According to keyctl(2), the return value for KEYCTL_READ is:
The amount of data that is available in the key,
irrespective of the provided buffer size
So, we could pass in a NULL buffer to query the size, then allocate the
exact right amount of space, then call keyctl again to get the key data.
However, we must still keep the for loop to avoid TOCTOU issues: the key
might have been replaced with something bigger while we're busy
allocating the buffer to store it.
Thus, we can actually save a syscall by picking some reasonable default
buffer size and skipping the NULL call to keyctl. If our default is big
enough, we're done and have saved a syscall! If not, then the first call
behaves essentially the same as the NULL call, and we use the size it
returns to reallocate the buffer appropriately.
owneridmap bind option will map the target directory owner from inside the
container to the owner of the directory bound from the host filesystem.
This will ensure files and directories created in the container will be owned
by the directory owner of the host filesystem. All other users will remain
unmapped. Files to be written as other users in the container will not be
allowed.
Resolves: #27037
To make the order matches with log_internal().
No functional change. Hopefully silence coverity issues like
CID#1534478, CID#1534479, CID#1534480, CID#1534482.
I understand that the original motivation to introduce the template
argument here is to make journal_file_open() and _reliabrly() take the
same arguments. But, yeah, that's completely unused, not necessary to
complicate the code even the difference is not big.
The linux kernel does not update several parameters, e.g. RTA_PREF.
Hence, when we configure routes based on a RA, we need to remove
existing conflicting routes.
Fixes#28426 and #28439.
As networkd does not configure nexthops, neighbors, or routing policy rules
dynamically, this only changes link_drop_managed_addresses(). Previously
it also removed dynamic addresses, but with this commit, now it only
removes static addresses. Note, link_drop_managed_routes() has removed
only static routes, so this commit only changes its name.
Note. as commented in the code, the dynamic addresses will be removed
on lease expiration or stopping client, and link_drop_managed_xyz() are
called on reconfiguration or so, and at that time, all dynamic
addressing clients are stopped or restarted, so this should not
effective behavior change.
Our function so far assumed that the LoaderEntries's last string is or
is not NUL terminated. But if it was, then we'd debug log about this,
claiming there was an invalid id. sd-boot actually ends the list in a
properly NUL-terminated string, hence we should just accept that. Handle
that case gracefully, and add comments explaining why we have two ways
why we exit the loop.
This is cosmetic only, just suppresses a misleading debug log message.
RefUnit() only succeeds it a home dir is fully up. We already dealt with
it not being up at all, but let's also cover the case where it is
currently busy with changing state, and in that case fall back to
RefUnrestricted(), with the usual implications.
This has the effect that two subsequent ssh logins one-after-the-other
will work correctly.
ssh runs PAM session hooks before they allocate a pty for the session.
(That's because they allow multiplexed connections, and hence might run
multiple ptys over the same same session). This means PAM modules cannot
interactively ask the user for additional information as they deem fit.
That's a problem for us, since generally during an SSH login no
authentication token (aka "password") is supplied to us which we could
use to unlock the user's home dir.
With this commit we implement a way out: we allow the login to proceed,
but without the home dir activated, and then override the user's shell
with our fallback shell, which will ask for the user's password and then
chainload the actual shell.
This will of course only work if the login actually involves invoking
the configured interactive shell of the user. For other logins (such as
sftp or so), this cannot work, and they'll see an empty home dir
instead.
This is very similar to ActivateHome() but will fail if the home
directory is not referenced yet. Or in other words, this doesn't add any
new reference, but simply is the other side of RefUnrestricted(): if we
allowed a home dir to be referenced without it actually being active,
then this can catch up with things and activated what was previously
referenced already.
This also relaxes access rights to that users can always activate their
own home dirs. This is useful once we allow user code to run without the
home dir being activated.
This is useful for allowing users to login without the ability to unlock
their home dir. Usecase is ssh: ssh might grant access without giving
us the chance to unlock the home dir for the user (because it doesn't
allow us asking questions during authentication), hence with this call
we can pin the home dir, but not activate it and then allow the
activation to be delayed until later.
This adds fields to the user record logic to allow a "fallback" home
directory and shell to be set as part of the "status" section of the
user record, i.e. supplied by the manager of the user record.
The idea is that if the fallback homedir/shell is set it will take
precedence over the real one in most ways.
Usecase: let's try to make ssh logins into homed directories work.
systemd-homed would set a fallback shell/homedir for inactive home dirs.
Thus, when ssh logins take place via key auth, we can allow them, and
these fallback session params would be used because the real home cannot
be activated just yet becasue we cannot acquire any password for it from
the user.
This allows earlier PAM modules (i.e. pam_systemd_home) to inform
pam_systemd that the session is not "complete" yet (i.e. doesn't have
the home dir set up properly yet).
An UKI final command line is not just made of the content of .cmdline,
but also from the addons that are inserted in
/boot/efi/EFI/Linux/<UKI_NAME>.efi.extra.d (local addons) and /boot/efi/loader/addons
(global addons).
Therefore bootclt "status" and "list" should also include these addons
when printing the UKI command line.
Support for /boot/efi/EFI/Linux/<UKI_NAME>.efi.extra.d is already present,
so now cover addons present in /boot/efi/loader/addons (global addons).
Example (assume UKI_NAME=devel):
$ bootctl
ukiCmdline: console=tty0 console=ttyS0
globalAddon: loader/addons/global.addon.efi
cmdline: └─quiet
localAddon: devel.efi.extra.d/rpm_addon.addon.efi
cmdline: └─this is a normal addon
finalCmdline: console=tty0 console=ttyS0 quiet this is a normal addon
An UKI final command line is not just made of the content of .cmdline,
but also from the addons that are inserted in
/boot/efi/EFI/Linux/<UKI_NAME>.efi.extra.d (local addons) and
/boot/efi/loader/addons (global addons).
Therefore bootclt "status" and "list" should also include these addons
when printing the UKI command line.
Right now, discover addons present in
/boot/efi/EFI/Linux/<UKI_NAME>.efi.extra.d.
Example output (assume UKI_NAME=devel):
$ bootctl
ukiCmdline: console=tty0 console=ttyS0
localAddon: devel.efi.extra.d/rpm_addon.addon.efi
cmdline: └─this is a normal addon
finalCmdline: console=tty0 console=ttyS0 this is a normal addon
Currently, hashmap_dump_sorted sorts by key and then returns the values
in order sorted by key. This commit adds another helper that does the
same but returns the sorted keys instead
This is useful after a fork but before an exec into a binary that uses
systemd's logging utilities. For example, this should be used in dbus
services that fork off worker processes: currently, the log level set by
the LogControl dbus API will be lost because of the exec, and the worker
process will not have the correct log level set.
