int
lfs_markv(
fsid_t *fsidp
, BLOCK_INFO *blkiov
, int blkcnt
)
)
rewrites the blocks specified in
blkiov
to new disk locations, for the purposes of grouping them next to one
another, or to move them out of a segment to clean it.
All fields of the BLOCK_INFO structure must be filled in, except for
bi_segcreate
.
If
bi_daddr
is not the correct current address for logical block
bi_lbn
of the file with inode number
bi_inode
,
or if the file's version number does not match
bi_version
,
the block will not be written to disk, but no error will be returned.
The
fsidp
argument contains the id of the filesystem to which the inodes and
blocks belong.
The
bi_bp
field contains
bi_size
bytes of data to be written into the appropriate block.
If
bi_lbn
is specified as LFS_UNUSED_LBN, the inode itself will be rewritten.
The
blkiov
argument is an array of BLOCK_INFO structures (see below).
The
blkcnt
argument determines the size of the
blkiov
array.
typedef struct block_info {
ino_t bi_inode; /* inode # */
ufs_daddr_t bi_lbn; /* logical block w/in file */
ufs_daddr_t bi_daddr; /* disk address of block */
time_t bi_segcreate; /* origin segment create time */
int bi_version; /* file version number */
void *bi_bp; /* data buffer */
int bi_size; /* size of the block (if fragment) */
} BLOCK_INFO;
)
returns 0 on success, or -1 on error.
)
indicates:
EFAULT
]
fsidp
points outside the process's allocated address space.
EINVAL
]
*fsidp
does not specify a valid filesystem.
EBUSY
]
)
function call appeared in
4.4BSD.
)
does not really belong in user space.
Among other things it could be used to work around the
SF_IMMUTABLE
and
SF_APPEND
file flags (see
chflags(2)).