Ready for 4.16

Removed trailing white space at end of lines

Changes: Expanded tabs

Changes: tfix

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

execve.2, fallocate.2, getrlimit.2, io_submit.2, membarrier.2, mmap.2, msgget.2, open.2, ptrace.2, readv.2, semget.2, shmget.2, shutdown.2, syscall.2, wait.2, wait4.2, crypt.3, encrypt.3, fseek.3, getcwd.3, makedev.3, pthread_create.3, puts.3, tsearch.3, elf.5, filesystems.5, group.5, passwd.5, sysfs.5, mount_namespaces.7, posixoptions.7, time.7, unix.7, vdso.7, xattr.7, ld.so.8: tstamp

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

Changes: Ready for 4.16

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

syscall.2: spfix

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

keyctl.2, sincos.3: spfix

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

bpf.2, sched_setattr.2, crypt.3, elf.5, proc.5, fanotify.7, feature_test_macros.7, sched.7: spfix

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

keyctl.2: tfix

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

fallocate.2, keyctl.2, bzero.3: spfix

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

copy_file_range.2: tfix

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

namespaces.7: Mention that device ID should also be checked when comparing NS symlinks

When comparing two namespaces symlinks to see if they refer to
the same namespace, both the inode number and the device ID
should be compared. This point was already made clear in
ioctl_ns(2), but was missing from this page.

Reported-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

open.2: Add missing argument for snprintf() in example code

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

unix.7: tfix

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

capabilities.7: tfix

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

vdso.7: VDSO symbols (system calls) are not visible to seccomp(2) filters

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

seteuid.2: grfix

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

setgid.2: wfix: Make wording consistent with setuid(2)

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

fseek.3: Improve EPIPE error text

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

fseek.3: ERRORS: EBADF should be ESPIPE

Reported-by: Andy Owen <andrew.owen@dolby.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

capabilities.7: tfix

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

msgget.2, semget.2, shmget.2: The purpose of "flags" == 0 is to obtain ID of an existing IPC object

This was implied in these pages, but the meaning of "flags" == 0
could be more explicit, as indicated by questions such as
https://stackoverflow.com/questions/49833569/flag-value-of-semget-function

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

msgget.2: ffix

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

shmget.2: ffix

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

capabilities.7: srcfix: FIXME

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

capabilities.7: Rework file capability versioning and namespaced file caps text

There was some confused missing of concepts between the
two subsections, and some other details that needed fixing up.

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

capabilities.7: Explain when VFS_CAP_REVISION_3 file capabilities have effect

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

capabilities.7: Explain rules that determine version of security.capability xattr

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

capabilities.7: Explain term "namespace root user ID"

Confirmed with Serge Hallyn that: "nsroot" means the UID 0
in the namespace as it would be mapped into the initial userns.

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

capabilities.7: Document namespaced-file capabilities

Cowritten-by: Serge E. Hallyn <serge@hallyn.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

capabilities.7: Describe file capability versioning

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

encrypt.3: Replace text that is duplicated in crypt(3) with a cross reference

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

crypt.3: tfix

Reported-by: Nikos Mavrogiannopoulos <nmavrogi@redhat.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

crypt.3: Tweaks to Carlos O'Donell's patch

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

encrypt.3: srcfix: rewrap source lines

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

crypt.3: srcfix: rewrap source lines

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

crypt.3, encrypt.3: Add notes about _XOPEN_CRYPT

The distribution may choose not to support _XOPEN_CRYPT in the
case that the distribution has transitioned from glibc crypt to
libxcrypt.

Signed-off-by: Carlos O'Donell <carlos@redhat.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

posixoptions.7: tfix

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

posixoptions.7: Use a more consistent, less cluttered layout for option lists

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

posixoptions.7: Make function lists more consistent and less cluttered

Use more consistent layout for lists of functions, and
remove punctuation from the lists to make them less cluttered.

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

posixoptions.7: tfix

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

posixoptions.7: wfix

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

posixoptions.7: ffix

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

posixoptions.7: tfix

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

posixoptions.7: ffix

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

posixoptions.7: Expand XSI Options groups

We define in detail the X/Open System Interfaces i.e. _XOPEN_UNIX
and all of the X/Open System Interfaces (XSI) Options Groups.

The XSI options groups include encryption, realtime, advanced
realtime, realtime threads, advanced realtime threads, tracing,
streams, and legacy interfaces.

Signed-off-by: Carlos O'Donell <carlos@redhat.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

mmap.2: Minor wording improvements to text on MAP_FIXED hazard

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

mmap.2: Further tweaks to mention of MAP_FIXED_NOREPLACE in MAP_FIXED notes

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

mmap.2: Move the text on MAP_FIXED to NOTES

This text has become rather long, making it it somewhat
unwieldy in the discussion of the mmap() flags. Therefore,
move it to NOTES, with a pointer in DESCRIPTION referring
the reader to NOTES.

