From patchwork Thu Aug 18 22:36:42 2016 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: srinivas pandruvada X-Patchwork-Id: 9288841 Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork.web.codeaurora.org (Postfix) with ESMTP id 79C30600CB for ; Fri, 19 Aug 2016 01:10:23 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 6923A29197 for ; Fri, 19 Aug 2016 01:10:23 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id 5D64E2925D; Fri, 19 Aug 2016 01:10:23 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-6.9 required=2.0 tests=BAYES_00,RCVD_IN_DNSWL_HI autolearn=unavailable version=3.3.1 Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id D787629197 for ; Fri, 19 Aug 2016 01:10:22 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753696AbcHSBKH (ORCPT ); Thu, 18 Aug 2016 21:10:07 -0400 Received: from mga04.intel.com ([192.55.52.120]:18890 "EHLO mga04.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754074AbcHSBJl (ORCPT ); Thu, 18 Aug 2016 21:09:41 -0400 Received: from orsmga002.jf.intel.com ([10.7.209.21]) by fmsmga104.fm.intel.com with ESMTP; 18 Aug 2016 15:39:03 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.28,542,1464678000"; d="scan'208";a="1038140024" Received: from spandruv-mobl3.jf.intel.com ([10.254.32.30]) by orsmga002.jf.intel.com with ESMTP; 18 Aug 2016 15:39:01 -0700 From: Srinivas Pandruvada To: mingo@redhat.com, tglx@linutronix.de, hpa@zytor.com, rjw@rjwysocki.net, peterz@infradead.org Cc: x86@kernel.org, bp@suse.de, sudeep.holla@arm.com, ak@linux.intel.com, linux-acpi@vger.kernel.org, linux-pm@vger.kernel.org, alexey.klimov@arm.com, viresh.kumar@linaro.org, akpm@linux-foundation.org, linux-kernel@vger.kernel.org, lenb@kernel.org, tim.c.chen@linux.intel.com, srinivas.pandruvada@linux.intel.com, paul.gortmaker@windriver.com, jpoimboe@redhat.com, mcgrof@kernel.org, jgross@suse.com, robert.moore@intel.com, dvyukov@google.com, jeyu@redhat.com Subject: [PATCH 01/11] sched, cpuset: Add regenerate_sched_domains function to rebuild all sched domains Date: Thu, 18 Aug 2016 15:36:42 -0700 Message-Id: <1471559812-19967-2-git-send-email-srinivas.pandruvada@linux.intel.com> X-Mailer: git-send-email 2.5.5 In-Reply-To: <1471559812-19967-1-git-send-email-srinivas.pandruvada@linux.intel.com> References: <1471559812-19967-1-git-send-email-srinivas.pandruvada@linux.intel.com> Sender: linux-pm-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-pm@vger.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP From: Tim Chen The current rebuild_sched_domains will only rebuild the sched domains unless the cpumask changes. However, in some scenarios when the topology flag value changes, it will not rebuild the sched domain. We create a regenerate_sched_domains function that will always rebuild all the sched domains to take care of this scenario. Signed-off-by: Tim Chen Signed-off-by: Srinivas Pandruvada --- include/linux/cpuset.h | 2 ++ include/linux/sched.h | 3 +++ kernel/cpuset.c | 32 +++++++++++++++++++++++++------- kernel/sched/core.c | 25 ++++++++++++++++++++++--- 4 files changed, 52 insertions(+), 10 deletions(-) diff --git a/include/linux/cpuset.h b/include/linux/cpuset.h index bfc204e..9f948fa 100644 --- a/include/linux/cpuset.h +++ b/include/linux/cpuset.h @@ -102,6 +102,8 @@ extern int current_cpuset_is_being_rebound(void); extern void rebuild_sched_domains(void); +extern void regenerate_sched_domains(void); + extern void cpuset_print_current_mems_allowed(void); /* diff --git a/include/linux/sched.h b/include/linux/sched.h index 62c68e5..3301959 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1152,6 +1152,9 @@ static inline struct cpumask *sched_domain_span(struct sched_domain *sd) extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[], struct sched_domain_attr *dattr_new); +extern void regen_partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[], + struct sched_domain_attr *dattr_new); + /* Allocate an array of sched domains, for partition_sched_domains(). */ cpumask_var_t *alloc_sched_domains(unsigned int