platform/upstream/systemd.git
6 years agocore: namespace: nitpick /dev/ptmx error handling
Alan Jenkins [Wed, 17 Jan 2018 13:28:04 +0000 (13:28 +0000)]
core: namespace: nitpick /dev/ptmx error handling

If /dev/tty did not exist, or had st_rdev == 0, we ignored it.  And the
same is true for null, zero, full, random, urandom.

If /dev/ptmx did not exist, we treated this as a failure.  If /dev/ptmx had
st_rdev == 0, we ignored it.

This was a very recent change, but there was no reason for ptmx creation
specifically to treat st_rdev == 0 differently from non-existence.  This
confuses me when reading it.

Change the creation of /dev/ptmx so that st_rdev == 0 is
treated as failure.

This still leaves /dev/ptmx as a special case with stricter handling.
However it is consistent with the immediately preceding creation of
/dev/pts/, which is treated as essential, and is directly related to ptmx.

I don't know why we check st_rdev.  But I'd prefer to have only one
unanswered question here, and not to have a second unanswered question
added on top.

6 years agoefivars: include errno.h when EFI support is disabled (#7900)
Yu Watanabe [Wed, 17 Jan 2018 09:25:42 +0000 (18:25 +0900)]
efivars: include errno.h when EFI support is disabled (#7900)

Fixes #7898.

6 years agoMerge pull request #7886 from gdamjan/fix-ptmx
Alan Jenkins [Wed, 17 Jan 2018 09:24:00 +0000 (09:24 +0000)]
Merge pull request #7886 from gdamjan/fix-ptmx

namespace: make /dev/ptmx a copy of the host not a symlink

6 years agoMerge pull request #7893 from poettering/parse-tweaks
Zbigniew Jędrzejewski-Szmek [Wed, 17 Jan 2018 09:22:17 +0000 (20:22 +1100)]
Merge pull request #7893 from poettering/parse-tweaks

parsing tweaks

6 years agoMerge pull request #7902 from yuwata/fix-warning-by-clang
Zbigniew Jędrzejewski-Szmek [Wed, 17 Jan 2018 09:17:23 +0000 (20:17 +1100)]
Merge pull request #7902 from yuwata/fix-warning-by-clang

network: small fixes

6 years agohwdb: 60-sensors: Add DMI strings for Trekstor Surftab 7.0 newer BIOS versions (...
Hans de Goede [Wed, 17 Jan 2018 09:15:41 +0000 (10:15 +0100)]
hwdb: 60-sensors: Add DMI strings for Trekstor Surftab 7.0 newer BIOS versions (#7904)

Some newer BIOS versions of the TrekStor SurfTab wintron 7.0 tablet use
different (better) DMI strings, update the existing 60-sensors.hwdb
entry for this tablet to also work with the newer BIOS.

6 years ago hwdb: HP vendor name for ZBooks in 60-keyboard.hwdb (#7905)
Jerónimo Borque [Wed, 17 Jan 2018 09:15:00 +0000 (06:15 -0300)]
 hwdb: HP vendor name for ZBooks in 60-keyboard.hwdb (#7905)

Added new HP vendor name to support Zbook's mic mute key mapping

6 years agoMerge pull request #7897 from yuwata/small-man-fixes
Zbigniew Jędrzejewski-Szmek [Wed, 17 Jan 2018 09:13:54 +0000 (20:13 +1100)]
Merge pull request #7897 from yuwata/small-man-fixes

Several man fixes

6 years agonamespace: only make the symlink /dev/ptmx if it was already a symlink
Дамјан Георгиевски [Tue, 16 Jan 2018 20:50:36 +0000 (21:50 +0100)]
namespace: only make the symlink /dev/ptmx if it was already a symlink

…otherwise try to clone it as a device node

On most contemporary distros /dev/ptmx is a device node, and
/dev/pts/ptmx has 000 inaccessible permissions. In those cases
the symlink /dev/ptmx -> /dev/pts/ptmx breaks the pseudo tty support.

In that case we better clone the device node.

OTOH, in nspawn containers (and possibly others), /dev/pts/ptmx has
normal permissions, and /dev/ptmx is a symlink. In that case make the
same symlink.

fixes #7878

6 years agonamespace: extract clone_device_node function from mount_private_dev
Дамјан Георгиевски [Tue, 16 Jan 2018 20:27:51 +0000 (21:27 +0100)]
namespace: extract clone_device_node function from mount_private_dev

6 years agodhcp6: fix warnings by clang with -Waddress-of-packed-member
Yu Watanabe [Tue, 16 Jan 2018 16:53:00 +0000 (01:53 +0900)]
dhcp6: fix warnings by clang with -Waddress-of-packed-member

