prctl.2: Rework the PR_SET_PDEATHSIG description a little, for easier readability

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

prctl.2: ffix

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

ptrace.2: BUGS: ptrace() may set errno to zero

ptrace() with requests PTRACE_PEEKTEXT, PTRACE_PEEKDATA and
PTRACE_PEEKUSER can set errno to zero. AFAICS this is for a good
reason (so that you can tell the difference between a successful
PEEK with a result of -1 and a failed PEEK, even if you forget to
clear errno yourself), but it technically violates the rules
described in the errno.3 manpage.

glibc snippet from sysdeps/unix/sysv/linux/ptrace.c:

  res = INLINE_SYSCALL (ptrace, 4, request, pid, addr, data);
  if (res >= 0 && request > 0 && request < 4)
    {
      __set_errno (0);
      return ret;
    }

reproducer:

$ cat ptrace_test.c
char foobar_data[4] = "ABCD";
int main(void) {
  pid_t child = fork();
  if (child == -1) err(1, "fork");
  if (child == 0) {
    if (prctl(PR_SET_PDEATHSIG, SIGKILL)) err(1, "prctl");
    while (1) sleep(1);
  }
  int status;
  if (ptrace(PTRACE_ATTACH, child, NULL, NULL)) err(1, "attach");
  if (waitpid(child, &status, 0) != child) err(1, "wait");
  errno = EINVAL;
  unsigned int res = ptrace(PTRACE_PEEKDATA, child, foobar_data, NULL);
  printf("errno after PEEKDATA: %d\n", errno);
  printf("PEEKDATA result: 0x%x\n", res);
}
$ gcc -o ptrace_test ptrace_test.c -Wall
$ ./ptrace_test
errno after PEEKDATA: 0
PEEKDATA result: 0x44434241

Signed-off-by: Jann Horn <jannh@google.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

clone.2: Pending CLONE_NEWPID prevents thread creation

See copy_process() in kernel/fork.c:

if (clone_flags & CLONE_THREAD) {
if ((clone_flags & (CLONE_NEWUSER | CLONE_NEWPID)) ||
    (task_active_pid_ns(current) !=
current->nsproxy->pid_ns_for_children))
return ERR_PTR(-EINVAL);
}

current->nsproxy->pid_ns_for_children is where unshare(CLONE_NEWPID)
stashes the pending namespace.

Signed-off-by: Jann Horn <jannh@google.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

fanotify.7: wfix

Use "FAN_OPEN_PERM" consistently rather than "FAN_PERM_OPEN".

Signed-off-by: Anthony Iliopoulos <ailiopoulos@suse.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

system.3: Note that system() can fail for the same reasons as fork(2)

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

system.3: ffix

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

system.3: Mention that 'errno' is set on error

Corresponding system.3p already mentions that.
Tested with glibc and musl.

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

proc.5: Minor clean-ups for Alan Jenkins' patch

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

proc.5: wfix

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

proc.5: Vmalloc information is no longer calculated (Linux 4.4)

See Linux commit a5ad88ce8c7fae7ddc72ee49a11a75aa837788e0,
"mm: get rid of 'vmalloc_info' from /proc/meminfo".

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

pid_namespaces.7: Clarify the semantics for the adoption of orphaned processes

Because of setns() semantics, the parent of a process may reside
in the outer PID namespace. If that parent terminates, then the
child is adopted by the "init" in the outer PID namespace (rather
than the "init" of the PID namespace of the child).

Thus, in a scenario such as the following, if process M
terminates, P is adopted by the init process in the initial
PID namespace, and if P terminates, Q is adopted by the init
process in the inner PID namespace.

