Stable API extensions

This introduces a new ZFS_IOC_LIBZFS_CORE ioctl and replaces the ioctl
calls in all but 1 function in libzfs_core.so and adds many more. It is
intended to be as stable kernel/userland interface that allows
basic userland functionality to continue working when an older userland
is used with newer kernel modules or vice versa by mistake.

At this time, it is a subset of the legacy functionality, but enough API
functions have been implemented that mismatched kernel/userland code
will not break the Gentoo's boot process (excluding zfs share and ZED,
which are excluded at this time) and ClusterHQ's flocker.

This sufficiently addresses the long standing /dev/zfs API instability
that it opens the possibility of stabilization of the ZFS packaging in
Gentoo on amd64 and other architectures, which makes this an improvement
in more ways than one.

To ease the transition to the new ABI, we include a shim library that
implements the new C library extensions using the old kernel ABI. The
--with-libzfs_core=default|all|legacy|stable configure switch determines
which is used and presently, the default is an alias for legacy that can
be changed in a future release. --with-libzfs_core=all will install
both, with libzfs_core.so.1 using the new ABI while a
libzfs_core-legacy.so.0 that can be loaded with LD_PRELOAD will be
installed containing the old one.

Signed-off-by: Richard Yao <ryao@gentoo.org>

Author: ryao

cmd/mount_zfs/Makefile.am

@@ -19,4 +19,4 @@ mount_zfs_LDADD = \
 	$(top_builddir)/lib/libuutil/libuutil.la \
 	$(top_builddir)/lib/libzpool/libzpool.la \
 	$(top_builddir)/lib/libzfs/libzfs.la \
-	$(top_builddir)/lib/libzfs_core/libzfs_core.la
+	$(top_builddir)/lib/libzfs_core-$(LIBZFS_CORE_VARIANT)/libzfs_core-$(LIBZFS_CORE_VARIANT).la

cmd/zdb/Makefile.am

@@ -17,4 +17,4 @@ zdb_LDADD = \
 	$(top_builddir)/lib/libuutil/libuutil.la \
 	$(top_builddir)/lib/libzpool/libzpool.la \
 	$(top_builddir)/lib/libzfs/libzfs.la \
-	$(top_builddir)/lib/libzfs_core/libzfs_core.la
+	$(top_builddir)/lib/libzfs_core-$(LIBZFS_CORE_VARIANT)/libzfs_core-$(LIBZFS_CORE_VARIANT).la

cmd/zed/Makefile.am

@@ -30,7 +30,7 @@ zed_LDADD = \
 	$(top_builddir)/lib/libspl/libspl.la \
 	$(top_builddir)/lib/libzpool/libzpool.la \
 	$(top_builddir)/lib/libzfs/libzfs.la \
-	$(top_builddir)/lib/libzfs_core/libzfs_core.la
+	$(top_builddir)/lib/libzfs_core-$(LIBZFS_CORE_VARIANT)/libzfs_core-$(LIBZFS_CORE_VARIANT).la
 
 zedconfdir = $(sysconfdir)/zfs/zed.d
 

cmd/zfs/Makefile.am

@@ -17,6 +17,6 @@ zfs_LDADD = \
 	$(top_builddir)/lib/libuutil/libuutil.la \
 	$(top_builddir)/lib/libzpool/libzpool.la \
 	$(top_builddir)/lib/libzfs/libzfs.la \
-	$(top_builddir)/lib/libzfs_core/libzfs_core.la
+	$(top_builddir)/lib/libzfs_core-$(LIBZFS_CORE_VARIANT)/libzfs_core-$(LIBZFS_CORE_VARIANT).la
 
 zfs_LDFLAGS = -pthread

cmd/zfs/zfs_iter.c

@@ -68,82 +68,37 @@ typedef struct callback_data {
 
 uu_avl_pool_t *avl_pool;
 