This fixes the following warnings:
```
[194/1521] Compiling C object 'src/libsystemd-network/systemd-network@sta/dhcp6-option.c.o'.
../../git/systemd/src/libsystemd-network/dhcp6-option.c:110:25: warning: taking address of packed member 'id' of class or structure 'ia_na' may result in an unaligned pointer value [-Waddress-of-packed-member]
                iaid = &ia->ia_na.id;
                        ^~~~~~~~~~~~
../../git/systemd/src/libsystemd-network/dhcp6-option.c:115:25: warning: taking address of packed member 'id' of class or structure 'ia_ta' may result in an unaligned pointer value [-Waddress-of-packed-member]
                iaid = &ia->ia_ta.id;
                        ^~~~~~~~~~~~
2 warnings generated.
```

6 years agonetworkd: fix wrong argument check
Yu Watanabe [Tue, 16 Jan 2018 16:29:13 +0000 (01:29 +0900)]
networkd: fix wrong argument check

6 years agoipvlan: fix wrong assignment in ipvlan_init()
Yu Watanabe [Tue, 16 Jan 2018 16:28:09 +0000 (01:28 +0900)]
ipvlan: fix wrong assignment in ipvlan_init()

6 years agoman: mention that systemctl is-active or is-failed do not load units
Yu Watanabe [Tue, 16 Jan 2018 14:25:56 +0000 (23:25 +0900)]
man: mention that systemctl is-active or is-failed do not load units

See the discussion in the issue #7875.

6 years agovirt: add comment that we need to use sscanf()
Yu Watanabe [Tue, 16 Jan 2018 14:00:39 +0000 (23:00 +0900)]
virt: add comment that we need to use sscanf()

Follow-up for 13e0f9fe8334859ee86f4ff725374d1d83f5baf7.
See PR #7890 and comment in PR #7581.

6 years agoman: remove duplicated line
Yu Watanabe [Tue, 16 Jan 2018 13:22:14 +0000 (22:22 +0900)]
man: remove duplicated line

Follow-up for c46bc7e2162d774f55847c1a8cb9d49085cf89bb.

6 years agoparse-util: detect overflows in parse_percent_unbounded()
Lennart Poettering [Tue, 16 Jan 2018 10:50:12 +0000 (11:50 +0100)]
parse-util: detect overflows in parse_percent_unbounded()

We shouldn't accept percentages beyon INT32_MAX and consider them
valid.

6 years agoparse-util: coding style fix
Lennart Poettering [Tue, 16 Jan 2018 10:49:24 +0000 (11:49 +0100)]
parse-util: coding style fix

Let's not rely on C's downgrade-to-bool feature to check for NUL bytes

6 years agolocale-util: add freelocale() cleanup helper
Lennart Poettering [Tue, 16 Jan 2018 10:48:25 +0000 (11:48 +0100)]
locale-util: add freelocale() cleanup helper

6 years agoMerge pull request #7885 from pfl/dhcp6_fixes
Lennart Poettering [Tue, 16 Jan 2018 09:44:35 +0000 (10:44 +0100)]
Merge pull request #7885 from pfl/dhcp6_fixes

Minor fixes

6 years agoman: fix broken kernel document links (#7892)
John Lin [Tue, 16 Jan 2018 09:29:35 +0000 (17:29 +0800)]
man: fix broken kernel document links (#7892)

6 years agoFix parsing of features in detect_vm_xen_dom0 (#7890)
Olaf Hering [Tue, 16 Jan 2018 09:24:37 +0000 (10:24 +0100)]
Fix parsing of features in detect_vm_xen_dom0 (#7890)

Use sscanf instead of the built-in safe_atolu because the scanned string
lacks the leading "0x", it is generated with snprintf(b, "%08x", val).
As a result strtoull handles it as octal, and parsing fails.

The initial submission already used sscanf, then parsing was replaced by
safe_atolu without retesting the updated PR.

Fixes 575e6588d ("virt: use XENFEAT_dom0 to detect the hardware domain
(#6442, #6662) (#7581)")

6 years agoMerge pull request #7540 from fbuihuu/systemd-delta-tweaks
Zbigniew Jędrzejewski-Szmek [Tue, 16 Jan 2018 09:22:25 +0000 (20:22 +1100)]
Merge pull request #7540 from fbuihuu/systemd-delta-tweaks

Systemd delta tweaks

6 years agomachined: use getent to get default shell for machinectl shell (#7684)
Shawn Landden [Mon, 15 Jan 2018 18:17:51 +0000 (10:17 -0800)]
machined: use getent to get default shell for machinectl shell (#7684)

Closes: https://github.com/systemd/systemd/issues/1395

6 years agosd-dhcp6-client: Use offsetof() instead of sizeof()
Patrik Flykt [Mon, 15 Jan 2018 15:37:52 +0000 (17:37 +0200)]
sd-dhcp6-client: Use offsetof() instead of sizeof()

The slightly modified review comments say that "...in theory
offsetof(DHCP6Option, data) is nicer than sizeof(DHCP6Option)
because the former removes alignment artifacts. In this
specific case there are no alignment whitespaces hence it's
fine, but out of a matter of principle offsetof() is preferred
over sizeof() in cases like this..."

