[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20120513231609.3566.23600.stgit@srivatsabhat>
Date: Mon, 14 May 2012 04:46:23 +0530
From: "Srivatsa S. Bhat" <srivatsa.bhat@...ux.vnet.ibm.com>
To: a.p.zijlstra@...llo.nl, mingo@...nel.org, pjt@...gle.com,
paul@...lmenage.org, akpm@...ux-foundation.org
Cc: rjw@...k.pl, nacc@...ibm.com, paulmck@...ux.vnet.ibm.com,
tglx@...utronix.de, seto.hidetoshi@...fujitsu.com, tj@...nel.org,
mschmidt@...hat.com, berrange@...hat.com,
nikunj@...ux.vnet.ibm.com, vatsa@...ux.vnet.ibm.com,
liuj97@...il.com, linux-kernel@...r.kernel.org,
linux-pm@...r.kernel.org, srivatsa.bhat@...ux.vnet.ibm.com
Subject: [PATCH v3 3/5] cpusets: Update tasks' cpus_allowed mask upon updates
to root cpuset
The root cpuset's cpus_allowed mask can get updated during CPU hotplug.
However, during those updates, the tasks belonging to the root cpuset
aren't touched at all - their cpus_allowed masks aren't updated to reflect
the change in the configuration of the cpuset they belong to.
Fix this by moving the update of top_cpuset.cpus_allowed to
scan_cpusets_upon_hotplug() and ensure that the call to update_tasks_cpumask()
is not missed out for the root cpuset, including for cpu online events.
Signed-off-by: Srivatsa S. Bhat <srivatsa.bhat@...ux.vnet.ibm.com>
Cc: stable@...r.kernel.org
---
kernel/cpuset.c | 35 +++++++++++++++++++++++++++--------
1 files changed, 27 insertions(+), 8 deletions(-)
diff --git a/kernel/cpuset.c b/kernel/cpuset.c
index 87b0048..0fb9bff 100644
--- a/kernel/cpuset.c
+++ b/kernel/cpuset.c
@@ -150,6 +150,7 @@ typedef enum {
/* the type of hotplug event */
enum hotplug_event {
CPUSET_CPU_OFFLINE,
+ CPUSET_CPU_ONLINE,
CPUSET_MEM_OFFLINE,
};
@@ -2058,10 +2059,14 @@ scan_cpusets_upon_hotplug(struct cpuset *root, enum hotplug_event event)
if (cpumask_subset(cp->cpus_allowed, cpu_active_mask))
continue;
- /* Remove offline cpus from this cpuset. */
mutex_lock(&callback_mutex);
- cpumask_and(cp->cpus_allowed, cp->cpus_allowed,
- cpu_active_mask);
+ /* top_cpuset.cpus_allowed must track cpu_active_mask */
+ if (cp == &top_cpuset)
+ cpumask_copy(cp->cpus_allowed, cpu_active_mask);
+ else
+ /* Remove offline cpus from this cpuset. */
+ cpumask_and(cp->cpus_allowed, cp->cpus_allowed,
+ cpu_active_mask);
mutex_unlock(&callback_mutex);
/* Move tasks from the empty cpuset to a parent */
@@ -2072,6 +2077,20 @@ scan_cpusets_upon_hotplug(struct cpuset *root, enum hotplug_event event)
}
break;
+ case CPUSET_CPU_ONLINE:
+ /*
+ * Restore only the top_cpuset because it has to track
+ * cpu_active_mask always.
+ */
+ if (root == &top_cpuset) {
+ mutex_lock(&callback_mutex);
+ cpumask_copy(root->cpus_allowed, cpu_active_mask);
+ mutex_unlock(&callback_mutex);
+ update_tasks_cpumask(root, NULL);
+ }
+ list_del(queue.next);
+ break;
+
case CPUSET_MEM_OFFLINE:
while (!list_empty(&queue)) {
cp = traverse_cpusets(&queue);
@@ -2105,7 +2124,8 @@ scan_cpusets_upon_hotplug(struct cpuset *root, enum hotplug_event event)
* but making no active use of cpusets.
*
* This routine ensures that top_cpuset.cpus_allowed tracks
- * cpu_active_mask on each CPU hotplug (cpuhp) event.
+ * cpu_active_mask on each CPU hotplug (cpuhp) event. (This maintenance
+ * is actually implemented in scan_cpusets_upon_hotplug().)
*
* Called within get_online_cpus(). Needs to call cgroup_lock()
* before calling generate_sched_domains().
@@ -2120,11 +2140,10 @@ void cpuset_update_active_cpus(bool cpu_online)
int ndoms;
cgroup_lock();
- mutex_lock(&callback_mutex);
- cpumask_copy(top_cpuset.cpus_allowed, cpu_active_mask);
- mutex_unlock(&callback_mutex);
- if (!cpu_online)
+ if (cpu_online)
+ scan_cpusets_upon_hotplug(&top_cpuset, CPUSET_CPU_ONLINE);
+ else
scan_cpusets_upon_hotplug(&top_cpuset, CPUSET_CPU_OFFLINE);
ndoms = generate_sched_domains(&doms, &attr);
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists