We should not keep @css_set_lock spinlock while removing group files which leads to
| BUG: sleeping function called from invalid context at kernel/locking/mutex.c:938 this is because cgroup_rm_file() function takes @kernfs_mutex itself. To fix this we should fetch @cset under the lock and grab a reference, then we can safely walk over set links to remove the files. https://jira.sw.ru/browse/PSBM-135460 Signed-off-by: Cyrill Gorcunov <gorcu...@gmail.com> --- kernel/cgroup/cgroup.c | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-) Index: vzkernel/kernel/cgroup/cgroup.c =================================================================== --- vzkernel.orig/kernel/cgroup/cgroup.c +++ vzkernel/kernel/cgroup/cgroup.c @@ -2163,19 +2163,27 @@ void cgroup_unmark_ve_roots(struct ve_st */ cset = rcu_dereference_protected(ve->ve_ns, lockdep_is_held(&ve->op_sem))->cgroup_ns->root_cset; + BUG_ON(!cset); + get_css_set(cset); + spin_unlock_irq(&css_set_lock); + /* + * Once we've grabbed a reference to cset the list + * entries won't be removed so traversing is safe. + */ list_for_each_entry(link, &cset->cgrp_links, cgrp_link) { cgrp = link->cgrp; if (!is_virtualized_cgroup(cgrp)) continue; - cgroup_rm_file(cgrp, cft); + if (!cgroup_is_dead(cgrp)) + cgroup_rm_file(cgrp, cft); rcu_assign_pointer(cgrp->ve_owner, NULL); clear_bit(CGRP_VE_ROOT, &cgrp->flags); } - spin_unlock_irq(&css_set_lock); + put_css_set(cset); mutex_unlock(&cgroup_mutex); /* ve_owner == NULL will be visible */ synchronize_rcu(); _______________________________________________ Devel mailing list Devel@openvz.org https://lists.openvz.org/mailman/listinfo/devel
