setarch: use personality() system call when it is available

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

setarch: reindent code

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

hwclock: remove referal to deprecated keyboard interface

The KDGHWCLK has been gone quite a while.

Reference: http://marc.info/?l=linux-kernel&m=104171103925897
Signed-off-by: Sami Kerola <kerolasa@iki.fi>

eject: make open_device() and select_speed() to use struct eject_control

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

eject: add struct eject_control to remove global variables

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

mountpoint: simplify if statement

Returning straight after print_devno() makes the code to be more obvious
and removes need for long else statement.

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

mkfs.minix: fix couple compiler warnings

disk-utils/mkfs.minix.c:366:3: warning: ISO C forbids 'return' with
expression, in function returning void [-Wpedantic]

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

mountpoint: add struct mountpoint_control

This unifies variable names in different functions, and removes redundant
stat() calls.

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

last: improve code readability by renaming variable names

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

last: make is_phantom() when kernel config does not include audit support

When kernel CONFIG_AUDIT is not set the /proc/<pid>/loginuid information
is not present resulting live sessions to be marked 'gone - no logout' in
last(1) print out.  To go-around this change makes last(1) to look
/dev/<tty> device ownership as a substitute of loginuid.

The go-around seems to work fairly well, but it has it short comings.
For example after closing a X window session the /dev/ttyN file seems to
be owned by root, not the user who had it before entering to the X
session.  While that is suboptimal it is still better than an attmempt to
determine uid_t by looking owner of the /proc/<struct utmp ut_pid>, that
is a login(1) process running as root.

The issue was found using Archlinux installation.

$ pacman -Qi linux
Name           : linux
Version        : 3.16-2
[...]
Build Date     : Mon Aug 4 18:06:51 2014

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

lib: remove xgetpass()

This function is not in use, and it has reference to obsoleted getpass().

Reference: http://man7.org/linux/man-pages/man3/getpass.3.html
Signed-off-by: Sami Kerola <kerolasa@iki.fi>

include: simplify fputc_careful() in carefulputc.h

New version of comparisions does the same thing as the old without being
quite as difficult to understand.

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

libuuid: add extern qualifiers to uuid/uuid.h system header

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

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

build-sys: use -lutil for BSD only

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

ipcrm: add hacks to avoid FTBFS

Signed-off-by: Andreas Henriksson <andreas@fatal.se>

agetty: fixing FTBFS on !linux (Debian util-linux 2.20.1-1.2)

Handle vc flags missing on FreeBSD
Fix tty creation on kFreeBSD taking patch from 2.19

Addresses-Debian-Bug: #650185

ipcs: remaining kFreeBSD hackery for building.

Addresses-Debian-Bug: #527384
Signed-off-by: LaMont Jones <lamont@debian.org>

lib/blkdev: report correct disk size on GNU/kFreeBSD. Thanks Tuco.

Addresses-Debian-Bug: 592292

libmount: hide details about failed search in fstab/mtab

The current code returns -errno when not found "mount /foo" in fstab
and mtab does not exist (or /etc/mtab points to non-mounted /proc).

This is problem because the return value is too low-level and maybe
misinterpreted by top level code. It's better to always return
MNT_ERR_NOFSTAB when not found in fstab/mtab.

Reported-by: Dylan Cali <calid1984@gmail.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

lscpu: fix cppcheck warnings

CppCheck founds a few wrong arguments in format strings and a NULL
pointer dereference.

Amended version with fixed strcmp() usage.

Signed-off-by: Boris Egorov <egorov@linux.com>

fdformat: fixing copy'n'paste issue in the manual

The end track default is media specific, not zero.

zramctl: fixed two -Wformat-security warnings

(they prevented compilation if -Werror=format-security is enabled)

Signed-off-by: Steven Honeyman <stevenhoneyman@users.noreply.github.com>

travis: whitelist branches

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

lsblk: add notes about udev to the man page

Reported-by: Francis Moreau <francis.moro@gmail.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

build-sys: create parent directory for $(PATHFILES)

When building outside the source tree there is nothing to guarantee the
target directory exists before writing to it. Most of the time this just
happens to work because something else creates the directory but not
always. For example:

    $ mkdir build
    $ cd build
    $ ../configure --disable-dependency-tracking
    $ make libuuid/uuid.pc
      GEN      libuuid/uuid.pc
    /bin/sh: line 2: libuuid/uuid.pc.tmp: No such file or directory
    make: *** [libuuid/uuid.pc] Error 1

libfdisk: add support for sfdisk-like script format

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

libfdisk: (gpt) follow label-id from script

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

libfdisk: add functions to apply partitiong scripts

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

libfdisk: (gpt) don't align last possible LBA

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

libfdisk: rename fdisk_dump_ to fdisk_script_

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

libfdisk: fix fdisk_delete_all_partitions()

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

libfdisk: add fdisk_apply_table() and fdisk_delete_all_partitions()

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

libfdisk: improve debug messages

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

libfdisk: accept dump with quotes

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

libfdisk: (dos) follow bootable setting when add partition

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

libfdisk: (gpt) allow to specify uuid and name when create new partition

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

libfdisk: (gpt) set_{name,uuid} functions refactoring

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

libfdisk: add reference to context to dump

 * add fdisk_ref_context() function

 * context is necessary to parse label specific partition types

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

libfdisk: use reference counting for context

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

libfdisk: extend dump parsing function

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

libfdisk: (dos) fix typo

Reported-by: https://github.com/Mikachu
Signed-off-by: Karel Zak <kzak@redhat.com>

include/optutils: fix typo

Reported-by: Rafael Ferreira <rafael.f.f1@gmail.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

cfdisk: fix typo in the previous commit

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

cfdisk: wrong interpretation of yes in non-english language

Reported-by: tnut@nutyx.org
Signed-off-by: Karel Zak <kzak@redhat.com>

libfdisk: (dos) fix logical partitions add/delete code

If you delete logical partition and then create a new one than fdisk
(and cfdisk) might write EBR to the first sector on the device. That's
wrong of course; because you will lost MBR (primary partitions).

(Probably introduced by commit bcddbe96882b88d53b6bc0495e7322c0820a5122
 where code completely clears EBR stuff.)

Reported-by: Michał Górny <mgorny@gentoo.org>
Signed-off-by: Karel Zak <kzak@redhat.com>

libfdisk: vary basic dump parsing functions

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

libblkid: move string trim function to strutils.h

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

libfdisk: basic fdisk_dump_* functions

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

libfdisk: remove debug dump function, cleanup API

The fdisk_dump_* prefix will be used for sfdisk-like functionality.
The patch also add FDISK_ prefix to fdisk_get_unit() options.

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

mountpoint: clarify that it works on files

fef4dfa174 removed the restriction that the argument must explicitly be
a directory, so let's fix up the documentation to reflect that.

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

mountpoint: drop S_ISDIR check

Files can be mountpoints, too.

libmount: use -t for type.subtype in helpers API

References: https://github.com/karelzak/util-linux/issues/116
Signed-off-by: Karel Zak <kzak@redhat.com>

tests: apply travis clang PATH workaround

As seen in Travis build log the clang is not found, causing one of the
tests to fail.  This change adds a workaround from Travis issue 2607 to
util-linux.

Reference: https://travis-ci.org/karelzak/util-linux/jobs/33056101#L3630
Reference: https://github.com/travis-ci/travis-ci/issues/2607
Signed-off-by: Sami Kerola <kerolasa@iki.fi>

logger: use generic --id=<id>

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

Merge remote-tracking branch 'sami/logger'

libfdisk: fix fdisk_get_optimal_iosize() and update tests

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

tests: fix fdisk/bsd for the two possible sectors/offsets

BSD disklabels depend on the endianess, but also on the architecture as
one can see in include/pt-bsd.h with the BSD_LABELSECTOR and
BSD_LABELOFFSET #define. This cause the testsuite to fail on the
affected architectures as the disk image is not the same as the expected
result.

Commit 180b3a7e tried to fix the endianess, but the special PowerPC case
has been chosen as a reference, so it still fails for example on MIPS BE
or S/390.

This patch fixes the testsuite by converting the md5sums to the expected
values, still depending on the endianess, but also for the two possible
sectors/offsets. This has been tested on Alpha, MIPS, PowerPC and x86-64,
so this should cover all 4 cases.

