Dan Streetman [Tue, 3 Jan 2017 19:37:59 +0000 (14:37 -0500)]
udev-event: add replace_whitespace param to udev_event_apply_format
If replace_whitespace is true, each substitution value has all its
whitespace removed/replaced by util_replace_whitespace (except the
SUBST_RESULT substitution - $result{} or %c{} - which handles spaces
itself as field separators). All existing callers are updated to
pass false, so no functional change is made by this patch.
This is needed so the SYMLINK assignment can replace any spaces
introduced through variable substitution, becuase the SYMLINK value is
a space-separated list of symlinks to create. Any variables that
contain spaces will thus unexpectedly change the symlink value from
a single symlink to multiple incorrectly-named symlinks.
This is used in the next patch, which enables the whitespace
replacement for SYMLINK variable substitution.
Dan Streetman [Tue, 3 Jan 2017 19:31:45 +0000 (14:31 -0500)]
libudev-util: change util_replace_whitespace to return number of chars in dest
Instead of returning 0, which is unhelpful, return the number of chars
copied into the dest string. This allows callers that care about that
to easily use it, instead of having to calculate the strlen.
No current users of the function check the return value, so this does not
break any existing code; it is used in the following patch.
The problem here is that the *i that is put into todo[] is later freed
in strv_uniq(), which is not directly visible from this patch. Somewhere
further in the code, the string that *i pointed to is freed again. That
happens only when multiple services with the same name/path are specified.
systemctl: permit "enable" and "add-wants" without any instances (#4992)
This permits "systemctl enable" and "systemctl add-wants" on template
units without any specifications of an instance name, neither specified
on the command line, nor specified in DefaultInstance= field of the
[install] section.
units: fix condition for systemd-journal-catalog-update.service (#4990)
The service is supposed to regenerate the catalog index whenever /usr is
updated, but /var is not. Hence the ConditionNeedsUpdate= line should
actually reference /var, as that's where the index file is located.
tmpfiles: automatically remove old private tmp dirs after reboot
Let's automatically destory per-unit private temporary directories, as
they are created by PrivateTmp=yes on each boot, if we notice them to be
around, in case they are left-overs from the last boot.
man: document that "systemd-run -M" propagates exit codes, and "machinectl shell" does not
This adds a brief explanation, suggesting the use of "systemd-run -M" to
acquire exit status/code information for the invoked process.
My original plan was to propagate the exit code/status in "machinectl
shell" too, but this would mean we'd have to actively watch the shell's
runtime status, and thus would need full, highly privileged and
continious access to the container's system manager, the way
"systemd-run" does it. This would be quite a departure from the
simplistic, low-priviliged OpenShell() bus call implementation of the
current code, that really just acquires a PTY device with a shell
connected.
Moreover it would blur the lines between the two commands even further,
which I think is not desirable. Hence, from now on:
"machinectl shell" is the full-session, interactive shell for human
users
"systemd-run -M …" is the low-level tool, that supports
on-interactive mode, and is more configurable and suitable for
streaming.
nss: when we encounter an invalid user/group name or UID/GID, don't return EINVAL
It's not our business to validate invalid user/group names or UID/GID.
Ideally, libc would filter these out, but they don't, hence we have to
filter, but let's not propagate this as error, but simply as "not found"
to the caller.
User name rules are pretty vaguely defined, and the rules defined by
POSIX clash with reality quite heavily (for example, utmp doesn't offer
enough room for user name length, and /usr/bin/chown permits separating
user/group names by a single dot, even though POSIX allows dots being
used in user/group names themselves.) We enforce stricter rules than
POSIX for good reason, and hence in doing so we should not categorically
return EINVAL on stuff we don't consider valid, but other components
might.
Martin Pitt [Mon, 26 Dec 2016 11:19:25 +0000 (12:19 +0100)]
tests: more precise negative check for dnsmasq log (#4982)
In test_resolved_domain_restricted_dns(), add dot domain separator to
negative .lab/.company tests, so that we don't catch these as part of
the host name (like "lxc-labjfr").
