There are many subtleties when recovering from errors in the event loop, and coldboot requires internals to drain the receive buffer as synthetic uevents are being arranged.
Signed-off-by: Ophestra <cat@gensokyo.uk>
This makes package check portable, and removes nonportable behaviour from package pkg, pipewire, and system. All other packages remain nonportable due to their nature. No latency increase was observed due to this change on amd64 and arm64 linux.
Signed-off-by: Ophestra <cat@gensokyo.uk>
This is done by both systemd sd-device and AOSP ueventd to improve robustness. Rosa OS will still handle ENOBUFS via coldboot but a big buffer should mitigate this as well.
Signed-off-by: Ophestra <cat@gensokyo.uk>
This requires the caller to arrange for a coldboot to happen, some time after this error is encountered, and to resume event processing.
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>
These are not possible to cover outside integration vm. Extreme care is required when dealing with this method, so keep it simple.
Signed-off-by: Ophestra <cat@gensokyo.uk>
The wire format and behaviour is entirely undocumented. This is implemented by reading lib/kobject_uevent.c, with testdata collected from the internal/rosa kernel.
Signed-off-by: Ophestra <cat@gensokyo.uk>
This runs in the Rosa OS init, so recover as much as possible, as otherwise it is likely to require a full system reboot to resume event processing. The caller is responsible for reporting the error.
Signed-off-by: Ophestra <cat@gensokyo.uk>
This type is used very frequently. The new type is much easier to type and can receive helper methods eventually if needed.
Signed-off-by: Ophestra <cat@gensokyo.uk>
Some python tools try to be clever and buffers output. This makes the build process appear to hang and is quite frustrating. Instead of trying to address this on a case-by-case basis, this is turned off globally for the interpreter.
Signed-off-by: Ophestra <cat@gensokyo.uk>
This is very useful for troubleshooting failing tests and such. The ephemeral state is cleaned up by internal/pkg.
Signed-off-by: Ophestra <cat@gensokyo.uk>
This also gets rid of the cached pid value for port since that prevents multiple sockets from being open at once.
Signed-off-by: Ophestra <cat@gensokyo.uk>
The behaviour of this was changed to include the stage2 toolchain instead, but the help text was never updated.
Signed-off-by: Ophestra <cat@gensokyo.uk>
