modify_ldt.2: Use syscall(SYS_...); for system calls without a wrapper

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

mmap2.2: Use syscall(SYS_...); for system calls without a wrapper

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

mknod.2: Remove unused includes

All of the constants used by mknod() are defined in <sys/stat.h>.

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

mincore.2: Remove unused include

AFAICS, there's no use for <unistd.h> here.  The prototype is
declared in <sys/mman.h>, and there are no constants needed.

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

membarrier.2: Use syscall(SYS_...); for system calls without a wrapper

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

lookup_dcookie.2: Use syscall(SYS_...); for system calls without a wrapper

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

llseek.2: Use syscall(SYS_...); for system calls without a wrapper

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

link.2: ffix

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

keyctl.2: Use syscall(SYS_...); for system calls without a glibc wrapper

Remove the libkeyutils prototype from the synopsis, which isn't
documented in the rest of the page, and as NOTES says, it's
probably better to use the various library functions.

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

scripts/bash_aliases: man_lsfunc(): Extract syscall name from syscall(SYS_...)

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

kexec_load.2: Use syscall(SYS_...); for system calls without a wrapper

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

kcmp.2: tfix

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

kcmp.2: Use syscall(SYS_...); for system calls without a wrapper

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

ipc.2: Add needed include

The constants needed for using this function are defined in
<linux/ipc.h>.  Add the include, even when those constants are not
mentioned in this manual page.

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

ipc.2: Use syscall(SYS_...); for system calls without a wrapper

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

ioprio_set.2: Use syscall(SYS_...); for system calls without a wrapper

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

ioperm.2: Remove obvious comment

Of course that is for the glibc wrapper.  As all of the other
pages that don't explicitly say otherwise.

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

io_getevents.2: Use syscall(SYS_...); for system calls without a wrapper

In this case there's a wrapper provided by libaio,
but this page documents the raw syscall.

Also remove <linux/time.h> from the includes: 'struct timespec'
is already documented in system_data_types(7), where the
information is more up to date.

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

io_destroy.2: Use syscall(SYS_...); for system calls without a wrapper

In this case there's a wrapper provided by libaio,
but this page documents the raw syscall.

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

ioctl_userfaultfd.2: SYNOPSIS: Add <linux/userfaultfd.h>

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

ioctl_tty.2: Fix includes

<sys/ioctl.h> is needed for the prototype of ioctl().  That header
also provides most of the constants used by the function.  Only a
few of those constants are not provided by that header, and need
<termios.h>; clarify which constants do need that include.

......

$ <man2/ioctl_tty.2 \
  sed -n '/^.SH DESCRIPTION/,/^.SH/p' \
  |grep -e '^\.B' -e TIOCM \
  |sed 's/^\.B[^ ]* //' \
  |awk '{print $1}' \
  |grep '^[[:upper:]]' \
  |grep -v -e '^CAP' -e '^E' -e '^SIG' -e '^O_' -e '^[TR]XD$' -e '^POLL' \
  |sort \
  |uniq \
  |while read f; do \
find /usr/include/ -type f \
|xargs grep -l "define\s$f" \
|grep -q ioctl.*.h \
||echo $f \
|while read ff; do \
echo "============ $ff"; \
find /usr/include/ -type f \
|xargs grep -n "define\s$ff"; \
done; \
  done;
============ CLOCAL
/usr/include/asm-generic/termbits.h:142:#define CLOCAL 0004000
/usr/include/gphoto2/gphoto2-port-portability.h:127:#   define CLOCAL  0x00000800
/usr/include/x86_64-linux-gnu/bits/termios-c_cflag.h:34:#define CLOCAL 0004000
============ TCIFLUSH
/usr/include/asm-generic/termbits.h:191:#define TCIFLUSH 0
/usr/include/x86_64-linux-gnu/bits/termios.h:70:#define TCIFLUSH 0
============ TCIOFF
/usr/include/asm-generic/termbits.h:187:#define TCIOFF 2
/usr/include/x86_64-linux-gnu/bits/termios.h:66:#define TCIOFF 2
============ TCIOFLUSH
/usr/include/asm-generic/termbits.h:193:#define TCIOFLUSH 2
/usr/include/x86_64-linux-gnu/bits/termios.h:72:#define TCIOFLUSH 2
============ TCION
/usr/include/asm-generic/termbits.h:188:#define TCION 3
/usr/include/x86_64-linux-gnu/bits/termios.h:67:#define TCION 3
============ TCOFLUSH
/usr/include/asm-generic/termbits.h:192:#define TCOFLUSH 1
/usr/include/x86_64-linux-gnu/bits/termios.h:71:#define TCOFLUSH 1
============ TCOOFF
/usr/include/asm-generic/termbits.h:185:#define TCOOFF 0
/usr/include/x86_64-linux-gnu/bits/termios.h:64:#define TCOOFF 0
============ TCOON
/usr/include/asm-generic/termbits.h:186:#define TCOON 1
/usr/include/x86_64-linux-gnu/bits/termios.h:65:#define TCOON 1
============ TIOCREMOTE
============ TIOCSTART
============ TIOCSTOP
============ TIOCTTYGSTRUCT
============ TIOCUCNTL

