Follow-up for 4d8b0f0f7a
After the mentioned commit, when the ExecCommand executable is missing,
and failure will be ignored by manager, we exit with EXIT_SUCCESS at executor
side too. The behavior however contradicts systemd.service(5), which states:
> If the executable path is prefixed with "-", an exit code of the command
> normally considered a failure (i.e. non-zero exit status or abnormal exit
> due to signal is _recorded_, but has no further effect and is considered
> equivalent to success.
and thus makes debugging unexpected failures harder. Therefore, let's still
exit with EXIT_EXEC, but just skip LOG_ERR level log.
This makes it possible to edit blob directories using homectl. The
following syntax is available:
* `--blob-directory=/path/somewhere`: Replaces the entire blob directory
with the contents of /path/somewhere
* `--blob-directory=foobar=/path/somewhere`: Replaces just the file
foobar in the blob directory with the contents of /path/somewhere
* `--blob-directory=foobar=`: Deletes the file foobar from the blob
directory
* `--blob-directory=`: Resets all previous flags
* `--avatar=`, etc: Shortcuts for `--blob-directory=FILENAME=` for the
known files in the blob directory
Introduces new extended variants of the various incarnations of
Create and Update, which take a map of filenames to FDs. This map is
then used to populate the bulk directory.
FDs are used to prevent the client from abusing homed's blob directory
permissions (everything is made world-readable by homed) to open files
that they normally aren't allowed to open. Passing along an FD ensures
that the client has read access to the file it wants homed to make
world-readable.
Internally, homework uses the map to overwrite the system blob dir.
Later, homework's existing blob dir reconciliation logic will propagate
the new contents from the system blob dir into the embedded blob
dir
Whenever the host & embedded records are reconciled, the host & embedded
blob directories are now reconciled too in the same direction.
Reconciling the blob directories serves exactly the same purpose as
reconciling the user records, and thus should behave in the same way.
This ensures that a user-specific blob directory exists in
/var/cache/systemd/homed for as long as the user exists, and gets
deleted if the user gets deleted.
It also advertises this blob directory via the user record, so that
clients can find and use it.
We're documenting the behavior of blob directories here. These docs
refer to things that aren't yet implemented at the time of the commit, but will be later in the same PR.
This is useful for situations where an array of FDs is to be passed into
a child process (i.e. by passing it through safe_fork). This function
can be called in the child (before calling exec) to pack the FDs to all
be next to each-other starting from SD_LISTEN_FDS_START (i.e. 3)
There's something very wrong going on when using btrfs for the test
images, namely:
- there's a significant performance hit, i.e. the Arch Linux run is
~20% slower, in the coverage run the situation is even worse
- intermittent boot failures
- intermittent "No space left on device" errors (even though there's
enough free space)
Since debugging this might take a while, let's temporarily revert back
to ext4 to make the CI stable again.
This reverts commit 7eb7e3ec4f.
During the boot process, systemd-vconsole-setup can be started when the only
allocated VC is already taken by plymouth.
This case is expected when a boot splash is displayed hence
systemd-vconsole-setup.service should not fail if it happens.
However rather than doing nothing, the sysfs utf8 flag is set before exiting
early.
Forward journal to console, since we won't have any journal from initrd
and shutdown/exit initrd phases. Also, mention
systemd.journald.max_level_console=debug that is very handy for
debugging initrd shenanigans, but don't use it by default since it
sends a _lot_ of stuff to the serial console, which slows down the test
a lot.
This does not change anything for DHCPv4, as a DHCPv4 address is always
requested anyway. However for DHCPv6, the client may not request IA_NA
addresses by UseAddress=no, or even if it is requested, the server may
not provide any IA_NA addresses. Even in such cases, here the check is
for delegated prefixes, hence it is not necessary to check if DHCPv6
IA_NA addresses are configured.
Fixes a bug introduced by 195b83edf8.
Fixes#31349.
Though, when unit_prepare_exec() is called, the unit should always
have the cgroup runtime context 'crt'. So, I think we can insert assert().
But, for consistency with other places that call unit_get_cgroup_runtime(),
here use the simple non-NULL check for 'crt' instead of using assert().
Follow-up for 9cc545447e.
Fixes CID#1534667.
The varlink_collect_full function did not set varlink client's state
when the reply was an error. The state was stuck in "collecting-reply".
I discovered that while hacking on network varlink interface (adding a
new varlink method). The debug logs shows the process of performing the
first query which replies with an error:
varlink: Setting state idle-client
network: Sending message: {"method":"io.systemd.Network.LLDPNeighbors","parameters":{"ifindex":1},"more":true}
network: Changing state idle-client → collecting
network: Received message: {"error":"org.varlink.service.MethodNotFound","parameters":{"method":"io.systemd.Network.LLDPNeighbors"}}
network: Changing state collecting → collecting-reply
Now another varlink_collect call is being made, but
network: Connection busy.
Failed to execute varlink call: Device or resource busy
This was not caught by the tests because there were no varlink_collect
calls that resulted in error reply.