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.
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).
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().
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.
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.
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.
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)
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.
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.
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.
We should be careful with errno in cleanup functions, and not alter it
under any circumstances. In the safe_close cleanup handlers we are
already safe in that regard, but let's add similar protections on other
cleanup handlers that invoke system calls.
Why bother? Cleanup handlers insert code at function return in
non-obvious ways. Hence, code that sets errno and returns should not be
confused by us overrding the errno from a cleanup handler.
This is a paranoia fix only, I am not aware where this actually mattered
in real-life situations.
Both netinet/icmp6.h and linux/in6.h will define struct in6_addr, and in
user space we want to use the netinet/icmp6.h variant.
Fixes build problem:
In file included from src/libsystemd-network/sd-radv.c:23:0:
/home/hegtvedt/work/os/product/sunrise/root/_build/v2/include/linux/in6.h:30:8:
error: redefinition of 'struct in6_addr'
Jan Klötzke [Mon, 23 Oct 2017 06:57:09 +0000 (08:57 +0200)]
shutdown: prevent core dumps in final shutdown stage
If the system is finally shutting down it makes no sense to write core
dumps as the last remaining processes are terminated / killed. This is
especially significant in case of a "force reboot" where all processes
are hit concurrently with a SIGTERM and no orderly shutdown of
processes takes place.
Jörg Thalheim [Mon, 18 Dec 2017 14:17:06 +0000 (15:17 +0100)]
sd-netlink: add generic netlink support
This also adds the ability to incorporate arrays into netlink messages
and to determine when a netlink message is too big, used by some generic
netlink protocols.
Susant Sahani [Tue, 9 Jan 2018 08:13:06 +0000 (13:43 +0530)]
networkd: call link_set_routing_policy_rule before setting routes (#7815)
commit 7715629 (networkd: Fix race condition in [RoutingPolicyRule] handling (#7615)).
Does not fix race. Still there is a race in case of bride because the
bride goes down and up .
calling route_configure then link_set_routing_policy_rule and the
link_check_ready makes a race between routing_policy_rule_messages and route_messages.
While bride comes up and we call the call again route_configure if finds
it self in the callback function LINK_STATE_CONFIGURED networkd dies.
Let's handle first routing policy rules then route_configure. This fixes
the crash.
Closes #7797
Thomas Huth [Mon, 8 Jan 2018 09:49:15 +0000 (10:49 +0100)]
rules: Mark the kvm device as static_node (for s390x and ppc64) (#7795)
On s390x and ppc64, the permissions of the /dev/kvm device are currently
not right as long as the kvm kernel module has not been loaded yet. The
kernel module is using MODULE_ALIAS("devname:kvm") there, so the module
will be loaded on the first access to /dev/kvm. In that case, udev needs
to apply the permission to the static node already (which was created via
devtmpfs), i.e. we have to specify the option "static_node=kvm" in the
udev rule.
Note that on x86, the kvm kernel modules are loaded early instead (via the
MODULE_DEVICE_TABLE(x86cpu, ...) feature checking), so that the right module
is loaded for the Intel or AMD hypervisor extensions right from the start.
Thus the "static_node=kvm" is not required on x86 - but it also should not
hurt here (and using it here even might be more future proof in case the
module loading is also done delayed there one day), so we just add the new
option to the rule here unconditionally.
Yu Watanabe [Sun, 7 Jan 2018 16:36:08 +0000 (01:36 +0900)]
sd-device: support the case that /sys is a symlink
When /sys is a symlink to the sysfs mountpoint, e.g. /path/to/sysfs.
Then, device->syspath was set to like /path/to/sysfs/devices/foo/baz.
This converts the path to /sys/devices/foo/baz.
rkolchmeyer [Fri, 5 Jan 2018 18:28:23 +0000 (10:28 -0800)]
core: Fix edge case when processing /proc/self/mountinfo (#7811)
Currently, if there are two /proc/self/mountinfo entries with the same
mount point path, the mount setup flags computed for the second of
these two entries will overwrite the mount setup flags computed for
the first of these two entries. This is the root cause of issue #7798.
This patch changes mount_setup_existing_unit to prevent the
just_mounted mount setup flag from being overwritten if it is set to
true. This will allow all mount units created from /proc/self/mountinfo
entries to be initialized properly.
