Merge rsync://rsync.kernel.org/pub/scm/linux/kernel/git/paulus/ppc64-2.6

[linux-2.6-omap-h63xx.git] / kernel / cpuset.c

diff --git a/kernel/cpuset.c b/kernel/cpuset.c
index 69792bbe22817b6289fc20d1669e1402ab36395e..79dd929f4084395ff82dd6fa8f8065dcc111b16f 100644 (file)
--- a/kernel/cpuset.c
+++ b/kernel/cpuset.c
@@ -166,9 +166,8 @@ static struct super_block *cpuset_sb = NULL;
  * The hooks from fork and exit, cpuset_fork() and cpuset_exit(), don't
  * (usually) grab cpuset_sem.  These are the two most performance
  * critical pieces of code here.  The exception occurs on exit(),
- * if the last task using a cpuset exits, and the cpuset was marked
- * notify_on_release.  In that case, the cpuset_sem is taken, the
- * path to the released cpuset calculated, and a usermode call made
+ * when a task in a notify_on_release cpuset exits.  Then cpuset_sem
+ * is taken, and if the cpuset count is zero, a usermode call made
  * to /sbin/cpuset_release_agent with the name of the cpuset (path
  * relative to the root of cpuset file system) as the argument.
  *
@@ -229,13 +228,7 @@ static struct dentry_operations cpuset_dops = {
 
 static struct dentry *cpuset_get_dentry(struct dentry *parent, const char *name)
 {
-       struct qstr qstr;
-       struct dentry *d;
-
-       qstr.name = name;
-       qstr.len = strlen(name);
-       qstr.hash = full_name_hash(name, qstr.len);
-       d = lookup_hash(&qstr, parent);
+       struct dentry *d = lookup_one_len(name, parent, strlen(name));
        if (!IS_ERR(d))
                d->d_op = &cpuset_dops;
        return d;
@@ -1404,6 +1397,18 @@ void cpuset_fork(struct task_struct *tsk)
  *
  * Description: Detach cpuset from @tsk and release it.
  *
+ * Note that cpusets marked notify_on_release force every task
+ * in them to take the global cpuset_sem semaphore when exiting.
+ * This could impact scaling on very large systems.  Be reluctant
+ * to use notify_on_release cpusets where very high task exit
+ * scaling is required on large systems.
+ *
+ * Don't even think about derefencing 'cs' after the cpuset use
+ * count goes to zero, except inside a critical section guarded
+ * by the cpuset_sem semaphore.  If you don't hold cpuset_sem,
+ * then a zero cpuset use count is a license to any other task to
+ * nuke the cpuset immediately.
+ *
  **/
 
 void cpuset_exit(struct task_struct *tsk)
@@ -1415,10 +1420,13 @@ void cpuset_exit(struct task_struct *tsk)
        tsk->cpuset = NULL;
        task_unlock(tsk);
 
-       if (atomic_dec_and_test(&cs->count)) {
+       if (notify_on_release(cs)) {
                down(&cpuset_sem);
-               check_for_release(cs);
+               if (atomic_dec_and_test(&cs->count))
+                       check_for_release(cs);
                up(&cpuset_sem);
+       } else {
+               atomic_dec(&cs->count);
        }
 }
 
@@ -1432,7 +1440,7 @@ void cpuset_exit(struct task_struct *tsk)
  * tasks cpuset.
  **/
 
-const cpumask_t cpuset_cpus_allowed(const struct task_struct *tsk)
+cpumask_t cpuset_cpus_allowed(const struct task_struct *tsk)
 {
        cpumask_t mask;