nixos/rtkit: Disable debug logging and fix sd_notify

[dasJ]

Description of changes

See: heftig/rtkit#22

Things done

• Built on platform(s)
  • x86_64-linux
  • aarch64-linux
  • x86_64-darwin
  • aarch64-darwin
• For non-Linux: Is sandbox = true set in nix.conf? (See Nix manual)
• Tested, as applicable:
  • NixOS test(s) (look inside nixos/tests)
  • and/or package tests
  • or, for functions and "core" functionality, tests in lib/tests or pkgs/test
  • made sure NixOS tests are linked to the relevant packages
• Tested compilation of all packages that depend on this change using nix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD". Note: all changes have to be committed, also see nixpkgs-review usage
• Tested basic functionality of all binary files (usually in ./result/bin/)
• 22.11 Release Notes (or backporting 22.05 Release notes)
  • (Package updates) Added a release notes entry if the change is major or breaking
  • (Module updates) Added a release notes entry if the change is significant
  • (Module addition) Added a release notes entry if adding a new NixOS module
  • (Release notes changes) Ran nixos/doc/manual/md-to-db.sh to update generated release notes
• Fits CONTRIBUTING.md.

[viraptor]

This feels like a terrible hack. Is this the right way to go considering this service is type dbus, so sd_notify doesn't really change the status anyway?
See https://github.com/heftig/rtkit/blob/c295fa849f52b487be6433e69e08b46251950399/rtkit-daemon.service.in#L23
According to service docs the notification is only interesting in case of Type=notify.
On fedora for example this works without log spam and without specific socket mapping.

Or from a different angle: What does having a "Status" line do for us in a service which is actively managed through a specified dbus topic?

[dasJ]

sd_notify works with non-notify services as well. The reason to use Type=notify is startup notifications, not whether sd_notify() works. That is controlled by NotifyAccess=.

The Status: line is Status: "Supervising 0 threads of 0 processes of 0 users.", which is the same as the debug message that is logged constantly.
Fixing the status line is pretty irrelevant to me, I don't care about it at all, I just implemented it so people who care about the info that is being logged constantly have a way of accessing it that isn't annoying.

[viraptor]

Digging into this more, adding notification won't influence the logs, since they're both sent anyway: https://github.com/heftig/rtkit/blob/c295fa849f52b487be6433e69e08b46251950399/rtkit-daemon.c#L1435-L1446

I think bumping up the log level is a good idea, but otherwise the behaviour should be either fixed upstream or patched. The whole chroot("/proc") could be replaced with a strict filesystem protection at service level anyway which would unbreak the status notification. (disable do_chroot) But messing with mounting over random /proc directories is just... yuck.

[dasJ]

Agree but it was disabling chroot() vs messing with /proc and while I agree with you that this is very ugly, it's the solution that doesn't lower security.

[dasJ]

Or to make it more clear: I'm not happy with this solution at all but it shuts up the daemon while not supressing info that was available previously. And upstream does not seem to care at all: heftig/rtkit#26

[mweinelt]

We can replicate the chroot with TemporaryFileSystem and BindPaths and then disable the chroot the application creates itself.

[viraptor]

Something like this should work:

ProtectSystem=strict
PrivateDevices=true
InaccessiblePaths=/etc
TemporaryFileSystem=/var
BindPaths=/run/systemd/notify
SystemCallFilter=~@mount

(not tested)

[rvl]

This change looks good to me because it provides a quick way to see how many threads rtkit thinks should be RT.

Despite all the debug logging, rtkit still fails to do its one job on my system (see heftig/rtkit#13).

I don't think it's too important exactly how the sd-notify socket is bind-mounted into a chroot.