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 "dn-" 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: #3500 Signed-off-by: Johann Lombardi <[email protected]> Signed-off-by: Jinshan Xiong <[email protected]> Change-Id: I899ff446cbf2aa7e355e7d98a83d614f1cd4624b
openzfs · Feb 13, 2016 · 76004c7 · 76004c7
1 parent eea9309
commit 76004c7
Show file tree

Hide file tree

Showing 18 changed files with 289 additions and 27 deletions.
diff --git a/cmd/zdb/zdb.c b/cmd/zdb/zdb.c
@@ -1937,11 +1937,13 @@ dump_object(objset_t *os, uint64_t object, int verbosity, int *print_header)
 	}
 
 	if (verbosity >= 4) {
-		(void) printf("\tdnode flags: %s%s%s\n",
+		(void) printf("\tdnode flags: %s%s%s%s\n",
 		    (dn->dn_phys->dn_flags & DNODE_FLAG_USED_BYTES) ?
 		    "USED_BYTES " : "",
 		    (dn->dn_phys->dn_flags & DNODE_FLAG_USERUSED_ACCOUNTED) ?
 		    "USERUSED_ACCOUNTED " : "",
+		    (dn->dn_phys->dn_flags & DNODE_FLAG_USEROBJUSED_ACCOUNTED) ?
+		    "USEROBJUSED_ACCOUNTED " : "",
 		    (dn->dn_phys->dn_flags & DNODE_FLAG_SPILL_BLKPTR) ?
 		    "SPILL_BLKPTR" : "");
 		(void) printf("\tdnode maxblkid: %llu\n",

diff --git a/cmd/zfs/zfs_main.c b/cmd/zfs/zfs_main.c
@@ -2817,11 +2817,15 @@ zfs_do_userspace(int argc, char **argv)
 		cb.cb_width[i] = strlen(gettext(us_field_hdr[i]));
 
 	for (p = 0; p < ZFS_NUM_USERQUOTA_PROPS; p++) {
-		if (((p == ZFS_PROP_USERUSED || p == ZFS_PROP_USERQUOTA) &&
+		if ((((p == ZFS_PROP_USERUSED || p == ZFS_PROP_USERQUOTA ||
+		    p == ZFS_PROP_USEROBJUSED || p == ZFS_PROP_USEROBJQUOTA) &&
 		    !(types & (USTYPE_PSX_USR | USTYPE_SMB_USR))) ||
-		    ((p == ZFS_PROP_GROUPUSED || p == ZFS_PROP_GROUPQUOTA) &&
-		    !(types & (USTYPE_PSX_GRP | USTYPE_SMB_GRP))))
+		    ((p == ZFS_PROP_GROUPUSED || p == ZFS_PROP_GROUPQUOTA ||
+		    p == ZFS_PROP_GROUPOBJUSED ||
+		    p == ZFS_PROP_GROUPOBJQUOTA) &&
+		    !(types & (USTYPE_PSX_GRP | USTYPE_SMB_GRP)))))
 			continue;
+
 		cb.cb_prop = p;
 		if ((ret = zfs_userspace(zhp, p, userspace_cb, &cb)) != 0)
 			return (ret);
@@ -3993,6 +3997,11 @@ zfs_do_receive(int argc, char **argv)
 #define	ZFS_DELEG_PERM_GROUPQUOTA	"groupquota"
 #define	ZFS_DELEG_PERM_USERUSED		"userused"
 #define	ZFS_DELEG_PERM_GROUPUSED	"groupused"
+#define	ZFS_DELEG_PERM_USEROBJQUOTA	"userobjquota"
+#define	ZFS_DELEG_PERM_GROUPOBJQUOTA	"groupobjquota"
+#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"
@@ -4023,6 +4032,10 @@ static zfs_deleg_perm_tab_t zfs_deleg_perm_tbl[] = {
 	{ ZFS_DELEG_PERM_USERPROP, ZFS_DELEG_NOTE_USERPROP },
 	{ ZFS_DELEG_PERM_USERQUOTA, ZFS_DELEG_NOTE_USERQUOTA },
 	{ ZFS_DELEG_PERM_USERUSED, ZFS_DELEG_NOTE_USERUSED },
+	{ ZFS_DELEG_PERM_USEROBJQUOTA, ZFS_DELEG_NOTE_USEROBJQUOTA },
+	{ ZFS_DELEG_PERM_USEROBJUSED, ZFS_DELEG_NOTE_USEROBJUSED },
+	{ ZFS_DELEG_PERM_GROUPOBJQUOTA, ZFS_DELEG_NOTE_GROUPOBJQUOTA },
+	{ ZFS_DELEG_PERM_GROUPOBJUSED, ZFS_DELEG_NOTE_GROUPOBJUSED },
 	{ NULL, ZFS_DELEG_NOTE_NONE }
 };
 
