blob: 270b3195560c497aef945fe5aee67f8c968adc5a [file] [log] [blame]
#ifndef __DBFS_H__
#define __DBFS_H__
* Copyright 2006 Jeff Garzik <>
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation.
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* GNU General Public License for more details.
* You should have received a copy of the GNU General Public License
* along with this program; see the file COPYING. If not, write to
* the Free Software Foundation, 675 Mass Ave, Cambridge, MA 02139, USA.
#include <glib.h>
#define ALIGN(x,a) (((x)+(a)-1)&~((a)-1))
enum {
DBFS_BLK_ID_LEN = 20, /* block id (sha1 hash) len */
DBFS_UNLINK_DIR = (1 << 0),
DBFS_ROOT_INO = 1, /* root inode number */
DBFS_DIRENT_ALIGN = 8, /* struct dbfs_dirent alignment */
DBFS_FILENAME_MAX = 256, /* max filename component size */
DBFS_XATTR_NAME_LEN = 256, /* extended attr limits */
DBFS_XATTR_MAX_LEN = (1024 * 1024),
DBFS_XLIST_ALIGN = 8, /* struct dbfs_xlist alignment */
/* our data items are small, so use the smallest possible page
* size. This is a guess, and should be verified by looking at
* overflow pages and other DB statistics.
/* another guess. must take into account small data items
* as well as large ones
DBFS_MAX_EXT_LEN = (4 * 1024 * 1024), /* max extent len */
enum {
/* dirent magic number */
DBFS_DE_MAGIC = 0xd4d4d4d4U,
enum dbfs_inode_type {
typedef struct {
char buf[DBFS_BLK_ID_LEN];
} dbfs_blk_id_t;
struct dbfs_dirent {
guint32 magic;
guint16 res2;
guint16 namelen;
guint64 ino;
char name[0];
} __attribute__ ((packed));
struct dbfs_xlist {
guint32 namelen;
char name[0];
} __attribute__ ((packed));
struct dbfs_extent {
guint32 off; /* offset into block */
guint32 len; /* length of fragment */
dbfs_blk_id_t id; /* block id */
} __attribute__ ((packed));
struct dbfs_hashref {
guint32 refs;
} __attribute__ ((packed));
struct dbfs_raw_inode {
guint64 ino;
guint64 version;
guint32 mode;
guint32 nlink;
guint32 uid;
guint32 gid;
guint64 rdev;
guint64 size;
guint64 ctime;
guint64 atime;
guint64 mtime;
struct dbfs_extent blocks[0];
} __attribute__ ((packed));
struct dbfs_inode {
enum dbfs_inode_type type;
unsigned int n_extents;
unsigned int raw_ino_size;
struct dbfs_raw_inode *raw_inode;
struct dbfs {
const char *home;
char *passwd;
DB_ENV *env;
DB *meta;
DB *data;
DB *hashref;
guint64 next_inode;
extern int debugging;
typedef int (*dbfs_dir_actor_t) (struct dbfs_dirent *, void *);
/* dbfs-backend.c */
extern int dbmeta_del(DB_TXN *txn, const char *key_str);
extern int dbfs_inode_read(DB_TXN *txn, guint64 ino_n, struct dbfs_inode **ino_out);
extern int dbfs_dir_read(DB_TXN *txn, guint64 ino, DBT *val);
extern int dbfs_symlink_read(DB_TXN *txn, guint64 ino, DBT *val);
extern int dbfs_dir_foreach(void *dir, dbfs_dir_actor_t func, void *userdata);
extern int dbfs_dir_lookup(DB_TXN *txn, guint64 parent, const char *name, guint64 *ino);
extern int dbfs_link(DB_TXN *txn, struct dbfs_inode *ino, guint64 ino_n, guint64 parent, const char *name);
extern int dbfs_unlink(DB_TXN *txn, guint64 parent, const char *name, unsigned long flags);
extern void dbfs_init(void *userdata);
extern void dbfs_exit(void *userdata);
extern int dbfs_mknod(DB_TXN *txn, guint64 parent, const char *name,
guint32 mode, guint64 rdev,
struct dbfs_inode **ino);
extern int dbfs_symlink_write(DB_TXN *txn, guint64 ino, const char *link);
extern int dbfs_inode_del(DB_TXN *txn, struct dbfs_inode *ino);
extern int dbfs_xattr_get(DB_TXN *TXN, guint64 ino_n, const char *name,
void **buf_out, size_t *buflen_out);
extern int dbfs_xattr_set(DB_TXN *TXN, guint64 ino_n, const char *name,
const void *buf, size_t buflen,
int flags);
extern int dbfs_xattr_remove(DB_TXN *TXN, guint64, const char *, gboolean);
extern int dbfs_xattr_list(DB_TXN *TXN, guint64 ino, void **buf_out, size_t *buflen_out);
extern int dbfs_read(DB_TXN *, guint64, guint64, size_t, void **);
extern int dbfs_write(DB_TXN *, guint64, guint64, const void *, size_t);
extern int dbfs_inode_resize(DB_TXN *txn, struct dbfs_inode *ino, guint64 new_size);
extern int dbfs_rename(DB_TXN *txn, guint64, const char *, guint64, const char *);
/* libdbfs.c */
extern int dbfs_open(struct dbfs *, unsigned int, unsigned int, const char *);
extern void dbfs_close(struct dbfs *fs);
extern struct dbfs *dbfs_new(void);
extern void dbfs_free(struct dbfs *fs);
extern struct dbfs *gfs;
extern int dbfs_inode_write(DB_TXN *txn, struct dbfs_inode *ino);
extern int dbfs_dir_new(DB_TXN *txn, guint64 parent, guint64 ino_n, const struct dbfs_inode *ino);
extern int dbfs_dir_write(DB_TXN *txn, guint64 ino, DBT *val);
extern void dbfs_inode_free(struct dbfs_inode *ino);
static inline size_t dbfs_dirent_next(guint16 namelen)
return ALIGN(sizeof(struct dbfs_dirent) + namelen, DBFS_DIRENT_ALIGN);
static inline size_t dbfs_xlist_next(guint16 namelen)
return ALIGN(sizeof(struct dbfs_xlist) + namelen, DBFS_XLIST_ALIGN);
#endif /* __DBFS_H__ */