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

mmap.2: Further wording tweaks to the discussion of MAP_FIXED hazards

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

mmap.2: Minor wording fix

Change "extremely hazardous" to "hazardous". The former phrasing
is a little overwrought; on its own "hazardous" is enough to
convey the sense of danger.

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

mmap.2: Minor wording fixes to to text describing MAP_FIXED hazards

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

mmap.2: Simplify mention of MAP_FIXED_NOREPLACE in the discussion of MAP_FIXED

After discussions with Jann Horn and John Hubbard.

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

mmap.2: wfix

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

mmap.2: wfix

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

mmap.2: tfix

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

mmap.2: ffix

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

mmap.2: MAP_FIXED is okay if the address range has been reserved

Clarify that MAP_FIXED is appropriate if the specified address
range has been reserved using an existing mapping, but shouldn't
be used otherwise.

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

mmap.2: Clarify rationale of MAP_SYNC somewhat

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

mmap.2: Minor tweaks to Jan Kara's patch

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

mmap.2: srcfix: rewrap source lines

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

mmap.2: Add description of MAP_SHARED_VALIDATE and MAP_SYNC

Reviewed-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

getcwd.3: Minor tweaks to Carlos O'Donell's patch

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

getcwd.3: srcfix

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

getcwd.3: Mention that "(unreachable)" is no longer returned for glibc >= 2.27.

With glibc fix 52a713fdd0a30e1bd79818e2e3c4ab44ddca1a94 for
CVE-2018-1000001 (Sourceware BZ #22679) the implementation in the
just released glibc 2.27 has been changed such that instead of
returning "(unreachable)" the implementation now returns ENOENT
as it would have if the current directory had been unlinked.

I see that in 2015 the quirk was documented in commit
a2ac97c78bf05a55f8f616fc39a4724372dcfa95, and this is no longer
true with glibc 2.27, but may continue to be true in other C libraries,
so I reference NOTES from the paragraph in the central text.

Signed-off-by: Carlos O'Donell <carlos@redhat.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

membarrier.2: wfix

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

membarrier.2: ffix code example

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

membarrier.2: ffix

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

membarrier.2: Some wording improvements

I'll pass these to Mathieu Desnoyers for checking

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

membarrier.2: Minor tweaks to Mathieu Desnoyers' patch

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

membarrier.2: Add kernel version for MEMBARRIER_CMD_QUERY and MEMBARRIER_CMD_SHARED

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

membarrier.2: Minor tweaks to MEMBARRIER_CMD_SHARED text

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

membarrier.2: tfix

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

membarrier.2: ffix

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

membarrier.2: srcfix

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

membarrier.2: Document new membarrier commands introduced in Linux 4.16

Document the following membarrier commands introduced in
Linux 4.16:

    MEMBARRIER_CMD_GLOBAL_EXPEDITED
        (the old enum label MEMBARRIER_CMD_SHARED is now an
        alias to preserve header backward compatibility)
    MEMBARRIER_CMD_REGISTER_GLOBAL_EXPEDITED
    MEMBARRIER_CMD_PRIVATE_EXPEDITED_SYNC_CORE
    MEMBARRIER_CMD_REGISTER_PRIVATE_EXPEDITED_SYNC_CORE

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
CC: Michael Kerrisk <mtk.manpages@gmail.com>
CC: Ingo Molnar <mingo@redhat.com>
CC: Peter Zijlstra <peterz@infradead.org>
CC: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

io_submit.2: Add kernel version numbers for various 'aio_rw_flags' flags

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

readv.2: wfix

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

readv.2: Remove redundant sentence

The sentence is out of place, and probably doesn't really add to
the understanding already provided by the rest of the text

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

readv.2: srcfix

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

io_submit.2: Place 'aio_rw_flags' in alphabetical order

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

io_submit.2: srcfix

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

readv.2, io_submit.2: Document RWF_APPEND added in Linux 4.16

Signed-off-by: Jürg Billeter <j@bitron.ch>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

unix.7: ERRORS: add EBADF for sending closed file descriptor with SCM_RIGHTS

As noted by Rusty Russell:

I was really surprised that sendmsg() returned EBADF on a valid fd;
turns out I was using sendmsg with SCM_RIGHTS to send a closed fd,
which gives EBADF (see test program below).

But this is only obliquely referenced in unix(7):

       SCM_RIGHTS
              Send or receive a set  of  open  file  descriptors
              from  another  process.  The data portion contains
              an integer array of  the  file  descriptors.   The
              passed file descriptors behave as though they have
              been created with dup(2).