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

getdents.2: Use syscall(SYS_...); for system calls without a wrapper

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

futex.2: Use syscall(SYS_...); for system calls without a wrapper

At the same time, document only headers that are required
for calling the function, or those that are specific to the
function:

<unistd.h> is required for the syscall() prototype.
<sys/syscall.h> is required for the syscall name SYS_xxx.
<linux/futex.h> is specific to this syscall.

However, uint32_t is generic enough that it shouldn't be
documented here.  The system_data_types(7) page already documents
it, and is more precise about it.  The same goes for timespec.

As a general rule a man[23] page should document the header that
includes the prototype, and all of the headers that define macros
that should be used with the call.  However, the information about
types should be restricted to system_data_types(7) (and that page
should probably be improved by adding types), except for types
that are very specific to the call.  Otherwise, we're duplicating
info and it's then harder to maintain, and probably outdated in
the future.

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

execveat.2: Remove unused include

This complements commit e3eba861bd966911b38b7ebc572f0c092ca7bdee.

Since we don't need syscall(2) anymore, we don't need SYS_* definitions.

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

Makefile: Remove shebang line

On 5/10/21 7:13 PM, Alejandro Colomar (man-pages) wrote:
> Hi Michael,
>
> On 5/10/21 1:39 AM, Michael Kerrisk (man-pages) wrote:
>>> - Specify shebang
>>
>> Why? It's not quite obvious to me, and the commit message
>> should really explain...
>
> Hmmm.  I have some minor reasons to add it, but not a really good one.
>
> * Some editors don't recognize 'Makefile' as a special name, so the
> shebang helps detecting which language the file is using (e.g., for
> coloring).
>
> * I tend to subdivide a big Makefile into a small Makefile and many
> submakefiles stored in <./libexec/>.  Those obviously need different
> names, and given that the makefile extension is not very standard (I use
> .mk), having a shebang helps knowing what the file is.  After that, I
> also have it on the main Makefile for consistency.   But here we only
> have one Makefile, so it doesn apply very much.

I think I'll remove it. It is kind of idiosyncratic, leaves the
reader asking "why?".

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

.gitignore: Tweaks to Alex's .gitignore

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

.gitignore: Add file

Ignore everything new by default.

This avoids having to update the .gitignore when we need to ignore
something new.  It also avoids accidents that may add an unwanted
temporary file.

Cc: Debian man-pages <manpages@packages.debian.org>
Cc: Dr. Tobias Quathamer <toddy@debian.org>
Signed-off-by: Alejandro Colomar <alx.manpages@gmail.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

fflush.3: SEE ALSO: Add fpurge(3)

fpurge(i_stream) does the same as fflush(i_stream), AFAIK.