6 years agodhcp6: Fix valgrind nitpick about returned test case value
Patrik Flykt [Mon, 15 Jan 2018 15:15:13 +0000 (17:15 +0200)]
dhcp6: Fix valgrind nitpick about returned test case value

Calling dhcp6_option_parse_address() will always return a value
< 0 on error even though lt_valid remains unset. This is more
than valgrind can safely detect, but let's fix the valgrind
nitpick anyway.

While fixing, use UINT32_MAX instead of ~0 on the same line.

6 years agoMerge pull request #7884 from yuwata/small-fixes
Lennart Poettering [Mon, 15 Jan 2018 15:40:41 +0000 (16:40 +0100)]
Merge pull request #7884 from yuwata/small-fixes

Small fixes

6 years agoman: fix typo
Дилян Палаузов [Mon, 15 Jan 2018 15:09:59 +0000 (00:09 +0900)]
man: fix typo

Closes #7866.

6 years agoman: logind: update reference
Yu Watanabe [Mon, 15 Jan 2018 15:03:29 +0000 (00:03 +0900)]
man: logind: update reference

Fixes #7858.

6 years agojournal: coding style fix
Yu Watanabe [Mon, 15 Jan 2018 14:53:05 +0000 (23:53 +0900)]
journal: coding style fix

This is originally pointed out by @cpsw.

6 years agonetwork: fix indentation
Yu Watanabe [Mon, 15 Jan 2018 14:49:30 +0000 (23:49 +0900)]
network: fix indentation

Follow-up for 56a23cb40aadea95f7e24a911ba973fe132878b8.

6 years agonetworkd: remove unnecessary parentheses
Yu Watanabe [Mon, 15 Jan 2018 14:47:55 +0000 (23:47 +0900)]
networkd: remove unnecessary parentheses

Follow-up for 56a23cb40aadea95f7e24a911ba973fe132878b8.

6 years agosystemctl: comment that the verb 'check' is deprecated
Yu Watanabe [Mon, 15 Jan 2018 14:44:23 +0000 (23:44 +0900)]
systemctl: comment that the verb 'check' is deprecated

6 years agotimesync: use current uid when started as non-root user
Yu Watanabe [Mon, 15 Jan 2018 14:40:23 +0000 (23:40 +0900)]
timesync: use current uid when started as non-root user

Fixes #7883.

6 years agotest: free memory
Yu Watanabe [Mon, 15 Jan 2018 14:36:58 +0000 (23:36 +0900)]
test: free memory

Fixes #7882.

6 years agoadd false option for tests (#7778)
Adam Duskett [Mon, 15 Jan 2018 11:25:46 +0000 (06:25 -0500)]
add false option for tests (#7778)

Currently there is no way to prevent tests from building using meson.
This introduces two problems:

1) It adds a extra 381 files to compile.
2) One of these tests explicitly requires libgcrypt to be built even if systemd
   is not using it.
3) It adds C++ to the requirements to build systemd.

When cross-compiling, this is uneccessary.

6 years agoMerge pull request #7582 from pfl/dhcp6_prefix_delegation
Lennart Poettering [Mon, 15 Jan 2018 11:02:37 +0000 (12:02 +0100)]
Merge pull request #7582 from pfl/dhcp6_prefix_delegation

DHCPv6 prefix delegation

6 years agoman: --this-boot is deprecated (#7880)
Zbigniew Jędrzejewski-Szmek [Mon, 15 Jan 2018 10:45:40 +0000 (14:45 +0400)]
man: --this-boot is deprecated (#7880)

This removes the last public reference. Follow-up for #7844.

6 years agoMerge pull request #7816 from poettering/chase-pid
Zbigniew Jędrzejewski-Szmek [Mon, 15 Jan 2018 10:14:34 +0000 (14:14 +0400)]
Merge pull request #7816 from poettering/chase-pid

Make MAINPID= and PIDFile= handling more restrictive (and other stuff)

6 years agoMerge pull request #7844 from yuwata/bash-completion
Zbigniew Jędrzejewski-Szmek [Mon, 15 Jan 2018 09:51:29 +0000 (13:51 +0400)]
Merge pull request #7844 from yuwata/bash-completion

update bash completions

6 years agoMerge pull request #7855 from poettering/log-h-includes
Zbigniew Jędrzejewski-Szmek [Mon, 15 Jan 2018 09:43:09 +0000 (13:43 +0400)]
Merge pull request #7855 from poettering/log-h-includes

log.h #include cleanups

6 years agoMerge pull request #7871 from sourcejedi/log_mounts
Zbigniew Jędrzejewski-Szmek [Mon, 15 Jan 2018 09:34:06 +0000 (13:34 +0400)]
Merge pull request #7871 from sourcejedi/log_mounts

core: fix output (logging) for mount units (#7603)

6 years agoMerge pull request #7872 from mineo/coredumpctl-zsh-completion
Zbigniew Jędrzejewski-Szmek [Mon, 15 Jan 2018 09:28:14 +0000 (13:28 +0400)]
Merge pull request #7872 from mineo/coredumpctl-zsh-completion

Fix the ZSH completion on `coredumpctl gdb <tab>`

6 years agocore: prevent spurious retries of umount
Alan Jenkins [Sat, 13 Jan 2018 17:22:46 +0000 (17:22 +0000)]
core: prevent spurious retries of umount

Testing the previous commit with `systemctl stop tmp.mount` logged the
reason for failure as expected, but unexpectedly the message was repeated
32 times.

