Add a new option `--network-namespace-path` to systemd-nspawn to allow
users to specify an arbitrary network namespace, e.g. `/run/netns/foo`.
Then systemd-nspawn will open the netns file, pass the fd to
outer_child, and enter the namespace represented by the fd before
running inner_child.
```
$ sudo ip netns add foo
$ mount | grep /run/netns/foo
nsfs on /run/netns/foo type nsfs (rw)
...
$ sudo systemd-nspawn -D /srv/fc27 --network-namespace-path=/run/netns/foo \
/bin/readlink -f /proc/self/ns/net
/proc/1/ns/net:[4026532009]
```
Note that the option `--network-namespace-path=` cannot be used together
with other network-related options such as `--private-network` so that
the options do not conflict with each other.
Fixes https://github.com/systemd/systemd/issues/7361
If user namespacing is used, let's make sure that the root user in the
container gets access to both /sys/fs/cgroup/systemd and
/sys/fs/cgroup/unified.
This matches similar logic in cg_set_access().
In -U mode we might need to re-chown() all files and directories to
match the UID shift we want for the image. That's problematic on fat
partitions, such as the ESP (and which is generated by mkosi's
--bootable switch), because fat of course knows no UID/GID file
ownership natively.
With this change we take benefit of the uid= and gid= mount options FAT
knows: instead of chown()ing all files and directories we can just
specify the right UID/GID to use at mount time.
This beefs up the image dissection logic in two ways:
1. First of all support for mounting relevant file systems with
uid=/gid= is added: when a UID is specified during mount it is used for
all applicable file systems.
2. Secondly, two new mount flags are added:
DISSECT_IMAGE_MOUNT_ROOT_ONLY and DISSECT_IMAGE_MOUNT_NON_ROOT_ONLY.
If one is specified the mount routine will either only mount the root
partition of an image, or all partitions except the root partition.
This is used by nspawn: first the root partition is mounted, so that
we can determine the UID shift in use so far, based on ownership of
the image's root directory. Then, we mount the remaining partitions
in a second go, this time with the right UID/GID information.
Also, add "cgroups.stat". It's read-only anyway, hence its UID/GID
ownership matters little, but it's probably a good idea to keep it
ownership in sync with the other read-only files such as
"cgroups.controllers".
Also, order the list of files alphabetically.
If we operate on a disk image (i.e. --image=) then it's pointless to
look into the mount directory before it is actually mounted to see which
systemd version is running inside...
Unfortunately we only mount the disk image in the child process, but the
parent needs to know the cgroup mode, hence add some IPC for this
purpose and communicate the cgroup mode determined from the image back
to the parent.
When "-U" is used we look for a UID range we can use for our container.
We start with the UID the tree is already assigned to, and if that
didn't work we'd pick random ranges so far. With this change we'll first
try to hash a suitable range from the container name, and use that if it
works, in order to make UID assignments more likely to be stable.
This follows a similar logic PID 1 follows when using DynamicUser=1.
Since time began, scope units had a concept of "Controllers", a bus peer
that would be notified when somebody requested a unit to stop. None of
our code used that facility so far, let's change that.
This way, nspawn can print a nice message when somebody invokes
"systemctl stop" on the container's scope unit, and then react with the
right action to shut it down.
Let's rename get_controllers() → get_process_controllers(), in order to
underline the difference to cg_kernel_controllers(). After all, one
returns the controllers available to the process, the other the
controllers enabled in the kernel at all).
Let's also update the code to use read_line() and set_put_strdup() to
shorten the code a bit, and make it more robust.
We shouldn't call alloca() as part of function calls, that's not really
defined in C. Hence, let's first do our stack allocations, and then
invoke functions.
Also, some coding style fixes, and minor shuffling around.
No functional changes.
So far I avoided adding license headers to meson files, but they are pretty
big and important and should carry license headers like everything else.
I added my own copyright, even though other people modified those files too.
But this is mostly symbolic, so I hope that's OK.
We currently use the ownership of the top-level directory as a hint
whether we need to descent into the whole tree to chown() it recursively
or not. This is problematic with the previous chown()ing algorithm, as
when descending into the tree we'd first chown() and then descend
further down, which meant that the top-level directory would be chowned
first, and an aborted recursive chowning would appear on the next
invocation as successful, even though it was not. Let's reshuffle things
a bit, to make the re-chown()ing safe regarding interruptions:
a) We chown() the dir we are looking at last, and descent into all its
children first. That way we know that if the top-level dir is
properly owned everything inside of it is properly owned too.
b) Before starting a chown()ing operation, we mark the top-level
directory as owned by a special "busy" UID range, which we can use to
recognize whether a tree was fully chowned: if it is marked as busy,
it's definitely not fully chowned, as the busy ownership will only be
fixed as final step of the chowning.
Fixes: #6292
This adds some simply detection logic for cases where dissection is
invoked on an externally created loop device, and partitions have been
detected on it, but partition scanning so far was off. If this is
detected we now print a brief message indicating what the issue is,
instead of failing with a useless EINVAL message the kernel passed to
us.
