Hi there, So we did tag a rc, but I have held of the release for a while on purpose. Glauber pinged me the other day to let me know that he had a bug to report with cgroup_delete_cgroup.
Basically if two subsystems are co-mounted, cgroup_delete_cgroup returns an error, despite actually succeeding. Now this is in my opinion due to the model we follow currently, where we delete the cgroup the moment we empty it, so the second subsystem cannot find its group. My proposed solution (after a short discussion with vgoyal) is simple. Let's move cgroup_delete_cgroup to a two phase model. in the first phase, we get a list of cgroups to be deleted, and therefore, don't have any duplicates in the list. In the second phase, we actually delete the cgroup. (which would entail moving the tasks away, and then deleting them). This should also lead to a cleaner error path, where we actually don't destroy the cgroup unless it actually exists for all mentioned subsystems as opposed to giving us a half done thing. However, as part of this effort, I have also realized that a lot of the code has become quite complicated. I think we should at some point in time look to simplify the code. I have to pay some attention to linsched at the moment (as I have a demo I am working on for LPC), but this rewrite is on my plate immediately after that, unless of course someone else gets to it before that (HINT! HINT!) Thanks! Dhaval ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ Libcg-devel mailing list Libcg-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/libcg-devel
