libblkid: fix EFI GPT uuid byte order

Intel uses little-endians for UUID, the rest of the sane world uses 16
byte big-endian array...

Reported-by: Andrew Lutomirski <luto@mit.edu>
Signed-off-by: Karel Zak <kzak@redhat.com>

libblkid: try to detect if PT is newer than LVM

LVM (pvcreate) wipes the begin of the device. If there is a PT in the
wiped area then LVM signature is obsolete.

 # pvcreate /dev/sdb
 # fdisk /dev/sdb

old version:
 # blkid -p -o udev /dev/sdb
 ID_FS_TYPE=LVM2_member

new version:
 # blkid -p -o udev /dev/sdb
 ID_PART_TABLE_TYPE=dos

Reported-by: Matej Cepl <mcepl@redhat.com>
Addresses: https://bugzilla.redhat.com/show_bug.cgi?id=679799
Signed-off-by: Karel Zak <kzak@redhat.com>

lsblk: bitwise or TT_FL_ASCII with tt_flags instead of setting

Signed-off-by: Dave Reisner <d@falconindy.com>

scriptreplay: bug in argument check

Number of arguments cannot be at the same time too many and few.

Signed-off-by: Sami Kerola <kerolasa@iki.fi>

script: fix compiler warnings

Signed-off-by: Sami Kerola <kerolasa@iki.fi>

lib: [strutils.c] more robust strtol checks

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: add tests for [u]mount regular file

Signed-off-by: Karel Zak <kzak@redhat.com>

umount: allow unmounting loopdev specified by associated file

Make it possible to unmount a filesystem on a loop device if it is
specified by associated backing file. It does not attempt to unmount
anything if there are more than one loop device associated with the
given file.

Umount looks for associated loopdevice(s) only if umount is called
with the regular file as an argument.

Before:
mount -o loop -t ext2 img mnt
umount -v img
> Could not find /home/puzel/upstream/util-linux/img in mtab
> umount: img: not mounted

After:
mount -o loop -t ext2 img mnt
umount -v img
> img is associated with /dev/loop0, trying to unmount it
> /dev/loop0 has been unmounted

[kzak@redhat.com: - fix memory leak in lomount.c]

Addresses: https://bugzilla.novell.com/show_bug.cgi?id=666161
Signed-off-by: Petr Uzel <petr.uzel@suse.cz>
Signed-off-by: Karel Zak <kzak@redhat.com>

lsblk: fix readlink() usage

Reported-by: Steve Grubb <sgrubb@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

chsh: fix gcc link() warn_unused_result warning

setpwnam.c:176:9: warning: ignoring return value of ‘link’, declared with attribute warn_unused_result

Signed-off-by: Karel Zak <kzak@redhat.com>

fsck: fix -C parsing

Reported-by: Steve Grubb <sgrubb@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

chsh: fix small memory leak

Reported-by: Steve Grubb <sgrubb@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

login: does not ignore setgid() return code for non-roots

Reported-by: Steve Grubb <sgrubb@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

fdisk: more robust whole-disk detection

In lib/wholedisk.c, i is set to 0. If fd was a -1, then at
line 18 geometry.start is used without it being initialized.

Reported-by: Steve Grubb <sgrubb@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

libuuid: do not use invalid file descriptor

Signed-off-by: Petr Uzel <petr.uzel@suse.cz>

libuuid: manpage spelling fixes

Signed-off-by: Petr Uzel <petr.uzel@suse.cz>

libuuid: fix test in get_clock()

Signed-off-by: Petr Uzel <petr.uzel@suse.cz>

uuidd: fix manpage: libuuid(3) does not exist

Signed-off-by: Petr Uzel <petr.uzel@suse.cz>

lib: [strutils] avoid integer overflow on large values

This is visible on a 2TB disk via lsblk, where a large partition
incorrectly displays as 1.171798692T instead of 1.8T. This is
corrected by using a uint64_t type instead of a simple int --
consistant with the type used in lsblk.c to represent the raw size in
bytes.