Felipe Sateler [Sat, 24 Dec 2016 10:34:26 +0000 (07:34 -0300)]
man: Align ExecStop= documentation with actual kill behavior (#4974)
The manpage claimed that ExecStop would be followed immediately by
SIGKILL, whereas the actual behavior is to go through KillMode= and
KillSignal= first.
Marcin Bachry [Fri, 23 Dec 2016 23:42:13 +0000 (00:42 +0100)]
journalctl: add remote log dir to search path when --merge is passed (#4970)
The journalctl man page says: "-m, --merge Show entries interleaved from all
available journals, including remote ones.", but current version of journalctl
doesn't live up to this promise. This patch simply adds
"/var/log/journal/remote" to search path if --merge flag is used.
Martin Pitt [Thu, 22 Dec 2016 06:58:02 +0000 (07:58 +0100)]
resolved: correctly handle non-address RR types with /etc/hosts lookups (#4808)
Fix wrong condition test in manager_etc_hosts_lookup(), which caused it to
return an IPv4 answer when an IPv6 question was asked, and vice versa.
Also only return success if we actually found any A or AAAA record.
In systemd-resolved.service(8), point out that /etc/hosts mappings only
affect address-type lookups, not other types.
The test case currently disables DNSSEC in resolved, as there is a bug
where "-t MX" fails due to "DNSSEC validation failed" even after
"downgrading to non-DNSSEC mode". This should be dropped once that bug
gets fixed.
Susant Sahani [Wed, 21 Dec 2016 18:10:36 +0000 (23:40 +0530)]
networkd: bond support primary slave and active slave (#4873)
active_slave:
Specifies the new active slave for modes that support it
(active-backup, balance-alb and balance-tlb).
primary slave:
systemd-networks currently lacks the capability to set the primary slave
in an
active-backup bonding. This is necessary if you prefer one interface
over the
other. A common example is a eth0-wlan0 bonding on a laptop where you'd
want to
switch to the wired connection whenever it's available.
Check if the parsed seconds value fits in an integer *after*
multiplying by USEC_PER_SEC, otherwise a large value can trigger
modulo by zero during normalization.
- linux does not accept prefixes for SLAAC unequal to 64 bits: http://lxr.free-electrons.com/source/net/ipv6/addrconf.c#L2741
- when networkd tries export such a route to the kernel it will get -EINVAL and
set the whole device into a failed state.
- this patch will make networkd ignore such prefixes for SLAAC,
but process other informations which may contain other prefixes.
- Note that rfc4862 does not forbid prefix length != 64 bit
systemctl: do not segfault when we cannot find template unit (#4915)
Core was generated by `systemctl cat test@.target test@.service'.
Program terminated with signal SIGSEGV, Segmentation fault.
32 movdqu (%rdi), %xmm0
(gdb) bt
-0 strrchr () at ../sysdeps/x86_64/strrchr.S:32
-1 0x00007f57fdf837fe in __GI___basename (filename=0x0) at basename.c:24
-2 0x000055b8a77d0d91 in unit_find_paths (bus=0x55b8a9242f90, unit_name=0x55b8a92428f0 "test@.service", lp=0x7ffdc9070400, fragment_path=0x7ffdc90703e0, dropin_paths=0x7ffdc90703e8) at src/systemctl/systemctl.c:2584
-3 0x000055b8a77dbae5 in cat (argc=3, argv=0x7ffdc9070678, userdata=0x0) at src/systemctl/systemctl.c:5324
-4 0x00007f57fe55fc6b in dispatch_verb (argc=5, argv=0x7ffdc9070668, verbs=0x55b8a77f1c60 <verbs>, userdata=0x0) at src/basic/verbs.c:92
-5 0x000055b8a77e477f in systemctl_main (argc=5, argv=0x7ffdc9070668) at src/systemctl/systemctl.c:8141
-6 0x000055b8a77e5572 in main (argc=5, argv=0x7ffdc9070668) at src/systemctl/systemctl.c:8412
The right behaviour is not easy in this case. Implement some "sensible" logic.
