strtoday.c: remove unused defines.h inclusion

Signed-off-by: Iker Pedrosa <ipedrosa@redhat.com>

strtoday.c: remove USE_GETDATE as it was always used

Signed-off-by: Iker Pedrosa <ipedrosa@redhat.com>

strtoday.c: remove POSIX 1995 conditional dependency

Since the project is supposed to be POSIX.1-2001 compliant it doesn't
make sense to have that added conditionally.

Signed-off-by: Iker Pedrosa <ipedrosa@redhat.com>

Assume struct tm is defined in <time.h>

It has been a requirement since at least C90, according to tm(3type).

Link: <https://github.com/shadow-maint/shadow/pull/600>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Cc: Serge Hallyn <serge@hallyn.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume struct stat has st_atim and st_mtim fields

That's required by POSIX.1-2008.

Link: <https://github.com/shadow-maint/shadow/pull/600>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Cc: Serge Hallyn <serge@hallyn.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Remove USE_SYSLOG preprocessor conditional, which was always defined

Reported-by: Iker Pedrosa <ipedrosa@redhat.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Remove dead code

Signed-off-by: Alejandro Colomar <alx@kernel.org>

Don't redefine errno(3)

It is Undefined Behavior to declare errno (see NOTES in its manual page).
Instead of using the errno dummy declaration, use one that doesn't need
a comment.

Signed-off-by: Alejandro Colomar <alx@kernel.org>

Fix typos in length calculations

Link: <https://github.com/shadow-maint/shadow/pull/607>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Use 'uintmax_t' to print 'gid_t'

This is shorter to write than 'unsigned long int', so we can collapse
some lines.  It is guaranteed by C99.

Link: <https://github.com/shadow-maint/shadow/pull/607>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Fix off-by-one mistakes

The buffers have a size of 512 (see xmalloc() above), which is what
snprintf(3) expects.

Link: <https://github.com/shadow-maint/shadow/pull/607>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Cosmetic fixes

Previous commits, to keep readability of the diffs, left the code that
was previously wrapped by preprocessor coditionals untouched.  Apply
some minor cosmetic changes to merge it in the surrounding code.

Signed-off-by: Alejandro Colomar <alx@kernel.org>

Remove traces of utmpx

-  USER_NAME_MAX_LENGTH was being calculated in terms of utmpx.  Do it
   in terms of utmp.
-  Remove utmpx support from the whishlist.
-  Remove unused tests about utmpx members.

Signed-off-by: Alejandro Colomar <alx@kernel.org>

Disable utmpx permanently

On Linux, utmpx and utmp are identical.  However, documentation (manual
pages) covers utmp, and just says about utmpx that it's identical to
utmp.  It seems that it's preferred to use utmp, at least by reading the
manual pages.

Moreover, we were defaulting to utmp (utmpx had to be explicitly enabled
at configuration time).  So, it seems safer to just make it permanent,
which should not affect default builds.

Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume <utmpx.h> always exists

We already made that assumption in commit b47aa1e9aaf4.  While the
header is not required by POSIX (it is an XSI extension), it is defined
in systems that are of interest to this project (GNU/Linux).

Fixes: b47aa1e9aaf4 ("Assume <utmpx.h> exists")
Cc: Serge Hallyn <serge@hallyn.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Remove pwdauth.c

We don't know what it was for.  If anyone cares, it's in git history.
In my distro, there seem to be no traces of it:

alx@debian:~$ apt-file find pwdauth
alx@debian:~$

Link: <https://codesearch.debian.net/search?q=pwdauth&literal=1>
Link: <https://github.com/shadow-maint/shadow/pull/612>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Iker Pedrosa <ipedrosa@redhat.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Add support for skeleton files from /usr/etc/skel

This patch is used by openSUSE to make useradd look for
skeleton files in /usr/etc/skel additionally to /etc/skel
in accordance with
https://uapi-group.org/specifications/specs/base_directory_specification/

Fix useradd audit event logging of ID field

When useradd sends its ADD_USER event, it is filling in the id field. This is not yet written to disk. When auditd sees the event and the log format is enriched, auditd tries to lookup the user name but it does not exist. This causes the event to never be resolvable since ausearch relies on the lookup information attached by auditd.

The fix is to not send the id information for any event until after close_files() is called. Just the acct field is all that is

Patch by Steve Grubb (afaik).

Reported at https://bugzilla.redhat.com/show_bug.cgi?id=1713432

Remove comments that survived the Helicoprion

The OSes that are referred to by these comments, are extinct, but
their comments survived, fossilized in amber.

Reported-by: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Ping? :)

Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Remove preprocessor conditionals that are always true

In a previous commit, we made USE_TERMIOS unconditionally defined.
Let's just remove it, and remove the condition everywhere.

Reported-by: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Remove code conditional on USE_TERMIO

The definition for this macro was removed in a previous commit.

