Support for longnames for files/directories (Linux part)

This patch adds the ability for zfs to support file/dir name up to 1023
bytes. This number is chosen so we can support up to 255 4-byte
characters. This new feature is represented by the new feature flag
feature@longname.

A new dataset property "longname" is also introduced to toggle longname
support for each dataset individually. This property can be disabled,
even if it contains longname files. In such case, new file cannot be
created with longname but existing longname files can still be looked
up.

Note that, to my knowledge native Linux filesystems don't support name
longer than 255 bytes. So there might be programs not able to work with
longname.

Note that NFS server may needs to use exportfs_get_name to reconnect
dentries, and the buffer being passed is limit to NAME_MAX+1 (256). So
NFS may not work when longname is enabled.

Note, FreeBSD vfs layer imposes a limit of 255 name lengh, so even
though we add code to support it here, it won't actually work.

Reviewed-by: Tony Hutter <[email protected]>
Reviewed-by: Brian Behlendorf <[email protected]>
Reviewed-by: Alexander Motin <[email protected]>
Signed-off-by: Chunwei Chen <[email protected]>
Closes openzfs#15921

Author: sanjeevbagewadinutanix

cmd/zdb/zdb.c

@@ -1117,7 +1117,7 @@ dump_zap(objset_t *os, uint64_t object, void *data, size_t size)
 {
 	(void) data, (void) size;
 	zap_cursor_t zc;
-	zap_attribute_t *attrp = zap_attribute_alloc();
+	zap_attribute_t *attrp = zap_attribute_long_alloc();
 	void *prop;
 	unsigned i;
 
@@ -1365,7 +1365,7 @@ dump_zpldir(objset_t *os, uint64_t object, void *data, size_t size)
 {
 	(void) data, (void) size;
 	zap_cursor_t zc;
-	zap_attribute_t *attrp = zap_attribute_alloc();
+	zap_attribute_t *attrp = zap_attribute_long_alloc();
 	const char *typenames[] = {
 		/* 0 */ "not specified",
 		/* 1 */ "FIFO",

cmd/zhack.c

@@ -203,7 +203,7 @@ static void
 dump_obj(objset_t *os, uint64_t obj, const char *name)
 {
 	zap_cursor_t zc;
-	zap_attribute_t *za = zap_attribute_alloc();
+	zap_attribute_t *za = zap_attribute_long_alloc();
 
 	(void) printf("%s_obj:\n", name);
 

include/os/freebsd/zfs/sys/zfs_vfsops_os.h

@@ -83,6 +83,7 @@ struct zfsvfs {
 	boolean_t	z_use_sa;	/* version allow system attributes */
 	boolean_t	z_xattr_sa;	/* allow xattrs to be stores as SA */
 	boolean_t	z_use_namecache; /* make use of FreeBSD name cache */
+	boolean_t	z_longname;	/* Dataset supports long names */
 	uint8_t		z_xattr;	/* xattr type in use */
 	uint64_t	z_version;	/* ZPL version */
 	uint64_t	z_shares_dir;	/* hidden shares dir */

include/os/freebsd/zfs/sys/zfs_znode_impl.h

@@ -179,7 +179,7 @@ extern void zfs_znode_free(struct znode *);
 extern zil_replay_func_t *const zfs_replay_vector[TX_MAX_TYPE];
 
 extern int zfs_znode_parent_and_name(struct znode *zp, struct znode **dzpp,
-    char *buf);
+    char *buf, uint64_t buflen);
 
 extern int zfs_rlimit_fsize(off_t fsize);
 #ifdef	__cplusplus

include/os/linux/zfs/sys/zfs_vfsops_os.h

@@ -118,6 +118,7 @@ struct zfsvfs {
 	boolean_t	z_xattr_sa;	/* allow xattrs to be stores as SA */
 	boolean_t	z_draining;	/* is true when drain is active */
 	boolean_t	z_drain_cancel; /* signal the unlinked drain to stop */
+	boolean_t	z_longname;	/* Dataset supports long names */
 	uint64_t	z_version;	/* ZPL version */
 	uint64_t	z_shares_dir;	/* hidden shares dir */
 	dataset_kstats_t	z_kstat;	/* fs kstats */