[kzak@redhat.com: - add EiB support
                  - add size_to_human_string() to regression tests]

Signed-off-by: Dave Reisner <d@falconindy.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

libmount: add fallbacks for old systems without umount2() syscall

Reported-by: Peter Breitenlohner <peb@mppmu.mpg.de>
Signed-off-by: Karel Zak <kzak@redhat.com>

include: [at.h]: include c.h for PATH_MAX

Reported-by: Peter Breitenlohner <peb@mppmu.mpg.de>
Signed-off-by: Karel Zak <kzak@redhat.com>

libmount: replace mkostmps() with more portable mkstemp()

Reported-by: Peter Breitenlohner <peb@mppmu.mpg.de>
Signed-off-by: Karel Zak <kzak@redhat.com>

build-sys: don't build lsblk on Linux without openat()

Reported-by: Peter Breitenlohner <peb@mppmu.mpg.de>
Signed-off-by: Karel Zak <kzak@redhat.com>

partx, lsblk: fix gettext calls

Reported-by: Peter Breitenlohner <peb@mppmu.mpg.de>
Signed-off-by: Karel Zak <kzak@redhat.com>

build-sys: release++ (v2.19)

Signed-off-by: Karel Zak <kzak@redhat.com>

docs: update v2.19 ReleaseNotes

Signed-off-by: Karel Zak <kzak@redhat.com>

docs: update AUTHORS file

Signed-off-by: Karel Zak <kzak@redhat.com>

po: merge changes

Signed-off-by: Karel Zak <kzak@redhat.com>

po: update pl.po (from translationproject.org)

po: update nl.po (from translationproject.org)

po: update fr.po (from translationproject.org)

po: update da.po (from translationproject.org)

po: update cs.po (from translationproject.org)

include: [c.h]: add unused attribute to ignore_result()

unfortunately, gcc 4.6

void fn (void) {
int foo;
foo = bar ();
}

evaluates that foo is never used. The __attribute__((unused)) is
necessary...

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: fix libmount test

Signed-off-by: Karel Zak <kzak@redhat.com>

libuuid: do not check for EAGAIN after flock()

As per flock() manpage, flock() does not set errno=EAGAIN.

Signed-off-by: Petr Uzel <petr.uzel@suse.cz>

libmount: do not check for EAGAIN after flock()

As per flock() manpage, flock() does not set errno=EAGAIN.

Signed-off-by: Petr Uzel <petr.uzel@suse.cz>

mount: add rootcontext= for tmpfs on SELinux boxes

Addresses: http://bugzilla.redhat.com/show_bug.cgi?id=476964
Addresses: http://bugzilla.redhat.com/show_bug.cgi?id=676100
Signed-off-by: Karel Zak <kzak@redhat.com>

include: [c.h]: fix __alloc_size__ test

Reported-by: Francesco Cosoleto <cosoleto@gmail.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

libmount: more robust flock() usage

Signed-off-by: Karel Zak <kzak@redhat.com>

libuuid: fix locking of clock state counter file

fcntl(2) does not work for synchronization of threads belonging
to the same process. Use flock(2) instead.

http://marc.info/?l=util-linux-ng&m=129708412829971&w=2

Signed-off-by: Petr Uzel <petr.uzel@suse.cz>

schedutils: remove unneeded header files

Signed-off-by: Davidlohr Bueso <dave@gnu.org>

lscpu: add description of --sysroot option to man page, plus grammar touches

Signed-off-by: Benno Schulenberg <bensberg@justemail.net>

lsblk: small grammar touches to the man page

Signed-off-by: Benno Schulenberg <bensberg@justemail.net>

partx: normalize the punctuation of man page, and improve grammar here and there

Signed-off-by: Benno Schulenberg <bensberg@justemail.net>

blkid: dynamically allocate devicename array

If more than 128 devices are specified on the blkid cmdline,
the devices[] array will overflow.

We can dynamically allocate the devices[] array based on number
of arguments to avoid this problem.

[kzak@redhat.com: - add "if (optind < argc)" check]

Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

wipefs: use write_all from include/writeall.h

Signed-off-by: Petr Uzel <petr.uzel@suse.cz>

swapon: use EXIT_SUCCESS constants

Signed-off-by: Davidlohr Bueso <dave@gnu.org>

dmesg: use strtol_or_err instead of atoi

We shouldn't be accepting things like 'dmesg -n 2crapinput'
This patch also changes the exit's value to use EXIT_* constants.

