Michal Schmidt [Mon, 4 Mar 2013 13:38:51 +0000 (14:38 +0100)]
core/manager: fix conditions to start and stop watching running jobs
Harald encountered division by zero in manager_print_jobs_in_progress.
Clearly we had the watch enabled when we shouldn't - there were no
running jobs in m->jobs, only waiting ones. This is either a deadlock,
or maybe some of them would be detected as runnable in the next dispatch
of the run queue. In any case we mustn't crash.
Fix it by starting and stopping the watch based on n_running_jobs
instead of the number of all jobs.
core/path: install inotify watches top-down instead of bottom-up
When watches are installed from the bottom, it is always possible
to race, and miss a file creation event. The race can be avoided
if a watch is first established for a parent directory, and then for
the file in the directory. If the file is created in the time between,
the watch on the parent directory will fire.
Some messages (mostly at debug level) are added to help diagnose
pidfile issues.
Should fix https://bugzilla.redhat.com/show_bug.cgi?id=917075.
Michael Biebl [Mon, 4 Mar 2013 01:13:34 +0000 (02:13 +0100)]
build-sys: be more tolerant if dbus directories do not exist
use readlink -m instead of -f since we might be building in a minimal
chroot where those directories do not actually exist and readlink -f
would return an empty string.
Rob Clark [Sun, 3 Mar 2013 17:55:08 +0000 (12:55 -0500)]
udev: fix segfault with android rndis
The android gadget driver for network tethering over rndis somehow has a
parent device with a null subsystem. Probably this is bug in android driver,
but it is easy enough to make systemd/udev behave gracefully and not
segfault. And this will help for making linux distros with systemd
(like fedora) work on android devices.
Kay Sievers [Sun, 3 Mar 2013 17:20:09 +0000 (18:20 +0100)]
udev: firmware - do not created /run/udev/firmware-missing/
The userspace firmware loader is deprecated now, and will be entirely
removed when we depend on a kernel version with the built-in firmware
loader available.
Michael Biebl [Sat, 2 Mar 2013 21:34:01 +0000 (22:34 +0100)]
bash-completion: split completions and move to new location
Split the large bash completion script into separate, smaller files each
named after the binary it is used for and move the files to
/usr/share/bash-completion/completions. This way the completions can be
loaded on demand and we only install the completions for the tools we
actually build. The old path /etc/bash_completion.d/ is deprecated and
will disappear in the future.
Michal Schmidt [Sat, 2 Mar 2013 12:16:27 +0000 (13:16 +0100)]
job: print the "OK" status messages in normal green
The "OK" status messages should not draw attention to themselves.
It's better if they're not printed in bright/bold. Leave that
to errors and warnings.
Tom Gundersen [Sun, 24 Feb 2013 16:40:36 +0000 (17:40 +0100)]
initrd: add unit files needed for basic systemd-in-initrd support
This will:
* mount all configured filesystems (typically the rootfs on /sysroot)
* reload the configuration to pick up anything from the mounted fs (typically
/sysroot/etc/fstab)
* mount any newly configured filesystems (typically /usr on /sysroot/usr, if
applicable)
* shut-down and clean-up any daemons running in the initramfs (typically udevd)
* switch-root to /sysroot and start the real init
For an example of what files should be included in an initramfs based on this
see
<https://mailman.archlinux.org/pipermail/arch-projects/2013-February/003628.html>.
Cc: Harald Hoyer <harald.hoyer@gmail.com> Cc: Dave Reisner <d@falconindy.com>
Tom Gundersen [Fri, 22 Feb 2013 10:52:36 +0000 (11:52 +0100)]
fstab-generator: initrd - mount selected entries from /sysroot/etc/fstab
We only mount "/usr" and entries marked with "x-initrd.mount".
This (together with the right unit files) is needed in the initramfs in order to
natively support mounting /usr (and friends) from the initramfs.
The way it is meant to work is:
* wait for sysroot.mount to be mounted
* do a daemon-reload to generate sysroot-usr.mount (++) from /sysroot/etc/fstab
* wait for sysroot-usr.mount to be mounted
* switch-root
Cc: Harald Hoyer <harald.hoyer@gmail.com> Cc: Dave Reisner <d@falconindy.com>
Tom Gundersen [Fri, 1 Mar 2013 20:15:51 +0000 (21:15 +0100)]
fstab-generator: drop rootwait support
I originally added this to stay as compatible as possible with the kernel, but
as Lennart argued it is not really useful in the initramfs, so let's drop it (we
already don't support 'rootdealy').