include/os/linux/zfs/sys/zfs_vnops_os.h

@@ -44,6 +44,7 @@ extern int zfs_write_simple(znode_t *zp, const void *data, size_t len,
     loff_t pos, size_t *resid);
 extern int zfs_lookup(znode_t *dzp, char *nm, znode_t **zpp, int flags,
     cred_t *cr, int *direntflags, pathname_t *realpnp);
+extern int zfs_get_name(znode_t *dzp, char *name, znode_t *zp);
 extern int zfs_create(znode_t *dzp, char *name, vattr_t *vap, int excl,
     int mode, znode_t **zpp, cred_t *cr, int flag, vsecattr_t *vsecp,
     zidmap_t *mnt_ns);

include/sys/fs/zfs.h

@@ -81,6 +81,7 @@ typedef enum dmu_objset_type {
  * All of these include the terminating NUL byte.
  */
 #define	ZAP_MAXNAMELEN 256
+#define	ZAP_MAXNAMELEN_NEW 1024
 #define	ZAP_MAXVALUELEN (1024 * 8)
 #define	ZAP_OLDMAXVALUELEN 1024
 #define	ZFS_MAX_DATASET_NAME_LEN 256
@@ -194,6 +195,7 @@ typedef enum {
 	ZFS_PROP_PREFETCH,
 	ZFS_PROP_VOLTHREADING,
 	ZFS_PROP_DIRECT,
+	ZFS_PROP_LONGNAME,
 	ZFS_NUM_PROPS
 } zfs_prop_t;
 

include/sys/zap.h

@@ -315,7 +315,7 @@ int zap_count(objset_t *ds, uint64_t zapobj, uint64_t *count);
  * match must be exact (ie, same as mask=-1ULL).
  */
 int zap_value_search(objset_t *os, uint64_t zapobj,
-    uint64_t value, uint64_t mask, char *name);
+    uint64_t value, uint64_t mask, char *name, uint64_t namelen);
 
 /*
  * Transfer all the entries from fromobj into intoobj.  Only works on
@@ -387,6 +387,7 @@ void zap_fini(void);
  * Alloc and free zap_attribute_t.
  */
 zap_attribute_t *zap_attribute_alloc(void);
+zap_attribute_t *zap_attribute_long_alloc(void);
 void zap_attribute_free(zap_attribute_t *attrp);
 
 /*

include/sys/zfs_ioctl.h

@@ -143,7 +143,7 @@ typedef enum drr_headertype {
  * default use of "zfs send" won't encounter the bug mentioned above.
  */
 #define	DMU_BACKUP_FEATURE_SWITCH_TO_LARGE_BLOCKS (1 << 27)
-/* flag #28 is reserved for a Nutanix feature */
+#define	DMU_BACKUP_FEATURE_LONGNAME		(1 << 28)
 
 /*
  * Mask of all supported backup features
@@ -154,7 +154,7 @@ typedef enum drr_headertype {
     DMU_BACKUP_FEATURE_COMPRESSED | DMU_BACKUP_FEATURE_LARGE_DNODE | \
     DMU_BACKUP_FEATURE_RAW | DMU_BACKUP_FEATURE_HOLDS | \
     DMU_BACKUP_FEATURE_REDACTED | DMU_BACKUP_FEATURE_SWITCH_TO_LARGE_BLOCKS | \
-    DMU_BACKUP_FEATURE_ZSTD)
+    DMU_BACKUP_FEATURE_ZSTD | DMU_BACKUP_FEATURE_LONGNAME)
 
 /* Are all features in the given flag word currently supported? */
 #define	DMU_STREAM_SUPPORTED(x)	(!((x) & ~DMU_BACKUP_FEATURE_MASK))

include/zfeature_common.h

@@ -83,6 +83,7 @@ typedef enum spa_feature {
 	SPA_FEATURE_REDACTION_LIST_SPILL,
 	SPA_FEATURE_RAIDZ_EXPANSION,
 	SPA_FEATURE_FAST_DEDUP,
+	SPA_FEATURE_LONGNAME,
 	SPA_FEATURES
 } spa_feature_t;