libuuid: move clock state file from /var/lib to /var/run

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

uuidd: move uuidd files from /var/lib/libuuid to /var/run/uuidd

Reported-by: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Karel Zak <kzak@redhat.com>

build-sys: release++ (v2.16-rc1)

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

docs: update v2.16-ReleaseNotes

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

docs: add v2.16 ReleaseNotes

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

build-sys: fix --disable-uuidd

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 vi.po (from translationproject.org)

build-sys: fix libuuid Makefile.am

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

mount: a little clean up info about loopdevs in man page

Addresses-Debian-Bug: #478600
Signed-off-by: Karel Zak <kzak@redhat.com>

losetup: suggest to use modprobe rather than insmod in losetup.8

Addresses-Debian-Bug: #334796
Signed-off-by: Karel Zak <kzak@redhat.com>

mount: add info about obsolete vfat options to mount.8

Addresses-Debian-Bug: #329856
Signed-off-by: Karel Zak <kzak@redhat.com>

umount: clean up help output

Addresses-Debian-Bug: #320500
Signed-off-by: Karel Zak <kzak@redhat.com>

docs: remove obsolete information from fstab example

Addresses-Debian-Bug: #507700
Signed-off-by: Karel Zak <kzak@redhat.com>

po: fix typo in French translation

Addresses-Mandriva-Bug: #42783
Signed-off-by: Olivier Blin <blino@mandriva.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

docs: refresh TODO list

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

switch_root: do recursiveRemove after our root is moved to avoid races.

This way there's no race between unlinking the /newroot directory and
the MS_MOVE/chroot() to get away from it.

Signed-off-by: Peter Jones <pjones@redhat.com>

switch_root: fork before cleaning up the filesystem.

Fork before cleaning up the old filesystem, so it becomes asyncronous,
which results in a faster boot time.

Signed-off-by: Peter Jones <pjones@redhat.com>

switch_root: use file descriptor instead of path for recursiveRemove()

This makes recursiveRemove() use fdopendir() instead of taking a path,
so we're always sure about which namespace we're starting from.

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

switch_root: add man page

Acked-by: Peter Jones <pjones@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

switch_root: use snprintf() rather tan str{cpy,cat}()

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

switch_root: clean up argv[] usage, add -h and -V

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

switch_root: use err.h, clean up return codes

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

build-sys: check for openat() and linux for switch_root

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

switch_root: rewrite to use fstatat() and unlinkat()

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

switch_root: fix coding style

Signed-off-by: Peter Jones <pjones@redhat.com>
Acked-by: Peter Jones <pjones@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

build-sys: add --disable-switch_root

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

switch_root: new command

Copied from Dracut project:
git://dracut.git.sourceforge.net/gitroot/dracut

switch_root history in dracut.git repository:

$ git shortlog switch_root.c
Harald Hoyer (5):
      replace switch_root shell script with binary
      add \n to switch_root
      use switch_root code from http://pjones.fedorapeople.org/mkstart/usr/lib/mkstart/switchroot.c
      mount move instead of umount and fix the search for fallback inits
      setsid() and set controlling terminal for real /sbin/init

Victor Lowther (2):
      Simplify switch_root.c a bit
      Remove all files on the initramfs before switching root

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

libblkid: don't require udev symlinks verification for non-root users

There is noway how to verify LABEL/UUID for non-root users, we have to
follow udev symlinks or use cached information from blkid.tab. Use
unverified symlinks is faster.

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

mount: move MS_{PROPAGATION,BIND,MOVE} detection

The previous commit does not properly fix the problem with "none"
fstype and MS_{PROPAGATION,BIND,MOVE} flags. The real fstype has to
be replaced with "none" before we try to use /sbin/mount.<type>
helper.