Reported-by: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume socket(2) exists

It is required by POSIX.1-2001.

Link: <https://github.com/shadow-maint/shadow/pull/600>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume inet_ntoa(3) exists

It is required by POSIX.1-2001.

Link: <https://github.com/shadow-maint/shadow/pull/600>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume F_* and SEEK_* macros are defined

They are required by POSIX.1-2001.

Link: <https://github.com/shadow-maint/shadow/pull/600>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Remove code conditional on S_SPLINT_S

I don't know for sure what that is, but it's redefining setlocale(3)
and LC_ALL, which is are defined by C99, so it's supect of being some
variety of an extinct dynosaur.  Maybe related to the Dodo.

Link: <https://github.com/shadow-maint/shadow/pull/600>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume strdup(3) exists

It is required by POSIX.1-2001.

Link: <https://github.com/shadow-maint/shadow/pull/600>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume strcasecmp(3) exists

It is required by POSIX.1-2001.

Link: <https://github.com/shadow-maint/shadow/pull/600>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume rmdir(2) exists

It is required by POSIX.1-2001.

Link: <https://github.com/shadow-maint/shadow/pull/600>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume mkdir(2) exists

It is required by POSIX.1-2001.

Link: <https://github.com/shadow-maint/shadow/pull/600>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume B[0-9]* macros are defined

All of the macros we're using are required by POSIX.1-2001.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume SIGTTOU is defined

It is required by POSIX.1-2001.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume SIGTSTP is defined

It is required by POSIX.1-2001.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume RLIMIT_STACK is defined

It is required by POSIX.1-2001.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume RLIMIT_NOFILE is defined

It is required by POSIX.1-2001.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume RLIMIT_FSIZE is defined

It is required by POSIX.1-2001.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume RLIMIT_DATA is defined

It is required by POSIX.1-2001.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume RLIMIT_CPU is defined

It is required by POSIX.1-2001.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume RLIMIT_AS is defined

It is required by POSIX.1-2001.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume RLIMIT_CORE is defined

It is required by POSIX.1-2001.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume getgrgid_r(3) exists

It is required by POSIX.1-2001.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume getgrnam_r(3) exists

It is required by POSIX.1-2001.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume getpwuid_r(3) exists

It is required by POSIX.1-2001.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume getpwnam_r(3) exists

It is required by POSIX.1-2001.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume fsync(2) exists

It is required by POSIX.1-2001.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume fchown(2) exists

It is required by POSIX.1-2001.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume fchmod(2) exists

It is required by POSIX.1-2001.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume l64a(3) exists

It is required by POSIX.1-2001.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume <netdb.h> exists

It is required by POSIX.1-2001.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Remove preprocessor conditionals that are always true

Since the last commit, LIMITS is always defined.  Remove the dummy
macro, and all conditionals on it.

Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume <sys/resource.h> exists

It is required by POSIX.1-2001.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Remove uses of ulimit(3)

The function is obsolete.  It is recommended to use getrlimit(2) instead
(see the manual page for ulimit(3) or the POSIX manual for it).  Since
getrlimit(2) is required by POSIX.1-2001, we can rely on it.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Add indentation to heavy use of preprocessor conditionals

This clarifies which code is under which conditions,
for further clenaup.

Signed-off-by: Alejandro Colomar <alx@kernel.org>

Remove unused check for <utime.h>

Signed-off-by: Alejandro Colomar <alx@kernel.org>

Remove unused check for <syslog.h>

Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume <termios.h> exists

It is required by POSIX.1-2001.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume <utmpx.h> exists

It is required by POSIX.1-2001.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Remove unused check for <sys/time.h>

Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume <unistd.h> exists

It is required by POSIX.1-2001.

Cc: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Remove unused check for <fcntl.h>

Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume strstr(3) exists

ISO C99 requires strstr(3).

Link: <https://github.com/shadow-maint/shadow/pull/600>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume snprintf(3) exists

ISO C99 requires snprintf(3).

Link: <https://github.com/shadow-maint/shadow/pull/600>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume rename(2) exists

ISO C99 requires rename(2).

Link: <https://github.com/shadow-maint/shadow/pull/600>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume NULL exists

ISO C99 requires NULL.

Link: <https://github.com/shadow-maint/shadow/pull/600>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume strerror(3) exists

ISO C99 requires strerror(3).

Link: <https://github.com/shadow-maint/shadow/pull/600>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume fputs(3) exists

ISO C99 requires fputs(3).

Link: <https://github.com/shadow-maint/shadow/pull/600>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume <locale.h> exists

ISO C99 requires <locale.h>.

Other files in the project already include <locale.h> unconditionally,
so it's reasonable to assume that it is always available.

