Hi Paul,
> Query for Cliff:
> 1) Can we narrow the scope of callback_mutex in scan_for_empty_cpusets()?
> 2) Can we avoid rewriting the cpus, mems of cpusets except when it is
> likely that we'll be changing them?
> 3) Should not remove_tasks_in_empty_cpuset() also check for empty mems?
> -pj
I agree with all of the above refinements.
And I just tested the below patch and find no problem.
So this is an ACK from me.
-Cliff
---
kernel/cpuset.c | 21 +++++++++++++--------
1 file changed, 13 insertions(+), 8 deletions(-)
Index: linux-2.6/kernel/cpuset.c
===================================================================
--- linux-2.6.orig/kernel/cpuset.c
+++ linux-2.6/kernel/cpuset.c
@@ -1748,7 +1748,8 @@ static void remove_tasks_in_empty_cpuset
* has online cpus, so can't be empty).
*/
parent = cs->parent;
- while (cpus_empty(parent->cpus_allowed))
+ while (cpus_empty(parent->cpus_allowed) ||
+ nodes_empty(parent->mems_allowed))
parent = parent->parent;
move_member_tasks_to_cpuset(cs, parent);
@@ -1780,7 +1781,6 @@ static void scan_for_empty_cpusets(const
list_add_tail((struct list_head *)&root->stack_list, &queue);
- mutex_lock(&callback_mutex);
while (!list_empty(&queue)) {
cp = container_of(queue.next, struct cpuset, stack_list);
list_del(queue.next);
@@ -1789,19 +1789,24 @@ static void scan_for_empty_cpusets(const
list_add_tail(&child->stack_list, &queue);
}
cont = cp->css.cgroup;
+
+ /* Continue past cpusets with all cpus, mems online */
+ if (cpus_subset(cp->cpus_allowed, cpu_online_map) &&
+ nodes_subset(cp->mems_allowed, node_states[N_HIGH_MEMORY]))
+ continue;
+
/* Remove offline cpus and mems from this cpuset. */
+ mutex_lock(&callback_mutex);
cpus_and(cp->cpus_allowed, cp->cpus_allowed, cpu_online_map);
nodes_and(cp->mems_allowed, cp->mems_allowed,
node_states[N_HIGH_MEMORY]);
+ mutex_unlock(&callback_mutex);
+
+ /* Move tasks from the empty cpuset to a parent */
if (cpus_empty(cp->cpus_allowed) ||
- nodes_empty(cp->mems_allowed)) {
- /* Move tasks from the empty cpuset to a parent */
- mutex_unlock(&callback_mutex);
+ nodes_empty(cp->mems_allowed))
remove_tasks_in_empty_cpuset(cp);
- mutex_lock(&callback_mutex);
- }
}
- mutex_unlock(&callback_mutex);
}
/*
Cliff wrote:
> I agree with all of the above refinements.
> And I just tested the below patch and find no problem.
>
> So this is an ACK from me.
I haven't looked, but I doubt this ended up in Andrew's tree.
Note that the Subject above says "RFC", not "PATCH".
Cliff -- could you see what of this, or the other two patches
that I sent in review of your work, made it into Andrew's tree:
[PATCH 1/3] hotplug cpu move tasks in empty cpusets to parent node_online_map fix
[PATCH 2/3] hotplug cpu move tasks in empty cpusets to parent various other fixes
--
I won't rest till it's the best ...
Programmer, Linux Scalability
Paul Jackson <[email protected]> 1.940.382.4214