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

open.2: Fix bug in linkat(2) call example

AT_EMPTY_PATH works with empty strings (""), but not with NULL
(or at least it's not obvious).

The relevant kernel code is the following:

linux$ sed -n 189,198p fs/namei.c
result->refcnt = 1;
/* The empty path is special. */
if (unlikely(!len)) {
if (empty)
*empty = 1;
if (!(flags & LOOKUP_EMPTY)) {
putname(result);
return ERR_PTR(-ENOENT);
}
}

Reported-by: Walter Harms <wharms@bfs.de>
Cc: Theodore Ts'o <tytso@mit.edu>
Cc: Adam Borowski <kilobyte@angband.pl>
Signed-off-by: Alejandro Colomar <alx.manpages@gmail.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

basename.3: wfix

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

bswap.3: bswap_*() are implemented using functions

See <bits/byteswap.h> in glibc.
These macros call functions of the form __bswap_N(),
which use uintN_t.

Even though it's true that they are macros,
it's transparent to the user.

The user will see their results casted to unsigned types
after the conversion due to the underlying functions,
so it's better to document these as the underlying functions,
specifying the types.

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

pthread_attr_setschedparam.3: SYNOPSIS: Use 'restrict' in prototypes

Both POSIX and glibc use 'restrict' in
pthread_attr_getschedparam(), pthread_attr_setschedparam().
Let's use it here too.

.../glibc$ grep_glibc_prototype pthread_attr_getschedparam
sysdeps/htl/pthread.h:102:
extern int pthread_attr_getschedparam (const pthread_attr_t *__restrict __attr,
       struct sched_param *__restrict __param)
__THROW __nonnull ((1, 2));
sysdeps/nptl/pthread.h:294:
extern int pthread_attr_getschedparam (const pthread_attr_t *__restrict __attr,
       struct sched_param *__restrict __param)
     __THROW __nonnull ((1, 2));
.../glibc$ grep_glibc_prototype pthread_attr_setschedparam
sysdeps/htl/pthread.h:107:
extern int pthread_attr_setschedparam (pthread_attr_t *__restrict __attr,
       const struct sched_param *__restrict
       __param) __THROW __nonnull ((1, 2));
sysdeps/nptl/pthread.h:299:
extern int pthread_attr_setschedparam (pthread_attr_t *__restrict __attr,
       const struct sched_param *__restrict
       __param) __THROW __nonnull ((1, 2));
.../glibc$

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

pthread_attr_setinheritsched.3: SYNOPSIS: Use 'restrict' in prototypes

Both POSIX and glibc use 'restrict' in
pthread_attr_getinheritsched().
Let's use it here too.

.../glibc$ grep_glibc_prototype pthread_attr_getinheritsched
sysdeps/htl/pthread.h:90:
extern int pthread_attr_getinheritsched (const pthread_attr_t *__restrict __attr,
 int *__restrict __inheritsched)
__THROW __nonnull ((1, 2));
sysdeps/nptl/pthread.h:313:
extern int pthread_attr_getinheritsched (const pthread_attr_t *__restrict
 __attr, int *__restrict __inherit)
     __THROW __nonnull ((1, 2));
.../glibc$

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

epoll_wait.2: Move subsection to NOTES from BUGS

'C library/kernel differences' was added to BUGS incorrectly.
Fix it

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

xdr.3: SYNOPSIS: Fix prototype types

Use the same types glibc uses, and add a missing 'const'.

.../glibc$ grep_glibc_prototype xdr_union
sunrpc/rpc/xdr.h:315:
extern bool_t xdr_union (XDR *__xdrs, enum_t *__dscmp, char *__unp,
 const struct xdr_discrim *__choices,
 xdrproc_t __dfault) __THROW;
.../glibc$

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

rpc.3: SYNOPSIS: Fix prototypes (misc.)