The retry is a special case for umount; it is only supposed to cover the
case where the umount command was _successful_, but there was still some
remaining mount(s) underneath.  Fix it by making sure to test the first
condition :).

Re-tested with and without a preceding `mount --bind /mnt /tmp`,
and using `findmnt` to check the end result.

6 years agozsh/coredumpctl: Never sort the completion candidates
Wieland Hoffmann [Sat, 13 Jan 2018 14:23:28 +0000 (15:23 +0100)]
zsh/coredumpctl: Never sort the completion candidates

That way, they're always sorted by date. I do not know how to make ZSH sort
them by PID through some option, but that doesn't seem very useful in the first
place.

6 years agozsh/coredumpctl: Fix the completion
Wieland Hoffmann [Sat, 13 Jan 2018 12:43:28 +0000 (13:43 +0100)]
zsh/coredumpctl: Fix the completion

An output from coredumpctl list is like

> TIME                            PID   UID   GID SIG COREFILE  EXE
> Sun 2016-05-29 18:44:03 CEST  14578  1000  1000   6 none      /tmp/pacaurtmp-wieland/python33/src/Python-3.3.6/python

  ^1  ^2         ^3       ^4    ^5

, but the previous sub() command turns that into

> TIMEPID   UID   GID SIG COREFILE  EXE
> Sun2016-05-29 18:44:03 CEST  14578  1000  1000   6 none      /tmp/pacaurtmp-wieland/python33/src/Python-3.3.6/python
  ^1            ^2       ^3    ^4     ^5

so the whole pipeline generated entries like

$UID:$DESCRIPTION

but that's not useful and probably not what was supposed to happen.

This now generates entries like

$PID:$DESCRIPTION

which make everything work.

Note that with this commmit, the completions will be sorted by PID by
ZSH.

6 years agocore: fix output (logging) for mount units (#7603)
Alan Jenkins [Sat, 13 Jan 2018 12:30:43 +0000 (12:30 +0000)]
core: fix output (logging) for mount units (#7603)

Documentation - systemd.exec - strongly implies mount units get logging.

It is safe for mounts to depend on systemd-journald.socket.  There is no
cyclic dependency generated.  This is because the root, -.mount, was
already deliberately set to EXEC_OUTPUT_NULL.  See comment in
mount_load_root_mount().  And /run is excluded from being a mount unit.

Nor does systemd-journald depend on /var.  It starts earlier, initially
logging to /run.

Tested before/after using `systemctl stop tmp.mount`.

6 years agodbus-execute: define bus_set_transient_errno() only if HAVE_SECCOMP (#7869)
0xAX [Fri, 12 Jan 2018 23:48:53 +0000 (00:48 +0100)]
dbus-execute: define bus_set_transient_errno() only if HAVE_SECCOMP (#7869)

in other way we will get a warning during build:

../src/core/dbus-util.h:55:13: warning: ‘bus_set_transient_errno’
defined but not used [-Wunused-function]

    int bus_set_transient_##function(

6 years agologin: avoid external process call
Max Harmathy [Fri, 12 Jan 2018 13:16:43 +0000 (14:16 +0100)]
login: avoid external process call

Use a shell builtin of posix shells insteaf of calling 'which'.
See also http://stackoverflow.com/a/677212

6 years agoprocess-util: make our freeze() routine do something useful
Michal Sekletar [Fri, 12 Jan 2018 12:05:48 +0000 (13:05 +0100)]
process-util: make our freeze() routine do something useful

When we crash we freeze() our-self (or possibly we reboot the machine if
that is configured). However, calling pause() is very unhelpful thing to
do. We should at least continue to do what init systems being doing
since 70's and that is reaping zombies. Otherwise zombies start to
accumulate on the system which is a very bad thing. As that can prevent
admin from taking manual steps to reboot the machine in somewhat
graceful manner (e.g. manually stopping services, unmounting data
volumes  and calling reboot -f).

Fixes #7783

6 years agoMerge pull request #7865 from fbuihuu/restore-termination-of-bus-connection-first
Lennart Poettering [Fri, 12 Jan 2018 16:24:56 +0000 (17:24 +0100)]
Merge pull request #7865 from fbuihuu/restore-termination-of-bus-connection-first

Restore termination of bus connection first

6 years agoComment the fact that some tools need to termintate their bus connect first
Franck Bui [Fri, 12 Jan 2018 15:20:36 +0000 (16:20 +0100)]
Comment the fact that some tools need to termintate their bus connect first

6 years agomount: make sure we terminate the bus connection first, and then close the pager
Franck Bui [Fri, 12 Jan 2018 15:11:46 +0000 (16:11 +0100)]
mount: make sure we terminate the bus connection first, and then close the pager

See cf647b69baee4c478d3909c327e3d917e1563f44 commit message for details.

