blockdev: allow for larger values for start sector

commit 9147d2ad8a ("blockdev: Don't fail on missing start sector") limits
the size of the start sector to 10 digits.

Multi-terrabyte devices can have partitions with a start sector larger than
10 digits, which will cause an sprintf() to abort due to overflowing the buffer.

It causes:
  # blockdev --report /dev/sda4
  RO    RA   SSZ   BSZ   StartSec            Size   Device
  *** buffer overflow detected ***: terminated
  Aborted (core dumped)

Fix typo

build-sys: release++ (v2.37.1)

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

docs: update v2.37.1-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>

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

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

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

lslogins: fix memory leak [asan]

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

lslogins: ask for supplementary groups only once [asan]

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

lslogins: use sd_journal_get_data() in proper way

man sd_journal_get_data:
    The returned data is in a read-only memory map and is only valid until the next invocation
    of sd_journal_get_data().

It means that use data after 3x sd_journal_get_data() is really bad
idea. It also seems better to not assume the fields are zero
terminated as journal API works with void* and size_t to address the
data.

Addresses: https://bugzilla.redhat.com/show_bug.cgi?id=1984704
Signed-off-by: Karel Zak <kzak@redhat.com>

login: add callback for close_range()

Let's make it more robust for old kernels where is no close_range() or
in cases when it's unsuccessful.

Reported-by: Florian Weimer <fweimer@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

lslogins: consolidate and optimize utmp files use

* the original read_utmp() code is from GPLv3+, but lslogins is GPLv2+
  (see https://github.com/karelzak/util-linux/commit/a6bf40ee77a85f5108208dc764b5d023a07998f5#commitcomment-53407151)

* remove redundant parse_btmp() and parse_wtmp() as it does not
  provide anything

* reduce realloc() calls, allocate all in one step

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

hardlink: remove pcre2posix.h support

It's deprecated for years. Let's rely on old good libc regex.h for now.

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

login: fix close_range() use

This new syscall comes with three arguments (see kernel commit
278a5fbaed89dacd04e9d052f4594ffd0e0585de). Not sure why util-linux
assume only two.

Addresses: https://bugzilla.redhat.com/show_bug.cgi?id=1981729
Signed-off-by: Karel Zak <kzak@redhat.com>

tests: update sfdisk reorder test

References: commit f05a5bb2c2e62d3753f8c9262b9f6cbf469e5b0
Signed-off-by: Karel Zak <kzak@redhat.com>

tests: skip if scsi_debug model file is not accessible

* do not print grep(1) info message

* use ts_skip rather than ts_die when scsi_debug is useless

Fixes: https://github.com/karelzak/util-linux/issues/1376
Signed-off-by: Karel Zak <kzak@redhat.com>

fdisk: move reorder diag messages to fdisk_reorder_partitions()

The function fdisk_reorder_partitions() is also used in sfdisk and cfdisk
and these commands assume info/warn messages from the library. So move all
messages from fdisk to fdisk_reorder_partitions().

Signed-off-by: Pali Rohár <pali.rohar@gmail.com>

verity: fix verity.roothashsig only working as last parameter

Parsing of verity.roothashsig did not take into consideration that other options
might follow, and used the whole string as a file path. But mnt_optstr_get_option
just returns a pointer in the mount option string, it doesn't extract it, so it
would have other subsequent options too. The length parameter has to be used.

Signed-off-by: Luca Boccassi <luca.boccassi@microsoft.com>

fdisk: do not print error message when partition reordering is not needed

Option 'f' currently prints following RED error message:

    Nothing to do. Ordering is correct already.
    Failed to fix partitions order.

This change removes RED error message when ordering is already correct.

Signed-off-by: Pali Rohár <pali.rohar@gmail.com>

libblkid: vfat: Fix reading FAT32 boot label

FAT32 can be formatted with boot sign 0x28 to indicate that only serial id
is present or with boot sign 0x29 which indicates that both boot label and
serial id is present.

Signed-off-by: Pali Rohár <pali.rohar@gmail.com>

libblkid: vfat: Fix reading FAT16 boot label and serial id

Older FAT16 variants do not have to contain boot label or serial id. Boot
sign 0x28 indicates that only serial id is present and boot sign 0x29
indicates that both boot label and serial id is present. Other boot sign
values (e.g. zero) indicates older FAT16 variant without boot label and
boot sign.