Fix the types for the following functions:

authunix_create(),
callrpc(),
clnt_create(),
clnt_pcreateerror(),
clnt_perror(),
clnt_spcreateerror(),
clnt_sperror(),
pmap_set(),
svc_register(),
svcerr_progvers().

.../glibc$ grep_glibc_prototype authunix_create
sunrpc/rpc/auth.h:158:
extern AUTH *authunix_create (char *__machname, __uid_t __uid, __gid_t __gid,
      int __len, __gid_t *__aup_gids);
.../glibc$ grep_glibc_prototype callrpc
sunrpc/rpc/clnt.h:359:
extern int callrpc (const char *__host, const u_long __prognum,
    const u_long __versnum, const u_long __procnum,
    const xdrproc_t __inproc, const char *__in,
    const xdrproc_t __outproc, char *__out) __THROW;
.../glibc$ grep_glibc_prototype clnt_create
sunrpc/rpc/clnt.h:291:
extern CLIENT *clnt_create (const char *__host, const u_long __prog,
    const u_long __vers, const char *__prot)
     __THROW;
.../glibc$ grep_glibc_prototype clnt_pcreateerror
sunrpc/rpc/clnt.h:368:
extern void clnt_pcreateerror (const char *__msg); /* stderr */
.../glibc$ grep_glibc_prototype clnt_perror
sunrpc/rpc/clnt.h:379:
extern void clnt_perror (CLIENT *__clnt, const char *__msg);
.../glibc$ grep_glibc_prototype clnt_spcreateerror
sunrpc/rpc/clnt.h:369:
extern char *clnt_spcreateerror(const char *__msg) __THROW; /* string */
.../glibc$ grep_glibc_prototype clnt_sperror
sunrpc/rpc/clnt.h:381:
extern char *clnt_sperror (CLIENT *__clnt, const char *__msg) __THROW;
.../glibc$ grep_glibc_prototype pmap_set
sunrpc/rpc/pmap_clnt.h:69:
extern bool_t pmap_set (const u_long __program, const u_long __vers,
int __protocol, u_short __port) __THROW;
.../glibc$ grep_glibc_prototype svc_register
sunrpc/rpc/svc.h:188:
extern bool_t svc_register (SVCXPRT *__xprt, rpcprog_t __prog,
    rpcvers_t __vers, __dispatch_fn_t __dispatch,
    rpcprot_t __protocol) __THROW;
.../glibc$ grep_glibc_prototype svcerr_progvers
sunrpc/rpc/svc.h:253:
extern void svcerr_progvers (SVCXPRT *__xprt, rpcvers_t __low_vers,
 rpcvers_t __high_vers) __THROW;
.../glibc$

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

cpow.3: Use 'complex' after the type consistently

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

system_data_types.7: tfix

Those pages didn't exist.  Fix the section number.
I noticed the typo thanks to the HTML pages on man7.org.

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

Makefile: Fix bug when running in parallel

Prerequisites can run in parallel.  This wouldn't make any sense
when uninstalling and installing again.

For that, use consecutive commands, which run one after the other
even with multiple cores.

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

Makefile: Use standard features (IMPORTANT: default prefix changed)

IMPORTANT for distributions:
This changes prefix to be '/usr/local' as is expected by default,
instead of the old '/usr' value.

- Use standard variables:
    - prefix should be '/usr/local'
    - mandir (instead of MANDIR)
    - htmldir (instead of HTDIR)
    - ...
    see <https://www.gnu.org/software/make/manual/html_node/Directory-Variables.html>

- Use standard targets:
    - html (build html files; don't install them)
    - install-html (instead of html)
    - installdirs (instead of 'mkdir -p'/'install -d' inside other targets)
    - ...
    see <https://www.gnu.org/software/make/manual/html_node/Standard-Targets.html#Standard-Targets>

- Use .PHONY

- ?= is not needed.  User input overrides any assignment.  Use =

- Use standard command variables, instead of directly calling commands.
    - $(INSTALL_DATA) (instead of install -m 644)
    - $(INSTALL_DIR) (instead of install -d -m 755 or mkdir -p)
    see <https://www.gnu.org/software/make/manual/html_node/Command-Variables.html#Command-Variables>

- Specify SHELL = /bin/bash

- Specify shebang

- Allow variable html extension (or no extension at all)

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

Makefile: html: Simplify target

mkdir -p doesn't fail if the directory already exists.
Remove redundant checks.