Jörg Thalheim [Mon, 19 Dec 2016 14:34:07 +0000 (15:34 +0100)]
networkd-ndisc: handle missing mtu gracefully (#4913)
At least bird's implementation of router advertisement does not
set MTU option by default (instead it supplies an option to the user).
In this case just leave MTU as it is.
We currently don't expect any warnings about format strings, on any
architecture (#4612 removed the last few warnings). Turn those warnings into
errors in the future.
As requested by Martin Pitt.
gcc documentation says that -Wformat=2 includes -Wformat-security and
-Wformat-nonliteral so don't include them explicitly.
core: downgrade "Time has been changed" to debug (#4906)
That message is emitted by every systemd instance on every resume:
Dec 06 08:03:38 laptop systemd[1]: Time has been changed
Dec 06 08:03:38 laptop systemd[823]: Time has been changed
Dec 06 08:03:38 laptop systemd[916]: Time has been changed
Dec 07 08:00:32 laptop systemd[1]: Time has been changed
Dec 07 08:00:32 laptop systemd[823]: Time has been changed
Dec 07 08:00:32 laptop systemd[916]: Time has been changed
-- Reboot --
Dec 07 08:02:46 laptop systemd[836]: Time has been changed
Dec 07 08:02:46 laptop systemd[1]: Time has been changed
Dec 07 08:02:46 laptop systemd[926]: Time has been changed
Dec 07 19:48:12 laptop systemd[1]: Time has been changed
Dec 07 19:48:12 laptop systemd[836]: Time has been changed
Dec 07 19:48:12 laptop systemd[926]: Time has been changed
...
Franck Bui [Sat, 17 Dec 2016 14:49:17 +0000 (15:49 +0100)]
coredumpctl: let gdb handle the SIGINT signal (#4901)
Even if pressing Ctrl-c after spawning gdb with "coredumpctl gdb" is not really
useful, we should let gdb handle the signal entirely otherwise the user can be
suprised to see a different behavior when gdb is started by coredumpctl vs when
it's started directly.
Indeed in the former case, gdb exits due to coredumpctl being killed by the
signal.
So this patch makes coredumpctl ignore SIGINT as long as gdb is running.
We should also mention this in NEWS before release. Suggested text:
> DBus policy files are now installed into /usr rather than /etc. Make sure
> your system has dbus = 1.9.18 running before upgrading to this version, or
> override the install path with --with-dbuspolicydir=
Franck Bui [Fri, 16 Dec 2016 16:13:58 +0000 (17:13 +0100)]
core: make mount units from /proc/self/mountinfo possibly bind to a device (#4515)
Since commit 9d06297, mount units from mountinfo are not bound to their devices
anymore (they use the "Requires" dependency instead).
This has the following drawback: if a media is mounted and the eject button is
pressed then the media is unconditionally ejected leaving some inconsistent
states.
Since udev is the component that is reacting (no matter if the device is used
or not) to the eject button, users expect that udev at least try to unmount the
media properly.
This patch introduces a new property "SYSTEMD_MOUNT_DEVICE_BOUND". When set on
a block device, all units that requires this device will see their "Requires"
dependency upgraded to a "BindTo" one. This is currently only used by cdrom
devices.
This patch also gives the possibility to the user to restore the previous
behavior that is bind a mount unit to a device. This is achieved by passing the
"x-systemd.device-bound" option to mount(8). Please note that currently this is
not working because libmount treats the x-* options has comments therefore
they're not available in utab for later application retrievals.
socket_find_symlink_target() returns a pointer to
p->address.sockaddr.un.sun_path when the first byte is non-zero without
checking that this is AF_UNIX socket. Since sockaddr is a union this
byte could be non-zero for AF_INET sockets.
Existing callers happen to be safe but is an accident waiting to happen.
Use socket_address_get_path() since it checks for AF_UNIX.
Daniel Drake [Thu, 15 Dec 2016 22:11:11 +0000 (16:11 -0600)]
rules: identify internal sound cards on platform bus (#4893)
We have a system which has the HDMI audio capability internally,
but pulseaudio is not giving it a very high priority compared
to e.g. USB sound cards.
