build-sys: release++ (v2.31-rc2)

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

docs: update v2.31-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>

rfkill: support old "list" output

.. just because we want to be replacement for the original version.

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

rfkill: remove duplicate code

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

rfkill: refactor actions

* convert action string to ID and use switch()
* add note about default output to the man page

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

rfkill: improve default output

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

docs: add Atari PT to the TODO list

Addresses: https://github.com/karelzak/util-linux/issues/517
Signed-off-by: Karel Zak <kzak@redhat.com>

libmount: add human compatible message for EBADMSG errno

 mount: /media/sdb5: mount(2) system call failed: Bad message.

is really ugly for end users. It seems XFS, extN (etc) use EBADMSG for
bad checksums. For network or pseudo filesystems continue to use "Bad
message" error...

Addresses: https://bugzilla.redhat.com/show_bug.cgi?id=1496764
Signed-off-by: Karel Zak <kzak@redhat.com>

Merge branch 'fix-python_libmount_get_propagation' of https://github.com/giuseppe/util-linux

* 'fix-python_libmount_get_propagation' of https://github.com/giuseppe/util-linux:
  libmount: fix Python get_propagation

Merge branch 'patch-1' of https://github.com/sgargel/util-linux

* 'patch-1' of https://github.com/sgargel/util-linux:
  Fix end translation

losetup: rename to --sector-size and LOG-SEC

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

lib/randutils: remove superfluous continue

Addresses: https://github.com/karelzak/util-linux/commit/fa949792072530b49ef5760d7ecd6d250c66e35f#commitcomment-24678670
Signed-off-by: Karel Zak <kzak@redhat.com>

Fix end translation

See also https://translate.google.it/#en/it/end

losetup: Add support for logical block size

Kernel since 4.14 supports setting of logical block size[1]. It allows to
create loop devices that report logical block size different from 512.

Add support for this feature to losetup.

References:
[1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/drivers/block/loop.c?id=89e4fdecb51cf5535867026274bc97de9480ade5

[kzak@redhat.com: - fix loopcxt_get_blocksize()
                  - remove lo_blocksize from loop_info64]

Signed-off-by: Stanislav Brabec <sbrabec@suse.cz>
Cc: Ming Lei <ming.lei@redhat.com>
Cc: Hannes Reinecke <hare@suse.com>
Cc: Omar Sandoval <osandov@fb.com>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Karel Zak <kzak@redhat.com>

docs: add exec() exit codes to TODO

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

cfdisk: cleanup die-on-signal code

Let's keep signal handler simple and don't cleanup terminal from the
handler.

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

cfdisk: make resize code more readable

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

setproctitle: fix out of boundary access

A program using setproctitle can trigger an out of boundary access
if an attacker was able to clear the environment before execution.

The check in setproctitle prevents overflows, but does not take into
account that the whole length of the arguments could be 1, which is
possible by supplying such a program name to execlp(3) or using a
symbolic link, e.g. argv[0] = "l", argv[1] = NULL.

Only login uses setproctitle, which is not affected by this
problem due to initializing the environment right before the call.

login: fix signal race

The functions warnx(3) and gettext(3) are not safe to use within signal
handlers and should be avoided. Preparing the message beforehand and
calling write(2) as well as calling _exit(2) solves the problem.

[kzak@redhat.com: - use program_invocation_short_name rather than argv[0],
                  - use ignore_result() to keep compiler happy]

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

docs: add 2.30.2 to ReleaseNotes

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

build-sys: release++ (v2.31-rc1)

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

docs: update v2.31 release notes

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

po: merge changes

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

po: Fix translation in ja.po

English message and Japanese translation are unmatch.

Signed-off-by: Masanari Iida <standby24x7@gmail.com>

po:Fix a typo

Signed-off-by: Masanari Iida <standby24x7@gmail.com>

Merge branch '170915' of github.com:jwpi/util-linux

* '170915' of github.com:jwpi/util-linux:
  hwclock: make debug implicit for test mode.
  hwclock: exit bug fix and single test mode message

libmount: fix Python get_propagation

The current implementation would return always 0, as it is the return
code of mnt_fs_get_propagation.  Change the implementation to raise an
exception on an error and return the propagation flags otherwise.

Signed-off-by: Giuseppe Scrivano <gscrivan@redhat.com>

hwclock: make debug implicit for test mode.

Calling --test without --debug is not useful, so
make it implicit.

Signed-off-by: J William Piggott <elseifthen@gmx.com>

hwclock: exit bug fix and single test mode message

Bug fix: hwclock returns success when saving /etc/adjtime fails.

Remove redundant test mode messages.

Signed-off-by: J William Piggott <elseifthen@gmx.com>

Docs: add hwclock 'highlights' v2.31-ReleaseNotes

Signed-off-by: J William Piggott <elseifthen@gmx.com>

Docs: grammar fixes for v2.31-ReleaseNotes

Signed-off-by: J William Piggott <elseifthen@gmx.com>

po: fix ipcrm translation error fr.po

Addresses: https://github.com/karelzak/util-linux/issues/512
Signed-off-by: Karel Zak <kzak@redhat.com>

docs: update v2.31-ReleaseNotes

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

docs: remove UUID v3 and v5 from TODO

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 pt_BR.po (from translationproject.org)

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

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

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

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

tests: update build-sys tests

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

build-sys: fix non-blkid compilation

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

build-sys: make rfkill optional (and enabled by default)

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

build-sys: add libtinfow check

It seems some systems differentiate between tinfo and tinfow. And it
seems that mix ncursesw and tinfo (wide vs. non-wide char) is problem
for the systems.