Signed-off-by: Pali Rohár <pali.rohar@gmail.com>

mount: fix roothash signature extension in manpage

The PKCS7 RFC recommends to use .p7s for detached signatures file extensions,
so use that in the example.

https://datatracker.ietf.org/doc/html/rfc5751#section-3.2.1

meson: fix dlopen support for cryptsetup

dlopen is treated as a dependency, but that's not quite right, it
should be an alternative way to link to libcryptsetup.
Search for it only if cryptsetup is not disabled, and if the cryptsetup-dlopen
is explicitly set to enabled. If it is, do not link to libcryptsetup.

Add cryptsetup support status to the meson summary.

meson: fix crypt_activate_by_signed_key detection

It should set HAVE_CRYPT_ACTIVATE_BY_SIGNED_KEY, rather than HAVE_CRYPTSETUP
which enables the verity features, as it needs to detect the availability of
the crypt_activate_by_signed_key API

build-sys: display cryptsetup status after ./configure

Signed-off-by: Luca Boccassi <bluca@debian.org>

lscpu: don't use DMI if executed with --sysroot

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

test/eject: guard asan LD_PRELOAD with use-system-commands check

This test tries to add asan to LD_PRELOAD because the just-built eject
will call the host /bin/umount, and apparently asan doesn't like this.

However, if ldd isn't present, this fails as the path to asan is the
error message saying that ldd isn't present.

As the asan workaround is only needed when executing the binaries that
have just been built and not the system binaries, only use it if the
test is on the built binaries.

Closes #1373.

Signed-off-by: Ross Burton <ross.burton@arm.com>

fstrim: clean return code on --quiet-unsupported

This feature is already supported for -a and -A. Let's support it also
when FS specified on command line.

Addresses: https://github.com/systemd/mkosi/pull/721
Signed-off-by: Karel Zak <kzak@redhat.com>

lsblk: normalize space in SERIAL and MODEL

These strings come from HW and udev does not modify it. It seems good
idea to clean up it before we print it.

Fixes: https://github.com/karelzak/util-linux/issues/1368
Signed-off-by: Karel Zak <kzak@redhat.com>

tests: mark ul/ul as a known failure

As with ul/basic, this test produces different output when ran under
different terminals, which isn't very useful.

Set TS_KNOWN_FAIL so that these problems don't cause the test to fail.

Signed-off-by: Ross Burton <ross.burton@arm.com>

tests: check correct log file for errors in blkdiscard test

When checking if the discard ioctl is supported, look in stderr not stdout as
that is where the error appears.

Signed-off-by: Ross Burton <ross.burton@arm.com>

tests: don't hardcode /bin/kill in the kill tests

If the 'kill' test is executed with --use-system-commands, it calls
/bin/kill to avoid the shell's own kill command being invoked.

However, this doesn't work if the kill we want to test isn't in fact in
/bin.  Instead, use $(which kill) to find a kill on the PATH and call
that directly.

Signed-off-by: Ross Burton <ross.burton@arm.com>

cfdisk: optimize mountpoint detection for PARTUUID

Don't check fstab (and udev symplinks) for new UUIDs.

Fixes: https://github.com/karelzak/util-linux/issues/1331
Signed-off-by: Karel Zak <kzak@redhat.com>

hexdump: do not use atoi()

Addresses: https://github.com/karelzak/util-linux/issues/1358
Signed-off-by: Karel Zak <kzak@redhat.com>

pg: do not use atoi()

Addresses: https://github.com/karelzak/util-linux/issues/1358
Signed-off-by: Karel Zak <kzak@redhat.com>

misc: improve string to number conversions

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

include/strutils: consolidate string to number conversion

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

more: fix null-pointer dereference

The command allows executing arbitrary shell commands while viewing a file by
entering '!' followed by the command. Entering a command that contains a '%',
'!', or '\' causes a segmentation violation.

The same more(1) function has a problem when not file is specified (cat
/etc/passwd | more) on command line.