When loading .netdev files we parse them twice: first we do one parsing
iteration to figure out their "kind", and then we do it again to parse
out the kind's parameters. The first iteration is run with a "short"
NetDev structure, that only covers the generic NetDev properties. Which
should be enough, as we don't parse the per-kind properties. However,
before this patch we'd still try to destruct the per-kind properties
which resulted in memory corruption. With this change we distuingish the
two iterations by the state field, so that the destruction only happens
when the state signals we are running with a full NetDev structure.
Since this is not obvious, let's add a lot of comments.
sd-bus: remove 'hint_sync_call' parameter from various function calls
This is unused since kdbus is gone, hence remove this too. This permits
us to get rid of sd_bus_send_internal() and just implement sd_bus_send()
directly.
pid1: set org.freedesktop.systemd1 as sender service name for direct connections
This way, clients can install the very same match on direct and broker
connections as in both cases the messages will originate from the o.f.s1
service.
sd-bus: add API to optionally set a sender field on all outgoing messages
This is useful on direct connections to generate messages with valid
sender fields.
This is particularly useful for services that are accessible both
through direct connections and the broker, as it allows clients to
install matches on the sender service name, and they work the same in
both cases.
sd-bus: accept NULL callbacks in sd_bus_call_async()
This way sd_bus_call_method_async() (which is just a wrapper around
sd_bus_call_async()) can be used to put method calls together that
expect no reply.
networkd,resolved: make use of watch_bind feature to connect to the bus
The changes both networkd and resolved to make use of the watch_bind
feature of sd-bus to connect to the system bus. This way, both daemons
can be started during early boot, and automatically and instantly
connect to the system bus as it becomes available.
This replaces prior code that used a time-based retry logic to connect
to the bus.
With this new API sd-bus can synthesize a local "Connected" signal when
the connection is fully established. It mirrors the local "Disconnected"
signal that is already generated when the connection is terminated. This
is useful to be notified when connection setup is done, in order to
start method calls then, in particular when using "slow" connection
methods (for example slow TCP, or most importantly the "watch_bind"
inotify logic).
Note that one could also use hook into the initial NameAcquired signal
received from the bus broker, but that scheme works only if we actually
connect to a bus. The benefit of "Connected" OTOH is that it works with
any kind of connection.
Ideally, we'd just generate this message unconditionally, but in order
not to break clients that do not expect this message it is opt-in.
This new call is much light sd_bus_is_open(), but returns true only if
the connection is fully set up, i.e. after we finished with the
authentication and Hello() phase. This API is useful for clients in
particular when using the "watch_bind" feature, as that way it can be
determined in advance whether it makes sense to sync on some operation.
sd-bus: drop references to legacy /var/run D-Bus socket
Let's directly reference /run instead, so that we can work without /var
being around, or with /var/run being incorrectly set up.
Note that we keep the old socket path in place when referencing the
system bus of containers, as they might be foreign operating systems,
that still don't have adopted /run, and where it makes sense to use the
standardized name instead. On local systems, we insist on /run being set
up properly however, hence this limitation does not apply.
Also, get rid of the UNIX_SYSTEM_BUS_ADDRESS and
UNIX_USER_BUS_ADDRESS_FMT defines. They had a purpose when we still did
kdbus, as we then had to support two different backends. But since
that's gone, we don't need this indirection anymore, hence settle on a
one define only.
Let's remove a number of synchronization points from our service
startups: let's drop synchronous match installation, and let's opt for
asynchronous instead.
Also, let's use sd_bus_match_signal() instead of sd_bus_add_match()
where we can.
sd-bus: add new API sd_bus_match_signal() + sd_bus_match_signal_asnyc()
These are convenience helpers that hide the match string logic (which we
probably should never have exposed), and instead just takes regular C
arguments.
sd-bus: add asynchronous version of sd_bus_match()
We usually enqueue a number of these calls on each service
initialization. Let's do this asynchronously, and thus remove
synchronization points. This improves both performance behaviour and
reduces the chances to deadlock.
sd-bus: when removing a server-side match, do so in "fire and forget" fashion
We currently wait for the RemoveMatch() reply, but then ignore what it
actually says. Let's optimize this a bit, and not even ask for an answer
back: just enqueue the RemoveMatch() operation, and do not request not
wait for any answer.
sd-bus: add APIs to request/release names asynchronously
They do the same thing as their synchronous counterparts, but only
enqueue the operation, thus removing synchronization points during
service initialization.
If the callback function is passed as NULL we'll fallback to generic
implementations of the reply handlers, that terminate the connection if
the requested name cannot be acquired, under the assumption that not
being able to acquire the name is a technical problem.