-/*
- * Include snaps if they were requested or if this a zfs list where types
- * were not specified and the "listsnapshots" property is set on this pool.
- */
-static boolean_t
-zfs_include_snapshots(zfs_handle_t *zhp, callback_data_t *cb)
-{
-	zpool_handle_t *zph;
-
-	if ((cb->cb_flags & ZFS_ITER_PROP_LISTSNAPS) == 0)
-		return (cb->cb_types & ZFS_TYPE_SNAPSHOT);
-
-	zph = zfs_get_pool_handle(zhp);
-	return (zpool_get_prop_int(zph, ZPOOL_PROP_LISTSNAPS, NULL));
-}
-
-/*
- * Called for each dataset.  If the object is of an appropriate type,
- * add it to the avl tree and recurse over any children as necessary.
- */
+/* Called for each dataset. */
 static int
 zfs_callback(zfs_handle_t *zhp, void *data)
 {
 	callback_data_t *cb = data;
 	boolean_t dontclose = B_FALSE;
-	boolean_t include_snaps = zfs_include_snapshots(zhp, cb);
-	boolean_t include_bmarks = (cb->cb_types & ZFS_TYPE_BOOKMARK);
-
-	if ((zfs_get_type(zhp) & cb->cb_types) ||
-	    ((zfs_get_type(zhp) == ZFS_TYPE_SNAPSHOT) && include_snaps)) {
-		uu_avl_index_t idx;
-		zfs_node_t *node = safe_malloc(sizeof (zfs_node_t));
-
-		node->zn_handle = zhp;
-		uu_avl_node_init(node, &node->zn_avlnode, avl_pool);
-		if (uu_avl_find(cb->cb_avl, node, cb->cb_sortcol,
-		    &idx) == NULL) {
-			if (cb->cb_proplist) {
-				if ((*cb->cb_proplist) &&
-				    !(*cb->cb_proplist)->pl_all)
-					zfs_prune_proplist(zhp,
-					    cb->cb_props_table);
-
-				if (zfs_expand_proplist(zhp, cb->cb_proplist,
-				    (cb->cb_flags & ZFS_ITER_RECVD_PROPS),
-				    (cb->cb_flags & ZFS_ITER_LITERAL_PROPS))
-				    != 0) {
-					free(node);
-					return (-1);
-				}
+	uu_avl_index_t idx;
+	zfs_node_t *node = safe_malloc(sizeof (zfs_node_t));
+
+	node->zn_handle = zhp;
+	uu_avl_node_init(node, &node->zn_avlnode, avl_pool);
+	if (uu_avl_find(cb->cb_avl, node, cb->cb_sortcol,
+	    &idx) == NULL) {
+		if (cb->cb_proplist) {
+			if ((*cb->cb_proplist) &&
+			    !(*cb->cb_proplist)->pl_all)
+				zfs_prune_proplist(zhp,
+				    cb->cb_props_table);
+
+			if (zfs_expand_proplist(zhp, cb->cb_proplist,
+			    (cb->cb_flags & ZFS_ITER_RECVD_PROPS),
+			    (cb->cb_flags & ZFS_ITER_LITERAL_PROPS))
+			    != 0) {
+				free(node);
+				return (-1);
 			}
-			uu_avl_insert(cb->cb_avl, node, idx);
-			dontclose = B_TRUE;
-		} else {
-			free(node);
 		}
-	}
-
-	/*
-	 * Recurse if necessary.
-	 */
-	if (cb->cb_flags & ZFS_ITER_RECURSE &&
-	    ((cb->cb_flags & ZFS_ITER_DEPTH_LIMIT) == 0 ||
-	    cb->cb_depth < cb->cb_depth_limit)) {
-		cb->cb_depth++;
-		if (zfs_get_type(zhp) == ZFS_TYPE_FILESYSTEM)
-			(void) zfs_iter_filesystems(zhp, zfs_callback, data);
-		if (((zfs_get_type(zhp) & (ZFS_TYPE_SNAPSHOT |
-		    ZFS_TYPE_BOOKMARK)) == 0) && include_snaps)
-			(void) zfs_iter_snapshots(zhp,
-			    (cb->cb_flags & ZFS_ITER_SIMPLE) != 0, zfs_callback,
-			    data);
-		if (((zfs_get_type(zhp) & (ZFS_TYPE_SNAPSHOT |
-		    ZFS_TYPE_BOOKMARK)) == 0) && include_bmarks)
-			(void) zfs_iter_bookmarks(zhp, zfs_callback, data);
-		cb->cb_depth--;
+		uu_avl_insert(cb->cb_avl, node, idx);
+		dontclose = B_TRUE;
+	} else {
+		free(node);
 	}
 
 	if (!dontclose)
@@ -378,6 +333,9 @@ zfs_for_each(int argc, char **argv, int flags, zfs_type_t types,
 	int ret = 0;
 	zfs_node_t *node;
 	uu_avl_walk_t *walk;
+	zfs_type_t argtype;
+	zfs_type_t opentype;
+	boolean_t limit_specified = !!(flags & ZFS_ITER_DEPTH_LIMIT);
 
