Daan De Meyer [Tue, 14 Mar 2023 17:37:55 +0000 (18:37 +0100)]
chase-symlinks: Add CHASE_EXTRACT_FILENAME flag
Useful in combination with CHASE_PARENT to get a directory file
descriptor to the parent directory of the resolved path and the
filename of the resolved path in the parent directory (which might
not be the same as the filename of the input path because of symlinks).
Thomas Blume [Thu, 2 Mar 2023 11:06:13 +0000 (12:06 +0100)]
udev-rules: fix nvme symlink creation on namespace changes
The nvme by-id symlink changes to the latest namespace when a new namespace gets
added, for example by connecting multiple NVMe/TCP host controllers via nvme
connect-all.
That is incorrect for persistent device links.
The persistent symbolic device link should continue to point to the same NVMe
namespace throughout the lifetime of the current boot.
Therefore the namespace id needs to be added to the link name.
Daan De Meyer [Tue, 21 Mar 2023 13:51:56 +0000 (14:51 +0100)]
log: Add key/value support to the log context
Now that we have reference counting, it's useful to be able to push
single key values onto the log context separately, so that we don't
have to allocate new storage to join the separate string together into
a single field which means we won't be able to reuse a context containing
the same field.
Daan De Meyer [Tue, 21 Mar 2023 13:06:21 +0000 (14:06 +0100)]
log: Avoid pushing the same fields more than once on the log context
Let's try to optimize against pushing the same fields multiple times
onto the log context. To achieve this we make the log context reference
counted and return an existing context object if it's using the same
fields.
A consequence of this is that we have to make sure attaching/detaching
is coupled to the lifetime of the context object, so we make the attach
and detach functions private for now. If we need independent attach/detach
in the future, we can make that work with some extra complexity but since
we don't need it yet, let's not support it for now.
man/network-generator: replace dracut.kernel reference with dracut.cmdline
`dracut.kernel.7` is just a symlink to `dracut.cmdline.7`, so the web reference
points to a non-existent URL
(https://man7.org/linux/man-pages/man7/dracut.kernel.7.html).
Dmitry V. Levin [Thu, 16 Mar 2023 08:00:00 +0000 (08:00 +0000)]
udev-rules: fix matching of token types that support alternative patterns
For those token types that support matching of alternative patterns,
their token values are interpreted as nulstr, so make sure the parser
does the right thing and makes these token values terminated by two
subsequent NULs so they could be safely interpreted as nulstr.
Before this fix, the following rules would result to "echo foo" invocation:
ENV{foo}=", RUN"
ENV{foo}=="bar", RUN+="echo foo"
because the value of `ENV{foo}` is treated as nulstr, and it used to match
against alternative patterns, in this case `bar`, `, RUN`, and `="echo foo`.
Yu Watanabe [Sat, 18 Mar 2023 13:24:53 +0000 (22:24 +0900)]
test-systemd-tmpfiles.py: create global temporary directory
For some unknown reasons, the temporary directory created by the test
below is not removed:
```
# Test the case that a valid symlink is in the path.
label = 'valid_symlink-deep'
test_content('f= {} - - - - ' + label, label, user=user, subpath='/deep/1/2', path_cb=valid_symlink)
```
To keep /tmp clean, let's create the global temprary directory.
Daan De Meyer [Fri, 17 Mar 2023 12:07:56 +0000 (13:07 +0100)]
units: Order user@.service after systemd-oomd.service
The user manager connects to oomd over varlink. Currently, during
shutdown, if oomd is stopped before any user manager, the user
manager will try to reconnect to the socket, leading to a warning
from pid 1 about a conflicting transaction.
Let's fix this by ordering user@.service after systemd-oomd.service,
so that user sessions are stopped before systemd-oomd is stopped,
which makes sure that the user sessions won't try to start oomd via
its socket after systemd-oomd is stopped.
Yu Watanabe [Thu, 16 Mar 2023 04:33:28 +0000 (13:33 +0900)]
kernel-install: check all entry candidates first, then check existence of $pref/loader/entries only once
If we have multiple entries in $ENTRY_TOKEN_SEARCH, and $pref/loader/entries
exists, then previously we would always exit after the first candidate and
ignore the second and later candidates.
> Hmm, this is highly problematic, no? if I boot from my ssd and then
> plug in a fedora live usb stick, then there will be two ESPs around,
> the one from my ssd and the one from the live usb one, and this code
> might find the wrong one and bad things will happen
The array is a union, aligned as uint16_t, so we were accessing fields at
offsets of two, so we didn't do actually do unaligned access. But let's make
this explicit.
This is the usual set of flags. O_CLOEXEC doesn't matter because we don't spawn
anything, but O_NOCTTY could possibly make a difference if the helper is called
on a wrong device type.
Those are separate binaries, and occasionally people will get a misplaced
binary that doesn't match the rest of the installed system and be confused, so
it good to be able to check the version. It is also nice to have the same
interface in all binaries.
Note that we usually use a separate 'enum ARG_VERSION = 0x100' for an option
without a short name. We can use a less verbose approach of simply taking any
unused letter, which works just as well and even the compiler would warn us
if we tried to use the letter in another place. This way we avoid a few lines
of boilerplate.
The help texts are adjusted to have an empty line between the synopsis and
option list, and no empty lines after the option list.
The code was setting errno, which we don't do, and which is hard to get right…
Rework the code to use the usual negative-errno convention and add some
debug logging.
disk_scsi_inquiry_command() was working partially by accident: the v3 fallback
would enter the v4 check path, and pass it because the v4 data would still be
initialized to 0.