EBADF is not mentioned in the unix(7) ERRORS (it's mentioned in
dup(2)).

int fdpass_send(int sockout, int fd)
{
/* From the cmsg(3) manpage: */
struct msghdr msg = { 0 };
struct cmsghdr *cmsg;
struct iovec iov;
char c = 0;
union {         /* Ancillary data buffer, wrapped in a union
   in order to ensure it is suitably aligned */
char buf[CMSG_SPACE(sizeof(fd))];
struct cmsghdr align;
} u;

msg.msg_control = u.buf;
msg.msg_controllen = sizeof(u.buf);
memset(&u, 0, sizeof(u));
cmsg = CMSG_FIRSTHDR(&msg);
cmsg->cmsg_level = SOL_SOCKET;
cmsg->cmsg_type = SCM_RIGHTS;
cmsg->cmsg_len = CMSG_LEN(sizeof(fd));
memcpy(CMSG_DATA(cmsg), &fd, sizeof(fd));

msg.msg_name = NULL;
msg.msg_namelen = 0;
msg.msg_iov = &iov;
msg.msg_iovlen = 1;
msg.msg_flags = 0;

/* Keith Packard reports that 0-length sends don't work, so we
 * always send 1 byte. */
iov.iov_base = &c;
iov.iov_len = 1;

return sendmsg(sockout, &msg, 0);
}

int fdpass_recv(int sockin)
{
/* From the cmsg(3) manpage: */
struct msghdr msg = { 0 };
struct cmsghdr *cmsg;
struct iovec iov;
int fd;
char c;
union {         /* Ancillary data buffer, wrapped in a union
   in order to ensure it is suitably aligned */
char buf[CMSG_SPACE(sizeof(fd))];
struct cmsghdr align;
} u;

msg.msg_control = u.buf;
msg.msg_controllen = sizeof(u.buf);

msg.msg_name = NULL;
msg.msg_namelen = 0;
msg.msg_iov = &iov;
msg.msg_iovlen = 1;
msg.msg_flags = 0;

iov.iov_base = &c;
iov.iov_len = 1;

if (recvmsg(sockin, &msg, 0) < 0)
return -1;

cmsg = CMSG_FIRSTHDR(&msg);
        if (!cmsg
    || cmsg->cmsg_len != CMSG_LEN(sizeof(fd))
    || cmsg->cmsg_level != SOL_SOCKET
    || cmsg->cmsg_type != SCM_RIGHTS) {
errno = -EINVAL;
return -1;
}

memcpy(&fd, CMSG_DATA(cmsg), sizeof(fd));
return fd;
}

static void child(int sockfd)
{
int newfd = fdpass_recv(sockfd);
assert(newfd < 0);
exit(0);
}

int main(void)
{
int sv[2];
int pid, ret;

assert(socketpair(AF_UNIX, SOCK_STREAM, 0, sv) == 0);

pid = fork();
if (pid == 0) {
close(sv[1]);
child(sv[0]);
}

close(sv[0]);
ret = fdpass_send(sv[1], sv[0]);
printf("fdpass of bad fd return %i (%s)\n", ret, strerror(errno));
return 0;
}

Reported-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

unix.7: wfix

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

perf_event_open.2: Minor tweak to Song Liu's kprobe/uprobe patch: fix kernel version

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

perf_event_open.2: Minor tweaks to Song Liu's patch

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

perf_event_open.2: Add type kprobe and uprobe

Two new types kprobe and uprobe are being added to
perf_event_open(), which allow creating kprobe or
uprobe with perf_event_open. This patch adds
information about these types.

Signed-off-by: Song Liu <songliubraving@fb.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

mmap.2: Explicitly mention MAP_FIXED_NOREPLACE in EEXIST description

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

mmap.2: Minor wording fix

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

mmap.2: Add a little historical detail on the obsolete MAP_DENYWRITE

In Linux 2..0, do_mmap() had the following check:

                if (flags & MAP_DENYWRITE) {
                        if (file->f_inode->i_writecount > 0)
                                return -ETXTBSY;
                }

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

mmap.2: Minor rewording of MAP_FIXED text

The paragraph break-up of the existing text was
somewhat unnatural.

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

mmap.2: ffix

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

mmap.2: Minor tweaks to Michal Hocko's patch

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

mmap.2: tfix

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

mmap.2: srcfix: rewrap source lines

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

mmap.2: ffix

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

mmap.2: Document new MAP_FIXED_NOREPLACE flag

4.17+ kernels offer a new MAP_FIXED_NOREPLACE flag which allows
the caller to atomically probe for a given address range.

[wording heavily updated by John Hubbard]

Cowritten-by: John Hubbard <jhubbard@nvidia.com>
Signed-off-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Michael Kerrisk <mtk.manpages@gmail.com>

keyring.7: wfix

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