From: Gautham R Shenoy <ego@in.ibm.com>
To: Gregory Haskins <ghaskins@novell.com>
Cc: suresh.b.siddha@intel.com, rjw@sisk.pl,
akpm@linux-foundation.org, dmitry.adamushko@gmail.com,
mingo@elte.hu, oleg@sign.ru, yi.y.yang@intel.com,
linux-kernel@vger.kernel.org, tglx@linutronix.de,
Srivatsa Vaddagiri <vatsa@in.ibm.com>
Subject: [PATCH] cpu-hotplug: Register update_sched_domains() notifier with higher prio
Date: Mon, 10 Mar 2008 18:43:55 +0530 [thread overview]
Message-ID: <20080310131355.GF17646@in.ibm.com> (raw)
In-Reply-To: <20080310081425.GA11031@in.ibm.com>
cpu-hotplug: Register update_sched_domains() notifier with higher prio
From: Gautham R Shenoy <ego@in.ibm.com>
The current -rt wake_up logic uses the rq->rd->online_map which is
updated on cpu-hotplug events by update_sched_domains(). Currently
update_sched_domains() is registered with a priority 0. It is
preferable that update_sched_domains() be the first notifier called on
a CPU_DEAD or CPU_ONLINE events inorder to ensure that the
rq->rd->online_map is in sync with the cpu_online_map.
This way on a CPU_DOWN_PREPARE,
we would destroy the sched_domains and reattach all the rq's to the
def_root_domain. This will be followed by a CPU_DOWN_PREPARE in
migration_call() which will clear the bit of the cpu going offline
from the corresponding rq.
This will ensure that no task will be woken up on the cpu which is
going to be offlined between CPU_DOWN_PREPARE and CPU_DEAD.
Ditto for the CPU_ONLINE path.
Tested on a 4-way Intel Xeon Box with cpu-hotplug tests running in parallel
with kernbench.
Signed-off-by: Gautham R Shenoy <ego@in.ibm.com>
---
kernel/sched.c | 12 ++++++++++--
1 files changed, 10 insertions(+), 2 deletions(-)
diff --git a/kernel/sched.c b/kernel/sched.c
index 69ecb1a..374c46f 100644
--- a/kernel/sched.c
+++ b/kernel/sched.c
@@ -7079,8 +7079,16 @@ void __init sched_init_smp(void)
if (cpus_empty(non_isolated_cpus))
cpu_set(smp_processor_id(), non_isolated_cpus);
put_online_cpus();
- /* XXX: Theoretical race here - CPU may be hotplugged now */
- hotcpu_notifier(update_sched_domains, 0);
+ /*
+ * XXX: Theoretical race here - CPU may be hotplugged now
+ *
+ * We register the notifier with priority 11, which means that
+ * update_sched_domains() will be called just before migration_call().
+ *
+ * This is necessary to ensure that the rt wake up logic works fine
+ * and the rq->rd->online_map remains in sync with the cpu_online_map.
+ */
+ hotcpu_notifier(update_sched_domains, 11);
/* Move init over to a non-isolated CPU */
if (set_cpus_allowed(current, non_isolated_cpus) < 0)
--
Thanks and Regards
gautham
next prev parent reply other threads:[~2008-03-10 13:14 UTC|newest]
Thread overview: 54+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-03-02 18:42 [BUG 2.6.25-rc3] scheduler/hotplug: some processes are dealocked when cpu is set to offline Yi Yang
2008-03-03 11:54 ` Dmitry Adamushko
2008-03-03 11:56 ` Ingo Molnar
2008-03-03 12:02 ` Dmitry Adamushko
2008-03-03 14:53 ` Yi Yang
2008-03-03 17:37 ` Yi Yang
2008-03-03 15:31 ` Gautham R Shenoy
2008-03-03 14:45 ` Yi Yang
2008-03-04 5:26 ` Gautham R Shenoy
2008-03-04 9:09 ` Gautham R Shenoy
2008-03-03 21:56 ` Yi Yang
2008-03-04 15:01 ` Oleg Nesterov
2008-03-04 14:37 ` Yi Yang
2008-03-06 20:05 ` Yi Yang
2008-03-05 10:05 ` Gautham R Shenoy
2008-03-05 13:53 ` Oleg Nesterov
2008-03-06 11:15 ` Gautham R Shenoy
2008-03-06 12:22 ` Gautham R Shenoy
2008-03-06 13:44 ` Gautham R Shenoy
2008-03-07 2:54 ` Oleg Nesterov
2008-03-07 9:10 ` Gautham R Shenoy
2008-03-07 10:51 ` Gautham R Shenoy
2008-03-06 23:20 ` Yi Yang
2008-03-07 13:02 ` Dmitry Adamushko
2008-03-07 13:55 ` Gautham R Shenoy
2008-03-07 15:50 ` Gautham R Shenoy
2008-03-07 19:14 ` [BUG 2.6.25-rc3] scheduler/hotplug: some processes aredealocked " Suresh Siddha
2008-03-07 20:18 ` [BUG 2.6.25-rc3] scheduler/hotplug: some processes are dealocked " Andrew Morton
2008-03-07 21:36 ` Rafael J. Wysocki
2008-03-07 23:01 ` Suresh Siddha
2008-03-07 23:29 ` Andrew Morton
2008-03-07 23:43 ` Rafael J. Wysocki
2008-03-08 1:50 ` Suresh Siddha
2008-03-08 2:09 ` Andrew Morton
2008-03-08 5:10 ` [PATCH] adjust root-domain->online span in response to hotplug event Gregory Haskins
2008-03-08 8:41 ` Ingo Molnar
2008-03-08 17:50 ` [PATCH] adjust root-domain->online span in response to hotplugevent Gregory Haskins
2008-03-09 0:31 ` Dmitry Adamushko
2008-03-10 14:12 ` Gregory Haskins
2008-03-09 2:35 ` [PATCH] adjust root-domain->online span in response to hotplug event Suresh Siddha
2008-03-10 12:41 ` Gregory Haskins
2008-03-10 8:14 ` Gautham R Shenoy
2008-03-10 13:13 ` Gautham R Shenoy [this message]
2008-03-10 22:25 ` [PATCH] cpu-hotplug: Register update_sched_domains() notifier with higher prio Andrew Morton
2008-03-10 13:39 ` [PATCH] keep rd->online and cpu_online_map in sync Gregory Haskins
2008-03-10 14:21 ` Gautham R Shenoy
2008-03-10 18:12 ` Suresh Siddha
2008-03-10 22:03 ` Rafael J. Wysocki
2008-03-10 22:00 ` Gregory Haskins
2008-03-10 22:10 ` Suresh Siddha
2008-03-10 21:59 ` [PATCH v2] " Gregory Haskins
2008-03-10 23:36 ` Andrew Morton
2008-03-11 1:34 ` Suresh Siddha
2008-03-11 4:39 ` Gautham R Shenoy
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20080310131355.GF17646@in.ibm.com \
--to=ego@in.ibm.com \
--cc=akpm@linux-foundation.org \
--cc=dmitry.adamushko@gmail.com \
--cc=ghaskins@novell.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@elte.hu \
--cc=oleg@sign.ru \
--cc=rjw@sisk.pl \
--cc=suresh.b.siddha@intel.com \
--cc=tglx@linutronix.de \
--cc=vatsa@in.ibm.com \
--cc=yi.y.yang@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.