    +---------------------------------------------+
    | Initial PID NS                              |
    |                           +---------------+ |
    |  +-+                      | inner PID NS  | |
    |  |1|                      |               | |
    |  +-+                      |    +-+        | |
    |                           |    |1|        | |
    |                           |    +-+        | |
    |                           |               | |
    |  +-+   setns(), fork()    |    +-+        | |
    |  |M|----------------------+--> |P|        | |
    |  +-+                      |    +-+        | |
    |                           |     | fork()  | |
    |                           |     v         | |
    |                           |    +-+        | |
    |                           |    |Q|        | |
    |                           |    +-+        | |
    |                           +---------------+ |
    +---------------------------------------------+

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

clone.2, prctl.2, st.4, proc.5: Change references to '2.6.0-test*' series kernels to just '2.6.0'

The extra detail has little of noting with -test 2.6.0
added a particular feature has little value these days,
and is likely to confuse some readers who don't know
(and probably don't care) about the historical details.

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

getgroups.2: Note that a process can drop all groups with: setgroups(0, NULL)

Checking the FreeBSD source code, there's explicit support for
this to accommodate non-BSD systems (such as Linux).

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

signal.7: Unify signal lists into a signal table that embeds standards info

Having the signals listed in three different tables reduces
readability, and would require more table splits if future
standards specify other signals.

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

signal.7: Insert standards info into tables

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

signal.7: Place signal numbers in a separate table

The current tables of signal information are unwieldy,
as they try to cram in too much information.

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

signal.7: ffix

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

futex.2: Make the example use C11 atomics rather than GCC builtins

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

getaddrinfo.3: Fix off-by-one error in example client program

Reported-by: Eric Sanchis <eric.sanchis@iut-rodez.fr>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

pthread_rwlockattr_setkind_np.3: tfix

Reported-by: G. Branden Robinson <g.branden.robinson@gmail.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

pthread_rwlockattr_setkind_np.3: Remove bug notes

The notes in pthread_rwlockattr_setkind_np.3 imply there is a bug
in glibc's implementation of PTHREAD_RWLOCK_PREFER_WRITER_NP (a
non-portable constant anyway), but this is not true. The
implementation of PTHREAD_RWLOCK_PREFER_WRITER_NP is made almost
impossible by the POSIX standard requirement that reader locks be
allowed to be recursive, and that requirement makes writer
preference deadlock without an impossibly complex requirement that
we track all reader locks. Therefore the only sensible solution
was to add PTHREAD_RWLOCK_PREFER_WRITER_NONRECURSIVE_NP and
disallow recursive reader locks if you want writer preference.

This patch removes the bug description and documents the current
state and recommendations for glibc. I have also updated bug 7057
with this information, answering Steven Munroe's almost 10 year
old question :-) I hope Steven is enjoying his much earned
retirement.

Should we move the glibc discussion to some footnote? Some libc
may be able to implement the requirement to avoid deadlocks in the
future, but I doubt it (fundamental CS stuff).

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

ioctl_userfaultfd.2, madvise.2, memfd_create.2, migrate_pages.2, mmap.2, shmget.2, subpage_prot.2, userfaultfd.2, malloc.3, proc.5, sysfs.5, tmpfs.5: Update paths for in-kernel memory management documentation files

Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

capabilities.7: Update URL for libcap tarballs

The previous location does not seem to be getting updated.
(For example, at the time of this commit, libcap-2.26
had been out for two months, but was not present at
http://www.kernel.org/pub/linux/libs/security/linux-privs.

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

prctl.2: Note libcap(3) APIs for operating on ambient capability set

(However, the libcap APIs do not yet seem to have
manual pages...)

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

prctl.2: Mention libcap APIs for operating on capability bounding set

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

syscalls.2: Update syscall list for Linux 4.18

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

system.3: Use '(char *) NULL' rather than '(char *) 0'

Reported-by: Jonny Grant <jg@jguk.org>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

ioctl_userfaultfd.2, userfaultfd.2: wfix

Use "UFFDIO_ZEROPAGE" consistently rather than "UFFDIO_ZERO".