Link: <https://github.com/shadow-maint/shadow/pull/600>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Remove unused check for <limits.h>

Link: <https://github.com/shadow-maint/shadow/pull/600>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume <errno.h> exists

ISO C99 requires <errno.h>.

Many files in the project already include <errno.h> unconditionally,
so it's reasonable to assume that it is always available.

Link: <https://github.com/shadow-maint/shadow/pull/600>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Assume <stdbool.h> exists

ISO C99 requires <stdbool.h>.

Many files in the project already include <stdbool.h> unconditionally,
so it's reasonable to assume that it is always available.

Link: <https://github.com/shadow-maint/shadow/pull/600>
Cc: Christian Göttsche <cgzones@googlemail.com>
Cc: Iker Pedrosa <ipedrosa@redhat.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Add '62_usermod_remove_supplementary_groups' test case to test runner scripts (run_some)

Add '62_usermod_remove_supplementary_groups' test case to test runner scripts (run_all, run_all.coverage)

Add regression test for 'usermod -rG' -- it should not add users to groups they did not previously belong to

Preparation / clarity: rename existing usermod test from 'remove_supplemental_groups' to 'clear_supplemental_groups'

agetpass: Hook into build-system

Signed-off-by: Guillem Jover <guillem@hadrons.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Hide [[gnu::malloc(deallocator)]] in a macro

Clang doesn't implement this attribute and reports an error.  Work
around it by hiding it in a macro that will be empty in clang.

Reported-by: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Replace the deprecated getpass(3) by our agetpass()

getpass(3) is broken in all implementations; in some, more than
others, but somewhat broken in all of them.  Check the immediate
previous commit, which added the functions, for more details.
Check also the Linux man-pages commit that marked it as
deprecated, for more details:
7ca189099d73bde954eed2d7fc21732bcc8ddc6b.

Link: <https://git.kernel.org/pub/scm/docs/man-pages/man-pages.git/commit?id=7ca189099d73bde954eed2d7fc21732bcc8ddc6b>
Reported-by: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

libmisc: agetpass(), erase_pass(): Add functions for getting passwords safely

There are several issues with getpass(3).

Many implementations of it share the same issues that the infamous
gets(3).  In glibc it's not so terrible, since it's a wrapper
around getline(3).  But it still has an important bug:

If the password is long enough, getline(3) will realloc(3) memory,
and prefixes of the password will be laying around in some
deallocated memory.

See the getpass(3) manual page for more details, and especially
the commit that marked it as deprecated, which links to a long
discussion in the linux-man@ mailing list.

So, readpassphrase(3bsd) is preferrable, which is provided by
libbsd on GNU systems.  However, using readpassphrase(3) directly
is a bit verbose, so we can write our own wrapper with a simpler
interface similar to that of getpass(3).

One of the benefits of writing our own interface around
readpassphrase(3) is that we can hide there any checks that should
be done always and which would be error-prone to repeat every
time.  For example, check that there was no truncation in the
password.

Also, use malloc(3) to get the buffer, instead of using a global
buffer.  We're not using a multithreaded program (and it wouldn't
make sense to do so), but it's nice to know that the visibility of
our passwords is as limited as possible.

erase_pass() is a clean-up function that handles all clean-up
correctly, including zeroing the entire buffer, and then
free(3)ing the memory.  By using [[gnu::malloc(erase_pass)]], we
make sure that we don't leak the buffers in any case, since the
compiler will be able to enforce clean up.

Link: <https://git.kernel.org/pub/scm/docs/man-pages/man-pages.git/commit?id=7ca189099d73bde954eed2d7fc21732bcc8ddc6b>
Reported-by: Christian Göttsche <cgzones@googlemail.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Don't 'else' after a 'noreturn' call

Signed-off-by: Alejandro Colomar <alx@kernel.org>

CI: add libbsd and pkg-config dependencies

Signed-off-by: Iker Pedrosa <ipedrosa@redhat.com>
Co-developed-by: Alejandro Colomar <alx@kernel.org>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Fix HAVE_SHADOWGRP configure check

The missing #include <gshadow.h> causes the configure check to fail
spuriously, resulting in HAVE_SHADOWGRP not being defined even
on systems that actually have sgetsgent (such as current glibc).

Allow supplementary groups to be added via config file

Allow supplementary groups to be set via the /etc/default/useradd config
file. Allowing an administrator to set additonal groups via the GROUPS
configurable and control the default behaviour of useradd.

useradd: check if subid range exists for user

Check if a user already has a subid range before assigning one.

Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=2012929

Signed-off-by: Iker Pedrosa <ipedrosa@redhat.com>

Release 4.13

Signed-off-by: Serge Hallyn <serge@hallyn.com>

update changelog

Don't warn when failed to open /etc/nsswitch.conf

Maybe we should have a debug mode where it's still printed, but
we don't, so let's be quieter.