Addresses: https://bugzilla.redhat.com/show_bug.cgi?id=1975153
Signed-off-by: Karel Zak <kzak@redhat.com>

agetty: do not use atol()

Addresses: https://github.com/karelzak/util-linux/issues/1358
Signed-off-by: Karel Zak <kzak@redhat.com>

lscpu: do not use atoi()

Addresses: https://github.com/karelzak/util-linux/issues/1358
Signed-off-by: Karel Zak <kzak@redhat.com>

utmpdump: do not use atoi()

Addresses: https://github.com/karelzak/util-linux/issues/1358
Signed-off-by: Karel Zak <kzak@redhat.com>

swapon: do not use atoi()

Addresses: https://github.com/karelzak/util-linux/issues/1358
Signed-off-by: Karel Zak <kzak@redhat.com>

ipcs: do not use atoi()

Addresses: https://github.com/karelzak/util-linux/issues/1358
Signed-off-by: Karel Zak <kzak@redhat.com>

eject: do not use atoi()

* remove unnecessary strtok() use
* remove atoi use()

Addresses: https://github.com/karelzak/util-linux/issues/1358
Signed-off-by: Karel Zak <kzak@redhat.com>

fsck: do not use atoi()

Addresses: https://github.com/karelzak/util-linux/issues/1358
Signed-off-by: Karel Zak <kzak@redhat.com>

include/strutils: cleanup strto..() functions

* add ul_strtos64() and ul_strtou64()
* add simple test

Addresses: https://github.com/karelzak/util-linux/issues/1358
Signed-off-by: Karel Zak <kzak@redhat.com>

blockdev: improve arguments parsing (remove atoi)

old version:
 # blockdev --setfra 4x096 /dev/sdc

new version:
 # blockdev --setfra 4x096 /dev/sdc
 blockdev: failed to parse command argument: '4x096'

Addresses: https://github.com/karelzak/util-linux/issues/1358
Signed-off-by: Karel Zak <kzak@redhat.com>

cfdisk: do not use atoi()

It's unnecessary to use atoi in this case.

Addresses: https://github.com/karelzak/util-linux/issues/1358
Signed-off-by: Karel Zak <kzak@redhat.com>

readprofile: check errno after strto..()

Addresses: https://github.com/karelzak/util-linux/issues/1356
Signed-off-by: Karel Zak <kzak@redhat.com>

wipefs: check errno after strto..()

Addresses: https://github.com/karelzak/util-linux/issues/1356
Signed-off-by: Karel Zak <kzak@redhat.com>

lslocks: check errno after strto..()

Addresses: https://github.com/karelzak/util-linux/issues/1356
Signed-off-by: Karel Zak <kzak@redhat.com>

kill: check errno after strto..()

Addresses: https://github.com/karelzak/util-linux/issues/1356
Signed-off-by: Karel Zak <kzak@redhat.com>

lslogins: check errno after strto..()

Addresses: https://github.com/karelzak/util-linux/issues/1356
Signed-off-by: Karel Zak <kzak@redhat.com>

fsck: check errno after strto..()

Addresses: https://github.com/karelzak/util-linux/issues/1356
Signed-off-by: Karel Zak <kzak@redhat.com>

lsmem: check errno after strto..()

Addresses: https://github.com/karelzak/util-linux/issues/1356
Signed-off-by: Karel Zak <kzak@redhat.com>

lscpu: check errno after strto..()

Addresses: https://github.com/karelzak/util-linux/issues/1356
Signed-off-by: Karel Zak <kzak@redhat.com>

hwclock: check errno after strto..()

Addresses: https://github.com/karelzak/util-linux/issues/1356
Signed-off-by: Karel Zak <kzak@redhat.com>

ipcs: check errno after strto..()

Addresses: https://github.com/karelzak/util-linux/issues/1356
Signed-off-by: Karel Zak <kzak@redhat.com>

libuuid: check errno after strto..()

Addresses: https://github.com/karelzak/util-linux/issues/1356
Signed-off-by: Karel Zak <kzak@redhat.com>

libmount: check errno after strto..()

Addresses: https://github.com/karelzak/util-linux/issues/1356
Signed-off-by: Karel Zak <kzak@redhat.com>

libblkid: check errno after strto..()

Addresses: https://github.com/karelzak/util-linux/issues/1356
Signed-off-by: Karel Zak <kzak@redhat.com>

mount.8.adoc: Remove context options exclusion

The exclusivity between the {fscontext, defcontext} and context options
was removed in kernel 2.6.25[1]. No specific verification on these
options is done in mount(8)[2].

