NEWS: reorder entries by subject, fix some typos and descriptions (#5511)
This doesn't add anything major, but moves some stuff around.
In particular changes which might require updates to the build
environment (new kernel requirements, cgroup stuff, dbus, etc)
are moved to the top, where it's most likely that people will
read them. In particular cgroup hierarchy changes are moved to the
top because they're most likely to be problematic.
Various items are grouped by subject where it's easy.
The description of list-jobs --after/--before was reversed.
udev: Use parent bus id for virtio disk builtin path-id (#5500)
The builtin path id for virtio block devices has been changed
to use the bus id without a prefix "virtio-pci" to be
compatible with all virtio transport types.
In order to not break existing setups, the by-path symlinks for
virtio block devices on the PCI bus are reintroduced by udev rules.
The virtio-pci symlinks are considered to be deprecated and
should be replaced by the native PCI symlinks.
Example output for a virtio disk in PCI slot 7:
$ ls /dev/disk/by-path
pci-0000:00:07.0
pci-0000:00:07.0-part1
virtio-pci-0000:00:07.0
virtio-pci-0000:00:07.0-part1
See also
[1] https://lists.freedesktop.org/archives/systemd-devel/2017-February/038326.html
[2] https://lists.freedesktop.org/archives/systemd-devel/2017-March/038397.html
This reverts f073b1b but keeps the same symlinks for compatibility.
Let'd revert this for now, see #5446 for discussions.
We want systemd-detect-virt --chroot to return true for all chroot-like stuff, for
example mock environments which have use a mount namespace. The downside
of this revert that systemctl will not work from our own namespaced services, anything
with RootDirectory=/RootImage= set.
coredump: define a macro for a "short bus call timeout"
I think it would be a good idea to move such fixed, picked values out of
the main sources into the head of a file, to make sure they are
ultimately tunables.
coredump: introduce is_journald_crash() and is_pid1_crash() helpers
We check these a number of times, hence let's unify these checks here.
This also allows us to make the PID 1 check more elaborate as we can
check both the PID and the cgroup. Checking the PID has the benefit that
we'll also cover cases where PID 1 might still be in the root cgroup, and
the cgroup check has the benefit that we also cover crashes in forked
off crasher processes (the way we actually do it in systemd)
coredump: normalize generation/parsing of COREDUMP_TRUNCATED=
Given that this is a field primarily processed by computers, and not so
much by humans, assign "1" instead of "yes". Also, use parse_boolean()
as we usually do for parsing it again.
This makes things more alike udev options (as one example), such as
SYSTEMD_READY where we also spit out "1" and "0", and parse with
parse_boolean().
mount-util: accept that name_to_handle_at() might fail with EPERM (#5499)
Container managers frequently block name_to_handle_at(), returning
EACCES or EPERM when this is issued. Accept that, and simply fall back
to to fdinfo-based checks.
Note that we accept either EACCES or EPERM here, as container managers
can choose the error code and aren't very good on agreeing on just one.
(note that this is a non-issue with nspawn, as we permit
name_to_handle_at() there, only block open_by_handle_at(), which should
be sufficiently safe).
core: when a unit's SourcePath points to API VFS pretend we are never out-of-date (#5487)
If the unit's SourcePath is below /proc then it's a unit genreated from
a kernel resource (such as a .mount or .swap unit). And those we watch
anyway, and hence should never be out-of-date.
automount: if an automount unit is masked, don't react to activation anymore (#5445)
Otherwise we'll hit an assert sooner or later.
This requires us to initialize ->where even if we come back in "masked"
mode, as otherwise we don't know how to operate on the automount and
detach it.
coredumpctl: print a hint when no journal files are found
[guest@fedora ~]$ coredumpctl
No coredumps found.
[guest@fedora ~]$ ./coredumpctl
Hint: You are currently not seeing messages from other users and the system.
Users in groups 'adm', 'systemd-journal', 'wheel' can see all messages.
Pass -q to turn off this notice.
No coredumps found.
The only functional change is that log_notice("No journal files were found.")
is not printed any more with --quiet. log_error("No journal files were opened
due to insufficient permissions.") is still printed.