Signed-off-by: Anthony Iliopoulos <ailiopoulos@suse.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

system.3: ffix

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

system.3: wfix

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

user_namespaces.7: tfix

Signed-off-by: Jakub Wilk <jwilk@jwilk.net>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

pivot_root.2: Minor fixes to Elvira Khabirova's patch

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

pivot_root.2: Explain the initramfs case and point to switch_root(8).

Based on text from Documentation/filesystems/ramfs-rootfs-initramfs.txt.

Signed-off-by: Elvira Khabirova <lineprinter@altlinux.org>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

resolver.3: Add documentation of res_nclose()

I found the neccessary information in the glibc sources.

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

strfry.3: Remove incorrect reference to rand(3)

The strfry(3) function does not use rand(). The original version
from 1995 did, but it was changed to use a different PRNG in glibc
commit 4770745624b7f7f25623f1f10d46a4c4d6aec25c, 1996-12-04.

This C program demonstrates the behavior. By not calling srand(),
it gets the same values for successive calls to rand(), but
strfry() returns a different value each time the program is run.
If strfry() called srand(), it would alter the sequence of numbers
return by rand().

int main(void) {
    printf("%d\n", rand());
    char alphabet[] = "abcdefghijklmnopqrstuvwxyz";
    puts(strfry(alphabet));
    printf("%d\n", rand());
}

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

signal.7: tfix

Reported-by: Helge Deller <deller@gmx.de>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

signal.7: Reorder the architectures in the signal number lists

x86 and ARM are the most common architectures, but currently
are in the second subfield in the signal number lists.
Instead, swap that info with subfield 1, so the most
common architectures are first in the list.

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

signal.7: Add signal numbers for parisc

This patch adds the signal numbers for parisc to the signal(7) man page.

Those parisc-specific values for the various signals are valid since the
Linux kernel upstream commit ("parisc: Reduce SIGRTMIN from 37 to 32 to
behave like other Linux architectures") during development of kernel 3.18:
http://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=1f25df2eff5b25f52c139d3ff31bc883eee9a0ab

Signed-off-by: Helge Deller <deller@gmx.de>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

syscalls.2: parisc Linux does not any longer emulate HP-UX

Initially it was planned that the parisc linux port would natively
support 32-bit HP-UX binaries, but this compatibility was never
reached and finally dropped with Linux kernel 3.14.

With that background, drop parisc from the list of of platforms
which supports it's proprietary operating-system.

Additional notes from mtk:

The most relevant commit from the Linux 3.14 change log was:

[[
commit f5a408d53edef3af07ac7697b8bc54a755628450
Author: Guy Martin <gmsoft@tuxicoman.be>
Date:   Thu Jan 16 17:17:53 2014 +0100

    parisc: Make EWOULDBLOCK be equal to EAGAIN on parisc

    On Linux, only parisc uses a different value for EWOULDBLOCK which
    causes a lot of troubles for applications not checking for both values.
    Since the hpux compat is long dead, make EWOULDBLOCK behave the same as
    all other architectures.
]]

Additional notes from Helge:

The patch above is the initial and most important one with which
we stopped the HP-UX compatibility.

Then, with this commit in kernel 3.18 there is no way back:
"parisc: Reduce SIGRTMIN from 37 to 32 to behave like
other Linux architectures"
commit 1f25df2eff5b25f52c139d3ff31bc883eee9a0ab

And in kernel 4.0 we finally dropped the HP-UX compat layer
from Linux kernel source code with the commit series
"parisc: hpux - Drop support for HP-UX binaries":
commit 04c1614977168fb8f002e2d81f704eeabe0c5ebd

Signed-off-by: Helge Deller <deller@gmx.de>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

syscall.2: parisc needs care with syscall parameters

On parisc one needs to take care of the 32-bit calling conventions
with 64-bit syscall parameters on a 32-bit kernel. So on parisc we
suffer from the same issues like ARM, PowerPC and Xtensa.