When kdbus was still around we always had two implementations of the
various control calls: one for dbus1 and one for kdbus. Let'sget rid of
this, simplify things, and just merge the wrappers that used to
multiplex this with the implementations.
No change in behaviour, just some merging of functions
sd-bus: start reply callback timeouts only when the connection is established
Currently, reply callback timeouts are started the instant the method
calls are enqueued, which can be very early on. For example, the Hello()
method call is enqueued right when sd_bus_start() is called, i.e. before
the socket connection and everything is established.
With this change we instead start the method timeout the moment we
actually leave the authentication phase of the connection. This way, the
timeout the kernel applies on socket connecting, and we apply on the
authentication phase no longer runs in parallel to the Hello() method
call, but all three run serially one after the other, which is
definitely a cleaner approach.
Moreover, this makes the "watch bind" feature a lot more useful, as it
allows enqueuing method calls while we are still waiting for inotify
events, without them timeouting until the connection is actually
established, i.e. when the method call actually has a chance of being
actually run.
This is a change of behaviour of course, but I think the new behaviour
is much better than the old one, since we don't race timeouts against
each other anymore...
sd-bus: optionally, use inotify to wait for bus sockets to appear
This adds a "watch-bind" feature to sd-bus connections. If set and the
AF_UNIX socket we are connecting to doesn't exist yet, we'll establish
an inotify watch instead, and wait for the socket to appear. In other
words, a missing AF_UNIX just makes connecting slower.
This is useful for daemons such as networkd or resolved that shall be
able to run during early-boot, before dbus-daemon is up, and want to
connect to dbus-daemon as soon as it becomes ready.
bus: touch() the AF_UNIX sockets we listen() on after the fact
We'd like to use inotify to get notified when AF_UNIX sockets become
connectable. That happens at the moment of listen(), but this is doesn't
necessarily create in a watchable inotify event. Hence, let's synthesize
one whenever we generically create a socket, or when we know we created
it for a D-Bus server.
Ideally we wouldn't have to do this, and the kernel would generate an
event anyway for this. Doing this explicitly isn't too bad however, as
the event is still nicely associated with the AF_UNIX socket node, and
we generate all D-Bus sockets in our code hence it's safe.
socket-util: explicitly ensure there's one trailing NUL byte on AF_UNIX socket addresses
AF_UNIX socket addresses aren't necessarily NUL terminated, however
they are usually used as strings which are assumed to be NUL terminated.
Let's hence add an extra byte to the end of the sockaddr_un structure,
that contains this NUL byte, simply for safety reasons.
Note that actually this patch changes exactly nothing IRL, as the other
sockaddr structures already are large enough to accomodate for an extra
NUL byte. The size of the union hence doesn't change at all by doing
this. The entire value of this patch is hence in the philosophical
feeling of safety, and by making something explicit that before was
implicit.
This changes two things when binding to AF_UNIX file system sockets:
1. When wethe socket already exists in the fs, and unlink() on it fails,
don't bother to bind() a second time: since nothing changed it won't
work either.
2. Also use SELinux-aware bind() for the second attempt.
fs-util: rework touch_file() so that it can touch socket file nodes
Let's rework touch_file() so that it works correctly on sockets, fifos,
and device nodes: let's open an O_PATH file descriptor first and operate
based on that, if we can. This is usually the better option as it this
means we can open AF_UNIX nodes in the file system, and update their
timestamps and ownership correctly. It also means we can correctly touch
symlinks and block/character devices without triggering their drivers.
Moreover, by operating on an O_PATH fd we can make sure that we
operate on the same inode the whole time, and it can't be swapped out in
the middle.
While we are at it, rework the call so that we try to adjust as much as
we can before returning on error. This is a good idea as we call the
function quite often without checking its result, and hence it's best to
leave the files around in the most "correct" fashion possible.
sd-bus: when attached to an sd-event loop, disconnect on processing errors
If we can't process the bus for some reason we shouldn't just disable
the event source, but log something and give up on the connection. Hence
do that, and disconnect.
sd-bus: propagate handling errors for Hello method reply directly
Currently, when sd-bus is used to issue a method call, and we get a
reply and the specified reply handler fails, we log this locally at
debug priority and proceed. The idea is that a bad server-side reply
should not be fatal for the program, except when the developer
explicitly terminates the event loop.
The reply to the initial Hello() method call we issue when joining a bus
should not be handled like that however. Instead, propagate the error
immediately, as anything that is wrong with the Hello() reply should be
considered a fatal connection problem.
projects / thirdparty / systemd.git / log
