From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751641AbcGNSbj (ORCPT ); Thu, 14 Jul 2016 14:31:39 -0400 Received: from bombadil.infradead.org ([198.137.202.9]:43600 "EHLO bombadil.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751175AbcGNSbd (ORCPT ); Thu, 14 Jul 2016 14:31:33 -0400 Message-Id: <20160714182545.786693675@infradead.org> User-Agent: quilt/0.63-1 Date: Thu, 14 Jul 2016 20:25:45 +0200 From: Peter Zijlstra To: mingo@kernel.org, linux-kernel@vger.kernel.org Cc: oleg@redhat.com, tj@kernel.org, paulmck@linux.vnet.ibm.com, john.stultz@linaro.org, dimitrysh@google.com, romlem@google.com, ccross@google.com, tkjos@google.com, peterz@infradead.org Subject: [PATCH 0/2] locking/percpu-rwsem: Optimizations/tweaks Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org As reported by John and Dmitry, Android tickles bad behaviour in cgroup_threadgroup_rwsem. These two patches attempt to address the issue by optimizing the percpu-rwsem slow path and second add a bias knob to the percpu-rwsem that allows disabling the fast path which improves writer latency by getting rid of a synchronize_sched() call. --- fs/super.c | 3 +- include/linux/percpu-rwsem.h | 96 +++++++++++++++-- kernel/cgroup.c | 2 +- kernel/locking/percpu-rwsem.c | 246 +++++++++++++++++++++++++----------------- 4 files changed, 239 insertions(+), 108 deletions(-)