On Mon 29-10-12 18:06:34, Glauber Costa wrote: > On 10/29/2012 06:04 PM, Glauber Costa wrote: > > On 10/26/2012 03:37 PM, Michal Hocko wrote: > >> Now that mem_cgroup_pre_destroy callback doesn't fail (other than a race > >> with a task attach resp. child group appears) finally we can safely move > >> on and forbit all the callbacks to fail. > >> The last missing piece is moving cgroup_call_pre_destroy after > >> cgroup_clear_css_refs so that css_tryget fails so no new charges for the > >> memcg can happen. > >> We cannot, however, move cgroup_call_pre_destroy right after because we > >> cannot call mem_cgroup_pre_destroy with the cgroup_lock held (see > >> 3fa59dfb cgroup: fix potential deadlock in pre_destroy) so we have to > >> move it after the lock is released. > >> > > > > If we don't have the cgroup lock held, how safe is the following > > statement in mem_cgroup_reparent_charges(): > > > > if (cgroup_task_count(cgrp) || !list_empty(&cgrp->children)) > > return -EBUSY; > > > > ? > > > > IIUC, although this is not generally safe, but it would be safe here > > because at this point we are expected to had already set the removed bit > > in the css. If this is the case, however, this condition is impossible > > and becomes useless - in which case you may want to remove it from Patch1. > > > Which I just saw you doing in patch5... =) Yes, I just wanted to keep this one cgroup core only to enable further cgroup clean ups easier. Dropping the earlier in the series could introduce regressions which I tried to avoid as much as possible. Thanks -- Michal Hocko SUSE Labs -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>