The sound device appears on the platform bus and it is not
currently tagged with any form factor information.
It seems safe to assume that any sound card that is directly on the
platform bus is of internal form factor, but we must be careful because
udev rules will match all parent devices, not just the immediate parent,
and you will frequently encounter setups such as:
Platform bus -> USB host controller -> USB sound card
In that case, SUBSYSTEMS==platform would match even though we're
clearly working with an external USB sound card.
In order to detect true platform devices here, we rely on the observation
that if any parent devices of the sound card are PCI, USB or firewire
devices, then this sound card cannot directly connected to the platform
bus. Otherwise, if we find a parent device on the platform bus, we assume
this is an internal sound card connected directly to the platform bus.
Let's start placing our D-Bus policy files in /usr rather than /etc. D-Bus
supports this since 1.9.18, and moving our files over means we continue to work
even if /etc is flushed out entirely (for example if systemd-nspawn's
--volatile= switch is used).
Since 1.9.18 was released summer 2015 it should be fine to require a newer
version like this for our builds.
build-sys: include the builddir in $PATH while testing
udev-test.pl shells out systemd-detect-virt, and it really should invoke the
version from the build tree instead of one supplied by the installed system,
hence let's add the builddir to $PATH while building.
util-lib: rework rename_process() to be able to make use of PR_SET_MM_ARG_START
PR_SET_MM_ARG_START allows us to relatively cleanly implement process renaming.
However, it's only available with privileges. Hence, let's try to make use of
it, and if we can't fall back to the traditional way of overriding argv[0].
This removes size restrictions on the process name shown in argv[] at least for
privileged processes.
nspawn: flush out environment block of the -a stub init process
The container detection code in virt.c we ship checks for /proc/1/environ,
looking for "container=" in it. Let's make sure our "-a" init stub exposes that
correctly.
Without this "systemd-detect-virt" run in a "-a" container won't detect that it
is being run in a container.
Previously, systemd-detect-virt was unable to detect "systemd-nspawn -a"
container environments, i.e. where PID 1 is a stub process running in host
context, as in that case /proc/1/environ was inherited from the host. Let's
improve that, and add an additional check for container environments where
/proc/1/environ is not cleaned up and does not contain the $container
environment variable:
The /proc/1/sched file shows the host PID in the first line. if this is not
1, we know we are running in a PID namespace (but not which implementation).
With these changes we should be able to detect container environments that
don't set $container at all.
core: rework logic to determine when we decide to add automatic deps for mounts
This adds a concept of "extrinsic" mounts. If mounts are extrinsic we consider
them managed by something else and do not add automatic ordering against
umount.target, local-fs.target, remote-fs.target.
Extrinsic mounts are considered:
- All mounts if we are running in --user mode
- API mounts such as everything below /proc, /sys, /dev, which exist from
earliest boot to latest shutdown.
- All mounts marked as initrd mounts, if we run on the host
- The initrd's private directory /run/initrams that should survive until last
reboot.
This primarily merges a couple of different exclusion lists into a single
concept.
core: make sure targets that get a default Conflicts=shutdown.target are also ordered against it
Let's tweak the automatic dependency generation of target units: not only add a
Conflicts= towards shutdown.target but also an After= line for it, so that we
can be sure the new target is not started when the old target is still up.
Discovered in the context of #4733
(Also, exclude dependency generation if for shutdown.target itself. — This is
strictly speaking redundant, as unit_add_two_dependencies_by_name() detects
that and becomes a NOP, but let's make this explicit for readability.)
Let's be a bit more careful when detecting chroot() environments, so that we
can discern them from namespaced environments.
Previously this would simply check if the root directory of PID 1 matches our
own root directory. With this commit, we also check whether the namespaces of
PID 1 and ourselves are the same. If not we assume we are running inside of a
namespaced environment instead of a chroot() environment.
This has the benefit that systemctl (which uses running_in_chroot()) will work
as usual when invoked in a namespaced service.
projects / thirdparty / systemd.git / log
