a fixlet for cgroups-make-cgroup_path-rcu-safe.patch
PATCH 1/1 a fixlet for cgroups-make-cgroup_path-rcu-safe.patch
With cgroups-make-cgroup_path-rcu-safe.patch applied, I can trigger kernel panic easily using this test program:
for ((; ;))
{
mount -t cgroup -o debug xxx /mnt
mkdir /mnt/0
rmdir /mnt/0
umount /mnt
}It's caused by the deactive_super() in rcu callback.
This patch restores the original code that call deactive_super() in cgroup_diput().
Also remove duplicated comment in cgroup.h added by the patch.
Signed-off-by Li Zefan <[email protected]>
--- include/linux/cgroup.h | 6 ------ kernel/cgroup.c | 11 ++++++----- 2 files changed, 6 insertions(+), 11 deletions(-)
include/linux/cgroup.h renamed from include/linux/cgroup.h.orig +0 -6
@@ -334,12 +334,6 @@ int cgroup_add_files(struct cgroup *cgrp,int cgroup_is_removed (const struct cgroup *cgrp); -/* -* cgroup_path() fills in a filesystem-like path for the given cgroup -* into "buf", up to "buflen" characters. Should be called with -* cgroup_mutex held, or else in an RCU section with an RCU-protected -* cgroup reference -*/ int cgroup_path (const struct cgroup *cgrp,char *buf,int buflen);int cgroup_task_count (const struct cgroup *cgrp);
kernel/cgroup.c renamed from kernel/cgroup.c.orig +6 -5
@@ -598,11 +598,6 @@ static void cgroup_call_pre_destroy(struct cgroup *cgrp)static void free_cgroup_rcu (struct rcu_head *obj) { struct cgroup *cgrp = container_of(obj, struct cgroup, rcu_head); -/* -* Drop the active superblock reference that we took when we -* created the cgroup -*/ -deactivate_super (cgrp->root->sb);kfree (cgrp); }
@@ -632,6 +627,12 @@ static void cgroup_diput(struct dentry *dentry, struct inode *inode) cgrp->root->number_of_cgroups--;mutex_unlock (&cgroup_mutex); +/* +* Drop the active superblock reference that we took when we +* created the cgroup +*/ +deactivate_super (cgrp->root->sb); +call_rcu (&cgrp->rcu_head, free_cgroup_rcu); }iput (inode);