df: do not print duplicate entries and rootfs by default

* src/df.c (struct devlist): Add new struct for storing already-
examined device numbers.
(devlist_head): Add new store of the above type.
(show_rootfs): Add new global boolean to not skip rootfs.
(dev_examined): Add new function to check if the device has
already been traversed.
(get_dev): Filter out rootfs unless "-t rootfs" or the -a
option is specified. Filter out duplicate entries by calling
the above new dev_examined unless the -a option is specified.
(main): Set the show_rootfs variable appropriately when the -t
option is specified for rootfs. Free device list (guarded by
IF_LINT).
* tests/df/skip-duplicates.sh: Add test to exercise the skipping
of duplicate entries.
* tests/df/skip-rootfs.sh: Add test to exercise the skipping
of the rootfs pseudo file system.
* tests/local.mk: Add the above new tests.
* NEWS (Changes in behavior): Mention the changes.
* doc/coreutils.texi (df invocation): Document df's behavior about
skipping rootfs and duplicate entries.

Co-authored-by: Bernhard Voelker.

1 files changed, 8 insertions, 0 deletions

diff --git a/doc/coreutils.texi b/doc/coreutils.texi
index 46d36801e..21400adb6 100644
--- a/doc/coreutils.texi
+++ b/doc/coreutils.texi
@@ -10600,6 +10600,14 @@ Normally the disk space is printed in units of
 1024 bytes, but this can be overridden (@pxref{Block size}).
 Non-integer quantities are rounded up to the next higher unit.
 
+For bind mounts and without arguments, @command{df} only outputs the statistics
+for the first occurence of that device in the list of file systems (@var{mtab}),
+i.e., it hides duplicate entries, unless the @option{-a} option is specified.
+
+By default, @command{df} omits the early-boot pseudo file system type
+@samp{rootfs}, unless the @option{-a} option is specified or that file system
+type is explicitly to be included by using the @option{-t} option.
+
 @cindex disk device file
 @cindex device file, disk
 If an argument @var{file} is a disk device file containing a mounted