Merge branch 'fix-lscpu-cache' of https://github.com/rudimeier/util-linux

* 'fix-lscpu-cache' of https://github.com/rudimeier/util-linux:
  tests: cleanup, remove unused lscpu output
  tests: add lscpu dump for ppc cpu with no cache size
  lscpu: cleanup, use _PATH_SYS_CPU/NODE
  lscpu: don't abort if cache size is unknown

findfs: add ability to work with PART{UUID,LABEL}= too

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

tests: cleanup, remove unused lscpu output

All these files are not used since a292bbde or 0d86c70a.

Signed-off-by: Ruediger Meier <ruediger.meier@ga-group.nl>

tests: add lscpu dump for ppc cpu with no cache size

Taken from qemu-system-ppc (qemu 1.7) machine with installed
debian 7.4 wheezy, kernel 3.2.0-4-powerpc.

Signed-off-by: Ruediger Meier <ruediger.meier@ga-group.nl>

lscpu: cleanup, use _PATH_SYS_CPU/NODE

This make most lines shorter than 80 chars again (tab-witdh 4).

Signed-off-by: Ruediger Meier <ruediger.meier@ga-group.nl>

lscpu: don't abort if cache size is unknown

There are systems where the size file does not exist. Most badly even
lscpu -p would abort allthough it does not use the size:
$ lscpu -p
lscpu: error: cannot open
/sys/devices/system/cpu/cpu0/cache/index0/size: No such file or directory

This patch does not abort in this case and prints "unknown size" in
human-readable case. For examle on this qemu pcc test machine:
$ lscpu
Architecture:          ppc
CPU op-mode(s):        32-bit
Byte Order:            Big Endian
CPU(s):                1
On-line CPU(s) list:   0
Thread(s) per core:    1
Core(s) per socket:    1
Socket(s):             1
Model:                 Power Macintosh
BogoMIPS:              33.25
L1d cache:             unknown size
L1i cache:             unknown size

Signed-off-by: Ruediger Meier <ruediger.meier@ga-group.nl>

findfs: use symbolic exit values, and tell about them in manual

[[kzak@redhat.com: - move return codes to findfs.c]

Signed-off-by: Sami Kerola <kerolasa@iki.fi>
Signed-off-by: Karel Zak <kzak@redhat.com>

mcookie: add --max-size option

Just in case someone wants to add entropy from device with invocation
demonstrated below.

$ mcookie --file /dev/urandom --max-size 64k

[kzak@redhat.com: - use all-io.h]

Signed-off-by: Sami Kerola <kerolasa@iki.fi>
Signed-off-by: Karel Zak <kzak@redhat.com>

mcookie: use control structure, and fix usage()

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

mcookie: use same variable type consistently

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

mcookie: allow --file option be defined multiple times

Earlier only the last option argument took effect.

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

mcookie: use lib/randutils

The mcookie should reuse existing code, and there is definitely no need
to prefer /dev/random for this utility.  See reference for explanation
about later statement.

References: http://www.2uo.de/myths-about-urandom/
Signed-off-by: Sami Kerola <kerolasa@iki.fi>

flock: use nfs4 fallback on EBADF too

The kernel regression (probably v3.4, commit 55725513) introduces a
new errno for O_RDONLY on NFS. Now it returns EBADF rather than EIO.

Addresses: https://bugzilla.redhat.com/show_bug.cgi?id=1078618
Signed-off-by: Karel Zak <kzak@redhat.com>

Merge branch 'fix-compile-color-cfdisk' of https://github.com/theonewolf/util-linux

* 'fix-compile-color-cfdisk' of https://github.com/theonewolf/util-linux:
  cfdisk: fix compile-time error fixes #59

Merge branch 'cfdisk-fix-typo' of https://github.com/rudimeier/util-linux

* 'cfdisk-fix-typo' of https://github.com/rudimeier/util-linux:
  cfdisk: fix typo (undeclared variable since b1f58330)

Merge branch 'master' of https://github.com/rudimeier/util-linux

* 'master' of https://github.com/rudimeier/util-linux:
  tests: try hard to create swaplabel's test image
  build-sys: libmount/python/__init__.py is always a dist file
  tests: return error if failures file not usable
  tests: write tests/failures to build- instead of srcdir

libfdisk: properly implement read-only mode

Don't use fallback to read-only mode in fdisk_context_assign_device(),
it's application responsibility open the device in the right mode.