Signed-off-by: Helge Deller <deller@gmx.de>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

cgroups.7: Minor fix: bump kernel version to 4.19 in a couple of points

The stated points still hold true as at Linux 4.1.

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

io_submit.2: tfix

Signed-off-by: Josh Gao <jmgao@google.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

iconv.1: SEE ALSO: add uconv(1)

Refer to uconv(1) in iconv(1) manual page, it is helpful
transliterating e.g. Cyrillic to Latin:

echo <some-cyrillic-text> | uconv -f UTF-8 -t UTF-8 -x cyrillic-latin

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

proc.5: ffix: break /proc/[pid]/task text into smaller paragraphs

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

proc.5: srcfix

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

proc.5: tfix

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

proc.5: Minor reworking of description of /proc/[tid] directories

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

proc.5: wfix

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

rename.2: tfix

Reported-by: G. Branden Robinson <g.branden.robinson@gmail.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

rename.2: wfix: s/shmem/tmpfs/ in discussion of RENAME_NOREPLACE

Reported-by: Jakub Wilk <jwilk@jwilk.net>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

rename.2: XFS support for RENAME_NOREPLACE was added in Linux 4.0

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

rename.2: Rework list of supported filesystems for RENAME_NOREPLACE

There was probably a little too much detail in
Lukas Werkmeister's patch. Simplify, by removing a few
file systems, and arrange the information as a bulleted
list for easier readability.

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

rename.2: Add kernel versions for RENAME_NOREPLACE support

The RENAME_NOREPLACE flag was added with the initial release of the
renameat2 syscall in Linux 3.15, but support for most filesystems was
only added in later versions, and some may still not support it.

Signed-off-by: Lucas Werkmeister <mail@lucaswerkmeister.de>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

vdso.7: tfix

Escape hyphens; use \(aq for ASCII apostrophes.

Signed-off-by: Jakub Wilk <jwilk@jwilk.net>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

getmntent.3: Clarify that endmntent() should be used rather than fclose()

This doesn't actually matter on any C library I know of --- they
all just do a NULL check and forward to fclose(3). (The actual
mistake I saw was someone not realizing that they had to call
*anything*.)

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

ferror.3: Warn about closing the result of fileno()

Since adding checking to Android's bionic for file descriptor
double-closes, we've found that the most common cause of these
bugs is incorrect use of fileno(3). There appears to be a common
misconception that it transfers ownership of the file descriptor
to the caller.

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

getrlimit.2: Resource limits are process-wide attributes shared by all threads

This was already noted in pthreads(7), but bears repeating here.

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

proc.5: Document /proc/PID/status CoreDumping field

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

proc.5: tfix

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

proc.5: /proc/[pid]/status VmPMD field was removed in Linux 4.15

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

proc.5: Document /proc/Meminfo LazyFree field

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

proc.5: Fix kernel source pathname for soft-dirty documentation

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

proc.5: Document /proc/meminfo KReclaimable field

Added in Linux 4.20.

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

mount.2: Document EROFS for read-only filesystems

See https://bugzilla.kernel.org/show_bug.cgi?id=200649

Reported-by: Harry Mallon <hjmallon@gmail.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

mount.2: EACCES: note some reasons why a filesystem may be read-only

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

prctl.2: Add some further historical details on PR_SET_MM_EXE_FILE

Also some minor tweaks to Benjamin Peterson's patch.

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

prctl.2: PR_SET_MM_EXE_FILE may now be used as many times as desired

The original implementation of PR_SET_MM_EXE_FILE only allowed it
to be used once in a process's lifetime. This restriction was
lifted in Linux commit 3fb4afd9a504c2386b8435028d43283216bf588e
("prctl: remove one-shot limitation for changing exe link").

