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

* 'fixes' of https://github.com/rudimeier/util-linux:
  wipefs: add missing ifdef
  lib/path: remove unused expression
  ipcs,chmem: fix access() usage

fallocate: add man page note about --collapse-range granularity

Reported-by: Radka Skvarilova <rskvaril@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

libblkid: Add very basic APFS recognition

wipefs: add missing ifdef

Seen on OSX:
misc-utils/wipefs.c:822:5: warning: implicit declaration of function 'rereadpt' is invalid in C99 [-Wimplicit-function-declaration]
                                rereadpt(fd, devname);

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

lib/path: remove unused expression

 error: expression result unused [-Werror,-Wunused-value]
                        prefix = optarg, "failed to parse range start";
                                         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Signed-off-by: Ruediger Meier <ruediger.meier@ga-group.nl>

ipcs,chmem: fix access() usage

Some mistakes happened lately when switching from path_exist()
to ul_path_access(). See f09a98de and 8ca31279.

This caused ipcs test failures when running i386 binaries on x86_64
hosts, because the syscall fallback was always used. That's why I
reviewed all similar changes and found another one in chmem.

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

exec_shell: prevent ".: applet not found" error when SHELL env is not set.

When SHELL env is not set, it cause xstrdup(NULL) be executed, and result in weird error message ".: applet not found" instead of /bin/sh being used.

docs: update TODO

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

include/path: make sure PATH_MAX is defined

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

Merge branch 'more' of https://github.com/kerolasa/util-linux

* 'more' of https://github.com/kerolasa/util-linux:
  more: remove function like preprocessor defines
  more: remove global variables, add struct more_control
  more: move couple functions

choom: use new ul_path_* API

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

lib/path: allow dir-path formatting

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

tests: add simple LVM lsblk dump

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

lsblk: fix ul_path_read_string() usage

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

tests: add simple ATA and NVME lsblk dump

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

tests: add lsblk test script

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

lsblk: fix COL_SCHED

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

tests: add simple & stupid script to generate lsblk /sys and /proc dump

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

lsblk: document --sysroot

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

lsblk: fix sysfs_devno_to_devname() usage

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

lsblk: follow --sysroot when read mountpoints

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

lsblk: use prefix for DM name

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

lsblk: add --sysroot

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

docs: add note about lsblk to TODO

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

lsblk: use new ul_path_* API

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

lib/sysfs: cleanup function names

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

blockdev: use new ul_path_* API

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

partx: use new ul_path_* API

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

resizepart: use new ul_path_* API

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

blkzone: use new ul_path_* API

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

fstrim: use new ul_path_* API

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

lib/path: improve debug message on redirect

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

fstrim: add --dry-run

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

fstrim: use new ul_path_* API, improve sysfs context usage

* improve has_discard() function to initialize sysfs context only once
* use new ul_path_* API

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

lib/sysfs: fix sysfs_blkdev_get_wholedisk()

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

lib/path: simplify debug message

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

lscpu: use new ul_path_* API

* use ul_path_* API for /sys/devices/system/cpu paths
* use ul_path_* API for /proc
* rename is_compatible() to is_devtree_compatible() as it works
  with the devices tree only

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

lib/path: add ul_prefix_fopen(), improve cpuset funcs

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

lib/path: don't use extra '/'

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

ipcs: use new ul_path_* API

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

lib/path: add more debug messages

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

chcpu: use new ul_path_* API

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

ctrlaltdel: use new ul_path_* API

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

lib/path: make ul_path_read_ usable with NULL handler

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

rfkill: don't use lib/path

It seems like crazy overkill for this trivial purpose.

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

chmem: use new ul_path_* API

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

lsmem: use new ul_path_* API

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

lib/path: add ul_path_read_buffer()

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

lib/path: add ul_path_get_abspath()

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

zramctl: use new ul_path_* API

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

lib/sysfs: fix ul_path_read_string() usage

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

libfdisk: use new ul_path_* API

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

libblkid: use new ul_path_* API

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

lib/sysfs: add sysfs_blkdev_get_parent()

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

lib/sysfs: make possible to call sysfs_blkdev_deinit_path() in loop

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

lib/path: fix read string to be backwardly compatible

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

lib/path lib/sysfs: add debug

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

lib/loopdev: remove obsolete macro

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

lib/loopdev: use new ul_path_* API

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

lib/sysfs: add ul_new_sysfs_path() shortcut

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

lib/sysfs: new implementation

* reuse ul_path_* API

* allow to use prefix for sysfs paths, so we can use snapshots from
  sysfs for regression tests

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

lib/path: new implementation

The goal is to avoid duplicate code in path.c and sysfs.c and make it
possible to define prefix for paths for all sysfs and procfs based
utils. Now we have /proc snapshots (for tests) for lscpu only. It
would be nice to have the same (for sysfs) for lsblk and another tools.