Use .html as default HTDIR.
Remove checks for undefined HTDIR.

Show what the target does, as with other targets (remove '@').

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

scripts/bash_aliases: srcfix

I clarified the code about two things:

- Checking how many arguments are being passed.
    Here, some functions didn't reject extra arguments when they
    weren't being used.  Fix that.
    I also changed the code to use $#, which is more explicit.
    And use arithmetic expressions, which better indicate that
    we're dealing with numbers.

- Remove unneeded options from sort.
Reported-by: Stefan Puiu <stefan.puiu@gmail.com>
    After Stefan asked about why am I using 'sort -V',
    I noticed that I really don't need '-V', and it may confuse
    people trying to understand the script, so even though I
    slightly prefer the output of 'sort -V', in this case, it's
    better to use the simpler 'sort' (yet I need 'sort', to
    maintain consistency in the results (find is quite random)).

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

scripts/bash_aliases: Fix error messages

Fix the error messages to clearly show that both dirs and manual
pages are accepted, and that more than one argument is accepted.

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

scripts/bash_aliases: man_section(): Accept multiple sections

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

scripts/bash_aliases: Add man_lsvar(), similar to man_lsfunc() but lists variables

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

scripts/bash_aliases: Make man_lsfunc() more robust; Add sed_rm_ccomments().

This patch makes man_lsfunc() search for the function prototypes,
instead of relying on the current manual page formatting,
which might change in the future, and break this function.

It also simplifies the code, by reusing man_section().

Create a new function sed_rm_ccomments(), which is needed by
man_lsfunc(), and may also be useful in other cases.

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

scripts/bash_aliases: man_gitstaged(): Script around a porcelain git command

The output of 'git status' is not stable.

The more stable 'git status --porcelain' is more complex,
and scripting around it would be more complex.

However, 'git diff --staged --name-only' produces
the output that we were lookiong for.

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

scripts/bash_aliases, scripts/modified_pages.sh: Move scripts/modified_pages.sh to a function man_gitstaged()

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

dladdr.3: SYNOPSIS: Add missing 'const'

Glibc uses 'const' for the 1st parameter of these functions.
Fix the prototypes.

......

.../glibc$ grep_glibc_prototype dladdr
dlfcn/dlfcn.h:98:
extern int dladdr (const void *__address, Dl_info *__info)
     __THROW __nonnull ((2));
.../glibc$ grep_glibc_prototype dladdr1
dlfcn/dlfcn.h:102:
extern int dladdr1 (const void *__address, Dl_info *__info,
    void **__extra_info, int __flags) __THROW __nonnull ((2));
.../glibc$

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

sched_get_priority_max.2, open_memstream.3: tfix

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

pthread_getattr_default_np.3: SYNOPSIS: Add missing 'const'

glibc uses 'const' in pthread_setattr_default_np().
Let's use it here too.

.../glibc$ grep_glibc_prototype pthread_setattr_default_np
sysdeps/nptl/pthread.h:406:
extern int pthread_setattr_default_np (const pthread_attr_t *__attr)
     __THROW __nonnull ((1));
.../glibc$

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

pthread_mutexattr_setrobust.3: SYNOPSIS: Remove incorrect 'const'

Neither POSIX or glibc use 'const' in
pthread_mutexattr_setrobust().
Remove it.