Cc: Andreas Henriksson <andreas@fatal.se>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

docs: bring the chrt and taskset man pages closer to standard formatting

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

textual: make the license of chrt and taskset slightly more explicit

This makes it match the license of the man pages.
While there, also tweak some other comment lines.

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

docs: make the man-page licenses of chrt and taskset match their intent

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

docs: improve the description of --size in the man page of zramctl

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

docs: mentintion default udp and tcp ports logger is using

The fact that 'syslog tcp/514' does not exist in RFS's, which has lead
to 'syslog-conn 601/tcp' be used in place could be a suprice and should
be told in manual.

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

logger: set function arguments read-only when possible

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

logger: optimize string initializations

Setting whole array to be completely full of nulls cannot be as quick as
making the only significant member of the array when needed.

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

logger: remove openlog(3) options

One variable less, and more importantly bit operations become unnecessary
in if statements.

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

logger: allow use of --id=ppid when logging locally

There is no obvious way to make syslog(3) to print both pid or ppid, so
duplicate the libc syslog() to logger.  Making the ppid printing work
using unix socket has side effect of local becoming capable to use both
rfc format output, which is hopefully seen as good thing.  The
syslog_local() is format wise one-to-one copy with glibc syslog(3)
format.

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

fdisk: add independent debug stuff

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

libfdisk: move fdisk_ask_partnum() to API

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

fdisk: use libfdisk public API only

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

libfdisk: add geometry to API, cleanup

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

cfdisk: use only public libfdisk API

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

libfdisk: fdisk_label_is_labeltype()

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

libfdisk: more docs

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

libfdisk: final parttype API cleanup

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

libfdisk: move get_parttypes to label API

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

libfdisk: (gpt) allow to specify attr bit by API

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

libfdisk: make it possible to get fields for all labes

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

libfdisk: cleanup label geometry API

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

libfdisk: cleanup fdisk_is_label stuff

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

libfdisk: clean up API (alignment.c)

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

libfdisk: clean up API (context.c)

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

cfdisk: add libfdisk independent debug stuff

Now it's possible to use
CFDISK_DEBUG=0xffff ./cfdisk
to debug only cfdisk UI.

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

tests: add debug tests

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

include/debug: allow to speficy empty masknames

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

include/debug: cleanup masks

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

include/debug: make masknames optional

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

libs/debug: accept human readable names for _DEBUG=

For example
$ LIBMOUNT_DEBUG=tab,cache findmnt

to debug only TAB and CACHE subsystem.

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

libfdisk: rename fdisk_column to fdisk_field

.. to make the API and fdisk(s) code more readable.

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

libfdisk: remove dependence on libsmartcols

It's application business to convert libfdisk_table to string.

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

lib/colors: add function to return color from scheme

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

libblkid: (gpt) fix PT offset

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

Fix bash-completion Makemodule options

tests: make ipcs limit calculations work when numbers are large

Test suite  failed when I ran it with kernel 3.16.  Error is the
following.

tests/ts/ipcs/limits2: line 31: [: 18446744073692774399: integer expression expected

Reference: http://lkml.iu.edu/hypermail/linux/kernel/1406.0/01869.html
Signed-off-by: Sami Kerola <kerolasa@iki.fi>

swapon: remove invalid error string print outs

To avoid messages like this:

$ swapon /dev/zram0
swapon: /dev/zram0: read swap header failed: Success

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

textual: remove some inconsistent periods from error messages

While doing so, also improve translatability and some wordings.

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

lscpu: clean up vmware inline asm

This code is not PIC clean which means it fails to build on hardened
32bit x86 systems (i.e. building as PIE).

While here, optimize the existing cpuid logic slightly.

URL: https://bugs.gentoo.org/518936
Signed-off-by: Mike Frysinger <vapier@gentoo.org>

zramctl: improve option combination error messaging

Use of --algorithm or --streams is not expected to work without
reseting device, that happens when --size is specified, so ensure the
former options to be combined with the later.

[kzak@redhat.com: - make it more generic]

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

zramctl: add bash completion script

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

zramctl: cannonicalize devname when no fullpath specified

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