@@ -4100,6 +4113,10 @@ deleg_perm_type(zfs_deleg_note_t note)
 	case ZFS_DELEG_NOTE_USERPROP:
 	case ZFS_DELEG_NOTE_USERQUOTA:
 	case ZFS_DELEG_NOTE_USERUSED:
+	case ZFS_DELEG_NOTE_USEROBJQUOTA:
+	case ZFS_DELEG_NOTE_USEROBJUSED:
+	case ZFS_DELEG_NOTE_GROUPOBJQUOTA:
+	case ZFS_DELEG_NOTE_GROUPOBJUSED:
 		/* other */
 		return (gettext("other"));
 	default:
@@ -4603,6 +4620,19 @@ deleg_perm_comment(zfs_deleg_note_t note)
 	case ZFS_DELEG_NOTE_USERUSED:
 		str = gettext("Allows reading any userused@... property");
 		break;
+	case ZFS_DELEG_NOTE_USEROBJQUOTA:
+		str = gettext("Allows accessing any userobjquota@... property");
+		break;
+	case ZFS_DELEG_NOTE_GROUPOBJQUOTA:
+		str = gettext("Allows accessing any \n\t\t\t\t"
+		    "groupobjquota@... property");
+		break;
+	case ZFS_DELEG_NOTE_GROUPOBJUSED:
+		str = gettext("Allows reading any groupobjused@... property");
+		break;
+	case ZFS_DELEG_NOTE_USEROBJUSED:
+		str = gettext("Allows reading any userobjused@... property");
+		break;
 		/* other */
 	default:
 		str = "";

diff --git 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;
@@ -172,6 +173,9 @@ 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);
+int dmu_objset_userobjspace_upgrade(objset_t *os);
+boolean_t dmu_objset_userobjspace_present(objset_t *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
@@ -120,6 +120,9 @@ enum dnode_dirtycontext {
 /* Does dnode have a SA spill blkptr in bonus? */
 #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 */
 	uint8_t dn_indblkshift;		/* ln2(indirect block size) */

diff --git 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
@@ -164,6 +164,10 @@ 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;
 
@@ -885,6 +889,7 @@ typedef enum zfs_ioc {
 	ZFS_IOC_BOOKMARK,
 	ZFS_IOC_GET_BOOKMARKS,
 	ZFS_IOC_DESTROY_BOOKMARKS,
+	ZFS_IOC_USEROBJSPACE_UPGRADE,
 
 	/*
 	 * Linux - 3/64 numbers reserved.

diff --git 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 */

diff --git a/include/zfeature_common.h b/include/zfeature_common.h
@@ -50,6 +50,7 @@ typedef enum spa_feature {
 	SPA_FEATURE_BOOKMARKS,
 	SPA_FEATURE_FS_SS_LIMIT,
 	SPA_FEATURE_LARGE_BLOCKS,
+	SPA_FEATURE_USEROBJ_ACCOUNTING,
 	SPA_FEATURES
 } spa_feature_t;
 

diff --git 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,

diff --git a/lib/libzfs/libzfs_dataset.c b/lib/libzfs/libzfs_dataset.c
@@ -947,7 +947,9 @@ zfs_valid_proplist(libzfs_handle_t *hdl, zfs_type_t type, nvlist_t *nvl,
 			}
 
 			if (uqtype != ZFS_PROP_USERQUOTA &&
-			    uqtype != ZFS_PROP_GROUPQUOTA) {
+			    uqtype != ZFS_PROP_GROUPQUOTA &&
+			    uqtype != ZFS_PROP_USEROBJQUOTA &&
+			    uqtype != ZFS_PROP_GROUPOBJQUOTA) {
 				zfs_error_aux(hdl,
 				    dgettext(TEXT_DOMAIN, "'%s' is readonly"),
 				    propname);
@@ -2736,8 +2738,12 @@ userquota_propname_decode(const char *propname, boolean_t zoned,
 		return (EINVAL);
 	*typep = type;
 