I wasn't quite sure where to put this function, but shared/ seems to be the
right place and none of the existing files seem to fit too well.
v2: rename journal_access_check to journal_access_check_and_warn.
coredump: process special crashes in an (almost) normal way
We would only log a terse message when pid1 or systemd-journald crashed.
It seems better to reuse the normal code paths as much as possible,
with the following differences:
- if pid1 crashes, we cannot launch the helper, so we don't analyze the
coredump, just write it to file directly from the helper invoked by the
kernel;
- if journald crashes, we can produce the backtrace, but we don't log full
structured messages.
With comparison to previous code, advantages are:
- we go through most of the steps, so for example vacuuming is performed,
- we gather and log more data. In particular for journald and pid1 crashes we
generate a backtrace, and for pid1 crashes we record the metadata (fdinfo,
maps, etc.),
- coredumpctl shows pid1 crashes.
A disavantage (inefficiency) is that we gather metadata for journald crashes
which is then ignored because _TRANSPORT=kernel does not support structued
messages.
Messages for the systemd-journald "crash" have _TRANSPORT=kernel, and
_TRANSPORT=journal for the pid1 "crash".
Feb 26 16:27:55 systemd[1]: systemd-journald.service: Main process exited, code=dumped, status=11/SEGV
Feb 26 16:27:55 systemd[1]: systemd-journald.service: Unit entered failed state.
Feb 26 16:37:54 systemd-coredump[18801]: Process 18729 (systemd-journal) of user 0 dumped core.
Feb 26 16:37:54 systemd-coredump[18801]: Coredump diverted to /var/lib/systemd/coredump/core.systemd-journal.0.36c14bf3c6ce4c38914f441038990979.18729.1488145074000000.lz4
Feb 26 16:37:54 systemd-coredump[18801]: Stack trace of thread 18729:
Feb 26 16:37:54 systemd-coredump[18801]: #0 0x00007f46d6a06b8d fsync (libpthread.so.0)
Feb 26 16:37:54 systemd-coredump[18801]: #1 0x00007f46d71bfc47 journal_file_set_online (libsystemd-shared-233.so)
Feb 26 16:37:54 systemd-coredump[18801]: #2 0x00007f46d71c1c31 journal_file_append_object (libsystemd-shared-233.so)
Feb 26 16:37:54 systemd-coredump[18801]: #3 0x00007f46d71c3405 journal_file_append_data (libsystemd-shared-233.so)
Feb 26 16:37:54 systemd-coredump[18801]: #4 0x00007f46d71c4b7c journal_file_append_entry (libsystemd-shared-233.so)
Feb 26 16:37:54 systemd-coredump[18801]: #5 0x00005577688cf056 write_to_journal (systemd-journald)
Feb 26 16:37:54 systemd-coredump[18801]: #6 0x00005577688d2e98 dispatch_message_real (systemd-journald)
Feb 26 16:37:54 kernel: systemd-coredum: 9 output lines suppressed due to ratelimiting
Feb 26 16:37:54 systemd-journald[18810]: Journal started
Feb 26 16:50:59 systemd-coredump[19229]: Due to PID 1 having crashed coredump collection will now be turned off.
Feb 26 16:51:00 systemd[1]: Caught <SEGV>, dumped core as pid 19228.
Feb 26 16:51:00 systemd[1]: Freezing execution.
Feb 26 16:51:00 systemd-coredump[19229]: Process 19228 (systemd) of user 0 dumped core.
David Herrmann [Tue, 28 Feb 2017 20:57:58 +0000 (21:57 +0100)]
hostname: detect detachable dmi chassis type (#5489)
Detect the 'Detachable' dmi chassis type properly. Use the new
'convertible' chassis class of hostnamed, instead of returning the
generic 'computer' chassis class.
cryptsetup-generator: run cryptsetup service before swap unit (#5480)
If the cryptsetup service unit and swap unit for a swap device
are not strictly ordered, it might happen that the swap unit
activates/mounts the swap device before its cryptsetup service unit
has a chance to run the 'mkswap' command (that it is programmed to).
This leads to the following error:
Starting Cryptography Setup for sda3_crypt...