Old version:

# grep test /proc/self/mountinfo
16 18 0:19 / /mnt/test rw,relatime - tmpfs tmpfs rw

# ./mount -v -v -v --make-private /mnt/test
mount: fstab path: "/etc/fstab"
mount: mtab path:  "/etc/mtab"
mount: lock path:  "/etc/mtab~"
mount: temp path:  "/etc/mtab.tmp"
mount: UID:        0
mount: eUID:       0
mount: spec:  "tmpfs"
mount: node:  "/mnt/test"
mount: types: "tmpfs"
mount: opts:  "rw"
mount: external mount: argv[0] = "/sbin/mount.tmpfs"
mount: external mount: argv[1] = "tmpfs"
mount: external mount: argv[2] = "/mnt/test"
mount: external mount: argv[3] = "-v"
mount: external mount: argv[4] = "-o"
mount: external mount: argv[5] = "rw"
tmpfs on /mnt/test type tmpfs (rw)

# grep test /proc/self/mountinfo
16 18 0:19 / /mnt/test rw,relatime - tmpfs tmpfs rw
17 16 0:20 / /mnt/test rw,relatime - tmpfs tmpfs rw

Fixed version:

# grep test /proc/self/mountinfo
16 18 0:19 / /mnt/test rw,relatime - tmpfs tmpfs rw

# ./mount -v -v -v --make-private /mnt/test
mount: fstab path: "/etc/fstab"
mount: mtab path:  "/etc/mtab"
mount: lock path:  "/etc/mtab~"
mount: temp path:  "/etc/mtab.tmp"
mount: UID:        0
mount: eUID:       0
mount: spec:  "tmpfs"
mount: node:  "/mnt/test"
mount: types: "tmpfs"
mount: opts:  "rw"
mount: mount(2) syscall: source: "tmpfs", target: "/mnt/test",
filesystemtype: "none", mountflags: 262144, data: (null)

# grep test /proc/self/mountinfo
16 18 0:19 / /mnt/test rw,relatime - tmpfs tmpfs rw

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

mount: use "none" fstype for MS_PROPAGATION mounts

Reported-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Karel Zak <kzak@redhat.com>

po: refresh POTFILES.in

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

mount: add ext4 to the list of filesystems in mount.8

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

mount: add ext4 to mount.8

Based on Documentation/filesystems/ext4.txt

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

mount: cleanup notes about -l option in mount.8

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

losetup: add --set-capacity

The LOOP_SET_CAPACITY allows to resize loop device size.

Example:

 # blockdev --getsize64 /dev/loop0
 10485760

 # dd if=/dev/zero of=/home/images/aaa.img count=10 bs=1M oflag=append conv=notrunc

 # blockdev --getsize64 /dev/loop0
 10485760

 # ./losetup --set-capacity /dev/loop0

 # blockdev --getsize64 /dev/loop0
 20971520

CC: J. R. Okajima <hooanon05@yahoo.co.jp>
Signed-off-by: Karel Zak <kzak@redhat.com>

mount: use TAG parsing function from libblkid

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

ionice: Allow setting the none class

Allow users to set the "none" class on processes. Using the
none class has the distict advantage that the io priority
is inherited from the cpu nice level. Update the man page
to reflect the change.

Signed-off-by: Jakob Unterwurzacher <jakobunt@gmail.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: add device-mapper snapshot cow device probe

The copy-on-write area of device-mapper snapshot has
SnAp identifier, disk header is written directly by kernel.

[kzak@redhat.com: - rename from "snap_cow" to "DM_snapshot_cow"]

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

blkid: fix LVM1 probe

Fix LVM1 probe and UUID parse.

Signed-off-by: Milan Broz <mbroz@redhat.com>

hwclock: fix mismatched popen/fclose.

date_child_fp is opened by popen, so should be closed with pclose.

Signed-off-by: John Keeping <john.keeping@lineone.net>
Reviewed-by: WANG Cong <xiyou.wangcong@gmail.com>

docs: add missing commands/projects to AUTHORS file

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

tests: fix script that creates lscpu dumps

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

tests: move lscpu /proc and /sys dumps to tarballs

It's easy to maintain few tarballs that tons of files.

