cgroup: use restart_syscall() for retries after offline waits in cgroup_subtree_contr...
authorTejun Heo <tj@kernel.org>
Tue, 13 May 2014 16:11:00 +0000 (12:11 -0400)
committerTejun Heo <tj@kernel.org>
Tue, 13 May 2014 16:11:00 +0000 (12:11 -0400)
After waiting for a child to finish offline,
cgroup_subtree_control_write() jumps up to retry from after the input
parsing and active protection breaking.  This retry makes the
scheduled locking update - removal of cgroup_tree_mutex - more
difficult.  Let's simplify it by returning with restart_syscall() for
retries.

Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Li Zefan <lizefan@huawei.com>
kernel/cgroup.c

index 250def0694b48cfa926b9c1d53a6455e3800e14d..3251cc9070fa33198b445f01f5f8f45d85fa651a 100644 (file)
@@ -2535,7 +2535,7 @@ out_finish:
 static int cgroup_subtree_control_write(struct cgroup_subsys_state *dummy_css,
                                        struct cftype *cft, char *buffer)
 {
-       unsigned int enable_req = 0, disable_req = 0, enable, disable;
+       unsigned int enable = 0, disable = 0;
        struct cgroup *cgrp = dummy_css->cgroup, *child;
        struct cgroup_subsys *ss;
        char *tok, *p;
@@ -2554,11 +2554,11 @@ static int cgroup_subtree_control_write(struct cgroup_subsys_state *dummy_css,
                                continue;
 
                        if (*tok == '+') {
-                               enable_req |= 1 << ssid;
-                               disable_req &= ~(1 << ssid);
+                               enable |= 1 << ssid;
+                               disable &= ~(1 << ssid);
                        } else if (*tok == '-') {
-                               disable_req |= 1 << ssid;
-                               enable_req &= ~(1 << ssid);
+                               disable |= 1 << ssid;
+                               enable &= ~(1 << ssid);
                        } else {
                                return -EINVAL;
                        }
@@ -2576,9 +2576,6 @@ static int cgroup_subtree_control_write(struct cgroup_subsys_state *dummy_css,
         */
        cgroup_get(cgrp);
        kernfs_break_active_protection(cgrp->control_kn);
-retry:
-       enable = enable_req;
-       disable = disable_req;
 
        mutex_lock(&cgroup_tree_mutex);
 
@@ -2608,7 +2605,9 @@ retry:
                                schedule();
                                finish_wait(&child->offline_waitq, &wait);
                                cgroup_put(child);
-                               goto retry;
+
+                               ret = restart_syscall();
+                               goto out_unbreak;
                        }
 
                        /* unavailable or not enabled on the parent? */
@@ -2692,6 +2691,7 @@ out_unlock:
        mutex_unlock(&cgroup_mutex);
 out_unlock_tree:
        mutex_unlock(&cgroup_tree_mutex);
+out_unbreak:
        kernfs_unbreak_active_protection(cgrp->control_kn);
        cgroup_put(cgrp);
        return ret;