Signed-off-by: Davidlohr Bueso <dave@gnu.org>

lscpu: cleanup usage()

Signed-off-by: Karel Zak <kzak@redhat.com>

uuidd: correctly report number of generated subsequent UUIDs

Before this change, "uuidd -t -n 1" reported
"6fd0672e-2f9b-11e0-b9c8-000feae00cb0 and subsequent 1 UUIDs".
That is two subsequent UUIDs, which is not correct.

Signed-off-by: Petr Uzel <petr.uzel@suse.cz>

uuidd: fix -n usage

It does not make sense to request zero UUIDs - report error.

Signed-off-by: Petr Uzel <petr.uzel@suse.cz>

uuidd: increase connection backlog size to SOMAXCONN

Signed-off-by: Petr Uzel <petr.uzel@suse.cz>

uuidd: use write_all from include/writeall.h

Signed-off-by: Petr Uzel <petr.uzel@suse.cz>

uuidd: use die() where possible

Signed-off-by: Petr Uzel <petr.uzel@suse.cz>

uuidd: use daemon(3) library function

Signed-off-by: Petr Uzel <petr.uzel@suse.cz>

uuidd: use UUID_STR_LEN and UUID_LEN

Use named constants for lentgh of UUID's binary and textual
representation, instead of magic numbers.

[kzak@redhat.com: - use sizeof(uuid_t) rather than hardcoded size]

Signed-off-by: Petr Uzel <petr.uzel@suse.cz>
Signed-off-by: Karel Zak <kzak@redhat.com>

uuidd: use UUIDD_OP_BULK_*_UUID named constants

Signed-off-by: Petr Uzel <petr.uzel@suse.cz>

uuidd: fix uuidd -r -n 2

uuidd -r -n 2 failed with "Unexpected reply length from server 36".
Fixed by this patch.

Signed-off-by: Petr Uzel <petr.uzel@suse.cz>

include: [xalloc.h] include string.h

xalloc.h needs string.h for declaration of strdup().  Otherwise we
can get implicit declaration warning.  This patch prevents it.

Signed-off-by: Marek Polacek <mpolacek@redhat.com>

include: [c.h] add fallback for alloc_size attributes

Reported-by: Davidlohr Bueso <dave@gnu.org>
Signed-off-by: Karel Zak <kzak@redhat.com>

fsck: include c.h, fallback for PATH_MAX

Reported-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Signed-off-by: Karel Zak <kzak@redhat.com>

include: [c.h] move ignore_result() to gcc-only block

Signed-off-by: Karel Zak <kzak@redhat.com>

libmount: add and copy FS internal flags

Signed-off-by: Karel Zak <kzak@redhat.com>

libmount: don't prepare update if syscall failed

Signed-off-by: Karel Zak <kzak@redhat.com>

docs: update ReleaseNotes

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add new return code 8 for ambivalent low-level probing result

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add note about PART_ENTRY_ tags to blkid.8

Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: don't return PART_ENTRY_* if partition is empty

The PART_ENTRY_* has nothing to do with the _contents_ of the block
device. So if the device is empty these values should not be printed
and blkid(8) has to return 2.

The PART_ENTRY_* for empty devices should be printed for "-o udev"
output only.

Reported-by: Thomas Bächler <thomas@archlinux.org>
Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: don't print devname if nothing detected

Signed-off-by: Karel Zak <kzak@redhat.com>

libmount: cleanup helper initialization API

Signed-off-by: Karel Zak <kzak@redhat.com>

libmount: add low-level API for umount

Signed-off-by: Karel Zak <kzak@redhat.com>

libmount: cleanup high-level mount API

Signed-off-by: Karel Zak <kzak@redhat.com>

libmount: copy mount attrs from utab to context

Signed-off-by: Karel Zak <kzak@redhat.com>

libblkid: check for swap or DM cow on small devices

The 1MiB limit for swap and COW is useless for Fedora LiveDVD where is
938KiB COW device.

Reported-by: Milan Broz <mbroz@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

libblkid: conditionalise parts from blkid/topology for Linux

Signed-off-by: Fabian Groffen <grobian@gentoo.org>
Signed-off-by: Karel Zak <kzak@redhat.com>

build-sys: shell syntax fix in autogen.sh

Use of == as synonym of single = sign backfires on systems where
/bin/sh is dash, like with Ubuntu.

Signed-off-by: Sami Kerola <kerolasa@iki.fi>

mount: fix parsing offset= followed by more options

mount does not parse parse offset= option if it is followed by other
options.  In umount, the parsing is done with the get_value() function.
This patch moves get_value to fstab.c (with new name get_option_value())
and fixes mount to use the function.