.../glibc$ grep_glibc_prototype pthread_mutexattr_setrobust
sysdeps/htl/pthread.h:355:
extern int pthread_mutexattr_setrobust (pthread_mutexattr_t *__attr,
int __robustness)
     __THROW __nonnull ((1));
sysdeps/nptl/pthread.h:888:
extern int pthread_mutexattr_setrobust (pthread_mutexattr_t *__attr,
int __robustness)
     __THROW __nonnull ((1));
.../glibc$

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

clone.2: tfix

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

tzset.3: ffix

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

strsignal.3: SYNOPSIS: Add missing 'const'

glibc uses 'const' in sigdescr_np(), sigabbrev_np().
Let's use it here too.

.../glibc$ grep_glibc_prototype sigdescr_np
string/string.h:469:
extern const char *sigdescr_np (int __sig) __THROW;
.../glibc$ grep_glibc_prototype sigabbrev_np
string/string.h:466:
extern const char *sigabbrev_np (int __sig) __THROW;
.../glibc$

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

strsignal.3: ffix

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

__ppc_get_timebase.3: ffix

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

perror.3: ffix

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

on_exit.3: ffix

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

getutmp.3: ffix

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

malloc_usable_size.3: ffix

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

signalfd.2: tfix

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

fanotify.7: ffix

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

semctl.2: ffix

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

move_pages.2: ffix

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

execve.2: tfix

Signed-off-by: Štěpán Němec <stepnem@gmail.com>
Signed-off-by: Alejandro Colomar <alx.manpages@gmail.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

shmop.2: tfix

Signed-off-by: Vishwajith K <vishuvikas1996@gmail.com>
Signed-off-by: Alejandro Colomar <alx.manpages@gmail.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

getopt.3: Minor tweaks to James' patch

The rest of the page writes the characters without naming them.
Follow that convention.

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

getopt.3: Clarify behaviour

Improved the `getopt(3)` man page in the following ways:

1) Defined the existing term "legitimate option character".
2) Added an additional NOTE stressing that arguments are parsed in strict
   order and the implications of this when numeric options are utilised.

Signed-off-by: James O. D. Hunt <jamesodhunt@gmail.com>
Signed-off-by: Alejandro Colomar <alx.manpages@gmail.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

sigwaitinfo.2: tfix

Fix wording issue introduced in commit
bf1298c9e5053f55dea43e74255dae5ec57f251e.

Reported-by: Chris Keilbart <keilbartchris@gmail.com>
Signed-off-by: Alejandro Colomar <alx.manpages@gmail.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

exit_group.2, getunwind.2: tfix

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

Changes.old: tfix

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

ptrace.2: mention PTRACE_GET_SYSCALL_INFO in RETURN VALUE section

Mirror the wording about PTRACE_GET_SYSCALL_INFO return value semantics
from "DESCRIPTION" section to "RETURN VALUE" section.

Reported-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Complements: fc91449cb "ptrace.2: Document PTRACE_GET_SYSCALL_INFO"
Signed-off-by: Dmitry V. Levin <ldv@altlinux.org>
Signed-off-by: Alejandro Colomar <alx.manpages@gmail.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

move_pages.2: ffix

Signed-off-by: Dmitry V. Levin <ldv@altlinux.org>
Signed-off-by: Alejandro Colomar <alx.manpages@gmail.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

clone.2: tfix

Despite my mention of this spawning a hilarious discussion
on IRC, this alignment restriction should be 128-bit, not
126-bit.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: Alejandro Colomar <alx.manpages@gmail.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

setbuf.3: tfix

Reported-by: Zhiheng Li <phoenix_lzh@sina.com>
Signed-off-by: Alejandro Colomar <alx.manpages@gmail.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

sigaltstack.2: tfix

Add a missing "to" in an "in order to" formulation.

Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Alejandro Colomar <alx.manpages@gmail.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

flock.2: add CIFS details

CIFS flock() locks behave differently than the standard. Give overview
of those differences.