Note that for example Fedora have ncursesw as well as ncurses, but
only one tinfo library. So, we need fallback this scenario.

Reported-by: Stanislav Brabec <sbrabec@suse.cz>
Signed-off-by: Karel Zak <kzak@redhat.com>

su: (pty) improve SIGSTOP/SIGCONT semantic

We want to use waitpid() only when child is terminated or stopped to
pick up child status, otherwise PTY proxy has to be active. This is
difference between "su" and "su --pty". For "su" we keep parent all
time in waitpid().

It would be possible to use separate code based on signalfd_siginfo,
but it seems better to keep all this stuff on one place -- it  means
wait_for_child().

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

su: (pty) save child status

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

su: (pty) simplify stdin usage in poll()

Not sure why I have problem with this years ago for script(1), but it
seems .fd=-1 is really enough to the ignore the FD.

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

script: simplify stdin usage in poll()

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

su: (pty) fix slave terminal attributes initialization

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

su: (pty) fix child signal mask usage

The signal mask is used by pty_init_slave(), but it has never been
uninitialized before fork(), so child gets 0 as a mask :-(

Note that script(1) has no this issue because it opens signal-fd
before fork().

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

runuser: fix linking

Add conditional -lutil to runuser when needed to avoid linking error.

login-utils/su-common.o: In function `pty_create':
login-utils/su-common.c:269: undefined reference to `openpty'
login-utils/su-common.c:273: undefined reference to `openpty'

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

su: fix non-pty compilation

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

su: enable TIOCSCTTY and minor changes

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

su: add more informartion to man page

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

su: add PTY support

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

su: add --pty option

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

su: move parent signals setup to separate function

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

su: make wait_for_child() usable in arbitrary situation

For example if called more than once; to keep PTY code simple and
robust.

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

su: keep old sigactions in control struct

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

su: fix refactoring bug in child PID usage

Fix child pid cleanup according to commit 0076012563ff34e294a6166d605118bcdd35f7e1.

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

su: add child to control struct

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

su: fix compiler warnings [-Wimplicit-fallthrough=]

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

su: add wait_for_child()

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

su: unblock signals is all initialized

This patch a little bit reorders signals initialization. The original
code unblocks SIGINT SIGQUIT before signal handler is set for the
signals. It means there is a small possible race.

It seems better to compose wanted mask, setup handlers and then
unblock all the wanted signals.

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

su: clean up signals usage

- don't use magic numbers to index old actions
- don't use if () if ()
- make if() conditions more readable

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

su: clean up const usage

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

su: fix debug message

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

su: properly clear child PID

The patch from master branch, somehow lost during su refactoring
rebase.

Reported-by: Tobias Stöckmann <tobias@stoeckmann.org>
Signed-off-by: Karel Zak <kzak@redhat.com>

su: improve some debug messages

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

su: add debug.h stuff

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

su: use switch() to split after fork

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

su: move PAM session initialization to separate function

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

su: rename evaluate_uid()

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

su: cosmetic code changes

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

su: cleanup setuid/gid related code

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

su: cleanup setenv related code

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

su: cleanup shell related code

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

su: cleanup usernames usage

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

lib/pwdutils: add xgetlogin()

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

su: cleanup pwd struct usage

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

login: add xgetpwnam()

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

su: consolidate tty name usage

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

su: reorder functions

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

su: cleanup PAM stuff

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

su: cleanup usage()

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

su: add control struct

* setup logindefs loader by function rather than by global pointer

* move basic booleans to the su_context struct

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

su: cosmetic changes after indent

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

su: fix indention

Yes, I know... this patch is horrible. We all hate this in git
history, but the original indention from coreutils has been so
crazy...

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

docs: add ASAN to TODO list

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

bash-completion: update uuidgen, wipefs, tunelp, setpriv, and hwclock

uuidgen: Add hash-based UUIDs to bash-completion.  These were added in
commit c6f1ec68a8b38863efff2a18e30b7272db4fb273.

wipefs: Command started to use libsmartcols, and it got some new options.
Commit d9921b2a128f17aa9a9a110444d5faa43e1b371c.

rename: New option was added in commit fabb90676af89bcb5f429793746f141c0.

tunelp: The --trust-irq was removed it being years broken.  Commit
d52eb4bd9062081c8420b22c029c694f29d036c7.

setpriv: Add --ambient-caps option from commit 0c92194eeee9c1fd58580ef852.
In same go fix 'bash set -u' issue, that is the same as mentioned commit
abbcec4fc9c8d7fb835b4eafd1bc9d82acbf0056.

hwclock: For some reason --get has always been missing from this file.

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

docs: remove repeated word from mount(8) man page

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

libsmartcols: fix scols_line_move_cells() n+1 error [asan]

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

Merge branch 'patch-1' of https://github.com/aner-perez/util-linux

libmount: export locking errors as MNT_EX_FILEIO

The mount man page assumes locking errors mapped to MNT_EX_FILEIO (16)
return code. Unfortunately, this is internally not exported as a
special error code, so it's returned as a generic (errno based)
stuff. This patch fixes this issue.

Note that we still use locking for example for utab or when enabled
/etc/mtab (disabled by default).

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

libsmartcols: fix heap-buffer-overflow when move columns

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

uuidparse:fix stack-buffer-overflow [asan]

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

swaplabel: fix --disable-libuuid compilation

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