[ OK ] Found device /dev/mapper/sda3_crypt.
Activating swap /dev/mapper/sda3_crypt...
[ OK ] Activated swap /dev/mapper/sda3_crypt.
[ OK ] Reached target Swap.
[FAILED] Failed to start Cryptography Setup for sda3_crypt.
See 'systemctl status systemd-cryptsetup@sda3_crypt.service' for
details.
[DEPEND] Dependency failed for Encrypted Volumes.
Which happens because the swap device is already mounted:
# systemctl status systemd-cryptsetup@sda3_crypt.service
<...>
Active: failed (Result: exit-code) since Mon 2017-02-27 14:21:43 CST;
54s ago
<...>
<...> systemd[1]: Starting Cryptography Setup for sda3_crypt...
<...> mkswap[2420]: mkswap: error: /dev/mapper/sda3_crypt is mounted;
will not make swapspace
<...>
So, modify cryptsetup-generator to include a 'Before=' option for the
respective 'dev-mapper-%i.swap' device in the cryptsetup service unit.
Now, correct ordering is ensured, and the error no longer occurs:
Starting Cryptography Setup for sda3_crypt...
[ OK ] Found device /dev/mapper/sda3_crypt.
[ OK ] Started Cryptography Setup for sda3_crypt.
Activating swap /dev/mapper/sda3_crypt...
[ OK ] Reached target Encrypted Volumes.
[ OK ] Activated swap /dev/mapper/sda3_crypt.
[ OK ] Reached target Swap.
Since 95f1d6bfecde60b245fae1ab0313b550201e7880 we'll subscribe to unit
signals to figure out when to disconnect the pty. But that can only work
correctly if we actually subscribe to the unit's signals. Hence,
explicitly pin (and thus subscribe to) the unit we just created not only
in --wait mode but also in --pty mode.
Or to say this differently: we need to pin the unit in the same cases as
we install the signal match. 95f1d6bfecde60b245fae1ab0313b550201e7880
forgot to do that.
This is relevant to make sure systemd-run works correctly in --user
mode, and correctly exits when the spawned service dies. To test:
systemd-run --user -t /bin/bash
And then press ^D. This will hang before this change, but exit cleanly
after it.
core: always consider clients that pinned a unit to be subscribers
If a client pins a unit, then it makes sense to also implicitly make it
a subscriber. This is useful for clients that just want to watch one
specific unit: they can pin it and receive its messages.
David Herrmann [Tue, 28 Feb 2017 10:58:14 +0000 (11:58 +0100)]
hostname: detect convertible dmi chassis type
Detect the 'Convertible' dmi chassis type properly. Use the new
'convertible' chassis class of hostnamed, instead of returning the
generic 'computer' chassis class.
Based on a patch by Jani Nikula <jani.nikula@intel.com>.
David Herrmann [Tue, 28 Feb 2017 10:54:52 +0000 (11:54 +0100)]
hostname: add 'convertible' chassis type
Add the 'convertible' type to the set of allowed chassis. This applies
to all devices that can be transformed by the user from laptop style to
tablet style.
This does not add any auto-detection, yet. It only makes 'set-chassis'
accept 'convertible' as valid input.
Mikko Ylinen [Mon, 27 Feb 2017 19:03:07 +0000 (21:03 +0200)]
sd-boot: stub: check LoadOptions contains data (#5467)
With some UEFI shells LoadOptionsSize is reported being > 0
but the corresponding LoadOptions does not contain any data
(the first element has value 0).
When that happens, the stub feature that allows .cmdline to be
replaced by what's in LoadOptions ends up copying nothing/random
data to the kernel cmdline resulting in different kinds of boot
problems.
To fix this, add a check to see if LoadOptions contains data
before replacing the .cmdline.
Signed-off-by: Mikko Ylinen <mikko.ylinen@linux.intel.com>
coredumpctl,man: mark truncated messages as such in output
Unit systemd-coredump@1-3854-0.service is failed/failed, not counting it.