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

tests: remove broken Xen dumps for lscpu

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

tests: don't run some mount tests for non-root users

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

tests: fix reiserfs test

For more details see 5b3f8e4f961298a9ca296f8362a717c1a4dd2ee0.

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

tests: add mdraid libblkid test

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

libblkid: fix reiserfs name

Reported-by: Jonas Karlsson <cj.karlsson@gmail.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

tests: fix 'delete extended partition' checksum

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

build-sys: add --disable-uuidd

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

uuidd: new command (UUID daemon from e2fsprogs)

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

libuuid: add .gitignore

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

uuidgen: new command (from e2fsprogs)

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

build-sys: add --disable-tls

 * add AX_TLS autoconf macro

 * add --disabletls option

 * cleanup gen_uuid.c code to remove gcc warnings when compiled
   without TLS macro.

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

build-sys: enable fsck by default

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

build-sys: add --disable-libblkid, remove volume_id support

Supported scenarios:

  * internal libblkid (--enable-libblkid, default)

  * external libblkid (--disable-libblkid)
     - systems without pkg-config
     - systems with pkg-config

  * systems without libblkid at all
    (requires --disable-mount and --disable-fsck)

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

mount: (and fsck) remove libvolume_id support

The libvolume_id is deprecated in favor of libblkid. The libblkid from
u-l-ng provides the same functionality as libvolume_id.

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

build-sys: cleanup libuuid stuff

Supported scenarios:

 * internal libuuid (--enable-libuuid, default)

 * external libuuid (--disable-libuuid)
   - systems without pkg-config
   - systems with pkg-config

 * systems without libuuid at all (the library is optional for u-l-ng)

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

build-sys: fix headers in mkswap and libblkid

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

build-sys: add UTIL_{SET,RESTORE}_FLAGS

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

libblkid: update man page

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

libuuid: add info about u-l-ng to man pages

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

libuuid: add --disable-libuuid and LIBUUID_VERSION

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

libuuid: import UUID library from e2fsprogs

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

mount: fix undefined reference to `security_get_initial_context'

> Gentoo-hardened stable have: sys-libs/libselinux-1.34.14
> I have:
> mount.c:(.text+0x12ce): undefined reference to `security_get_initial_context'

Reported-by: Максим Бритов <maxim.britov@gmail.com>
Addresses-Gentoo-Bug: #270168
Signed-off-by: Karel Zak <kzak@redhat.com>

fdisk: add simple test for doslabel stuff

Initial tests/fdisk patch to test most of the doslabel-related stuff.
Coverage of dos-related functions is about 60%, does not test the
verify/fix-PT code.

[kzak@redhat.com: - fix test design to match to
                    output/expect/diff idea]

Signed-off-by: Zdenek Behan <rain@matfyz.cz>
Signed-off-by: Karel Zak <kzak@redhat.com>

tests: add functions for work withdisk images

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

fdisk: (and cfdisk) fix to be consistent about maximum heads

Fix fdisk and cfdisk to be consistent about maximum heads

Without this patch, fdisk allows up to 256 heads to be set via the menu, but
only up to 255 heads by command line -H. Similarly, cfdisk's -H option also
enforces heads <= 255 instead of heads <= 256.

The strange limit of 256 comes about because (heads - 1) should fit in a
single byte for legacy CHS. Geometries of 256 heads and 32 sectors are seen
in disks from other operating systems where cylinder-aligned partitions
which are also aligned to a 4MB boundary are wanted.

Signed-off-by: Chris Webb <chris@arachsys.com>

cal: uClibc has langinfo.h but not _NL_TIME_WEEK_1STDAY.

