tdb.c

samba服务器！

	if (tdb->max_dead_records != 0) {

		/*
		 * Allow for some dead records per hash chain, mainly for
		 * tdb's with a very high create/delete rate like locking.tdb.
		 */

		if (tdb_lock(tdb, BUCKET(hash), F_WRLCK) == -1)
			return -1;

		if (tdb_count_dead(tdb, hash) >= tdb->max_dead_records) {
			/*
			 * Don't let the per-chain freelist grow too large,
			 * delete all existing dead records
			 */
			tdb_purge_dead(tdb, hash);
		}

		if (!(rec_ptr = tdb_find(tdb, key, hash, &rec))) {
			tdb_unlock(tdb, BUCKET(hash), F_WRLCK);
			return -1;
		}

		/*
		 * Just mark the record as dead.
		 */
		rec.magic = TDB_DEAD_MAGIC;
		ret = tdb_rec_write(tdb, rec_ptr, &rec);
	}
	else {
		if (!(rec_ptr = tdb_find_lock_hash(tdb, key, hash, F_WRLCK,
						   &rec)))
			return -1;

		ret = tdb_do_delete(tdb, rec_ptr, &rec);
	}

	if (ret == 0) {
		tdb_increment_seqnum(tdb);
	}

	if (tdb_unlock(tdb, BUCKET(rec.full_hash), F_WRLCK) != 0)
		TDB_LOG((tdb, TDB_DEBUG_WARNING, "tdb_delete: WARNING tdb_unlock failed!\n"));
	return ret;
}

int tdb_delete(struct tdb_context *tdb, TDB_DATA key)
{
	u32 hash = tdb->hash_fn(&key);
	return tdb_delete_hash(tdb, key, hash);
}

/*
 * See if we have a dead record around with enough space
 */
static tdb_off_t tdb_find_dead(struct tdb_context *tdb, u32 hash,
			       struct list_struct *r, tdb_len_t length)
{
	tdb_off_t rec_ptr;
	
	/* read in the hash top */
	if (tdb_ofs_read(tdb, TDB_HASH_TOP(hash), &rec_ptr) == -1)
		return 0;

	/* keep looking until we find the right record */
	while (rec_ptr) {
		if (tdb_rec_read(tdb, rec_ptr, r) == -1)
			return 0;

		if (TDB_DEAD(r) && r->rec_len >= length) {
			/*
			 * First fit for simple coding, TODO: change to best
			 * fit
			 */
			return rec_ptr;
		}
		rec_ptr = r->next;
	}
	return 0;
}

/* store an element in the database, replacing any existing element
   with the same key 

   return 0 on success, -1 on failure
*/
int tdb_store(struct tdb_context *tdb, TDB_DATA key, TDB_DATA dbuf, int flag)
{
	struct list_struct rec;
	u32 hash;
	tdb_off_t rec_ptr;
	char *p = NULL;
	int ret = -1;

	if (tdb->read_only || tdb->traverse_read) {
		tdb->ecode = TDB_ERR_RDONLY;
		return -1;
	}

	/* find which hash bucket it is in */
	hash = tdb->hash_fn(&key);
	if (tdb_lock(tdb, BUCKET(hash), F_WRLCK) == -1)
		return -1;

	/* check for it existing, on insert. */
	if (flag == TDB_INSERT) {
		if (tdb_exists_hash(tdb, key, hash)) {
			tdb->ecode = TDB_ERR_EXISTS;
			goto fail;
		}
	} else {
		/* first try in-place update, on modify or replace. */
		if (tdb_update_hash(tdb, key, hash, dbuf) == 0) {
			goto done;
		}
		if (tdb->ecode == TDB_ERR_NOEXIST &&
		    flag == TDB_MODIFY) {
			/* if the record doesn't exist and we are in TDB_MODIFY mode then
			 we should fail the store */
			goto fail;
		}
	}
	/* reset the error code potentially set by the tdb_update() */
	tdb->ecode = TDB_SUCCESS;

	/* delete any existing record - if it doesn't exist we don't
           care.  Doing this first reduces fragmentation, and avoids
           coalescing with `allocated' block before it's updated. */
	if (flag != TDB_INSERT)
		tdb_delete_hash(tdb, key, hash);

	/* Copy key+value *before* allocating free space in case malloc
	   fails and we are left with a dead spot in the tdb. */

	if (!(p = (char *)malloc(key.dsize + dbuf.dsize))) {
		tdb->ecode = TDB_ERR_OOM;
		goto fail;
	}

	memcpy(p, key.dptr, key.dsize);
	if (dbuf.dsize)
		memcpy(p+key.dsize, dbuf.dptr, dbuf.dsize);

	if (tdb->max_dead_records != 0) {
		/*
		 * Allow for some dead records per hash chain, look if we can
		 * find one that can hold the new record. We need enough space
		 * for key, data and tailer. If we find one, we don't have to
		 * consult the central freelist.
		 */
		rec_ptr = tdb_find_dead(
			tdb, hash, &rec,
			key.dsize + dbuf.dsize + sizeof(tdb_off_t));

		if (rec_ptr != 0) {
			rec.key_len = key.dsize;
			rec.data_len = dbuf.dsize;
			rec.full_hash = hash;
			rec.magic = TDB_MAGIC;
			if (tdb_rec_write(tdb, rec_ptr, &rec) == -1
			    || tdb->methods->tdb_write(
				    tdb, rec_ptr + sizeof(rec),
				    p, key.dsize + dbuf.dsize) == -1) {
				goto fail;
			}
			goto done;
		}
	}

