From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S934486AbbEMUf3 (ORCPT ); Wed, 13 May 2015 16:35:29 -0400 Received: from mail-qk0-f169.google.com ([209.85.220.169]:34522 "EHLO mail-qk0-f169.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932417AbbEMUf1 (ORCPT ); Wed, 13 May 2015 16:35:27 -0400 From: Tejun Heo To: lizefan@huawei.com Cc: cgroups@vger.kernel.org, mingo@redhat.com, peterz@infradead.org, linux-kernel@vger.kernel.org Subject: [PATCHSET] cgroup, sched: restructure threadgroup locking and replace it with a percpu_rwsem Date: Wed, 13 May 2015 16:35:15 -0400 Message-Id: <1431549318-16756-1-git-send-email-tj@kernel.org> X-Mailer: git-send-email 2.1.0 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org threadgroup locking was added because cgroup needs threadgroups to stay stable across attach operations. It was implemented as a per-signal_struct generic locking mechanism so that other users which require threadgroups stable across blocking operations can use it too; however, it hasn't grown any other use cases and still conditionalized on CONFIG_CGROUPS. It turns out that a single percpu_rwsem would serve cgroup's use case better as it's lighter on the thread operations and allows atomic operations across multiple processes. We can change the generic threadgroup lock mechanism to do that but it's pointless to put this in core task code. What's necessary from the task code is a place which subsystems needing synchronization against threadgroup operations can hook into so that they can implement the necessary exclusion. This patchset moves the specific locking details into cgroup proper leaving threadgroup_changes_begin/end() as the section markers and then converts the cgroup specific locking to use a percp_rwsem. This patchset contains the following three patches. 0001-sched-cgroup-reorganize-threadgroup-locking.patch 0002-sched-cgroup-replace-signal_struct-group_rwsem-with-.patch 0003-cgroup-simplify-threadgroup-locking.patch 0001 moves threadgroup locking details into cgroup proper. 0002-0003 convert per-signal_struct group_rwsem with a global percpu_rwsem. This patchset is on top of cgroup/for-4.2 d0f702e648dc ("cgroup: fix some comment typos") + [1] [PATCH] cgroup: separate out include/linux/cgroup-defs.h + [2] [PATCH] cgroup: reorganize include/linux/cgroup.h git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup.git review-threadgroup-percpu-rwsem diffstat follows. Thanks. include/linux/cgroup-defs.h | 33 ++++++++++++++++++++ include/linux/init_task.h | 8 ---- include/linux/sched.h | 65 ++++++++++------------------------------ init/Kconfig | 1 kernel/cgroup.c | 71 ++++++++++++++++---------------------------- kernel/fork.c | 4 -- 6 files changed, 78 insertions(+), 104 deletions(-) -- tejun [1] http://lkml.kernel.org/g/20150513193840.GC11388@htj.duckdns.org [2] http://lkml.kernel.org/g/20150513202416.GE11388@htj.duckdns.org From mboxrd@z Thu Jan 1 00:00:00 1970 From: Tejun Heo Subject: [PATCHSET] cgroup, sched: restructure threadgroup locking and replace it with a percpu_rwsem Date: Wed, 13 May 2015 16:35:15 -0400 Message-ID: <1431549318-16756-1-git-send-email-tj@kernel.org> Return-path: DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=sender:from:to:cc:subject:date:message-id; bh=kCEk7bvEo8sM7U5l8TfVQbw2cK2RfNJdg2lUWz/StK0=; b=Ns4aZKl3Hd6l73Ju5miU+X0XmmrTZvpJpihiFPBkgNFGkrfm3Dee2HHtCBcsSzG83z iUuM1KzQAP6Lf/3ZYgmm6if/H4/J7MJ6TlaVTUjsh+tfqgUUO12CY7mZRFrH+v0pwQ6/ 9wCPhx6ZvEnTJIVOxA/6WIahgjADGzCQEasb26xrUqXfL8m61sBPIu6TGsV2PZdmM+oE X0tGIpcb8mzZWROgBl32k3NrCqFJgpGvsfhZ5mquUs92+tpZg1NNcq3ZXWQj1g8I2q5T u7gt9xk6LPdTLj5sl32kAAvcqeTTN8RriazDpzaDc8ZOTzS8eM7u1ENyxRbJtcJMAeYP 9cEA== Sender: cgroups-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: lizefan-hv44wF8Li93QT0dZR+AlfA@public.gmane.org Cc: cgroups-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, mingo-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org, peterz-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org, linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org threadgroup locking was added because cgroup needs threadgroups to stay stable across attach operations. It was implemented as a per-signal_struct generic locking mechanism so that other users which require threadgroups stable across blocking operations can use it too; however, it hasn't grown any other use cases and still conditionalized on CONFIG_CGROUPS. It turns out that a single percpu_rwsem would serve cgroup's use case better as it's lighter on the thread operations and allows atomic operations across multiple processes. We can change the generic threadgroup lock mechanism to do that but it's pointless to put this in core task code. What's necessary from the task code is a place which subsystems needing synchronization against threadgroup operations can hook into so that they can implement the necessary exclusion. This patchset moves the specific locking details into cgroup proper leaving threadgroup_changes_begin/end() as the section markers and then converts the cgroup specific locking to use a percp_rwsem. This patchset contains the following three patches. 0001-sched-cgroup-reorganize-threadgroup-locking.patch 0002-sched-cgroup-replace-signal_struct-group_rwsem-with-.patch 0003-cgroup-simplify-threadgroup-locking.patch 0001 moves threadgroup locking details into cgroup proper. 0002-0003 convert per-signal_struct group_rwsem with a global percpu_rwsem. This patchset is on top of cgroup/for-4.2 d0f702e648dc ("cgroup: fix some comment typos") + [1] [PATCH] cgroup: separate out include/linux/cgroup-defs.h + [2] [PATCH] cgroup: reorganize include/linux/cgroup.h git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup.git review-threadgroup-percpu-rwsem diffstat follows. Thanks. include/linux/cgroup-defs.h | 33 ++++++++++++++++++++ include/linux/init_task.h | 8 ---- include/linux/sched.h | 65 ++++++++++------------------------------ init/Kconfig | 1 kernel/cgroup.c | 71 ++++++++++++++++---------------------------- kernel/fork.c | 4 -- 6 files changed, 78 insertions(+), 104 deletions(-) -- tejun [1] http://lkml.kernel.org/g/20150513193840.GC11388-piEFEHQLUPpN0TnZuCh8vA@public.gmane.org [2] http://lkml.kernel.org/g/20150513202416.GE11388-piEFEHQLUPpN0TnZuCh8vA@public.gmane.org