Signed-off-by: Tom Prince <cougar@hermes>

blkid: cleanup debug messages and return codes in blkid_do_probe()

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

blkid: blkid_do_safeprobe() has to be tolerant to RAIDs

It's pretty generic that RAID signature is on the same device as FS
signature. We have to be tolerant and ignore filesystems signatures
on such devices. The RAID signature is always preferred.

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

blkid: fix "hangs forever with partition type mdraid"

> #:> blkid /dev/sda7
> /dev/sda7: UUID="c193d159-1fd4-1a4b-4037-f3433e016f3d" TYPE="mdraid"
>
> The blkid binary build from the util-linux-ng sources just hangs
> forever with 100% cpu load with the same command.

Reported-by: Juergen Daubert <jue@jue.li>
Signed-off-by: Karel Zak <kzak@redhat.com>

build-sys: complete /libs to /shlibs rename

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

build-sys: rename /libs to /shlibs

On Wed, May 13, 2009 at 05:38:49AM -0400, Christoph Hellwig wrote:
> Also the lib vs libs toplevel directories probably aren't a too good
> choice.

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

blkid: make libuuid optional

read.c:28:23: error: uuid/uuid.h: No such file or directory
make[3]: *** [read.lo] Error 1
make[3]: *** Waiting for unfinished jobs....

Reported-by: Olaf <mailinglists@ban-solms.de>
Signed-off-by: Karel Zak <kzak@redhat.com>

build-sys: fix blkid.h include for old e2fsprogs

   When building util-linux-ng-2.15, I get this:
   ../lib/fsprobe.c:10:19: error: blkid.h: No such file or directory

Unfortunately, the blkid.pc uses -I${includedir}/blkidi since
e2fsprogs v1.41.2. The old versions use -I${includedir} only.

Reported-by: Robert Schwebel <r.schwebel@pengutronix.de>
Signed-off-by: Karel Zak <kzak@redhat.com>

build-sys: fix "make -C" bug

Running:

./configure --with-fsprobe=builtin
make -C mount mount

causes the following error:

gcc -std=gnu99 -DHAVE_CONFIG_H -I. -I..  -include ../config.h
-I../include -DLOCALEDIR=\"/home/matthew/util-linux-ng/share/locale\"
-fsigned-char -I../libs/blkid/src   -g -O2 -MT mount-setproctitle.o
-MD -MP -MF .deps/mount-setproctitle.Tpo -c -o mount-setproctitle.o
`test -f '../lib/setproctitle.c' || echo './'`../lib/setproctitle.c mv
-f .deps/mount-setproctitle.Tpo .deps/mount-setproctitle.Po
make -C
make: option requires an argument -- 'C'
Usage: make [options] [target] ...
...

Reported-by: Matthew Burgess <matthew@linuxfromscratch.org>
Signed-off-by: Karel Zak <kzak@redhat.com>

losetup: handle symlinks in /dev/loop/

The loop_scandir() expects all the files in /dev/loop/ to be actual
devices and not symlinks to devices.  However, udev by default sets up
symlinks in /dev/loop/ thereby breaking the fallback scan logic when
looking for devices more than the default number of 8.

Simple way to reproduce is:
 - <detach all devices>
 - cd /dev/loop
 - rm -f [2-7]
 - losetup 0 autogen.sh
 - mv 1 8
 - for n in {1..7} ; do ln -s 0 $n ; done
 - losetup -f
<error but should have found /dev/loop/8>

[kay.sievers@vrfy.org:
       just a note, udev does not do that by default. It is
       devfs-nonsense, which the "default" never did.]

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Reported-by: Mario Bachmann <grafgrimm77@gmx.de>

lscpu: fix cpuid code on x86/PIC

If we build lscpu as PIE, we currently get a build failure:
lscpu.c: In function 'main':
lscpu.c:333: error: can't find a register in class 'BREG' while reloading 'asm'
lscpu.c:333: error: 'asm' operand has impossible constraints
make[2]: *** [lscpu.o] Error 1

So we need a little bit of register shuffling to keep gcc happy.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>

chrt: don't assume SCHED_BATCH and SCHED_IDLE exist

SCHED_FIFO, SCHED_OTHER, SCHED_RR are part of POSIX 1003.1b Process
Scheduling, so it is correct to assume they always exists.