TIME PID UID GID SIG COREFILE EXE
Fri 2017-02-24 11:11:00 EST 10002 1000 1000 6 none /home/zbyszek/src/systemd-work/.libs/lt-Sat 2017-02-25 00:49:32 EST 26921 0 0 11 error /usr/libexec/fprintd
Sat 2017-02-25 11:56:30 EST 30703 1000 1000 - - /usr/bin/python3.5
Sat 2017-02-25 13:16:54 EST 3275 1000 1000 11 present /usr/bin/bash
Sat 2017-02-25 17:25:40 EST 4049 1000 1000 11 truncated /usr/bin/bash
For info and gdb output, the filename is marked in red and "(truncated)" is
appended. (Red is necessary because the annotation is hard to see when running
under a pager.)
coredumpctl: add debug information which services count towards the warning
A few times I have seen the hint unexpectedly. Add this so debug info
so it's easier to see what's happening.
...
Unit systemd-coredump@0-3119-0.service is failed/failed, not counting it.
Unit systemd-coredump@1-3854-0.service is activating/start-pre, counting it.
...
-- Notice: 1 systemd-coredump@.service unit is running, output may be incomplete.
coredump: when storing an incomplete coredump, add COREDUMP_TRUNCATED=yes
We logged about this, but did not attach information directly to the log
entry. It *would* be nice to log the full untruncated size, but afaict, to do
this, we would have to read the full data from the kernel. Doing this just to
log that information seems a bit excessive, in particular when the limit could
be set quite low. So for now let's just add a boolean field.
coredump: do not try to access unitialized CONTEXT_COMM field
Most of the fields in the context array come from the kernel (passed
through argv), but two are special: comm and exe. We allocate them
ourselves. We forgot to initialize context[CONTEXT_COMM] with the value
we allocated (introduced in 9aa820231414baa28e6bf02a033932cb69ff6b8b).
To simplify things, just set context[CONTEXT_COMM] and context[CONTEXT_EXE],
and free those two fields at the end.
coredumpctl: implement --since/--until (-S/-U) for info/list verbs
Implement --since/--until (-S/-U) in the same fashion as journalctl.
This lets the user filter the results a bit so it would be easier to
find relevant info in case there were many core dumps.
Let's just check the unified level, directly. There's really no value in
wrapping cg_unified_controllers() with this, i.e. potentially do string
comparison when there's no reason to.
Also, this makes the clal more alike cg_hybrid_unified().
cgroup: change cg_unified() to possibly return errors again
We use our cgroup APIs in various contexts, including from our libraries
sd-login, sd-bus. As we don#t control those environments we can't rely
that the unified cgroup setup logic succeeds, and hence really shouldn't
assert on it.
Namhyung Kim [Thu, 23 Feb 2017 12:04:57 +0000 (21:04 +0900)]
journal: avoid duplicated call to get cgroup path (#5404)
The cg_pid_get_path_shifted() is called twice during
server_dispatch_message(). We can get rid of the second by passing the
path to dispatch_message_real().
cgroup-util: cache all cg_is_*_wanted answers, disable /sys/fs/cgroups/unified on unified
If we encounter an error in proc cmdline parsing, just treat that as permanent,
i.e. the same as if the option was not specified. Realistically, it is better
to use the same condition for all related mounts, then to have e.g.
/sys/fs/cgroup mounted and /sys/fs/cgroup/unified not. If we find something is
mounted and base our answer on that, cache that result too.
Fix the conditions so that if "unified" is used, make sure any "hybrid" mounts
are not mounted.
man: update descriptions of argument-less kernel cmdline args
This updates the man page for the changes introduced in 1d84ad944520fc3e062ef518c4db4e1d3a1866af.
"=" is kep if the option is predominantly used with an argument, and dropped
otherwise.
v2:
- update also description of log_color
- drop '=' in all cases where it is optional
(previous rule of dropping it only in some cases was just too arbitrary.)
It is expected that general-purpose distributions might want to override this.
This commit is made separate from grandparent to make it easy to revert if
needed.
v2:
- use hybrid as the default
(We tested that the default of unified seems boot correctly everywhere and behave
correctly in general, but it is incompatible with docker/lxc and probably some
other tools, so for now we default to hybrid. The new "hybrid" mode should work
be OK for those tools.)
projects / thirdparty / systemd.git / log