[kzak@redhat.com: - rename to get_option_value()
  - use fstab.c rather than sundries.c]

Novell bugzilla: #666150

Reported-by: Ludwig Nussel <ludwig.nussel@suse.de>
Signed-off-by: Petr Uzel <petr.uzel@suse.cz>
Signed-off-by: Karel Zak <kzak@redhat.com>

build-sys: provide missing strnlen from strutils

Solaris lacks strnlen, provide it from strutils.c.

Signed-off-by: Fabian Groffen <grobian@gentoo.org>

swapon: make needlessly global variables static

Also constify the option array.

Signed-off-by: Tobias Klauser <tklauser@distanz.ch>

provide a workaround if program_invocation_short_name is missing

Try some replacements, such as getexecname() on Solaris and __progname
on BSDs and Darwin.  When not found, base program_invocation_short_name
on the source filename it is used in, as not to require argv[0] to be
passed along.  This latter approach is not dynamic, but doesn't require
code changes for all places where program_invocation_short_name is used
now.

Signed-off-by: Fabian Groffen <grobian@gentoo.org>
Signed-off-by: Karel Zak <kzak@redhat.com>

build-sys: enable lsblk and libmount for Linux only

Signed-off-by: Karel Zak <kzak@redhat.com>

mount: fix atime description in mount.8 man page

Addresses: https://bugzilla.redhat.com/show_bug.cgi?id=665376
Signed-off-by: Karel Zak <kzak@redhat.com>

build-sys: release++ (v2.19-rc3)

Signed-off-by: Karel Zak <kzak@redhat.com>

docs: update v2.19 ReleaseNotes

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: skip non-compiled libmount tests

Signed-off-by: Karel Zak <kzak@redhat.com>

libmount: don't compile tests by default

Signed-off-by: Karel Zak <kzak@redhat.com>

libmount: fix debug message

Signed-off-by: Karel Zak <kzak@redhat.com>

build-sys: release++ (v2.19-rc2)

Signed-off-by: Karel Zak <kzak@redhat.com>

docs: update v2.19 ReleaseNotes

Signed-off-by: Karel Zak <kzak@redhat.com>

docs: update AUTHORS file

Signed-off-by: Karel Zak <kzak@redhat.com>

po: merge changes

Signed-off-by: Karel Zak <kzak@redhat.com>

tests: update lscpu tests on x86_64

Signed-off-by: Karel Zak <kzak@redhat.com>

libmount: compile tests

Signed-off-by: Karel Zak <kzak@redhat.com>

column: data type mismatch compiler warning fixes

Following warnings will longer appear when one will compile with
gcc flags -Wall -Wextra -pedantic

column.c:364:2: warning: comparison of unsigned expression < 0 is always false
column.c:369:2: warning: comparison of unsigned expression < 0 is always false

Signed-off-by: Sami Kerola <kerolasa@iki.fi>

cfdisk: data type mismatch, and other, compiler warning fixes

Following warnings will longer appear when one will compile with
gcc flags -Wall -Wextra -pedantic

cfdisk.c:475:3: warning: comparison of unsigned expression < 0 is always false
cfdisk.c:487:16: warning: comparison between signed and unsigned integer expressions
cfdisk.c:492:14: warning: comparison between signed and unsigned integer expressions
cfdisk.c:565:19: warning: comparison between signed and unsigned integer expressions
cfdisk.c:569:19: warning: comparison between signed and unsigned integer expressions
cfdisk.c:1070:14: warning: comparison between signed and unsigned integer expressions
cfdisk.c:1568:5: warning: missing initializer
cfdisk.c:1568:5: warning: (near initialization for 'tmp_ext.volume_label')

mbsalign.c:131:2: warning: comparison of unsigned expression >= 0 is always true

Signed-off-by: Sami Kerola <kerolasa@iki.fi>

libmount: fix user= usage

Signed-off-by: Karel Zak <kzak@redhat.com>

libmount: improve status check, minor cleanups

Signed-off-by: Karel Zak <kzak@redhat.com>

include: [tt] enlarge output buffer

Signed-off-by: Karel Zak <kzak@redhat.com>

libmount: rename mount.sym -> libmount.sym

Signed-off-by: Karel Zak <kzak@redhat.com>