	/*
	 * We have to allocate some space from the freelist, so this means we
	 * have to lock it. Use the chance to purge all the DEAD records from
	 * the hash chain under the freelist lock.
	 */

	if (tdb_lock(tdb, -1, F_WRLCK) == -1) {
		goto fail;
	}

	if ((tdb->max_dead_records != 0)
	    && (tdb_purge_dead(tdb, hash) == -1)) {
		tdb_unlock(tdb, -1, F_WRLCK);
		goto fail;
	}

	/* we have to allocate some space */
	rec_ptr = tdb_allocate(tdb, key.dsize + dbuf.dsize, &rec);

	tdb_unlock(tdb, -1, F_WRLCK);

	if (rec_ptr == 0) {
		goto fail;
	}

	/* Read hash top into next ptr */
	if (tdb_ofs_read(tdb, TDB_HASH_TOP(hash), &rec.next) == -1)
		goto fail;

	rec.key_len = key.dsize;
	rec.data_len = dbuf.dsize;
	rec.full_hash = hash;
	rec.magic = TDB_MAGIC;

	/* write out and point the top of the hash chain at it */
	if (tdb_rec_write(tdb, rec_ptr, &rec) == -1
	    || tdb->methods->tdb_write(tdb, rec_ptr+sizeof(rec), p, key.dsize+dbuf.dsize)==-1
	    || tdb_ofs_write(tdb, TDB_HASH_TOP(hash), &rec_ptr) == -1) {
		/* Need to tdb_unallocate() here */
		goto fail;
	}

 done:
	ret = 0;
 fail:
	if (ret == 0) {
		tdb_increment_seqnum(tdb);
	}

	SAFE_FREE(p); 
	tdb_unlock(tdb, BUCKET(hash), F_WRLCK);
	return ret;
}


/* Append to an entry. Create if not exist. */
int tdb_append(struct tdb_context *tdb, TDB_DATA key, TDB_DATA new_dbuf)
{
	u32 hash;
	TDB_DATA dbuf;
	int ret = -1;

	/* find which hash bucket it is in */
	hash = tdb->hash_fn(&key);
	if (tdb_lock(tdb, BUCKET(hash), F_WRLCK) == -1)
		return -1;

	dbuf = tdb_fetch(tdb, key);

	if (dbuf.dptr == NULL) {
		dbuf.dptr = (char *)malloc(new_dbuf.dsize);
	} else {
		dbuf.dptr = (char *)realloc(dbuf.dptr,
					    dbuf.dsize + new_dbuf.dsize);
	}

	if (dbuf.dptr == NULL) {
		tdb->ecode = TDB_ERR_OOM;
		goto failed;
	}

	memcpy(dbuf.dptr + dbuf.dsize, new_dbuf.dptr, new_dbuf.dsize);
	dbuf.dsize += new_dbuf.dsize;

	ret = tdb_store(tdb, key, dbuf, 0);
	
failed:
	tdb_unlock(tdb, BUCKET(hash), F_WRLCK);
	SAFE_FREE(dbuf.dptr);
	return ret;
}


/*
  return the name of the current tdb file
  useful for external logging functions
*/
const char *tdb_name(struct tdb_context *tdb)
{
	return tdb->name;
}

/*
  return the underlying file descriptor being used by tdb, or -1
  useful for external routines that want to check the device/inode
  of the fd
*/
int tdb_fd(struct tdb_context *tdb)
{
	return tdb->fd;
}

/*
  return the current logging function
  useful for external tdb routines that wish to log tdb errors
*/
tdb_log_func tdb_log_fn(struct tdb_context *tdb)
{
	return tdb->log.log_fn;
}


/*
  get the tdb sequence number. Only makes sense if the writers opened
  with TDB_SEQNUM set. Note that this sequence number will wrap quite
  quickly, so it should only be used for a 'has something changed'
  test, not for code that relies on the count of the number of changes
  made. If you want a counter then use a tdb record.

  The aim of this sequence number is to allow for a very lightweight
  test of a possible tdb change.
*/
int tdb_get_seqnum(struct tdb_context *tdb)
{
	tdb_off_t seqnum=0;

	tdb_ofs_read(tdb, TDB_SEQNUM_OFS, &seqnum);
	return seqnum;
}

int tdb_hash_size(struct tdb_context *tdb)
{
	return tdb->header.hash_size;
}

size_t tdb_map_size(struct tdb_context *tdb)
{
	return tdb->map_size;
}

int tdb_get_flags(struct tdb_context *tdb)
{
	return tdb->flags;
}