It is pretty hard to figure out what the problem actually is, esp. when the rule
is long.
On my machine:
systemd[1]: Starting udev Kernel Device Manager...
systemd-udevd[217399]: /usr/lib/udev/rules.d/11-dm-lvm.rules:40 Invalid value for OPTIONS key, ignoring: 'event_timeout=180'
systemd-udevd[217399]: /usr/lib/udev/rules.d/11-dm-lvm.rules:40 The line takes no effect, ignoring.
systemd-udevd[217399]: /etc/udev/rules.d/60-ipath.rules:4 Invalid value "kcopy/%02n" for NAME (char 7: invalid substitution type), ignoring, but please fix it.
systemd-udevd[217399]: /usr/lib/udev/rules.d/65-md-incremental.rules:28 Invalid value "/sbin/mdadm -I $env{DEVNAME} --export $devnode --offroot ${DEVLINKS}" for IMPORT (char 58: invalid substitution type), ignoring, but please fix it.
systemd-udevd[217399]: /etc/udev/rules.d/73-special-net-names.rules:14 Invalid value "/bin/sh -ec 'D=${DEVPATH#*/vio/}; D=${D%%%%/*}; D=${D#????}; D=${D#0}; D=${D#0}; D=${D#0}; D=${D#0}; echo ${D:-0}'" for PROGRAM (char 16: invalid substitution type), ignoring, but please fix it.
systemd-udevd[217399]: /usr/lib/udev/rules.d/84-nm-drivers.rules:10 Invalid value "/bin/sh -c 'ethtool -i $1 | sed -n s/^driver:\ //p' -- $env{INTERFACE}" for PROGRAM (char 24: invalid substitution type), ignoring, but please fix it.
systemd-udevd[217399]: /usr/lib/udev/rules.d/90-libgpod.rules:19 IMPORT key takes '==' or '!=' operator, assuming '==', but please fix it.
systemd-udevd[217399]: /usr/lib/udev/rules.d/90-libgpod.rules:23 IMPORT key takes '==' or '!=' operator, assuming '==', but please fix it.
systemd-udevd[217399]: /usr/lib/udev/rules.d/99-vmware-scsi-udev.rules:5 Invalid value "/bin/sh -c 'echo 180 >/sys$DEVPATH/device/timeout'" for RUN (char 27: invalid substitution type), ignoring, but please fix it.
systemd-udevd[217399]: /usr/lib/udev/rules.d/99-vmware-scsi-udev.rules:6 Invalid value "/bin/sh -c 'echo 180 >/sys$DEVPATH/device/timeout'" for RUN (char 27: invalid substitution type), ignoring, but please fix it.
systemd[1]: Started udev Kernel Device Manager.
Whenever I see EXTRACT_QUOTES, I'm always confused whether it means to
leave the quotes in or to take them out. Let's say "unquote", like we
say "cunescape".
This reverts a part of commit 49fe5c0996 ('tree-wide: port various places
over to STARTSWITH_SET()') that replaced a pair of startswith() calls
with STARTSWITH_SET().
They were in fact for a different strings (device vs. name), botching
the crypttap parsing.
Previous code was allocating an array of iovecs big enough to store all the
fields added later by various functions.
This forced us to calculate the size of the array in advance which is too error
prone if for example one wants to add new fields or simply rework the
code. Various assertions were added to make sure there's no overflow but it's
still more code for no good reasons.
Instead, this patch switches to the new iovec array handling interface so the
array is grown dynamically when needed.
The other contraint was that some iovecs were supposed to be freed whereas some
others were not. This makes the code hard to (re)organize. The new code always
allocates fields so it becomes easier to rework the code.
We'd skip any whitespace immediately after "=", but then we'd treat whitespace
that is between "|" or "!" and the value as significant. This is rather
confusing, let's ignore it too.
If hibernation is requested but /sys/power/resume and
/sys/power/resume_offset are not configured correctly, systemd-sleep
attempts to calculate swapfile offset using fstat and fiemap.
Btrfs returns virtual device number for stat and a virtual offset
for fiemap which results in incorrect offset calculations. In the
case where offset would be calculated and the user is using Btrfs,
log a debug message and fail to write device and offset values.
Use hibernation configuration as defined in
/sys/power/resume and /sys/power/resume_offset
if present before inspecting /proc/swaps and
attempting to calculate swapfile offset
job_compare return value is undefined in case the jobs have a loop
between them, so better make a test to make sure transaction cycle
detection catches it.
The test-engine Test2 tests the cycle detection when units a, b and d
all start at once
,-------------------after-----------------,
v |
a/start ---after---> d/start ---after---> b/start
Extend the test with Test11 that adds i.service which causes a and d
stop (by unordered Conflicts=) while starting b. Because stops precede
starts, we effectively eliminate the job cycle and all transaction jobs
should be applicable.
,-------------------after-----------------,
v |
a/stop <---after--- d/stop <---after--- b/start
. . ^
. . |
'. . . . . . . . . i/start ---after------'
When we are validating a transaction, we take into account declared
ordering between job units. However, since JOB_STOP goes always first
regardless of the ordering constraint between respective units, we may
detect some false cycles in the transaction which would not prevent the
execution though.
Use the same logic in transaction checking as we use for job execution.
In this mode we are not supposed to "interact with the environment", so loading
all units and printing warnings about syntax errors and /var/run usage seems
inappropriate.
Originally the description of conditions was brief, so it was acceptable
to put this part at the end. But now we have a myriad conditions, and
this crucial bit of information is easy to miss.
I couldn't see any reason why the kernel could provide COMM to the coredump
handler via the core_pattern command line but could not make it available in
/proc. So let's assume that this info is always available in /proc.
For "backtrace" mode (when --backtrace option is passed), I assumed that the
crashing process still exists at the time systemd-coredump is called.
Also changing the core_pattern line is an API breakage for any users of the
backtrace mode but given that systemd-coredump is installed in
/usr/lib/systemd, it's a private tool which has no internal users. At least no
one complained when the hostname was added to the core_pattern line
(f45b801551)...
Indeed it's much easier to get it from /proc since the kernel substitutes '%e'
specifier with multiple strings if the process name contains spaces (!).
The job ordering logic is spread at multiple places of the code, making
it hard to maintain and also a bit to understand. The actual execution
order of two jobs always depends on their types and the ordering
contraint between their units. Extract this logic to a new function
job_compare. The second change is simplification of the order
evaluation, JOB_STOP takes always precedence (as documented), unless two
units are both stopping, then the ordering constraint is taken into
account.
Yu Watanabe
Zbigniew Jędrzejewski-Szmek
Lubomir Rintel
Franck Bui
Michael Prokop
Michael Stapelberg
Zach Smith
Evgeny Vereshchagin
Michal Koutný
Michal Sekletar