This removes the requirement to call fmsg.Exit on every exit path, and enables direct use of the "log" package. However, fmsg.BeforeExit is still encouraged when possible to catch exit on suspended output.
Signed-off-by: Ophestra <cat@gensokyo.uk>
App is no longer just a simple [exec.Cmd] wrapper, so exposing these steps separately no longer makes sense and actually hinders proper error handling, cleanup and cancellation. This change removes the five-second wait when the shim dies before receiving the payload, and provides caller the ability to gracefully stop execution of the confined process.
Signed-off-by: Ophestra <cat@gensokyo.uk>
The primary use case for extra perms is app-specific state directories, which may or may not exist (first run of any app).
Signed-off-by: Ophestra <cat@gensokyo.uk>
Fortify state store instances was specific to aids due to outdated design decisions carried over from the ego rewrite. That no longer makes sense in the current application, so the interface now enables a single store object to manage all transient state.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
The socket-based approach is no longer necessary as fsu allows extra files and sudo compatibility is no longer relevant.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
This value is used for passwd generation. Bad input can cause very confusing issues. This is not a security issue, however validation will improve user experience.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
This removes the dependency on external user switchers like sudo/machinectl and decouples fortify user ids from the passwd database.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
Also moved /dev/fortify to /fortify since it is impossible to create new directories in /dev from the init namespace and bind mounting its contents has undesirable side effects.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
Chromium's D-Bus client implementation refuses to work when its getuid call returns a different value than what the D-Bus server is running as. The reason behind this is not fully understood, but this workaround is implemented to support chromium and electron apps. This is not used by default since it has many side effects that break many other programs, like SSH on NixOS.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
Pointing xdg-dbus-proxy to stdout/stderr makes a huge mess. This change enables app to neatly print out prefixed xdg-dbus-proxy messages after output is resumed.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
This implements lazy loading of the systemd marker (they are not accessed in init and shim) and ensures consistent behaviour when running with a stub.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
This change moves all user switcher and shim management to the shim package and withholds output while shim is alive. This also eliminated all exit scenarios where revert is skipped.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
This test seals App against a deterministic os stub and checks the resulting sys and bwrap values against known correct ones. The effects of sys and bwrap on the OS and sandbox is deterministic and tested in their own respective packages.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
Fortify overrides /etc/passwd and /etc/group in the sandbox. Bind mounting /etc results in them being replaced when the passwd database is updated on host.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
This is already supported by the underlying bwrap helper. This change exposes access to it in Config.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
This prevents a hang when setup faults but the shim keeps waiting on the socket. Setup is automatically aborted when the shim is killed.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
This change helps tests stub out and simulate OS behaviour during the sealing process. This also removes dependency on XDG_RUNTIME_DIR as the internal.System implementation provided to App provides a compat directory inside the tmpdir-based share when XDG_RUNTIME_DIR is unavailable.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
Shim setup listens on a socket in the process share, if shim setup hasn't happened on exit revert will fail. This change makes sure shim setup is aborted on a doomed launch.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
The behaviour of print functions from package fmt is not thread safe. Functions provided by fmsg wrap around Logger methods. This makes prefix much cleaner and makes it easy to deal with future changes to logging.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
Values used in the Wayland mediation implementation is stored in various struct fields strewn across multiple app structs and checks are messy and confusing. This commit unifies them into a single struct and access it using much better looking methods.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
shareTmpdirChild happened to request an ephemeral dir within SharePath and was called before shareRuntime which ensures that path. This commit moves SharePath initialisation to shareSystem and moves shareTmpdirChild into ShareSystem. Further cleanup and tests are desperately needed for the app package but for now this fix will have to do.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
This commit does away with almost all baggage left over from the Ego port. Error wrapping also got simplified. All API changes happens to be internal which means no changes to main except renaming of the BaseError type.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
The argument builder was written based on the incorrect assumption that bwrap arguments are unordered. The argument builder is replaced in this commit to correct that mistake.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
Bubblewrap as init is a bit awkward and don't support a few setup actions fortify will need, such as starting/supervising nscd.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
Both machinectl and sudo launch methods launch shim as shim is now responsible for setting up the sandbox. Various app structures are adapted to accommodate bwrap configuration and mediated wayland access.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
ACL operations are now tagged with the enablement causing them. At the end of child process's life, enablements of all remaining launchers are resolved and inverted. This allows Wait to only revert operations targeting resources no longer required by other launchers.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
Launch methods serve the primary purpose of setting UID in the init namespace, which bubblewrap does not do. Furthermore, all applications will start within a bubblewrap sandbox once it has been implemented.
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
There was an earlier attempt of cleaning up the app package however it ended up creating even more of a mess and the code structure largely still looked like Ego with state setup scattered everywhere and a bunch of ugly hacks had to be implemented to keep track of all of them. In this commit the entire app package is rewritten to track everything that has to do with an app in one thread safe value.
In anticipation of the client/server split also made changes:
- Console messages are cleaned up to be consistent
- State tracking is fully rewritten to be cleaner and usable for multiple process and client/server
- Encapsulate errors to easier identify type of action causing the error as well as additional info
- System-level setup operations is grouped in a way that can be collectively committed/reverted
and gracefully handles errors returned by each operation
- Resource sharing is made more fine-grained with PID-scoped resources whenever possible,
a few remnants (X11, Wayland, PulseAudio) will be addressed when a generic proxy is available
- Application setup takes a JSON-friendly config struct and deterministically generates system setup operations
Signed-off-by: Ophestra Umiker <cat@ophivana.moe>
