NeilBrown [Wed, 8 Nov 2017 08:29:32 +0000 (19:29 +1100)]
umount: always use MNT_FORCE in umount_all() (#7213)
The linux umount2() systemcall accepts a MNT_FORCE flags
which some filesystems honor, particularly FUSE and various
network filesystems such as NFS.
These filesystems can sometimes wait for an indefinite period
for a response from an external service, and the wait if
sometimes "uninterruptible" meaning that the process cannot be
killed.
Using MNT_FORCE causes any such request that are outstanding to
be aborted. This normally allows the waiting process to
be killed. It will then realease and reference it has to the
filesytem, this allowing the filesystem to be unmounted.
If there remain active references to the filesystem, MNT_FORCE
is *not* forcefull enough to unmount the filesystem anyway.
By the time that umount_all() is run by systemd-shutdown, all
filesystems *should* be unmounted, and sync() will have been
called. Anything that remains cannot be unmounted in a
completely clean manner and just nees to be dealt with as firmly
as possible. So use MNT_FORCE and try to explain why in the
comment.
Also enhance an earlier comment to explain why umount2() is
safe even though mount(MNT_REMOUNT) isn't.
Yu Watanabe [Wed, 8 Nov 2017 06:50:58 +0000 (15:50 +0900)]
core/execute: do not create RuntimeDirectory= under private/ sub-directory
RuntimeDirectory= often used for sharing files or sockets with other
services. So, if creating them under private/ sub-directory, we cannot
set DynamicUser= to service units which want to share something through
RuntimeDirectory=.
This makes the directories given by RuntimeDirectory= are created under
/run/ even if DynamicUser= is set.
Yu Watanabe [Fri, 13 Oct 2017 12:22:25 +0000 (21:22 +0900)]
core: ReadWritePaths= and friends assume '+' prefix when BindPaths= or freinds are set
When at least one of BindPaths=, BindReadOnlyPaths=, RootImage=,
RuntimeDirectory= or their friends are set, systemd prepares
a namespace under /run/systemd/unit-root. Thus, ReadWritePaths=
or their friends without '+' prefix is completely meaningless.
So, let's assume '+' prefix when one of them are set.
Alan Jenkins [Tue, 7 Nov 2017 16:14:15 +0000 (16:14 +0000)]
core: simplify - don't add jobs to dbus queue if we immediately remove them (#7251)
job_finish_and_invalidate() calls job_free() to destroy jobs (and remove
them from the dbus queue). So we don't need to add them to the dbus queue
first.
We only want to add jobs to the dbus queue if they're a restart job, which
we're transmogrifying into a start job and putting back into the system.
Susant Sahani [Thu, 2 Nov 2017 12:36:03 +0000 (18:06 +0530)]
networkd: remove route - drop route type from netlink message. (#7240)
During startup of networkd we try to drop the configs. While droping
routes we filling ip route type and because of which message like
```
host: Could not drop route: Invalid argument
host: Could not drop route: Invalid argument
```
are shown.
util-lib: mark variable with _unused_ to silence clang warning
_unused_ means "the variable is meant to be possible unused and gcc
will not generate a warning about it", which is exactly what we need here,
since we're only declaring it for the side effect of _cleanup_.
journal: disable -Waddress-of-packed-member under clang
clang warns about a few sites like this:
../src/journal/journal-file.c:1780:48: warning: taking address of packed member 'entry_offset' of class or structure 'DataObject' may result in an unaligned pointer value [-Waddress-of-packed-member]
&o->data.entry_offset,
^~~~~~~~~~~~~~~~~~~~
but DataObject.entry_offset will always be 8-byte aligned as long as
the DataObject structure is aligned. Similarly in other cases, the
field is always aligned. Let's just silence the warning to avoid noise.
gcc does not know -Waddress-of-packed-member, and would warn about an unknown
warning, so we need to conditionalize on __clang__.
../src/network/networkd-link.c:3577:84: warning: format specifies type 'unsigned char' but the argument has type 'uint32_t' (aka 'unsigned int') [-Wformat]
route->dst_prefixlen, route->tos, route->priority, route->table, route->lifetime);
^~~~~~~~~~~~
../src/network/networkd-manager.c:1146:132: warning: format specifies type 'unsigned char' but the argument has type 'uint32_t' (aka 'unsigned int') [-Wformat]
rule->from_prefixlen, space ? " " : "", to_str, rule->to_prefixlen, rule->tos, rule->fwmark, rule->fwmask, rule->table);
^~~~~~~~~~~
Also add some line breaks to make it easier to see which argument is for which
part of the format string.
clang warns:
../src/import/importd.c:254:70: warning: 'break' is bound to current loop, GCC binds it to the enclosing loop [-Wgcc-compat]
while ((e < t->log_message + t->log_message_size) && IN_SET(*e, 0, '\n'))
^
Let's just play it safe and not use IN_SET here.
