| Age | Commit message (Collapse) | Author |
|---|
|
NixOS switched to systemd initramfs by default.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
Was unfortunately not able to implement vm test suite before this release. Hopefully the last nixos update we have to follow.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This works around the page size limit at the cost of negligible performance regressions.
Closes #34.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
These were useful during development because timing out is often the only indication of failure due to the terrible design of nixos vm test harness. This has become a nuisance however especially when the system is under load, so remove explicit values and fall back to the ludicrously high default.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
The run command was a legacy holdover from very early days and is only useful for testing and demonstration these days. This change also renames it to exec.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This asserts that no shim private dir was left behind after all containers terminate.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This no longer exposes the pipewire socket to the container, and instead mediates access via pipewire-pulse. This makes insecure parts of the protocol inaccessible as explained in the doc comment in hst.
Closes #29.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
NixOS 25.11 introduces a crash in cage and an intermittent crash in foot.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This fully replaces PulseAudio with PipeWire and enforces the PulseAudio check and error message. The pipewire-pulse daemon is handled in the NixOS module.
Closes #26.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This should never be used outside vm tests.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
While fd created from this side always has the FD_CLOEXEC flag, the same is not true for files left open by the parent. This change prevents those files from leaking into the container.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This prevents leaking the setup pipe.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This leaves slots available for additional uid ranges in Rosa OS.
This breaks all existing installations! Users are required to fix ownership manually.
Closes #18.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
Sharing and persisting these directories do not always make sense. Make it optional here.
Closes #16.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This enables configured filesystems to cover system mount points.
Closes #8.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This sets XDG_SESSION_TYPE to the corresponding values specified in pam_systemd(8) according to enablements.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
No idea what systemd is doing with this to cause its options to change.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This change also moves pd behaviour to cmd/hakurei, as this does not belong in the hst API.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This can seriously break things, and exposes extra host attack surface, so include it here.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
The integration vm is being very slow for some reason. This change should reduce spurious timeouts.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This prepares the test suite for private TMPDIR.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
Since this is going to be placed in a shared directory, it needs to be unique to the identity. Instead of trying to figure out identity from mountinfo, just have the test script pass hardcoded values.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This is not always obvious from mountinfo.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This makes it possible to place markers with private tmpdir.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This avoids covering /dev/shm mounts from hst.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This helps work around faulty software that misinterprets unix: DISPLAY string.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
Programs expect /dev/shm to be a writable tmpfs.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
The privileged user is identifier to hakurei through its hsu userid. Using the kernel uid here makes little sense and is a leftover design choice from before hsu was implemented.
Closes #7.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
There is no reason to give the home directory special treatment, as this behaviour can be quite confusing. The home directory also does not necessarily require its own mount point, it could be provided by a parent or simply be ephemeral.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This no longer needs special treatment since it can be specified as a generic filesystem entry.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
|
|
This is the first hakurei run in the test, if the container outright fails to start this is often where it happens, so throwing away the output is very unhelpful.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
The socket is typically owned by the priv-user, and inaccessible by the target user, so just allowing access to the directory is not enough. This change fixes this oversight and add checks that will also be useful for merging https://git.gensokyo.uk/security/hakurei/pulls/1.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This finally exposes overlay mounts in the high level hakurei API.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This is unfortunately the only feasible way of doing this in nix.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
There is usually no good reason to write to /dev. This however doesn't work in internal/app because FilesystemConfig supplied by ContainerConfig might add entries to /dev, so internal/app follows DevWritable with Remount instead.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This does nothing for security, but should help avoid hiding bugs of programs developed in a hakurei container.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This is a pd workaround that does nothing in the nixos module.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
There's often good reason to mount tmpfs in the container.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
Doing this instead of mounting directly on / because it's impossible to ensure a parent is available for every path hakurei wants to mount to. This situation is similar to autoetc hence the similar name, however a symlink mirror will not work in this case.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
Most of these are differences in qemu.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This change also makes seccomp hashes cross-platform.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This tool should not show up when building hakurei normally.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This syscall is not wired on all platforms. This test barely does anything anyway and seccomp is covered by the privileged test instrumentation.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
Fortify makes little sense for a container tool.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
|
This allows apps with the same identity to access the same runtime dir.
Signed-off-by: Ophestra <cat@gensokyo.uk>
|