6 years agoRevert "tree-wide: use _cleanup_(sd_bus_flush_close_unrefp) at various appropriate...
Franck Bui [Fri, 12 Jan 2018 15:09:21 +0000 (16:09 +0100)]
Revert "tree-wide: use  _cleanup_(sd_bus_flush_close_unrefp) at various appropriate places"

This reverts commit 0b3c84eb7da3a8c28ac248a68228f6a7edbb2e19.

The removal of _cleanup_() usages was done on purpose, see
cf647b69baee4c478d3909c327e3d917e1563f44.

Fixes: #3543

6 years agosupport LZMA concatenated files
Matthew Thode [Thu, 11 Jan 2018 16:48:43 +0000 (10:48 -0600)]
support LZMA concatenated files

The xz format supports concatenated files, images are sometimes created
and shipped with it to increase compression.

https://bugs.launchpad.net/ubuntu/+source/systemd/+bug/1742744 is
the source bug for this issue.

6 years agoMerge pull request #7851 from msekletar/dbus-error-logging
Lennart Poettering [Thu, 11 Jan 2018 22:37:04 +0000 (23:37 +0100)]
Merge pull request #7851 from msekletar/dbus-error-logging

dbus: propagate errors from bus_init_system() and bus_init_api()

6 years agobootctl: add newline after printf (#7856)
winnerman-pythian [Thu, 11 Jan 2018 17:01:57 +0000 (12:01 -0500)]
bootctl: add newline after printf (#7856)

Upon error condition, a newline is not printed, leading to the shell prompt jamming against this line.

6 years agoMerge pull request #7833 from keszybz/netdev-relaxed
Lennart Poettering [Thu, 11 Jan 2018 16:14:49 +0000 (17:14 +0100)]
Merge pull request #7833 from keszybz/netdev-relaxed

Avoid spurious warnings about unknown sections in .netdev files

6 years agococci: there's not ENOTSUP, there's only EOPNOTSUPP
Lennart Poettering [Fri, 5 Jan 2018 12:41:33 +0000 (13:41 +0100)]
cocci: there's not ENOTSUP, there's only EOPNOTSUPP

On Linux the former is a compat alias to the latter, and that's really
weird, as inside the kernel the two are distinct. Which means we really
should stay away from it.

6 years agoethtool-util: don't pass fds as pointers if we don't have to
Lennart Poettering [Fri, 5 Jan 2018 12:36:38 +0000 (13:36 +0100)]
ethtool-util: don't pass fds as pointers if we don't have to

Passing them as pointers is just weird, hence don't do it

6 years agoethtool-util: fix weird whitespace
Lennart Poettering [Fri, 5 Jan 2018 12:33:22 +0000 (13:33 +0100)]
ethtool-util: fix weird whitespace

6 years agoethtool-util: order includes properly
Lennart Poettering [Fri, 5 Jan 2018 12:33:10 +0000 (13:33 +0100)]
ethtool-util: order includes properly

6 years agoethtool-util: no need for memcpy() where normal assignment works too
Lennart Poettering [Fri, 5 Jan 2018 12:32:39 +0000 (13:32 +0100)]
ethtool-util: no need for memcpy() where normal assignment works too

6 years agocore: be stricter when handling PID files and MAINPID sd_notify() messages
Lennart Poettering [Fri, 5 Jan 2018 11:20:22 +0000 (12:20 +0100)]
core: be stricter when handling PID files and MAINPID sd_notify() messages

Let's be more restrictive when validating PID files and MAINPID=
messages: don't accept PIDs that make no sense, and if the configuration
source is not trusted, don't accept out-of-cgroup PIDs. A configuratin
source is considered trusted when the PID file is owned by root, or the
message was received from root.

This should lock things down a bit, in case service authors write out
PID files from unprivileged code or use NotifyAccess=all with
unprivileged code. Note that doing so was always problematic, just now
it's a bit less problematic.

When we open the PID file we'll now use the CHASE_SAFE chase_symlinks()
logic, to ensure that we won't follow an unpriviled-owned symlink to a
privileged-owned file thinking this was a valid privileged PID file,
even though it really isn't.

Fixes: #6632

6 years agonotify: add new --uid= command
Lennart Poettering [Fri, 5 Jan 2018 12:26:38 +0000 (13:26 +0100)]
notify: add new --uid= command

The new --uid= switch allows selecting the UID from which the
notificaiton messages shall originate.

This is primarily useful for testing purposes, but might have other
uses.