Here is the rendered text:

CIFS details
  In Linux kernels up to 5.4, flock() is not propagated over SMB.  A file
  with such locks will not appear locked for remote clients.

  Since Linux 5.5, flock() locks are emulated with SMB  byte-range  locks
  on  the  entire  file.   Similarly to NFS, this means that fcntl(2) and
  flock() locks interact with one another.  Another important side-effect
  is  that  the  locks  are not advisory anymore: any IO on a locked file
  will always fail with EACCES when done from a separate file descriptor.
  This  difference  originates from the design of locks in the SMB proto-
  col, which provides mandatory locking semantics.

  Remote and mandatory locking semantics  may  vary  with  SMB  protocol,
  mount options and server type.  See mount.cifs(8) for additional infor-
  mation.

Signed-off-by: Aurelien Aptel <aaptel@suse.com>
Discussion: linux-man <https://lore.kernel.org/linux-man/20210302154831.17000-1-aaptel@suse.com/>
Signed-off-by: Alejandro Colomar <alx.manpages@gmail.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

Various pages: Remove unused <sys/ipc.h> (and <sys/types.h>)

In b0b19983d9a2001ce94b908a99f4c05a50fd47ee we removed
<sys/types.h>.  For the same reasons there, remove now <sys/ipc.h>
from many pages.

If someone wonders why <sys/ipc.h> was needed, the reason was to
get all the definitions of IPC_* constants.  However, that header
is now included by <sys/msg.h>, so it's not needed anymore to
explicitly include it.  Quoting POSIX: "In addition, the
<sys/msg.h> header shall include the <sys/ipc.h> header."

There were some remaining cases where I forgot to remove
<sys/types.h>; remove them now too.

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

open_by_handle_at.2: Remove unused <sys/stat.h>

AFAICS, all types and constants used by these functions are
defined in <fcntl.h>.

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

syscalls.2: perfmonctl(2) was removed in Linux 5.10

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

perfmonctl.2: This system call was removed in Linux 5.10

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

shm_overview.7: ffix

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

pthread_yield.3: Note that this function is deprecated since glibc 2.34

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

pthread_mutexattr_setrobust.3: Note that the *_np() APIs are deprecated since glibc 2.34

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

pthread_mutex_consistent.3: Note that pthread_mutexattr_setrobust() is now deprecated

Deprecated since glibc 2.34.

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

execveat.2: Library support has been added in glibc 2.34

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

pthread_mutex_consistent.3: tfix

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

syscalls.2: SEE ALSO: add ausyscall(1)

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

get_mempolicy.2, mq_getsetattr.2, poll.2: ffix

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

pipe.2: Rearrange SYNOPSIS so that minority version pipe() is at end

A few architectures have a different call signature for pipe().
Since those architectures are the minority, place the prototype
at the end of the SYNOPSIS, rather than the start.

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

mount.2: ffix

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

userfaultfd.2: tfix

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

ioctl_userfaultfd.2: tfix

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

mount_namespaces.7, namespaces.7: Relocate reference to pam_namespace(8)

SEE ALSO: move pam_namespace(8) from namespaces(7) to
mount_namespaces(7) (since pam_namespace(8) makes use of
mount namespaces specifically).

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

vsock.7: ffix

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

vsock.7: ioctls are on /dev/vsock, not sockets

Signed-off-by: Alyssa Ross <hi@alyssa.is>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

ioctl_userfaultfd.2, userfaultfd.2: Minor tweaks to Peter Xu's patches

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

ioctl_userfaultfd.2: Add write-protect mode docs

Userfaultfd write-protect mode is supported starting from Linux 5.7.

Acked-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
[alx: ffix + srcfix]
Signed-off-by: Alejandro Colomar <alx.manpages@gmail.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

ioctl_userfaultfd.2: Add UFFD_FEATURE_THREAD_ID docs

UFFD_FEATURE_THREAD_ID is supported in Linux 4.14.

Acked-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Alejandro Colomar <alx.manpages@gmail.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>