Add two new options to quota:

-f path
Only print quota information for the file system that path resides on.
-r
Display the quota information in a raw format.

Reviewed by: freebsd-hackers

If a user is over both the soft block limit and soft i-node
limit, quota will report one of the grace times incorrectly.
This is due to it storing the result in a static buffer, and the
routine being

If a user is over both the soft block limit and soft i-node
limit, quota will report one of the grace times incorrectly.
This is due to it storing the result in a static buffer, and the
routine being called like:
printf("....", ..., timeprnt(btime), timeprnt(itime), ...)

The problem becomes very obvious if you change one of the default
grace periods to be much larger than the other one.

Changed timeprnt to dynamically allocate the string to be displayed.

show more ...

If two files systems, /a and /b are marked as having quotas enabled
in fstab and they are normally mounted as /a/b, if /b is not mounted,
the various quota utilities will incorrectly operate with the

If two files systems, /a and /b are marked as having quotas enabled
in fstab and they are normally mounted as /a/b, if /b is not mounted,
the various quota utilities will incorrectly operate with the quotas on
/a (silently) when operations are attemted on /b.

Sync up all the hasquota() routines between all the different
quota utilities and change it to detect if the file system we are
attempting to perform quota operations on is not currently mounted
and warn the user accordingly.

PR: bin/38918

show more ...

Make quota exit with a non-zero status if one more more file
systems are over quota, as documented in the man page.

PR: bin/77918

- Document the default disk block usage unit.
- Add option -h, human-readable output.

PR: docs/53732

Remove an unused #define. md5's with and without this commit match.

Approved by: murray (mentor)

Assign the result of getopt() to an int rather than to a char.

Remove redundant declarations.

Don't overflow the block quota calculations for NFS mounts.

PR: 47607
Submitted by: Eric van Gyzen <vangyzen@stat.duke.edu>

Now that the kernel access control for quotactl(2) appears to work
properly, clean up quota(1). quota(1) has the ability to query
quotas either directly from the kernel, or if that fails, by reading

Now that the kernel access control for quotactl(2) appears to work
properly, clean up quota(1). quota(1) has the ability to query
quotas either directly from the kernel, or if that fails, by reading
the quota.user or quota.group files specified for the file system
in /etc/fstab. The setuid bit existed solely (apparently) to let
non-operator users query their quotas and consumption when quotas
weren't enabled for the file system.

o Remove the setuid bit from quota(1).

o Remove the logic used by quota(1) when running setuid to prevent
users from querying the quotas of other users or groups. Note
that this papered over previously broken kernel access control;
if you queried directly using the system call, you could access
some of the data "restricted" by quota(1).

In the new world order, the ability to inspect the (live) quotas of
other uids and gids via the kernel is controlled by the privilege
requirement sysctl. The ability to query via the file is controlled
by the file permissions on the quota database backing files
(root:operator, group readable by default).

show more ...

Now that the kernel access control for quotactl(2) appears to work
properly, clean up quota(1). quota(1) has the ability to query
quotas either directly from the kernel, or if that fails, by reading

Now that the kernel access control for quotactl(2) appears to work
properly, clean up quota(1). quota(1) has the ability to query
quotas either directly from the kernel, or if that fails, by reading
the quota.user or quota.group files specified for the file system
in /etc/fstab. The setuid bit existed solely (apparently) to let
non-operator users query their quotas and consumption when quotas
weren't enabled for the file system.

o Remove the setuid bit from quota(1).

o Remove the logic used by quota(1) when running setuid to prevent
users from querying the quotas of other users or groups. Note
that this papered over previously broken kernel access control.

show more ...

Add a new flag (-l) that suppresses quota checks on NFS filesystems.

PR: bin/12939
Submitted by: Neil Blakey-Milner <nbm@rucus.ru.ac.za>

Fix a few compiler warnings, sort options and make the usage()
output match the synopsis.

Call getmntinfo with MNT_NOWAIT to avoid hanging if any NFS servers
are down. Avoid trying to send RPCs to the pidXXX@machine names
used in the special amd(8) mounts.

PR: bin/6183
Submitted by: Pet

Call getmntinfo with MNT_NOWAIT to avoid hanging if any NFS servers
are down. Avoid trying to send RPCs to the pidXXX@machine names
used in the special amd(8) mounts.

PR: bin/6183
Submitted by: Petr Lampa <lampa@fee.vutbr.cz>

show more ...

Fix some low-hanging fruit in the warnings department; ANSIfy functions,
sort out some const issues.

remove __P

Changed 'blocks' to 'usage' because we may not be getting quotas from
a ufs filesystem and it may be reporting in K instead of 512b blocks.
This is true when using a vxfs filesystem (on a solaris box

Changed 'blocks' to 'usage' because we may not be getting quotas from
a ufs filesystem and it may be reporting in K instead of 512b blocks.
This is true when using a vxfs filesystem (on a solaris box) for instance.

PR: bin/14545
Submitted by: Jim Pirzyk
Reviewed by: jkh
MFC after: 1 week

show more ...

Remove incorrect use of .Nm. Wrap long lines, Unwrap short lines
Options: -> The following options are available:

$Id$ -> $FreeBSD$

Local filesystems are ufs not ffs
Submitted by: "D. Rock" <rock@cs.uni-sb.de>

Don't assume that hp->h_lenght == 4. Be conservative in its use.
Submitted by: J. Assange a long time ago.

Converted to Lite2 mount interface - decide filesystem types by name
instead of by number.

Use err(3). Cosmetic in usage string. Typo in man page.

Finish importing Lite2's src/usr.bin, except for ex, diff, grep, mail,
pascal and vmstat.sparc. All changed files on the vendor branch should
already have been imported.

compare return value from getopt against -1 rather than EOF, per the final
posix standard on the topic.