 	avl_pool = uu_avl_pool_create("zfs_pool", sizeof (zfs_node_t),
 	    offsetof(zfs_node_t, zn_avlnode), zfs_sort, UU_DEFAULT);
@@ -386,6 +344,10 @@ zfs_for_each(int argc, char **argv, int flags, zfs_type_t types,
 		nomem();
 
 	cb.cb_sortcol = sortcol;
+	/*
+	 * XXX: We are phasing out the legacy recursive interface in
+	 * favor of the new stable list API.
+	 */
 	cb.cb_flags = flags;
 	cb.cb_proplist = proplist;
 	cb.cb_types = types;
@@ -431,39 +393,52 @@ zfs_for_each(int argc, char **argv, int flags, zfs_type_t types,
 	if ((cb.cb_avl = uu_avl_create(avl_pool, NULL, UU_DEFAULT)) == NULL)
 		nomem();
 
+	/*
+	 * zfs_iter_generic() lets the kernel worry about default types.
+	 */
+	argtype = types * !!(flags & ZFS_ITER_TYPES_SPECIFIED);
+	opentype = argtype;
+	if ((flags & ZFS_ITER_RECURSE) != 0 || limit_specified)
+		opentype |= ZFS_TYPE_FILESYSTEM | ZFS_TYPE_VOLUME;
 	if (argc == 0) {
 		/*
 		 * If given no arguments, iterate over all datasets.
 		 */
-		cb.cb_flags |= ZFS_ITER_RECURSE;
-		ret = zfs_iter_root(g_zfs, zfs_callback, &cb);
+		ret = zfs_iter_generic(g_zfs, NULL, argtype, 0,
+		    (limit_specified) ? limit : -1, limit_specified,
+		    zfs_callback, &cb);
 	} else {
 		int i;
 		zfs_handle_t *zhp;
-		zfs_type_t argtype;
-
-		/*
-		 * If we're recursive, then we always allow filesystems as
-		 * arguments.  If we also are interested in snapshots, then we
-		 * can take volumes as well.
-		 */
-		argtype = types;
-		if (flags & ZFS_ITER_RECURSE) {
-			argtype |= ZFS_TYPE_FILESYSTEM;
-			if (types & ZFS_TYPE_SNAPSHOT)
-				argtype |= ZFS_TYPE_VOLUME;
-		}
 
 		for (i = 0; i < argc; i++) {
+			/*
+			 * Special case for bookmarks
+			 */
+			if (argv[i][0] != '/' &&
+			    strncmp(argv[i], "./", strlen("./")) != 0 &&
+			    strchr(argv[i], '#') != NULL) {
+				ret = zfs_iter_generic(g_zfs, argv[i], argtype,
+				    0, 0, B_FALSE, zfs_callback, &cb);
+				continue;
+			}
+
 			if (flags & ZFS_ITER_ARGS_CAN_BE_PATHS) {
 				zhp = zfs_path_to_zhandle(g_zfs, argv[i],
-				    argtype);
+				    opentype);
 			} else {
-				zhp = zfs_open(g_zfs, argv[i], argtype);
+				zhp = zfs_open(g_zfs, argv[i],
+				    opentype);
 			}
-			if (zhp != NULL)
-				ret |= zfs_callback(zhp, &cb);
-			else
+			if (zhp != NULL) {
+				ret |= zfs_iter_generic(zfs_get_handle(zhp),
+				    zfs_get_name(zhp), argtype, 0,
+				    (limit_specified) ? limit : (flags &
+				    ZFS_ITER_RECURSE) ? -1 : 0,
+				    !(flags & ZFS_ITER_RECURSE), zfs_callback,
+				    &cb);
+				zfs_close(zhp);
+			} else
 				ret = 1;
 		}
 	}

cmd/zfs/zfs_iter.h

@@ -47,6 +47,7 @@ typedef struct zfs_sort_column {
 #define	ZFS_ITER_RECVD_PROPS	   (1 << 4)
 #define	ZFS_ITER_LITERAL_PROPS	   (1 << 5)
 #define	ZFS_ITER_SIMPLE		   (1 << 6)
+#define	ZFS_ITER_TYPES_SPECIFIED   (1 << 7)
 
 int zfs_for_each(int, char **, int options, zfs_type_t,
     zfs_sort_column_t *, zprop_list_t **, int, zfs_iter_f, void *);