 @@ struct objset {
 	kmutex_t os_user_ptr_lock;
 	void *os_user_ptr;
 	sa_os_t *os_sa;
+
+	/* kernel thread to upgrade this dataset */
+	kmutex_t os_upgrade_lock;
+	taskqid_t os_upgrade_id;
+	dmu_objset_upgrade_cb_t os_upgrade_cb;
+	boolean_t os_upgrade_exit;
+	int os_upgrade_status;
 };
 
 #define	DMU_META_OBJSET		0
@@ -173,6 +183,17 @@ void dmu_objset_userquota_get_ids(dnode_t *dn, boolean_t before, dmu_tx_t *tx);
 boolean_t dmu_objset_userused_enabled(objset_t *os);
 int dmu_objset_userspace_upgrade(objset_t *os);
 boolean_t dmu_objset_userspace_present(objset_t *os);
+boolean_t dmu_objset_userobjused_enabled(objset_t *os);
+void dmu_objset_userobjspace_upgrade(objset_t *os);
+boolean_t dmu_objset_userobjspace_present(objset_t *os);
+
+static inline boolean_t dmu_objset_userobjspace_upgradable(objset_t *os)
+{
+	return (dmu_objset_type(os) == DMU_OST_ZFS &&
+	    dmu_objset_userobjused_enabled(os) &&
+	    !dmu_objset_userobjspace_present(os));
+}
+
 int dmu_fsname(const char *snapname, char *buf);
 
 void dmu_objset_evict_done(objset_t *os);
diff --git a/include/sys/dnode.h b/include/sys/dnode.h
index fe36e5989..fe4cd3e4b 100644
--- a/include/sys/dnode.h
+++ b/include/sys/dnode.h
@@ -126,11 +126,14 @@ enum dnode_dirtycontext {
 };
 
 /* Is dn_used in bytes?  if not, it's in multiples of SPA_MINBLOCKSIZE */
-#define	DNODE_FLAG_USED_BYTES		(1<<0)
-#define	DNODE_FLAG_USERUSED_ACCOUNTED	(1<<1)
+#define	DNODE_FLAG_USED_BYTES			(1 << 0)
+#define	DNODE_FLAG_USERUSED_ACCOUNTED		(1 << 1)
 
 /* Does dnode have a SA spill blkptr in bonus? */
-#define	DNODE_FLAG_SPILL_BLKPTR	(1<<2)
+#define	DNODE_FLAG_SPILL_BLKPTR			(1 << 2)
+
+/* User/Group dnode accounting */
+#define	DNODE_FLAG_USEROBJUSED_ACCOUNTED	(1 << 3)
 
 typedef struct dnode_phys {
 	uint8_t dn_type;		/* dmu_object_type_t */
diff --git a/include/sys/dsl_deleg.h b/include/sys/dsl_deleg.h
index 59e8e0555..d399d1da9 100644
--- a/include/sys/dsl_deleg.h
+++ b/include/sys/dsl_deleg.h
@@ -51,8 +51,12 @@ extern "C" {
 #define	ZFS_DELEG_PERM_VSCAN		"vscan"
 #define	ZFS_DELEG_PERM_USERQUOTA	"userquota"
 #define	ZFS_DELEG_PERM_GROUPQUOTA	"groupquota"
+#define	ZFS_DELEG_PERM_USEROBJQUOTA	"userobjquota"
+#define	ZFS_DELEG_PERM_GROUPOBJQUOTA	"groupobjquota"
 #define	ZFS_DELEG_PERM_USERUSED		"userused"
 #define	ZFS_DELEG_PERM_GROUPUSED	"groupused"
+#define	ZFS_DELEG_PERM_USEROBJUSED	"userobjused"
+#define	ZFS_DELEG_PERM_GROUPOBJUSED	"groupobjused"
 #define	ZFS_DELEG_PERM_HOLD		"hold"
 #define	ZFS_DELEG_PERM_RELEASE		"release"
 #define	ZFS_DELEG_PERM_DIFF		"diff"
diff --git a/include/sys/fs/zfs.h b/include/sys/fs/zfs.h
index 9d3008934..5c93f53de 100644
--- a/include/sys/fs/zfs.h
+++ b/include/sys/fs/zfs.h
@@ -171,6 +171,10 @@ typedef enum {
 	ZFS_PROP_USERQUOTA,
 	ZFS_PROP_GROUPUSED,
 	ZFS_PROP_GROUPQUOTA,
+	ZFS_PROP_USEROBJUSED,
+	ZFS_PROP_USEROBJQUOTA,
+	ZFS_PROP_GROUPOBJUSED,
+	ZFS_PROP_GROUPOBJQUOTA,
 	ZFS_NUM_USERQUOTA_PROPS
 } zfs_userquota_prop_t;
 
