[00:39] I'm looking at a problem involving LXD, Apparmor and mount rules and I'd appreciate if someone could help me :) [00:39] what I'm seeing suggest that mount options *ordering* would be relevant [00:39] you should pick easier problems :) [00:39] https://github.com/lxc/lxd/blob/master/lxd/apparmor/instance_lxc.go#L177: mount options=(ro,remount,bind,noatime) /[^spd]*{,/**}, [00:39] https://github.com/lxc/lxd/blob/master/lxd/apparmor/instance_lxc.go#L194: mount options=(ro,remount,noatime,bind) /[^spd]*{,/**}, [00:40] sarnold: haha [00:40] are the above 2 rules redundant and LXD devs were overly zealous? Or is AppArmor really that picky? [00:41] or maybe a past version of AppArmor was had a bug forcing this kind of workaround? [00:42] sarnold: I don't think I pick problems, they seem to pick me [00:46] sdeziel: I'm pretty sure those compile to the same thing.. [00:47] oh ffs .. I overlooked that these dump output to stderr, no stdout.. [00:48] sarnold: if you are positive those compile to the same, I'll send a PR killing this madness ;) [00:48] https://termbin.com/qpfjo [00:49] sarnold: thanks for checking and teaching me how to do, much appreciated (as always) [00:50] sdeziel: it'd be reassuring if you fiddled with it a bit and saw similar things to convince you :) hehe [00:54] sarnold: I intend to compare 2 dumps of the lxd generated profiles, one as-is and the other with the order alternations removed. I'd do this once on 18.04 and another on 20.04. Does that sound like a good test? Or should I throw in something more modern in terms of AA version? [00:55] sdeziel: that sounds like an excellent approach :) [00:55] awesome, thanks again! [00:55] sdeziel: it might also be possible / better to compare the binary blobs from the cache directories, but I don't know for sure that they're going to be deterministic ;( [00:55] I sure expect them to be.. [00:58] yeah, good point [03:36] sarnold: let's hope I did it right: https://github.com/lxc/lxd/pull/9342 [03:36] Pull 9342 in lxc/lxd "Apparmor simplification" [Open] [17:07] sdeziel: woot! nice :) thanks