* very simple API to read numbers, strings and symlinks

* based on openat()

     pc = ul_new_path("/sys/block/sda");
     ul_path_read_u64(pc, &size, "size");
     ul_path_read_u64(pc, &lsz, "queue/logical_block_size");

* printf-like API to generate paths, for example:

     ul_path_readf_u64(pc, &num, "sda%d/size", partno)

* allow to define prefix to redirect hardcoded paths to another
  location, for example:

     pc = ul_new_path("/sys/block/sda");
     ul_path_set_prefix(pc, "/my/regression/dump");
     ul_path_read_u64(pc, &num, "size");

  to read /my/regression/dump/sys/block/sda/size

* allow to extend the API by "dialects", for example for sysfs:

     pc = ul_new_path(NULL);
     sysfs_blkdev_init_path(pc, devno, NULL);

  and use ul_path_* functions to read from @pc initialized by
  sysfs_blkdev_init_path()

* add test_path binary

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

lsblk: add partition table UUID and type fields.

This patch adds PTUUID and PTTYPE fields to lsblk, that are corresponding
fields to ID_PART_TABLE_UUID and ID_PART_TABLE_TYPE in udev database.

[kzak@redhat.com: - small change in PTUUID description]

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

more: remove function like preprocessor defines

Inlining code using preprocessor function like macros is bad for
readability, and prone to errors.  Besides this is a pager, who cares if
code is few milliseconds slower.

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

more: remove global variables, add struct more_control

This is rather big change, but that is the only way to do this sort job.  To
keep this change relatively understandable only moves global variables to a
state structure.

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

more: move couple functions

Earlier commit moved lots of functions to work without declarations, but to
be able to get rid of global variables few more moves is needed.

Reference: a8f98304e6d2f4627ca31f6c661934c92b1095f7
Signed-off-by: Sami Kerola <kerolasa@iki.fi>

partx: exit with error code when partition read failed

Make sure partx exits with a non-0 return code when
it runs into either code-path where getting the partition
table failed (or wasn't even attempted because of previous
error condition).

Change was tested using:
touch /tmp/foobar
partx -s - /tmp/foobar

Previously that was only printing an error/warning message
and then exiting with 0, but after this change it exits
with 1.

Signed-off-by: Andreas Henriksson <andreas@fatal.se>
Reported-by: Juan Céspedes <cespedes@debian.org>
Addresses: https://bugs.debian.org/898426

setarch: add another PER_LINUX32 aliases for ppcle

The big-endian version maps generic "ppc" and "ppc32" aliases to
PER_LINUX32.  It seems that we can do the same for little-endian.

Note that SYS_personality does not care about LE/BE, it's 32 or 64 bit.

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

Merge branch 'hardlink-import' into hardlink

Import hardlink source and manpage version 1.3 from
https://pagure.io/hardlink.git
revision 24da8615f7a9c67648bfe9b6709ba7786d61b89a

fixes for the fixes

Signed-off-by: Francisco Javier Tsao Santín <tsao@gpul.org>

temporal fix before re-patch (updates from Fedora repo)

Update hardlink.1

The example regex in the BUGS section needs its backslashes backslashed.

Fixed version number, added changelog about Todd Lewis' patch

exclude files via pcre

Fixed 32 bit build with gcc7 (RH Bugzilla ID 1422989)

spec file reflects the atomic hardlinking patch; removed cleaning buildroot (redundant); update FSF address at .c source file

Revert "spec file reflects the atomic hardlinking patch; removed cleaning buildroot (redundant); current FSF address at .c source file"

This reverts commit bb9e76ae339794c2243ae294207942b7ea278364.

spec file reflects the atomic hardlinking patch; removed cleaning buildroot (redundant); current FSF address at .c source file

Mention -f option in the man page

do not allow to hardlink files across filesystems by default (#786719)
(use -f option to override)

fix possible buffer overflows, integer overflows, update man page

fix URL and remove mmap() (#676962, #672917)

- update docs to describe highest verbosity -vv option (#210816)
- use dist Resolves: 210816

mostly spec cleanup

add -h option

Document hardlink command line options. (Ville Skytta) (#161738)

- don't try to hardlink 0 byte files (#154404)

kill warning

auto-import hardlink-1.0-1.1 on branch devel from hardlink-1.0-1.1.src.rpm

docs: update TODO

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

mount: add -N to man page helpers section

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

umount: fix behaviour of -A and -R with --namespace

Load /proc/self/mountinfo from correct namespace.

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

umount: allow PID as --namespace argument

[kzak@redhat.com: - update code]

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

mount: allow PID as --namespace argument

[[kzak@redhat.com: - update code]

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

mount: document --namespace in man/help

Signed-off-by: Vaclav Dolezal <vdolezal@redhat.com>

libmount: switch namespace when appropriate

Signed-off-by: Vaclav Dolezal <vdolezal@redhat.com>

libmount: support for namespaces for helpers

Signed-off-by: Vaclav Dolezal <vdolezal@redhat.com>

libmount: make errno usable after mnt_context_set_target_ns()

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

umount: add support for namespaces

Signed-off-by: Vaclav Dolezal <vdolezal@redhat.com>

mount: add support for namespaces

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