-	isuser = (type == ZFS_PROP_USERQUOTA || type == ZFS_PROP_USERUSED);
-	isgroup = (type == ZFS_PROP_GROUPQUOTA || type == ZFS_PROP_GROUPUSED);
+	isuser = (type == ZFS_PROP_USERQUOTA || type == ZFS_PROP_USERUSED ||
+		    type == ZFS_PROP_USEROBJQUOTA ||
+		    type == ZFS_PROP_USEROBJUSED);
+	isgroup = (type == ZFS_PROP_GROUPQUOTA || type == ZFS_PROP_GROUPUSED ||
+		    type == ZFS_PROP_GROUPOBJQUOTA ||
+		    type == ZFS_PROP_GROUPOBJUSED);
 
 	cp = strchr(propname, '@') + 1;
 
@@ -2870,7 +2876,8 @@ zfs_prop_get_userquota(zfs_handle_t *zhp, const char *propname,
 		(void) snprintf(propbuf, proplen, "%llu",
 		    (u_longlong_t)propvalue);
 	} else if (propvalue == 0 &&
-	    (type == ZFS_PROP_USERQUOTA || type == ZFS_PROP_GROUPQUOTA)) {
+	    (type == ZFS_PROP_USERQUOTA || type == ZFS_PROP_GROUPQUOTA ||
+	    type == ZFS_PROP_USEROBJQUOTA || type == ZFS_PROP_GROUPOBJQUOTA)) {
 		(void) strlcpy(propbuf, "none", proplen);
 	} else {
 		zfs_nicenum(propvalue, propbuf, proplen);

diff --git a/man/man5/zpool-features.5 b/man/man5/zpool-features.5
@@ -432,5 +432,26 @@ set larger than 128KB, and will return to being \fBenabled\fR once all
 filesystems that have ever had their recordsize larger than 128KB are destroyed.
 .RE
 
+.sp
+.ne 2
+.na
+\fB\fBuserobj_accounting\fR\fR
+.ad
+.RS 4n
+.TS
+l l .
+GUID	org.zfsonlinux:user_object_accounting
+READ\-ONLY COMPATIBLE	yes
+DEPENDENCIES	none
+.TE
+
+This feature allows administrators to account the object usage information
+by user and group.
+
+This feature becomes \fBactive\fR as soon as it is enabled and will
+never return to being \fBenabled\fR.
+
+.RE
+
 .SH "SEE ALSO"
 \fBzpool\fR(8)
diff --git a/module/zcommon/zfs_deleg.c b/module/zcommon/zfs_deleg.c
@@ -62,6 +62,10 @@ zfs_deleg_perm_tab_t zfs_deleg_perm_tab[] = {
 	{ZFS_DELEG_PERM_GROUPQUOTA},
 	{ZFS_DELEG_PERM_USERUSED},
 	{ZFS_DELEG_PERM_GROUPUSED},
+	{ZFS_DELEG_PERM_USEROBJQUOTA},
+	{ZFS_DELEG_PERM_GROUPOBJQUOTA},
+	{ZFS_DELEG_PERM_USEROBJUSED},
+	{ZFS_DELEG_PERM_GROUPOBJUSED},
 	{ZFS_DELEG_PERM_HOLD},
 	{ZFS_DELEG_PERM_RELEASE},
 	{NULL}

diff --git a/module/zcommon/zfs_prop.c b/module/zcommon/zfs_prop.c
@@ -51,7 +51,11 @@ const char *zfs_userquota_prop_prefixes[] = {
 	"userused@",
 	"userquota@",
 	"groupused@",
-	"groupquota@"
+	"groupquota@",
+	"userobjused@",
+	"userobjquota@",
+	"groupobjused@",
+	"groupobjquota@"
 };
 
 zprop_desc_t *