diff --git a/include/sys/spa_impl.h b/include/sys/spa_impl.h
index 7b9e1ee0c..cb1d16ad5 100644
--- a/include/sys/spa_impl.h
+++ b/include/sys/spa_impl.h
@@ -277,6 +277,8 @@ struct spa {
 	 */
 	spa_config_lock_t spa_config_lock[SCL_LOCKS]; /* config changes */
 	refcount_t	spa_refcount;		/* number of opens */
+
+	taskq_t		*spa_upgrade_taskq;	/* taskq for upgrade jobs */
 };
 
 extern char *spa_config_path;
diff --git a/include/sys/zfs_vfsops.h b/include/sys/zfs_vfsops.h
index efaefdacc..b59ace5b1 100644
--- a/include/sys/zfs_vfsops.h
+++ b/include/sys/zfs_vfsops.h
@@ -110,6 +110,8 @@ typedef struct zfs_sb {
 	kmutex_t	z_lock;
 	uint64_t	z_userquota_obj;
 	uint64_t	z_groupquota_obj;
+	uint64_t	z_userobjquota_obj;
+	uint64_t	z_groupobjquota_obj;
 	uint64_t	z_replay_eof;	/* New end of file - replay only */
 	sa_attr_type_t	*z_attr_table;	/* SA attr mapping->id */
 	uint64_t	z_hold_size;	/* znode hold array size */
@@ -190,6 +192,8 @@ extern boolean_t zfs_owner_overquota(zfs_sb_t *zsb, struct znode *,
     boolean_t isgroup);
 extern boolean_t zfs_fuid_overquota(zfs_sb_t *zsb, boolean_t isgroup,
     uint64_t fuid);
+extern boolean_t zfs_fuid_overobjquota(zfs_sb_t *zsb, boolean_t isgroup,
+    uint64_t fuid);
 extern int zfs_set_version(zfs_sb_t *zsb, uint64_t newvers);
 extern int zfs_get_zplprop(objset_t *os, zfs_prop_t prop,
     uint64_t *value);
diff --git a/include/zfeature_common.h b/include/zfeature_common.h
index f05480181..acf76381b 100644
--- a/include/zfeature_common.h
+++ b/include/zfeature_common.h
@@ -54,6 +54,7 @@ typedef enum spa_feature {
 	SPA_FEATURE_SHA512,
 	SPA_FEATURE_SKEIN,
 	SPA_FEATURE_EDONR,
+	SPA_FEATURE_USEROBJ_ACCOUNTING,
 	SPA_FEATURES
 } spa_feature_t;
 
diff --git a/include/zfs_deleg.h b/include/zfs_deleg.h
index 16133c59f..95db9921f 100644
--- a/include/zfs_deleg.h
+++ b/include/zfs_deleg.h
@@ -63,6 +63,10 @@ typedef enum {
 	ZFS_DELEG_NOTE_GROUPQUOTA,
 	ZFS_DELEG_NOTE_USERUSED,
 	ZFS_DELEG_NOTE_GROUPUSED,
+	ZFS_DELEG_NOTE_USEROBJQUOTA,
+	ZFS_DELEG_NOTE_GROUPOBJQUOTA,
+	ZFS_DELEG_NOTE_USEROBJUSED,
+	ZFS_DELEG_NOTE_GROUPOBJUSED,
 	ZFS_DELEG_NOTE_HOLD,
 	ZFS_DELEG_NOTE_RELEASE,
 	ZFS_DELEG_NOTE_DIFF,