Alan Jenkins [Mon, 14 Aug 2017 10:15:38 +0000 (11:15 +0100)]
core: failure to spawn ExecStartPost should not run ExecStop
Failure to spawn ExecStartPost was being handled differently to e.g.
EXIT_FAILURE returned by ExecStartPost. It looks like this was an
oversight. Fix to match documented behaviour.
`man systemd.service`:
> Note that if any of the commands specified in ExecStartPre=, ExecStart=,
> or ExecStartPost= fail (and are not prefixed with "-", see above) or time
> out before the service is fully up, execution continues with commands
> specified in ExecStopPost=, the commands in ExecStop= are skipped.
Alan Jenkins [Mon, 2 Oct 2017 13:40:15 +0000 (14:40 +0100)]
core: distinguish "Killing"/"Terminating"/"Stopping" for mount unit timeout
Update the timeout warnings for remount and unmount. For consistency with
mount, for accuracy, and for consistency with their equivalents in
service.c.
Michal Sekletar [Wed, 1 Nov 2017 09:25:48 +0000 (02:25 -0700)]
manager: fix connecting to bus when dbus is actually around (#7205)
manager_connect_bus() is called *before* manager_coldplug(). As a last
thing in service_coldplug() we set service state to
s->deserialized_state, and thus before we do that all services are
inactive and try_connect always evaluates to false. To fix that we must
look at deserialized state instead of current unit state.
Alan Jenkins [Mon, 30 Oct 2017 16:10:37 +0000 (16:10 +0000)]
core: remove "misuse" of getpgid() in systemd-shutdown
Using `kill()` with a signal of 0 is a slightly more documented idiom for
checking whether a process still exists. It is mentioned explicitly in
man pages. This avoids the need to comment the call as "misuse".
A comment is still necessary - in fact this idiom is even more confusing if
you don't know how it works. But it's easy enough to explain.
Hans de Goede [Mon, 30 Oct 2017 13:33:54 +0000 (14:33 +0100)]
hwdb: Update GP-electronic T701 accel mount settings (#7220)
The GP-electronic T701 has its LCD panel mounted upside-down, initially
my plan was to fix this by transparently rotating the image in the i915
driver (my "drm/i915: Deal with upside-down mounted LCD" patch), but
that approach has been rejected instead the kernel will now export
a "panel orientation" property on the drm-connector for the panel and
let userspace deal with it.
Since the upside-down-ness of the panel is now no longer transparently
hidden from userspace, the current accel mount quirk for the T701 needs
to be updated to take the upside-down-ness into account.
rules: add Freescale IMX serial ports name to serial port udev rule (#7203)
Freescale IMX SoCs serial ports driven by kernel "imx-uart" driver have
names of "ttymxcN", let's add this pattern to an udev rule for serial
ports so they will have proper ownership applied.
Peter Hutterer [Mon, 30 Oct 2017 00:42:29 +0000 (10:42 +1000)]
hwdb: add a hwdb file to override ID_INPUT assignments
The input_id builtin assigns the various ID_INPUT based on the exported evdev
bits. In some cases, the device may not have the properties required to label
a device as one specific type but the physical form factor is clear.
e.g. in the case of #7197 it's a tablet pad that does not have x/y axes which
the kernel exports for pads for historical reasons.
A custom override is needed, best to be solved with a hwdb entry.
Lars Karlitski [Fri, 27 Oct 2017 03:10:47 +0000 (05:10 +0200)]
journalctl: add --output-fields= (#7181)
This option allows restricting the shown fields in the output modes that
would normally show all fields. It allows clients that are only
interested in a subset of the fields to access those more efficiently.
Also, it makes the resulting size of the output more predictable.
It has no effect on the various `short` output modes, because those
already only show a subset of the fields.
core: add support for expanding state/cache/log directory root in unit files
This augments %t which already resolves to the runtime directory root, and
should be useful for units that want to pass any of these paths in
command line arguments.
Example:
ExecStart=/usr/bin/mydaemon --datadir=%S/mydaemon
Why not expose a specifier resolving directly to the configured
state/runtime/cache/log dir? Three reasons:
1. Specifiers should be independent of configuration of the unit itself,
and StateDirectory= and friends are unit configuration. See 03fc9c723cfc59467a7fccc305f34273f8564b25 and related work.
2. We permit multiple StateDirectory= values per unit, and it hence
wouldn't be clear which one is passed.