This adds some basic discovery of block device images for nspawn and
friends. Note that this doesn't add searching for block devices using
udev, but instead expects users to symlink relevant block devices into
/var/lib/machines. Discovery is hence done exactly like for
dir/subvol/raw file images, except that what is found may be a (symlink
to) a block device.
For now, we do not support cloning these images, but removal, renaming
and read-only flags are supported to the point where that makes sense.
Fixe: #6990
This fixes --read-only with --private-users. mkdir_userns_p may return
-EROFS if either mkdir or lchown fails; lchown failing is fine as the
mount point will just be overmounted, and if mkdir fails then the
following mount() will also fail (with ENOENT).
The current code shifting an integer 1 failed for capabilities like
CAP_MAC_ADMIN (numerical value 33). This caused issues when specifying
them in the nspawn configuration file. Using an uint64_t 1 instead.
The similar code for processing the --capability command line option
was already correctly working.
Also, move prlimit64() out of @resources.
prlimit64() may be used both for getting and setting resource limits, and
is implicitly called by glibc at various places, on some archs, the same
was as getrlimit(). SImilar, igetrlimit() is an arch-specific
replacement for getrlimit(), and hence should be whitelisted at the same
place as getrlimit() and prlimit64().
Also see: https://lists.freedesktop.org/archives/systemd-devel/2017-September/039543.html
The advantage is that is the name is mispellt, cpp will warn us.
$ git grep -Ee "conf.set\('(HAVE|ENABLE)_" -l|xargs sed -r -i "s/conf.set\('(HAVE|ENABLE)_/conf.set10('\1_/"
$ git grep -Ee '#ifn?def (HAVE|ENABLE)' -l|xargs sed -r -i 's/#ifdef (HAVE|ENABLE)/#if \1/; s/#ifndef (HAVE|ENABLE)/#if ! \1/;'
$ git grep -Ee 'if.*defined\(HAVE' -l|xargs sed -i -r 's/defined\((HAVE_[A-Z0-9_]*)\)/\1/g'
$ git grep -Ee 'if.*defined\(ENABLE' -l|xargs sed -i -r 's/defined\((ENABLE_[A-Z0-9_]*)\)/\1/g'
+ manual changes to meson.build
squash! build-sys: use #if Y instead of #ifdef Y everywhere
v2:
- fix incorrect setting of HAVE_LIBIDN2
This removes the '@credentials' syscall set that was added in commit
v234-468-gcd0ddf6f75.
Most of these syscalls are so simple that we do not want to filter them.
They work on the current calling process, doing only read operations,
they do not have a deep kernel path.
The problem may only be in 'capget' syscall since it can query arbitrary
processes, and used to discover processes, however sending signal 0 to
arbitrary processes can be used to discover if a process exists or not.
It is unfortunate that Linux allows to query processes of different
users. Lets put it now in '@process' syscall set, and later we may add
it to a new '@basic-process' set that allows most basic process
operations.
keyring material should not leak into the container. So far we relied on
seccomp to deny access to the keyring, but given that we now made the
seccomp configurable, and access to keyctl() and friends may optionally
be permitted to containers now let's make sure we disconnect the callers
keyring from the keyring of PID 1 in the container.
Let's lock things down a bit, and maintain a list of what's permitted
rather than a list of what's prohibited in nspawn (also to make things a
bit more like Docker and friends).
Note that this slightly alters the effect of --system-call-filter=, as
now the negative list now takes precedence over the positive list.
However, given that the option is just a few days old and not included
in any released version it should be fine to change it at this point in
time.
Note that the whitelist is good chunk more restrictive thatn the
previous blacklist. Specifically:
- fanotify is not permitted (given the buffer size issues it's
problematic in containers)
- nfsservctl is not permitted (NFS server support is not virtualized)
- pkey_xyz stuff is not permitted (really new stuff I don't grok)
- @cpu-emulation is prohibited (untested legacy stuff mostly, and if
people really want to run dosemu in nspawn, they should use
--system-call-filter=@cpu-emulation and all should be good)
Now that we have ported nspawn's seccomp code to the generic code in
seccomp-util, let's extend it to support whitelisting and blacklisting
of specific additional syscalls.
This uses similar syntax as PID1's support for system call filtering,
but in contrast to that always implements a blacklist (and not a
whitelist), as we prepopulate the filter with a blacklist, and the
unit's system call filter logic does not come with anything
prepopulated.
(Later on we might actually want to invert the logic here, and
whitelist rather than blacklist things, but at this point let's not do
that. In case we switch this over later, the syscall add/remove logic of
this commit should be compatible conceptually.)
Fixes: #5163
Replaces: #5944
glibc appears to propagate different errors in different ways, let's fix
this up, so that our own code doesn't get confused by this.
See #6752 + #6737 for details.
Fixes: #6755
Given that we set NOTIFY_SOCKET unconditionally it's not surprising that
processes way down the process tree think it's smart to send us a
notification message.
It's still useful to keep this message, for debugging things, but it
shouldn't be generated by default.
Zbigniew Jędrzejewski-Szmek
Lennart Poettering
Dongsu Park
Yu Watanabe
Daniel Lockyer
Shawn Landden
Lauri Tirkkonen
myrkr
Alan Jenkins
Djalal Harouni
Andreas Rammhold