From ab3f5faa6255a0eb4f832675507d9e295ca7e9ba Mon Sep 17 00:00:00 2001 From: Hugh Dickins Date: Thu, 6 Feb 2014 15:56:01 -0800 Subject: cgroup: use an ordered workqueue for cgroup destruction Sometimes the cleanup after memcg hierarchy testing gets stuck in mem_cgroup_reparent_charges(), unable to bring non-kmem usage down to 0. There may turn out to be several causes, but a major cause is this: the workitem to offline parent can get run before workitem to offline child; parent's mem_cgroup_reparent_charges() circles around waiting for the child's pages to be reparented to its lrus, but it's holding cgroup_mutex which prevents the child from reaching its mem_cgroup_reparent_charges(). Just use an ordered workqueue for cgroup_destroy_wq. tj: Committing as the temporary fix until the reverse dependency can be removed from memcg. Comment updated accordingly. Fixes: e5fca243abae ("cgroup: use a dedicated workqueue for cgroup destruction") Suggested-by: Filipe Brandenburger Signed-off-by: Hugh Dickins Cc: stable@vger.kernel.org # 3.10+ Signed-off-by: Tejun Heo --- kernel/cgroup.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) (limited to 'kernel/cgroup.c') diff --git a/kernel/cgroup.c b/kernel/cgroup.c index e2f46ba37f72..aa95591c1430 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -4845,12 +4845,16 @@ static int __init cgroup_wq_init(void) /* * There isn't much point in executing destruction path in * parallel. Good chunk is serialized with cgroup_mutex anyway. - * Use 1 for @max_active. + * + * XXX: Must be ordered to make sure parent is offlined after + * children. The ordering requirement is for memcg where a + * parent's offline may wait for a child's leading to deadlock. In + * the long term, this should be fixed from memcg side. * * We would prefer to do this in cgroup_init() above, but that * is called before init_workqueues(): so leave this until after. */ - cgroup_destroy_wq = alloc_workqueue("cgroup_destroy", 0, 1); + cgroup_destroy_wq = alloc_ordered_workqueue("cgroup_destroy", 0); BUG_ON(!cgroup_destroy_wq); /* -- cgit v1.2.3 From eb46bf89696972b856a9adb6aebd5c7b65c266e4 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Sat, 8 Feb 2014 10:26:33 -0500 Subject: cgroup: fix error return value in cgroup_mount() When cgroup_mount() fails to allocate an id for the root, it didn't set ret before jumping to unlock_drop ending up returning 0 after a failure. Fix it. Signed-off-by: Tejun Heo Acked-by: Li Zefan Cc: stable@vger.kernel.org --- kernel/cgroup.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel/cgroup.c') diff --git a/kernel/cgroup.c b/kernel/cgroup.c index aa95591c1430..793f37176077 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -1566,10 +1566,10 @@ static struct dentry *cgroup_mount(struct file_system_type *fs_type, mutex_lock(&cgroup_mutex); mutex_lock(&cgroup_root_mutex); - root_cgrp->id = idr_alloc(&root->cgroup_idr, root_cgrp, - 0, 1, GFP_KERNEL); - if (root_cgrp->id < 0) + ret = idr_alloc(&root->cgroup_idr, root_cgrp, 0, 1, GFP_KERNEL); + if (ret < 0) goto unlock_drop; + root_cgrp->id = ret; /* Check for name clashes with existing mounts */ ret = -EBUSY; -- cgit v1.2.3 From b58c89986a77a23658682a100eb15d8edb571ebb Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Sat, 8 Feb 2014 10:26:33 -0500 Subject: cgroup: fix error return from cgroup_create() cgroup_create() was returning 0 after allocation failures. Fix it. Signed-off-by: Tejun Heo Acked-by: Li Zefan Cc: stable@vger.kernel.org --- kernel/cgroup.