6 years agosd-dameon: also sent ucred when our UID differs from EUID
Lennart Poettering [Fri, 5 Jan 2018 12:24:58 +0000 (13:24 +0100)]
sd-dameon: also sent ucred when our UID differs from EUID

Let's be explicit, and always send the messages from our UID and never
our EUID. Previously this behaviour was conditionalized only on whether
the PID was specified, which made this non-obvious.

6 years agomanager: make use of pid_is_valid() where appropriate
Lennart Poettering [Fri, 5 Jan 2018 11:19:22 +0000 (12:19 +0100)]
manager: make use of pid_is_valid() where appropriate

6 years agomanager: make use of NEWLINE macro where appropriate
Lennart Poettering [Fri, 5 Jan 2018 11:18:32 +0000 (12:18 +0100)]
manager: make use of NEWLINE macro where appropriate

6 years agodbus-util: properly parse timeout values
Lennart Poettering [Fri, 5 Jan 2018 11:17:35 +0000 (12:17 +0100)]
dbus-util: properly parse timeout values

This makes transient TimeoutStopSec= properties work. After all they are
64bit entitites, not 32bit ones.

6 years agomanager: swap order in which we ellipsize/escape sd_notify() messages for debugging
Lennart Poettering [Thu, 4 Jan 2018 20:00:10 +0000 (21:00 +0100)]
manager: swap order in which we ellipsize/escape sd_notify() messages for debugging

If we have to chose between truncated escape sequences and strings
exploded to 4 times the desried length by fully escaping, prefer the
latter.

It's for debug only, hence doesn't really matter much.

6 years agofs-util: add new chase_symlinks() flag CHASE_OPEN
Lennart Poettering [Thu, 4 Jan 2018 19:00:28 +0000 (20:00 +0100)]
fs-util: add new chase_symlinks() flag CHASE_OPEN

The new flag returns the O_PATH fd of the final component, which may be
converted into a proper fd by open()ing it again through the
/proc/self/fd/xyz path.

Together with O_SAFE this provides us with a somewhat safe way to open()
files in directories potentially owned by unprivileged code, where we
want to refuse operation if any symlink tricks are played pointing to
privileged files.

6 years agofs-util: add new CHASE_SAFE flag to chase_symlinks()
Lennart Poettering [Thu, 4 Jan 2018 18:44:27 +0000 (19:44 +0100)]
fs-util: add new CHASE_SAFE flag to chase_symlinks()

When the flag is specified we won't transition to a privilege-owned
file or directory from an unprivileged-owned one. This is useful when
privileged code wants to load data from a file unprivileged users have
write access to, and validates the ownership, but want's to make sure
that no symlink games are played to read a root-owned system file
belonging to a different context.

6 years agoupdate TODO
Lennart Poettering [Thu, 11 Jan 2018 14:10:29 +0000 (15:10 +0100)]
update TODO

6 years agounit: log when we cannot add a watch on a specific PID
Lennart Poettering [Thu, 11 Jan 2018 14:07:14 +0000 (15:07 +0100)]
unit: log when we cannot add a watch on a specific PID

6 years agonetworkd: initcwn/initwnd use the right parsers
Susant Sahani [Tue, 2 Jan 2018 09:20:15 +0000 (14:50 +0530)]
networkd: initcwn/initwnd use the right parsers

Closes #7765

6 years agohwdb: Add accelerometer orientation entry for Asus TP300LA
Samuel Dionne-Riel [Wed, 27 Dec 2017 20:42:38 +0000 (15:42 -0500)]
hwdb: Add accelerometer orientation entry for Asus TP300LA

6 years agolog: assign the realm to a variable first in log_full_errno_realm()
Lennart Poettering [Thu, 11 Jan 2018 12:15:19 +0000 (13:15 +0100)]
log: assign the realm to a variable first in log_full_errno_realm()

Let's avoid duplicate evaluation (which might have side-effects) and
assign the realm to an internal variable first, as we need it twice.

6 years agolog: rework log_syntax_invalid_utf8() a bit
Lennart Poettering [Thu, 11 Jan 2018 12:13:02 +0000 (13:13 +0100)]
log: rework log_syntax_invalid_utf8() a bit

The macro used utf8.h functions without including that. Let's clean this
up, by moving that code inside of log.c.

Let's also make the call return -EINVAL in all cases. This is in line
with log_oom() which also returns a well-defined error code even though
it doesn#t take one.

6 years agolog: minor modernizations to log_received_signal()
Lennart Poettering [Thu, 11 Jan 2018 12:09:30 +0000 (13:09 +0100)]
log: minor modernizations to log_received_signal()

6 years agoUIDS-GIDS: document early on, that 32bit uids are a linux 2.4 thing
Lennart Poettering [Thu, 11 Jan 2018 12:08:57 +0000 (13:08 +0100)]
UIDS-GIDS: document early on, that 32bit uids are a linux 2.4 thing

6 years agotty-ask-password-agent: add (void) cast to mkdir/mkfifo calls we knowingly ignore
Lennart Poettering [Thu, 11 Jan 2018 12:08:18 +0000 (13:08 +0100)]
tty-ask-password-agent: add (void) cast to mkdir/mkfifo calls we knowingly ignore

6 years agolog: minimize includes in log.h
Lennart Poettering [Wed, 10 Jan 2018 23:39:12 +0000 (00:39 +0100)]
log: minimize includes in log.h

log.h really should only include the bare minimum of other headers, as
it is really pulled into pretty much everything else and already in
itself one of the most basic pieces of code we have.

