df - report number of free disk blocks
df [ -befhiklmnPqrtV ] [ -w fieldwidth ] [ -F FStype ] [ filesystem ...]
df reports the number of total, used, and available disk blocks (one disk
block equals 512 bytes) in filesystems. The filesystem argument is a
device special file containing a disk filesystem, a mounted NFS
filesystem of the form hostname:pathname, or any file, directory, or
special node in a mounted filesystem. If no filesystem arguments are
specified, df reports on all mounted filesystems.
The options to df are:
-b Causes df to report usage in 512-byte units, which is the default.
-e Causes only the device and the number of free inodes to be printed.
Causes filesystems of types other than FStype to be skipped.
-f Normally, the free block information is gleaned from the
filesystem's superblock. The -f flag forces a scan of the free
-h Causes df to report usage in ``human'' blocks. Each size is
converted to kilobytes, megabytes, gigabytes, or terabytes and
printed with a postfix indicating the units. Units are in powers of
two, i.e., a megabyte is 2 to the 20th. This option also affects the
inode style listing.
-i Reports the number and percentage of used inodes and the number of
-k Causes df to report usage in 1024-byte units.
-l Restricts the report to local disk filesystems only. This option is
supported only with EFS and XFS filesystems.
-m Causes df to report usage in 1048576-byte (megabyte) units. This
option also affects the inode style listing.
-n Prints only the device name and filesystem type for each filesystem.
-P When both the -P and -k options are specified, the following header
line will be written:
Filesystem 1024-blocks Used Available Capacity Mounted on
When the -P is specified without the -k option, the following header
line will be written:
Filesystem 512-blocks Used Available Capacity Mounted on
-q Recognized but ignored. Provided for compatibility with previous
-r For XFS filesystems, adds the realtime portion of the filesystem,
which is normally excluded.
-t Recognized but ignored. Provided for compatibility with previous
-V Causes a command line to be constructed from the defaults and
echoed. Additional arguments are ignored.
Causes the width of the first field (the Filesystem field) to be
padded to that value. This allows control of the output, so that
systems with long pathnames can still have columnar output. In
earlier releases, this field was truncated, in an attempt to keep
the output from wrapping on an 80 column display (which often failed
anyway, except for very short mount point names). Now it is never
To report usage in the root filesystem, use either of the following:
Report on the filesystem containing the current directory:
statfs(2), efs(4), xfs(4).
If the environment variable HUMAN_BLOCKS is set, it implies -h.
Free counts may be incorrect, with or without the -f flag.
In previous IRIX releases, usage was reported in 1024-byte units.
The interpretation of megabyte et al as 1,000,000 or 2^20 is a matter of
debate. The current reasoning is that kbytes are 1024, so megabytes
should be 1024*1024.
The proc filesystem (normally mounted under /proc) is not printed by
default, but can be explicitly specified. This filesystem consumes no
actual disk space, but is an interface to the virtual space of running
processes. The total and free blocks reported represent the total
virtual memory (real memory plus swap space) present and the amount
currently free, respectively.
The -i option applied to filesystems of type nfs reports a free inode
count of 0. Future versions of NFS will support useful inode counts.
For the proc filesystem type, -i reports the number of active process
slots in the iuse column and the number of available slots in the ifree
For XFS filesystems, there is no way to see the space used by the log
portion of the filesystem.
In earlier releases, df silently right truncated long device names and
NFS server pathnames. df now left truncates, since the left portion is
more likely to be non-unique than the right.
PPPPaaaaggggeeee 3333 [ Back ]