3. We already have %t for the runtime directory root, and we should
continue with the same scheme.
dissect: when we invoke dissection on a loop device with partscan help the user
This adds some simply detection logic for cases where dissection is
invoked on an externally created loop device, and partitions have been
detected on it, but partition scanning so far was off. If this is
detected we now print a brief message indicating what the issue is,
instead of failing with a useless EINVAL message the kernel passed to
us.
machine-image: add partial discovery of block devices as images
This adds some basic discovery of block device images for nspawn and
friends. Note that this doesn't add searching for block devices using
udev, but instead expects users to symlink relevant block devices into
/var/lib/machines. Discovery is hence done exactly like for
dir/subvol/raw file images, except that what is found may be a (symlink
to) a block device.
For now, we do not support cloning these images, but removal, renaming
and read-only flags are supported to the point where that makes sense.
btrfs-util: when opening subvolume fds, always set O_NOFOLLOW
Some of the btrfs utility functions already used O_NOFOLLOW others
didn't. Let's streamline this, and refuse operation when we are called
for symlinks on "remove" and "snapshot" too.
In particular in the "remove" case following symlinks is a bad idea, and
is quite different from how unlink() and friends work, which always
remove the symlink, and not the destination, a logic we should follow
here too.
Lauri Tirkkonen [Tue, 24 Oct 2017 17:40:50 +0000 (20:40 +0300)]
nspawn: EROFS for chowning mount points is not fatal (#7122)
This fixes --read-only with --private-users. mkdir_userns_p may return
-EROFS if either mkdir or lchown fails; lchown failing is fine as the
mount point will just be overmounted, and if mkdir fails then the
following mount() will also fail (with ENOENT).
After previous output from systemd-shutdown indicated a bug, my attention
was drawn to redundant output lines. Did they indicate an anomaly?
It turns out to be an expected, harmless result of the current code. But
we don't have much justification to run such redundant operations. Let's
remove the confusing redundant message.
We can stop trying to remount a directory read-only once its mount entry
has successfully been changed to "ro". We can simply let the kernel keep
track of this for us. I don't bother to try and avoid re-parsing the
mountinfo. I appreciate snappy shutdowns, but this code is already
intricate and buggy enough (see issue 7131).
(Disclaimer: At least for the moment, you can't _rely_ on always seeing
suspicious output from systemd-shutdown. By default, you can expect the
kernel to truncate the log output of systemd-shutdown. Ick ick ick!
Because /dev/kmsg is rate-limited by default. Normally it prints a message
"X lines supressed", but we tend to shut down before the timer expires
in this case).
I also tested with `systemctl reboot --force`, plus a loopback mount to
cause one of the umounts to fail initially. In this case another 2 lines
of output are removed (out of a larger number of lines).
core: in --user mode, report READY=1 as soon as basic.target is reached (#7102)
When a user logs in, systemd-pam will wait for the user manager instance to
report readiness. We don't need to wait for all the jobs to finish, it
is enough if the basic startup is done and the user manager is responsive.
systemd --user will now send out a READY=1 notification when either of two
conditions becomes true:
- basic.target/start job is gone,
- the initial transaction is done.
myrkr [Tue, 24 Oct 2017 07:56:40 +0000 (09:56 +0200)]
nspawn: Fix calculation of capabilities for configuration file (#7087)
The current code shifting an integer 1 failed for capabilities like
CAP_MAC_ADMIN (numerical value 33). This caused issues when specifying
them in the nspawn configuration file. Using an uint64_t 1 instead.
The similar code for processing the --capability command line option
was already correctly working.
Peter Hutterer [Mon, 23 Oct 2017 08:20:16 +0000 (18:20 +1000)]
hwdb: invert the USB touchpad integration assumption (#7161)
The vast majority of touchpads are internal, so let's assume that any USB
touchpad is internal by default (exception: bluetooth) and manually mark the
ones that are external. That's a lot more future-proof than having to mark all
internal touchpads that use USB as internal - that number is only going to
increase.
Michael Biebl [Sat, 21 Oct 2017 06:32:50 +0000 (08:32 +0200)]
modprobe.d: fix directory of modprobe configuration file (#7149)
kmod upstream uses /lib/modprobe.d which means we need to use rootprefix
instead of prefix for installing the modprobe configuration file as
otherwise split-usr systems are broken.
See https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=879191
Alan Jenkins [Thu, 19 Oct 2017 14:00:57 +0000 (15:00 +0100)]
core: systemd-shutdown: add missing check for umount_changed
The assumption was that nothing changes in the final attempt. This
would be confusing if a filesystem with a process in uninterruptible
sleep suddenly became un-stuck for the final attempt, but we still give
up and don't try to e.g. unmount any parent mounts.
I don't know how possible that is. But the code will be easier to read
without an assumption that it does not attempt to justify.
projects / thirdparty / systemd.git / log
