VFS: Integrate zfs_znode_alloc()

Under Linux the convention for filesystem specific data structure is
to embed it along with the generic vfs data structure.  This differs
significantly from Solaris.

Since we want to integrates as cleanly with the Linux VFS as possible.
This changes modifies zfs_znode_alloc() to allocate a znode with an
embedded inode for use with the generic VFS.  This is done by calling
iget_locked() which will allocate a new inode if needed by calling
sb->alloc_inode().  This function allocates enough memory for a
znode_t by returns a pointer to the inode structure for Linux's VFS.
This function is also responsible for setting the callback
znode->z_set_ops_inodes() which is used to register the correct
handlers for the inode.

include/sys/zfs_znode.h

@@ -211,6 +211,8 @@ typedef struct znode {
 	list_node_t	z_link_node;	/* all znodes in fs link */
 	sa_handle_t	*z_sa_hdl;	/* handle to sa data */
 	boolean_t	z_is_sa;	/* are we native sa? */
+	void (*z_set_ops_inode) (struct inode *); /* set inode ops */
+	struct inode	z_inode;	/* generic vfs inode */
 } znode_t;
 
 
@@ -231,13 +233,18 @@ typedef struct znode {
  */
 
 /*
- * Convert between znode pointers and vnode pointers
+ * Convert between znode pointers and inode pointers
  */
+#define	ZTOI(ZP)	(&((ZP)->z_inode))
+#define	ITOZ(IP)	(container_of((IP), znode_t, z_inode))
+
+/* XXX - REMOVE ME ONCE THE OTHER BUILD ISSUES ARE RESOLVED */
 #define	ZTOV(ZP)	((ZP)->z_vnode)
 #define	VTOZ(VP)	((znode_t *)(VP)->v_data)
 
+
 /*
- * ZFS_ENTER() is called on entry to each ZFS vnode and vfs operation.
+ * ZFS_ENTER() is called on entry to each ZFS inode and vfs operation.
  * ZFS_EXIT() must be called before exitting the vop.
  * ZFS_VERIFY_ZP() verifies the znode is valid.
  */

module/zfs/zfs_znode.c

@@ -425,7 +425,7 @@ zfs_znode_dmu_fini(znode_t *zp)
 }
 
 /*
- * Construct a new znode/vnode and intialize.
+ * Construct a new znode+inode and initialize.
  *
  * This does not do a call to dmu_set_user() that is
  * up to the caller to do, in case you don't want to
@@ -436,16 +436,19 @@ zfs_znode_alloc(zfsvfs_t *zfsvfs, dmu_buf_t *db, int blksz,
     dmu_object_type_t obj_type, sa_handle_t *hdl)
 {
 	znode_t	*zp;
-#ifdef HAVE_ZPL
+	struct inode *inode;
-	vnode_t *vp;
-	uint64_t mode;
 	uint64_t parent;
 	sa_bulk_attr_t bulk[9];
 	int count = 0;
-#endif /* HAVE_ZPL */
 
-	zp = kmem_cache_alloc(znode_cache, KM_SLEEP);
+	ASSERT(zfsvfs != NULL);
+	ASSERT(zfsvfs->z_vfs != NULL);
+	ASSERT(zfsvfs->z_vfs->mnt_sb != NULL);
 
+	inode = iget_locked(zfsvfs->z_vfs->mnt_sb, db->db_object);
+	zp = ITOZ(inode);
+
+	ASSERT(inode->i_state & I_NEW);
 	ASSERT(zp->z_dirlocks == NULL);
 	ASSERT(!POINTER_IS_VALID(zp->z_zfsvfs));
 	zp->z_moved = 0;
@@ -463,21 +466,20 @@ zfs_znode_alloc(zfsvfs_t *zfsvfs, dmu_buf_t *db, int blksz,
 	zp->z_seq = 0x7A4653;
 	zp->z_sync_cnt = 0;
 
-#ifdef HAVE_ZPL
-	vp = ZTOV(zp);
-	vn_reinit(vp);
-
 	zfs_znode_sa_init(zfsvfs, zp, db, obj_type, hdl);
 