Closes #557

useradd: Fix buffer overflow when using a prefix

The buffer length did not count the string's trailing null byte.

Signed-off-by: David Michael <fedora.dm0@gmail.com>

adduser-old.c: Remove program

This program has 10 calls to gets(3) according to grep(1).  That
makes it a very unsafe program which should not be used at all.
Let's kill the program already.

See what gets(3) has to say:

SYNOPSIS
       #include <stdio.h>

       [[deprecated]] char *gets(char *s);

DESCRIPTION
       Never use this function.

       ...

BUGS
       Never use gets().  Because it is impossible to tell with‐
       out knowing the  data  in  advance  how  many  characters
       gets()  will  read,  and  because gets() will continue to
       store characters past the end of the buffer,  it  is  ex‐
       tremely dangerous to use.  It has been used to break com‐
       puter security.  Use fgets() instead.

       For more information, see CWE‐242 (aka "Use of Inherently
       Dangerous Function") at http://cwe.mitre.org/data/defini‐
       tions/242.html

Acked-by: "Serge E. Hallyn" <serge@hallyn.com>
Signed-off-by: Alejandro Colomar <alx@kernel.org>

Raise limit for passwd and shadow entry length

Moreover, include checks to prevent writing entries longer than the
length limit.

Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1422497

Signed-off-by: Tomáš Mráz <tm@t8m.info>
Signed-off-by: Iker Pedrosa <ipedrosa@redhat.com>

lastlog: check for localtime() return value

Signed-off-by: Tomáš Mráz <tm@t8m.info>
Signed-off-by: Iker Pedrosa <ipedrosa@redhat.com>

man: add missing space in useradd.8.xml

Resolves: https://github.com/shadow-maint/shadow/issues/580

Signed-off-by: Iker Pedrosa <ipedrosa@redhat.com>

fix `usermod -rG x y` while user `y` is not in group `x` will cause user `y` add into group `x`

libmisc: minimum id check for system accounts

The minimum id allocation for system accounts shouldn't be 0 as this is
reserved for root.

Signed-off-by: Tomáš Mráz <tm@t8m.info>
Signed-off-by: Iker Pedrosa <ipedrosa@redhat.com>

usermod: report error if homedir does not exist

Report error if usermod asked for moving homedir and it does not exist.

Signed-off-by: Tomáš Mráz <tm@t8m.info>
Signed-off-by: Iker Pedrosa <ipedrosa@redhat.com>

chage: Fix regression in print_date

Introduced by c6c8130db4319613a91dd07bbb845f6c33c5f79f

After removing snprintf, the format string should get unescaped once.

Fixes #564

Reporter and patch author: DerMouse (github.com/DerMouse)

Use libc MAX() and MIN()

glibc, musl, FreeBSD, and OpenBSD define the MAX() and MIN()
macros in <sys/param.h> with the same definition that we use.
Let's not redefine it here and use the system one, as it's
effectively the same as we define (modulo whitespace).

See:

shadow (previously):

alx@asus5775:~/src/shadow/shadow$ grepc -ktm MAX
./lib/defines.h:318:#define MAX(x,y) (((x) > (y)) ? (x) : (y))

glibc:

alx@asus5775:~/src/gnu/glibc$ grepc -ktm -x 'sys/param.h$' MAX
./misc/sys/param.h:103:#define MAX(a,b) (((a)>(b))?(a):(b))

musl:

alx@asus5775:~/src/musl/musl$ grepc -ktm -x 'sys/param.h$' MAX
./include/sys/param.h:19:#define MAX(a,b) (((a)>(b))?(a):(b))

OpenBSD:

alx@asus5775:~/src/bsd/openbsd/src$ grepc -ktm -x 'sys/param.h$' MAX
./sys/sys/param.h:193:#define MAX(a,b) (((a)>(b))?(a):(b))

FreeBSD:

alx@asus5775:~/src/bsd/freebsd/freebsd-src$ grepc -ktm -x 'sys/param.h$' MAX
./sys/sys/param.h:333:#define MAX(a,b) (((a)>(b))?(a):(b))

Signed-off-by: Alejandro Colomar <alx@kernel.org>

Don't test for NULL before calling free(3)

free(3) accepts NULL, since the oldest ISO C.  I guess the
paranoid code was taking care of prehistoric implementations of
free(3).  I've never known of an implementation that doesn't
conform to this, so let's simplify this.

Remove xfree(3), which was effectively an equivalent of free(3).

Signed-off-by: Alejandro Colomar <alx@kernel.org>

updated Dutch translation

lib: use memzero where applicable

Use memzero when operating in a buffer of known size to clear all bytes
and avoid leaking the size of the stored data.

lib: use strzero where applicable

Replace `memzero (s, strlen(s))` with just the internal wrapper
`strzero (s)` where the underlying allocated size is not known.