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) (limited to 'kernel/cgroup.c') diff --git a/kernel/cgroup.c b/kernel/cgroup.c index 793f37176077..0eb7b868e1ab 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -4158,7 +4158,7 @@ static long cgroup_create(struct cgroup *parent, struct dentry *dentry, struct cgroup *cgrp; struct cgroup_name *name; struct cgroupfs_root *root = parent->root; - int ssid, err = 0; + int ssid, err; struct cgroup_subsys *ss; struct super_block *sb = root->sb; @@ -4168,8 +4168,10 @@ static long cgroup_create(struct cgroup *parent, struct dentry *dentry, return -ENOMEM; name = cgroup_alloc_name(dentry); - if (!name) + if (!name) { + err = -ENOMEM; goto err_free_cgrp; + } rcu_assign_pointer(cgrp->name, name); /* @@ -4177,8 +4179,10 @@ static long cgroup_create(struct cgroup *parent, struct dentry *dentry, * a half-baked cgroup. */ cgrp->id = idr_alloc(&root->cgroup_idr, NULL, 1, 0, GFP_KERNEL); - if (cgrp->id < 0) + if (cgrp->id < 0) { + err = -ENOMEM; goto err_free_name; + } /* * Only live parents can have children. Note that the liveliness -- cgit v1.2.3 From 48573a893303986e3b0b2974d6fb11f3d1bb7064 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Sat, 8 Feb 2014 10:26:34 -0500 Subject: cgroup: fix locking in cgroup_cfts_commit() cgroup_cfts_commit() walks the cgroup hierarchy that the target subsystem is attached to and tries to apply the file changes. Due to the convolution with inode locking, it can't keep cgroup_mutex locked while iterating. It currently holds only RCU read lock around the actual iteration and then pins the found cgroup using dget(). Unfortunately, this is incorrect. Although the iteration does check cgroup_is_dead() before invoking dget(), there's nothing which prevents the dentry from going away inbetween. Note that this is different from the usual css iterations where css_tryget() is used to pin the css - css_tryget() tests whether the css can be pinned and fails if not. The problem can be solved by simply holding cgroup_mutex instead of RCU read lock around the iteration, which actually reduces LOC. Signed-off-by: Tejun Heo Acked-by: Li Zefan Cc: stable@vger.kernel.org --- kernel/cgroup.c | 11 ++--------- 1 file changed, 2 insertions(+), 9 deletions(-) (limited to 'kernel/cgroup.c') diff --git a/kernel/cgroup.c b/kernel/cgroup.c index 0eb7b868e1ab..3edf7163b84f 100644 --- a/kernel/cgroup.c +++ b/kernel/cgroup.c @@ -2763,10 +2763,7 @@ static int cgroup_cfts_commit(struct cftype *cfts, bool is_add) */ update_before = cgroup_serial_nr_next; - mutex_unlock(&cgroup_mutex); - /* add/rm files for all cgroups created before */ - rcu_read_lock(); css_for_each_descendant_pre(css, cgroup_css(root, ss)) { struct cgroup *cgrp = css->cgroup; @@ -2775,23 +2772,19 @@ static int cgroup_cfts_commit(struct cftype *cfts, bool is_add) inode = cgrp->dentry->d_inode; dget(cgrp->dentry); - rcu_read_unlock(); - dput(prev); prev = cgrp->dentry; + mutex_unlock(&cgroup_mutex); mutex_lock(&inode->i_mutex); mutex_lock(&cgroup_mutex); if (cgrp->serial_nr < update_before && !cgroup_is_dead(cgrp)) ret = cgroup_addrm_files(cgrp, cfts, is_add); - mutex_unlock(&cgroup_mutex); mutex_unlock(&inode->i_mutex); - - rcu_read_lock(); if (ret) break; } - rcu_read_unlock(); + mutex_unlock(&cgroup_mutex); dput(prev); deactivate_super(sb); return ret; -- cgit v1.2.3