-	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zfsvfs), NULL, &mode, 8);
+	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zfsvfs), NULL,
-	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GEN(zfsvfs), NULL, &zp->z_gen, 8);
+	    &zp->z_mode, 8);
+	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GEN(zfsvfs), NULL,
+	    &zp->z_gen, 8);
 	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zfsvfs), NULL,
 	    &zp->z_size, 8);
 	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_LINKS(zfsvfs), NULL,
 	    &zp->z_links, 8);
 	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs), NULL,
 	    &zp->z_pflags, 8);
-	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_PARENT(zfsvfs), NULL, &parent, 8);
+	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_PARENT(zfsvfs), NULL,
+	    &parent, 8);
 	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ATIME(zfsvfs), NULL,
 	    &zp->z_atime, 16);
 	SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zfsvfs), NULL,
@@ -488,57 +490,23 @@ zfs_znode_alloc(zfsvfs_t *zfsvfs, dmu_buf_t *db, int blksz,
 	if (sa_bulk_lookup(zp->z_sa_hdl, bulk, count) != 0 || zp->z_gen == 0) {
 		if (hdl == NULL)
 			sa_handle_destroy(zp->z_sa_hdl);
-		kmem_cache_free(znode_cache, zp);
+		iput(inode);
 		return (NULL);
 	}
 
-	zp->z_mode = mode;
+	inode->i_mode = (umode_t)zp->z_mode;
-	vp->v_vfsp = zfsvfs->z_parent->z_vfs;
+	if ((S_ISCHR(inode->i_mode)) || (S_ISBLK(inode->i_mode))) {
-
+		uint64_t rdev;
-	vp->v_type = IFTOVT((mode_t)mode);
+		VERIFY(sa_lookup(zp->z_sa_hdl, SA_ZPL_RDEV(zfsvfs),
-
+		    &rdev, sizeof (rdev)) == 0);
-	switch (vp->v_type) {
+		inode->i_rdev = zfs_cmpldev(rdev);
-	case VDIR:
-		if (zp->z_pflags & ZFS_XATTR) {
-			vn_setops(vp, zfs_xdvnodeops);
-			vp->v_flag |= V_XATTRDIR;
-		} else {
-			vn_setops(vp, zfs_dvnodeops);
-		}
-		zp->z_zn_prefetch = B_TRUE; /* z_prefetch default is enabled */
-		break;
-	case VBLK:
-	case VCHR:
-		{
-			uint64_t rdev;
-			VERIFY(sa_lookup(zp->z_sa_hdl, SA_ZPL_RDEV(zfsvfs),
-			    &rdev, sizeof (rdev)) == 0);
-
-			vp->v_rdev = zfs_cmpldev(rdev);
-		}
-		/*FALLTHROUGH*/
-	case VFIFO:
-	case VSOCK:
-	case VDOOR:
-		vn_setops(vp, zfs_fvnodeops);
-		break;
-	case VREG:
-		vp->v_flag |= VMODSORT;
-		if (parent == zfsvfs->z_shares_dir) {
-			ASSERT(zp->z_uid == 0 && zp->z_gid == 0);
-			vn_setops(vp, zfs_sharevnodeops);
-		} else {
-			vn_setops(vp, zfs_fvnodeops);
-		}
-		break;
-	case VLNK:
-		vn_setops(vp, zfs_symvnodeops);
-		break;
-	default:
-		vn_setops(vp, zfs_evnodeops);
-		break;
 	}
-#endif /* HAVE_ZPL */
+
+	/* zp->z_set_ops_inode() must be set in sb->alloc_inode() */
+	ASSERT(zp->z_set_ops_inode != NULL);
+	zp->z_set_ops_inode(inode);
+	unlock_new_inode(inode);
+
 	mutex_enter(&zfsvfs->z_znodes_lock);
 	list_insert_tail(&zfsvfs->z_all_znodes, zp);
 	membar_producer();