Mantas Mikulėnas [Sun, 17 Feb 2013 19:00:34 +0000 (21:00 +0200)]
inhibit: make the output more readable
The columnar output can become pretty horrible. When GNOME inhibits
power/suspend/hibernate keys, the "WHAT" column's text extends to "WHO"
and even "WHY". At the same time, all texts in "WHY" are of the form
Michal Schmidt [Fri, 1 Mar 2013 13:47:46 +0000 (14:47 +0100)]
core: fix running jobs counters after reload/reexec
All active units will call unit_notify() during coldplug, so we just
make sure we're counting from zero again and get the correct result for
n_on_console.
For n_running_jobs we likewise reset it to zero and then count
the running jobs as we encounter them in deserialization.
Oleksii Shevchuk [Wed, 13 Feb 2013 22:49:26 +0000 (00:49 +0200)]
unit: fix the race in deserialization.
unit_notify is fired in deserelization code (particulary in
service_set_state). Units passed in random order, and there is possibility,
that unit with StopWhenUnneeded=yes passed before it actual dependecies. In
that case unit will be stopped as unneeded, because deps in UNIT_INACTIVE state
yet.
So, reuse similar logic (unit.c:1421) to avoid this race
Harald Hoyer [Fri, 1 Mar 2013 10:33:14 +0000 (11:33 +0100)]
cryptsetup-generator: fix the kernel command line strategy for luks.uuid
If rd.luks.uuid or luks.uuid is specified on the kernel command, only
generate units for these UUIDs. Additionally use the information in
/etc/crypttab unless rd.luks.crypttab=0 or luks.crypttab=0 is specified.
After that commit commands such as "systemctl enable" and friends
printed the search path information multiple times in its output, which
is ugly.
If we want the search paths to be printed at a higher log level, then we
should do this in PID 1 only, i.e. split the printing out of the normal
path lookup logic and invoke that explicitly from PID 1 but not in the
auxiliary tools.
* python-systemd-reader:
python-systemd: rename Journal to Reader
build-sys: upload python documentation to freedesktop.org
systemd-python: add Journal class for reading journal
python: build html docs using sphinx
journalct: also print Python code in --new-id
python: utilize uuid.UUID in logging
python: add systemd.id128 module
... and 34 other commits
In short: python module systemd.id128 is added, and existing
systemd.journal gains a new class systemd.journal.Reader, which can be
used to iterate over journal entries. Documentation is provided, and
accessible under e.g.
pydoc3 systemd.journal.Reader
or
firefox http://www.freedesktop.org/software/systemd/man/python-systemd/
It seems inevitable that we'll also grow a writing interface,
and then it'll be cumbersome to have a "Journal" for reading,
and a "Writer" for writing.
Michal Schmidt [Wed, 27 Feb 2013 22:31:35 +0000 (23:31 +0100)]
core/transaction: replace a bare status_printf()
Like other status messages, this one too should not be printed
unconditionally, but it should take the manager state into account.
unit_status_printf() does that.
Michal Schmidt [Wed, 27 Feb 2013 21:54:14 +0000 (22:54 +0100)]
core: redefine unit_status_printf()
Take advantage of the fact that almost all callers want to pass unit
description as the last parameter. Those who don't can use the more
flexible manager_status_printf().
This introduces a new static list of known attributes and their special
semantics. This means that cgroup attribute values can now be
automatically translated from user to kernel notation for command line
set settings, too.
This also adds proper support for multi-line attributes.
This patch broke LOG_TARGET_AUTO, i.e. automatic selection of STDERR if
it is a TTY with a fallback on the journal and kmsg otherwise.
The general rule should probably be:
log_open() -- open the "best" possible logging channel according to
log_target configuration.
log_dispatch() -- don't open any log channels ever, with the exception
of kmsg since that has no drawbacks. And do this only on true errors of
the better log channel, not just when it wasn't opened.
Logs written by journald from the initramfs may be written to a
directory with the name created from a random machine-id. Afterwards,
when the root filesystem has been mounted and machine-id reinitalized,
logs will be written to the directory with a name created from the
proper machine-id. When logs are flushed to /var/log/journal,
everything is copied to one output directory.
When journalctl without '-m' is run after the logs have been flushed
to /var/log/journal, all messages are shown. However, when run while
logs are still in /run/log/journal, those stored under the random
machine-id will not be shown.
Make journalctl behave the same regardless whether persistent storage
has been enabled or not, and slurp all files from /run/log/journal
even without '-m'.
bash-completion: journalctl query by binary and device
The approach taken is different between the two:
- since there are many files in /usr, but messages appear
only for a tiny subset, the completion is performed
only for stuff shown by journalctl -F _EXE. This makes
sense because the list is already in proper form.
- since it is hard to convert _KERNEL_DEVICE to device
file name, simply all files in /dev/ are used as possible
completions.
Unfortunately zsh completion requires more work and is not
covered by this commit.