[1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=c9180a57a9ab2d5525faf8815a332364ee9e89b7
[2] https://github.com/karelzak/util-linux/blob/master/libmount/src/context_mount.c#L202

fsck.cramfs: use open+fstat rather than stat+open

Fixes: https://github.com/karelzak/util-linux/issues/1353
Signed-off-by: Karel Zak <kzak@redhat.com>

include/c: add drop_permissions(), consolidate UID/GID reset

Fixes: https://github.com/karelzak/util-linux/issues/1354
Signed-off-by: Karel Zak <kzak@redhat.com>

dmesg: remove  condition [lgtm scan]

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

tools: report and use LDFLAGS in tools/config-gen

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

build-sys: use $LIBS rather than LDFLAGS

Fixes: https://github.com/karelzak/util-linux/pull/1349
Signed-off-by: Karel Zak <kzak@redhat.com>

lib/path: (test) fix ul_new_path() use

more: add __format__ attribute

Reported-by: Jan Pazdziora <jpazdziora@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

wall: add __format__ attribute

Reported-by: Jan Pazdziora <jpazdziora@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

script: add __format__ attribute

Reported-by: Jan Pazdziora <jpazdziora@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

ldattach: add __format__ attribute

Reported-by: Jan Pazdziora <jpazdziora@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

hwclock: fix ul_path_scanf() use

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

eject: add __format__ attribute

Reported-by: Jan Pazdziora <jpazdziora@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

logger: add __format__ attribute

Reported-by: Jan Pazdziora <jpazdziora@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

libfdisk: add and fix __format__ attributes

Reported-by: Jan Pazdziora <jpazdziora@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

findmnt: add __format__ attribute

Reported-by: Jan Pazdziora <jpazdziora@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

libmount: add __format__ attribute

Reported-by: Jan Pazdziora <jpazdziora@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

include/strutils: fix __format__attribute

Reported-by: Jan Pazdziora <jpazdziora@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

include/path: add __format__attribute

Reported-by: Jan Pazdziora <jpazdziora@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

include/c: add __format__ attribute

Reported-by: Jan Pazdziora <jpazdziora@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

tests: fix lsns test on kernels without USER namespaces

Reported-by: Bruce Dubbs <bruce.dubbs@gmail.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

mount.8: fix overlayfs nfs_export= indention

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

include/strv: fix format attributes

Reported-by: Jan Pazdziora <jpazdziora@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

Fix some brackets in synopses

mount: man-page; add all overlayfs options

The section in man (8) mount for overlay is missing nine options which
aren't documented elsewhere either and are useful features to be aware
of and use.

Fixes #1350.

build-sys: Update configure.ac

1. the test incorrectly used AC_COMPILE_IFELSE instead of
   AC_LINK_IFELSE, defeating the purpose of checking -lcrypt.
2. the test did not properly restore LIBS, causing later checks to all
   fail if libcrypt wasn't found.
3. HAVE_LIBCRYPT only controls whether to use -lcrypt, it is not
   needed or used in any source files.

[kzak@redhat.com: - improve commit message
                  - use UL_{SET,RESTORE}_FLAGS() rather than directly
    modify $LIBS]

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

lib/path: improve ul_path_readlink() to be more robust

According to POSIX, readlink() makes no effort to null-terminate buffer
with the result. It seems better to hide this disadvantage in the
ul_path_...() API rather than assume buf[sz] = '\0' everywhere.

Reported-by: Reported-by: Jan Pazdziora <jpazdziora@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

mkfs.cramfs: add comment to explain readlink() use

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

rename: use readlink() in more robust way

Reported-by: Jan Pazdziora <jpazdziora@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

namei: simplify code

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

more: fix setuid/setgid order

The rule is pretty simple, always use setgid() before setuid().

Reported-by: Jan Pazdziora <jpazdziora@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

sulogin: add missing ifdefs

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

sulogin: use explicit_bzero() for buffer with password

Reported-by: Jan Pazdziora <jpazdziora@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

login: remove obsolete and confusing comment

Reported-by: Jan Pazdziora <jpazdziora@redhat.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

mkswap: fix holes detection (infinite loop and/or stack-buffer-underflow)

Reported-by: Brian Lane <bcl@redhat.com>
Addresses: https://bugzilla.redhat.com/show_bug.cgi?id=1971877
Fix: https://github.com/karelzak/util-linux/issues/1348
Signed-off-by: Karel Zak <kzak@redhat.com>

lib/loopdev: perform retry on EAGAIN

Suggested-by: Ingo Franzki <ifranzki@linux.ibm.com>
Signed-off-by: Karel Zak <kzak@redhat.com>

mount.8.adoc: document SELinux use of nosuid mount flag

Using mount flag `nosuid` also affects SELinux domain transitions but
this has not been documented well.

Signed-off-by: Topi Miettinen <toiwoton@gmail.com>