The commands fdisk and cfdisk check (and report) read-only mode now.

Reported-by: Maciej Małecki <me@mmalecki.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

libfdisk: clean up debug output

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

libfdisk: use new debug functions

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

libblkid: use new debug functions

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

libmount: use new debug functions

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

include/debug: add another debug macros

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

include/debug: fix coding style

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

Merge branch 'common_debug' of https://github.com/ooprala/util-linux

* 'common_debug' of https://github.com/ooprala/util-linux:
  clean up redundant macros and defines
  libfdisk: use include/debug.h
  libblkid: use include/debug.h
  libmount: further debug.h integration
  libcommon: don't mention lib versions in debug macros
  libcommon: define more debugging macros
  libmount: use macros from include/debug.h
  libcommon: add common debugging routines

build-sys: gettexts 0.18 -> 0.18.2 due to MKDIR_P

The autopoint is designed to not use the latest installed bug fix
release, if you specify

AM_GNU_GETTEXT_VERSION([0.18])

in your configure.ac then it really uses 0.18, rather than also
installed 0.18.2 or 0.18.3. It means that bug fix update has no any
effect. Oh...

The AM_PROG_MKDIR_P is deprecated for years, unfortunately still used
in 0.18 gettext m4 stuff. This problem should be fixed in 0.18.2 where
is the correct AC_PROG_MKDIR_P macro.

If you do not have gettext 0.18.2 (or 0.18.3) then update.

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

Merge branch 'fix-unshare-defines' of https://github.com/rudimeier/util-linux

* 'fix-unshare-defines' of https://github.com/rudimeier/util-linux:
  unshare: include libmount.h to provide missing MS_* defines

blkid: convert superblocks to new calling convention

With the previous patch the scanning functions were
modified to return a differentiated return code.
To correctly abort scanning when an I/O error occurs
we need to differentiate between 'error' (= -1) and
'not found' (= 1) in the probe functions.
This patch updates all superblock scanning functions
to the new calling convention.

Signed-off-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Karel Zak <kzak@redhat.com>

blkid: stop scanning on I/O error

Whenever we fail to read from a device it's pointless to
continue with probing; we should be failing immediately.
Otherwise the system will continue logging I/O errors.

This patch updates the probe functions to return the
negative error number on error and BLKID_PROBE_NONE
if not found.

[kzak@redhat.com: - fix s/return errno/return -errno/]

Signed-off-by: Hannes Reinecke <hare@suse.de>
Signed-off-by: Karel Zak <kzak@redhat.com>

cfdisk: fix compile-time error fixes #59

+ For compile to work also needs pull request #56 applied

unshare: include libmount.h to provide missing MS_* defines

Since 6728ca10 we are using MS_PRIVATE and MS_REC which are not defined
in some systems's sys/mount.h.

Signed-off-by: Ruediger Meier <ruediger.meier@ga-group.nl>

cfdisk: fix typo (undeclared variable since b1f58330)

Signed-off-by: Ruediger Meier <ruediger.meier@ga-group.nl>

tests: try hard to create swaplabel's test image

Signed-off-by: Ruediger Meier <ruediger.meier@ga-group.nl>

build-sys: libmount/python/__init__.py is always a dist file

"make dist" and "make distcheck" should work after a bare
./configure to give us a full featured tar ball.

Signed-off-by: Ruediger Meier <ruediger.meier@ga-group.nl>

tests: return error if failures file not usable

In past this could happen when doing make distcheck.

Signed-off-by: Ruediger Meier <ruediger.meier@ga-group.nl>

tests: write tests/failures to build- instead of srcdir

All other test output goes to builddir too and moreover "make distcheck"
complains about writing to it's read-only srcdir.

Signed-off-by: Ruediger Meier <ruediger.meier@ga-group.nl>

include/closestream: don't wipe errno on EPIPE

... the code in close_stdout() is sensitive to EPIPE, so
wipe errno is close_stream() is probably bad idea.