ndoms); void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms); diff --git a/kernel/cpuset.c b/kernel/cpuset.c index c7fd277..f6f7c17 100644 --- a/kernel/cpuset.c +++ b/kernel/cpuset.c @@ -794,10 +794,12 @@ done: * which has that flag enabled, or if any cpuset with a non-empty * 'cpus' is removed, then call this routine to rebuild the * scheduler's dynamic sched domains. + * If forced flag is set, then we will always regenerate all new + * sched domains. * * Call with cpuset_mutex held. Takes get_online_cpus(). */ -static void rebuild_sched_domains_locked(void) +static void rebuild_sched_domains_locked(bool rebuild_all) { struct sched_domain_attr *attr; cpumask_var_t *doms; @@ -818,12 +820,17 @@ static void rebuild_sched_domains_locked(void) ndoms = generate_sched_domains(&doms, &attr); /* Have scheduler rebuild the domains */ - partition_sched_domains(ndoms, doms, attr); + if (rebuild_all) + /* Will rebuild a complete set of all sched domains */ + regen_partition_sched_domains(ndoms, doms, attr); + else + /* Rebuild only sched domains with changed cpu masks */ + partition_sched_domains(ndoms, doms, attr); out: put_online_cpus(); } #else /* !CONFIG_SMP */ -static void rebuild_sched_domains_locked(void) +static void rebuild_sched_domains_locked(bool forced) { } #endif /* CONFIG_SMP */ @@ -831,7 +838,18 @@ static void rebuild_sched_domains_locked(void) void rebuild_sched_domains(void) { mutex_lock(&cpuset_mutex); - rebuild_sched_domains_locked(); + rebuild_sched_domains_locked(false); + mutex_unlock(&cpuset_mutex); +} + +/* + * Similar to rebuild_sched domains, but will force + * all sched domains to be always rebuilt. + */ +void regenerate_sched_domains(void) +{ + mutex_lock(&cpuset_mutex); + rebuild_sched_domains_locked(true); mutex_unlock(&cpuset_mutex); } @@ -919,7 +937,7 @@ static void update_cpumasks_hier(struct cpuset *cs, struct cpumask *new_cpus) rcu_read_unlock(); if (need_rebuild_sched_domains) - rebuild_sched_domains_locked(); + rebuild_sched_domains_locked(false); } /** @@ -1267,7 +1285,7 @@ static int update_relax_domain_level(struct cpuset *cs, s64 val) cs->relax_domain_level = val; if (!cpumask_empty(cs->cpus_allowed) && is_sched_load_balance(cs)) - rebuild_sched_domains_locked(); + rebuild_sched_domains_locked(true); } return 0; @@ -1333,7 +1351,7 @@ static int update_flag(cpuset_flagbits_t bit, struct cpuset *cs, spin_unlock_irq(&callback_lock); if (!cpumask_empty(trialcs->cpus_allowed) && balance_flag_changed) - rebuild_sched_domains_locked(); + rebuild_sched_domains_locked(false); if (spread_flag_changed) update_tasks_flags(cs); diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 2a906f2..ec752da 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -7069,8 +7069,9 @@ static int dattrs_equal(struct sched_domain_attr *cur, int idx_cur, * * Call with hotplug lock held */ -void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[], - struct sched_domain_attr *dattr_new) +static void __partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[], + struct sched_domain_attr *dattr_new, + int need_domain_rebuild) { int i, j, n; int new_topology; @@ -7081,7 +7082,7 @@ void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[], unregister_sched_domain_sysctl(); /* Let architecture update cpu core mappings. */ - new_topology = arch_update_cpu_topology(); + new_topology = arch_update_cpu_topology() | need_domain_rebuild; n = doms_new ? ndoms_new : 0; @@ -7132,6 +7133,24 @@ match2: mutex_unlock(&sched_domains_mutex); } +/* + * Generate sched domains only when the cpumask or domain attr changes + */ +void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[], + struct sched_domain_attr *dattr_new) +{ + __partition_sched_domains(ndoms_new, doms_new, dattr_new, 0); +} + +/* + * Generate new sched domains always + */ +void regen_partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[], + struct sched_domain_attr *dattr_new) +{ + __partition_sched_domains(ndoms_new, doms_new, dattr_new, 1); +} + static int num_cpus_frozen; /* used to mark begin/end of suspend/resume */ /*