SCHED_BATCH and SCHED_IDLE are Linux specific, we should not assume
they exists.

Defining SCHED_BATCH and SCHED_IDLE to random values (ie the ones found
on Linux systems) is not an option as they may *collide* with the one of
other systems. For example on GNU/kFreeBSD we have:

   #define SCHED_RR        3

and on Linux we have:

   #define SCHED_BATCH     3

[kzak@redhat.com: - add "Linux specific" notes to chrt.1
                  - add a note about BATCH and PR conflict to
                    this commit message]

Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Signed-off-by: Karel Zak <kzak@redhat.com>

cal: Highlight today even when month or year specified

* misc-utils/cal.c: Unconditionally call time()
and highlight today where possible.

Signed-off-by: Rajeev V. Pillai <rajeevvp@gmail.com>
Signed-off-by: Pádraig Brady <P@draigBrady.com>

build-sys: release++ (v2.15)

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

docs: update v2.15 ReleaseNotes

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

docs: update AUTHORS file

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

blkid: use /sys/block/dm-<N>/dm/name

The Linux kernel (since 2.6.29, patch 784aae735d9b0bba3f8b9faef4c8b30df3bf0128)
exports the real DM device names in /sys/block/<ptname>/dm/name.

The sysfs based solution is nicer and faster than scan for devno in
/dev/mapper/.

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

blkid: use /dev/mapper/<name> rather than /dev/dm-<N>

The libblkid (since v1.41.1) returns private device-mapper names (e.g.
/dev/dm-0). It's because the probe_one() function scans /dev before
/dev/mapper.

brw-rw---- 1 root disk 253, 0 2009-04-27 13:41 /dev/dm-0
brw-rw---- 1 root disk 253, 0 2009-04-27 13:41 /dev/mapper/TestVolGroup-TestLogVolume

Old version:
  # blkid -t LABEL="TEST-LABEL" -o device
  /dev/dm-0

Fixed version:
  # blkid -t LABEL="TEST-LABEL" -o device
  /dev/mapper/TestVolGroup-TestLogVolume

Addresses-Red-Hat-Bug: #497259
Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: linux_raid - fix logic for volumes with size == 0

 Based on

    commit 7643819062985d9fc6c7664072576e71d3822b10
    Author: Kay Sievers <kay.sievers@vrfy.org>
    Date:   Sat Sep 6 16:23:21 2008 +0200

 from udev upstream tree.

Reported-by: Scott James Remnant <scott@ubuntu.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

tests: fix -regex in run.sh

Reported-by: Scott James Remnant <scott@ubuntu.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

po: merge changes

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

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

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

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

blkid: rename blkid_evaluate_spec to blkid_evaluate_tag

"SPEC" is usually used in mount(8) as a generic term for device name or
tag (LABEL, UUID). The function blkid_evaluate_* works with TAGs only.

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

docs: add entry about /proc/partitions parsing

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

sfdisk: fix possible buffer overflow

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

fdisk: fix max. ptname

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

blockdev: fix possible buffer overflow

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

blkid: split SONAME and LIBBLKID_VERSION

It seems better to split SONAME and the public library version. The
library version will be the same as util-linux-ng PACKAGE_VERSION.

PACKAGE_VERSION:               <maj>.<min>[-<suffix>]  e.g. 2.15-rc2
Symbols versioning:            BLKID_<maj>.<min>       e.g. BLKID_2.15
blkid_get_library_version():   <maj>.<min>.0           e.g. 2.15.0
SONAME:                        libblkid.so.1

See also the original patch a0487b1cb5beffb8f6783476664b01e8833e0ea8
where was introduced library versioning.

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