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

dmesg: don't report EPIPE

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

clean up redundant macros and defines

Signed-off-by: Ondrej Oprala <ooprala@redhat.com>

libfdisk: use include/debug.h

Signed-off-by: Ondrej Oprala <ooprala@redhat.com>

libblkid: use include/debug.h

Signed-off-by: Ondrej Oprala <ooprala@redhat.com>

libfdisk: (gpt) use entry attributes as bitmap

 * care about endianess
 * don't use by compiler packed struct with bit array for GPT
   attributes, it's more robust use it as bitmap stored in uint64_t.

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

libmount: further debug.h integration

Signed-off-by: Ondrej Oprala <ooprala@redhat.com>

libcommon: don't mention lib versions in debug macros

Signed-off-by: Ondrej Oprala <ooprala@redhat.com>

libcommon: define more debugging macros

Signed-off-by: Ondrej Oprala <ooprala@redhat.com>

libblkid: convert GPT attributes to host endianess

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

libmount: use macros from include/debug.h

Signed-off-by: Ondrej Oprala <ooprala@redhat.com>

libcommon: add common debugging routines

Signed-off-by: Ondrej Oprala <ooprala@redhat.com>

libfdisk: update GPT partition types

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

mount: update XFS mount.8 allocsize=

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

libfdisk: avoid two compiler warnings about wrong print formats

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

docs: add a new command boilerplate code

Having howtos is good, having an working code that can be used as-is to
kickstart a write of a new command is better.

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

build-sys: sort gitignore files

Use 'LC_ALL=C sort -fu' order for all lists.

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

tests: check swaplabel, and perform more swapon boundary tests

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

mkswap: remove unnecessary usage output

The usage output changes too often making unnecessarily expected output
in test unstable.  Besides in this case the usage ouput does not even
help user, because invalid swap target size error cannot be fixed by
specificing an option.

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

tests: cleanup fallocate

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

Merge branch 'master' of https://github.com/rudimeier/util-linux

* 'master' of https://github.com/rudimeier/util-linux:
  tests, skip fallocate for unsupported filesystems

lib/colors: care about colors only on tty()

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

libblkid: improve comment

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

tests, skip fallocate for unsupported filesystems

tests: refresh fdisk outputs

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

lib/strutils: fix parse_size() return code

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

tests: update fdisk MBR and SUN tests

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

libfdisk: (bsd) fix output

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

cfdisk: follow terminal-colors.d

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

libfdisk: follow requested size rather than use all available space

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

lib/colors: fix colors test

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

cfdisk: fix colors initialization

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

fdisk: add note about colors to the man page

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

cfdisk: fix slang usage

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

build-sys: ove fdisks to disk-utils

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

fdisk: merge fdisk-ask.c to fdisk.c

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

sfdisk: merge partname.c

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

cfdisk: refresh man page

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

cfdisk: update table index when out of range

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

libfdisk: (sun) add info about whole-disk partition

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

cfdisk: fix header

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

libfdisk: (sgi) cleanup, add infor about whole-disk partition

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

libfdisk: don't count special whole-disk partitions

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

cfdisk: coding style, comments, etc.

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

cfdisk: cleanup, rename menuitem struct

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

cfdisk: add usage() and some basic main() stuff

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

cfdisk:add page down/up

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

cfdisk: support partitions list longer than screen LINES

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

cfdisk: improve write dialog

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

cfdisk: add help screen

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

cfdisk: improve add parttype UI

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

cfdisk: add parttype dialog

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

fdisk: remove unnecessary code

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

cfdisk: consolidate menu descriptions

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

cfdisk: menu actions refactoring

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

cfdisk: add create lebel UI

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

cfdisk: consolidate menu move code

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

cfdisk: supports also vertical menus

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

cfdisk: reorder abstract PT according to output

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

lib/tt: keep track about output order

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

lib/tt: add tt_line_get_userdata()

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

libfdisk: add dump functions, make freespace code generic

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

libfdisk: (dos) consolidate partition end-offset code

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

libfdisk: (dos) fix ext_index usage

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

cfdisk: improve tree output

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