Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753025Ab2KTIdU (ORCPT ); Tue, 20 Nov 2012 03:33:20 -0500 Received: from mailhub.sw.ru ([195.214.232.25]:28663 "EHLO relay.sw.ru" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752756Ab2KTIcq (ORCPT ); Tue, 20 Nov 2012 03:32:46 -0500 From: Glauber Costa To: Cc: , Peter Zijlstra , Paul Turner , Balbir Singh , Tejun Heo , , Peter Zijlstra , Mel Gorman Subject: [PATCH 0/6] Automatic NUMA placement of tasks in cpu cgroup Date: Tue, 20 Nov 2012 12:31:58 +0400 Message-Id: <1353400324-10897-1-git-send-email-glommer@parallels.com> X-Mailer: git-send-email 1.7.11.7 CC: Tejun Heo Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 2110 Lines: 52 Hi, This patchset has absolutely nothing to do with NUMA. But now that I got your attention: This is an attempt to ressurect a patchset that Tejun Heo sent a while ago, aiming at deprecation of cpuacct. He only went as far as publishing the files in the cpu cgroup, but the final work would require us to take advantage of it by not incurring in hierarchy walks more times than necessary. It is trivial to do it in the case where we have SCHEDSTATS enabled: we already record a statistic that is exactly the same as cpuusage: exec_clock. That is not collected for rt tasks, so the only thing we need to do is to also collect it for them, and print them back for cpuusage. In theory, it would also be possible to avoid hierarchy walks even without SCHEDSTATS: we could modify task_group_charge() to stop walking, and then update cpuusage only for the current group, during the walk we already do. I didn't do so, because I believe we care more about setups that would enable a bunch of options anyway - which is likely to include SCHEDSTATS. Custom setups can take a much easier route and just compile out the whole thing! But let me know if I should do it. Glauber Costa (3): don't call cpuacct_charge in stop_task.c sched: adjust exec_clock to use it as cpu usage metric cpuacct: don't actually do anything. Tejun Heo (3): cgroup: implement CFTYPE_NO_PREFIX cgroup, sched: let cpu serve the same files as cpuacct cgroup, sched: deprecate cpuacct include/linux/cgroup.h | 1 + init/Kconfig | 11 +- kernel/cgroup.c | 57 ++++++++++- kernel/sched/core.c | 262 ++++++++++++++++++++++++++++++++++++++++++++++- kernel/sched/fair.c | 1 + kernel/sched/rt.c | 2 + kernel/sched/sched.h | 14 ++- kernel/sched/stop_task.c | 1 - 8 files changed, 341 insertions(+), 8 deletions(-) -- 1.7.11.7 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/