Let's hence drop inclusion of:

1. sd-id128.h because it's entirely unneeded in current log.h
2. errno.h, dito.
3. sys/signalfd.h which we can replace by a simple struct forward
   declaration
4. process-util.h which was needed for getpid_cached() which we now hide
   in a funciton log_emergency_level() instead, which nicely abstracts
   the details away.
5. sys/socket.h which was needed for struct iovec, but a simple struct
   forward declaration suffices for that too.

Ultimately this actually makes our source tree larger (since users of
the functionality above must now include it themselves, log.h won't do
that for them), but I think it helps to untangle our web of includes a
tiny bit.

(Background: I'd like to isolate the generic bits of src/basic/ enough
so that we can do a git submodule import into casync for it)

6 years agosd-bus: drop bloom stuff, it's not needed anymore since kdbus is gone
Lennart Poettering [Wed, 10 Jan 2018 23:28:49 +0000 (00:28 +0100)]
sd-bus: drop bloom stuff, it's not needed anymore since kdbus is gone

6 years agodbus: propagate errors from bus_init_system() and bus_init_api()
Michal Sekletar [Wed, 10 Jan 2018 16:22:12 +0000 (17:22 +0100)]
dbus: propagate errors from bus_init_system() and bus_init_api()

The aim of this change is to make sure that we properly log about all
D-Bus connection problems. After all, we only ever attempt to get on the
bus if dbus-daemon is around, so any failure in the process should be
treated as an error.

bus_init_system() is only called from bus_init() and in
bus_init() we have a bool flag which governs whether we should attempt
to connect to the system bus or not.
Hence if we are in bus_init_system() then it is clear we got called from
a context where connection to the bus is actually required and therefore
shouldn't be treated as the "best effort" type of operation. Same
applies to bus_init_api().

We make use of those error codes in bus_init() and log high level
message that informs admin about what is going on (and is easy to spot
and makes sense to an end user).

Also "retrying later" bit is actually a lie. We won't retry unless we
are explicitly told to reconnect via SIGUSR1 or re-executed. This is
because bus_init() is always called from the context where dbus-daemon
is already around and hence bus_init() won't be called again from
unit_notify().

Fixes #7782

6 years agobus-socket: slight simplification in bus_get_peercred()
Zbigniew Jędrzejewski-Szmek [Thu, 4 Jan 2018 14:28:57 +0000 (15:28 +0100)]
bus-socket: slight simplification in bus_get_peercred()

6 years agonetdev: use "relaxed" parsing
Zbigniew Jędrzejewski-Szmek [Tue, 9 Jan 2018 13:32:44 +0000 (14:32 +0100)]
netdev: use "relaxed" parsing

We parse each netdev file twice: once to determine the type and match conditions,
and then the second time properly. In bcde742e78ac3b8e8ea348cfb022c820c11800e2
the flags for the first parsing were (inadvertently I assume) were changed to
emit warnings. But this first pass is called with only [Match] and [NetDev] sections,
so we'd get warnings about all other section types. The obvious solution would be
to remove CONFIG_PARSE_WARN again, but I think it's better to keep the warnings
and set CONFIG_PARSE_RELAXED: we do want to get warnings about malformed lines and
such, and _RELAXED is enough to kill warnings about unknown sections.

6 years agoresolved.service: set DefaultDependencies=no
Dimitri John Ledkov [Mon, 11 Dec 2017 18:27:49 +0000 (18:27 +0000)]
resolved.service: set DefaultDependencies=no

On systems that only use resolved for name resolution, there are usecases that
require resolved to be started before sysinit target, such that network name
resolution is available before network-online/sysinit targets. For example,
cloud-init for some datasources hooks into the boot process ahead of sysinit
target and may need network name resolution at that point already.

systemd-resolved already starts pretty early in the process, thus starting it
slightly earlier should not have negative side effects.

However, this depends on resolved ability to connect to system DBus once that
is up.

6 years agoFix Mount Matrix for Chuwi HiBook
Split7fire [Sun, 24 Dec 2017 14:20:07 +0000 (21:20 +0700)]
Fix Mount Matrix for Chuwi HiBook

6 years agohwdb: Add accelerometer orientation entry for Chuwi Hi10 Pro
Samuel Dionne-Riel [Wed, 27 Dec 2017 22:49:14 +0000 (17:49 -0500)]
hwdb: Add accelerometer orientation entry for Chuwi Hi10 Pro

6 years agotest-dnssec: Allow builds without libgcrypt (#7850) (#7850)
Michal Koutný [Thu, 11 Jan 2018 11:28:32 +0000 (12:28 +0100)]
test-dnssec: Allow builds without libgcrypt (#7850) (#7850)

Fixes: #7845

6 years agoIntegration of Travis CI and Coverity Scan Analysis (#7691)
Marek Čermák [Thu, 11 Jan 2018 10:41:35 +0000 (11:41 +0100)]
Integration of Travis CI and Coverity Scan Analysis (#7691)

- Coverity scan analysis tasks run as scheduled cron jobs

- Stage separation for Build, Test and Coverity scan phase
- Travis CI now uses Fedora container to build and run tests

- Containers are accessible from Docker Hub and failed builds
  can be reproduced and examined

- coverity.sh: separate build and upload

6 years agotmpfiles: consider /etc uninitialized also when /etc/machine-id is present but empty...
Franck Bui [Wed, 10 Jan 2018 22:28:44 +0000 (23:28 +0100)]
tmpfiles: consider /etc uninitialized also when /etc/machine-id is present but empty (#7849)

6 years agoMerge pull request #7846 from poettering/nobody-getenv
Zbigniew Jędrzejewski-Szmek [Wed, 10 Jan 2018 19:18:51 +0000 (20:18 +0100)]
Merge pull request #7846 from poettering/nobody-getenv

some assorted fixes and additions, in particular a way to turn off "nobody" synthesizing on a specific system

6 years agosd-netlink: let's make things compile on certain old glibc's and kernel headers again...
Lennart Poettering [Wed, 10 Jan 2018 19:08:18 +0000 (20:08 +0100)]
sd-netlink: let's make things compile on certain old glibc's and kernel headers again (#7848)

Let's include netinet/in.h instead of linux/in6.h, as the former is the
official libc location for these definitions, and the latter is a
linux-specific version that conflicts.

This hopefully makes systemd compile on current Semaphore again.

This takes e410b07d2aa64a653bc0e93b77856af41297b84d into consideration,
but makes us use glibc rather than kernel headers.

While we are at it, let's also sort our #include lines. Since kernel
headers are notoriously crappy we won't strictly order them globally,
but first include non-kernel headers in a sorted way, and then include
kernel headers in a somewhat sorted way (i.e. generic stuff first and
somewhat alphabetical, and specific stuff last)

6 years agoshutdown: make kill timeout configurable (#7835)
Jan Klötzke [Wed, 10 Jan 2018 18:00:20 +0000 (19:00 +0100)]
shutdown: make kill timeout configurable (#7835)

By default systemd-shutdown will wait for 90s after SIGTERM was sent
for all processes to exit. This is way too long and effectively defeats
an emergency watchdog reboot via "reboot-force" actions. Instead now
use DefaultTimeoutStopSec which is configurable.

6 years agoMerge pull request #7843 from hbakken/work/old-kernel-support
Lennart Poettering [Wed, 10 Jan 2018 17:59:14 +0000 (18:59 +0100)]
Merge pull request #7843 from hbakken/work/old-kernel-support

Work/old kernel support

6 years agoutil: minor tweaks to disable_core_dumps()
Lennart Poettering [Wed, 10 Jan 2018 17:37:54 +0000 (18:37 +0100)]
util: minor tweaks to disable_core_dumps()

First, let's rename it to disable_coredumps(), as in the rest of our
codebase we spell it "coredump" rather than "core_dump", so let's stick
to that.

However, also log about failures to turn off core dumpling on LOG_DEBUG,
because debug logging is always a good idea.

6 years agomanager: rework manager_clean_environment()
Lennart Poettering [Wed, 10 Jan 2018 17:28:42 +0000 (18:28 +0100)]
manager: rework manager_clean_environment()

Let's rename it manager_sanitize_environment() which is a more precise
name. Moreover, sort the environment implicitly inside it, as all our
callers do that anyway afterwards and we can save some code this way.

Also, update the list of env vars to drop, i.e. the env vars we manage
ourselves and don't want user code to interfear with. Also sort this
list to make it easier to update later on.

6 years agonss-systemd,user-util: add a way how synthesizing "nobody" can be turned off
Lennart Poettering [Wed, 10 Jan 2018 17:26:03 +0000 (18:26 +0100)]
nss-systemd,user-util: add a way how synthesizing "nobody" can be turned off

This is quite ugly, but provides us with an avenue for moving
distributions to define the "nobody" user properly without breaking legacy
systems that us the name for other stuff.

The idea is basically, that the distribution adopts the new definition
of "nobody" (and thus recompiles systemd with it) and then touches
/etc/systemd/dont-synthesize-nobody on legacy systems to turn off
possibly conflicting synthesizing of the nobody name by systemd.

6 years agotmpfiles: modernize load_unix_sockets() a bit
Lennart Poettering [Wed, 10 Jan 2018 16:28:03 +0000 (17:28 +0100)]
tmpfiles: modernize load_unix_sockets() a bit

Let's log in case of error.

Let's use read_line() instead of a fixed-size buffer

Let's make use of set_free_free()'s return value.