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

prctl.2: wfix: Remove a redundant sentence

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

proc.5: Document /proc/[pid]

See also https://bugzilla.kernel.org/show_bug.cgi?id=201441

Reported-by: Philip Dumont <phil@solidstatescientific.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

proc.5: Add an overview section describing the groups of files under /proc

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

lirc.4: Clarify the description LIRC_SET_REC_TIMEOUT

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

msgctl.2, semctl.2, shmctl.2: Some small wording improvements for Davidlohr Bueso's patch

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

msgctl.2: tfix

Reported-by: Joe Lawrence <joe.lawrence@redhat.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

msgctl.2, semctl.2, shmctl.2: ffix

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

msgctl.2, semctl.2, shmctl.2: Document STAT_ANY commands

Reported-by: Joe Lawrence <joe.lawrence@redhat.com>
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

lirc.4: Minor language fix-ups

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

lirc.4: Remove crufty text

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

lirc.4: Minor wording and formatting fixes

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

lirc.4: tfix

Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

lirc.4: Document remaining ioctl (LIRC_GET_REC_TIMEOUT)

Now all ioctls are documented.

Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

cgroups.7: tfix

Reported-by: Alan Jenkins <alan.christopher.jenkins@gmail.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

user_namespaces.7: ffix

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

lp.4: ffix

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

namespaces.7: tfix

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

lirc.4: wfix

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

lirc.4: wfix

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

lirc.4: Document error returns more explicitly

Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

lirc.4: Minor fix-ups for Sean Young's previous patch

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

lirc.4: Both ioctls are supported if LIRC_CAN_SET_REC_CARRIER_RANGE is set

Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

lirc.4: LIRC_CAN_SET_REC_DUTY_CYCLE_RANGE was never supported

No driver ever supported such a thing.

Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

lirc.4: Minor reworking of Sean Young's LIRC_SET_REC_TIMEOUT patch

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

lirc.4: Timeout reports are enabled by default

Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

lockf.3: ERRORS: add EINTR

Ian Turner: The exact return calls are at the discretion of the
underlying VFS, but I'm pretty sure that EINTR is a possibility.
Or, if it's not, then the flock() manpage should be amended
accordingly, since the two share the same underlying
implementation.

mtk: lockf(3) is implemented on top of fcntl() locking, so
EINTR is of course a possibility.

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

bpf.2: SEE ALSO: add bpf-helpers(7)

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

capabilities.7: Minor fixes to Marcus Gelderie's patch

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

capabilities.7: Mention header for SECBIT constants

Mention that the named constants (SECBIT_KEEP_CAPS and others)
are available only if the linux/securebits.h user-space header
is included.

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

mmap.2: tfix

Reported-by: Thomas Posch <kernel.org@online.posch.name>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

zic.8: Sync from tzdb upstream

Make zic.8 a copy of the upstream tzdb version, except that
the tzdb version's first line is replaced by man-pages
boilerplate, and omit features introduced after 2017b
(the most recent merge to glibc).

This has the following effect:

Document --version, --help.

Document new -v warnings.

Remove -y.

Document that input should be text files, and similar restrictions
on names.

Document negative DST.

Document what is meant by "white space".

Do some minor reformatting.

Use .B for as-is keywords, like commands.

New section "EXTENDED EXAMPLE".

Omit some changes that were made on the man-pages side, notably by
changing some "timezone"s back to the preferred-upstream "time
zone" when talking about traditional time zones as opposed to
POSIX timezone settings.  Also, fix some formatting glitches.

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

zdump.8: Sync from tzdb upstream

Make zdump.8 a copy of the upstream tzdb version, except that
the tzdb version's first line is replaced by man-pages
boilerplate.

This has the following effect:

Document new options -i, -t, -V.

New section LIMITATIONS.

Do some minor reformatting.

Omit some changes that were made on the man-pages side, notably by
changing some "timezone"s back to the preferred-upstream "time
zone" when talking about traditional time zones as opposed to
POSIX timezone settings.  Also, fix some formatting glitches.

Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>