From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755964AbZDTN5d (ORCPT ); Mon, 20 Apr 2009 09:57:33 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1755634AbZDTN5X (ORCPT ); Mon, 20 Apr 2009 09:57:23 -0400 Received: from lutt.itea.ntnu.no ([129.241.18.234]:58847 "EHLO lutt.itea.ntnu.no" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755607AbZDTN5V (ORCPT ); Mon, 20 Apr 2009 09:57:21 -0400 From: Henrik Austad To: Ingo Molnar Cc: LKML , Peter Ziljstra , Henrik Austad Subject: [PATCH 2/2] sched: Let sched_class handle effective_prio calculation Date: Mon, 20 Apr 2009 15:59:01 +0200 Message-Id: <1240235941-1273-3-git-send-email-henrik@austad.us> X-Mailer: git-send-email 1.6.1.2 In-Reply-To: <1240235941-1273-2-git-send-email-henrik@austad.us> References: <1240235941-1273-1-git-send-email-henrik@austad.us> <1240235941-1273-2-git-send-email-henrik@austad.us> Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Instead of testing for sched_class in the core scheduler, the calculation of the class-specific effective priority should be handled by the scheduling class. Signed-off-by: Henrik Austad Acked-by: Peter Zijlstra --- include/linux/sched.h | 1 + kernel/sched.c | 11 +++-------- kernel/sched_fair.c | 1 + kernel/sched_idletask.c | 1 + kernel/sched_rt.c | 6 ++++++ 5 files changed, 12 insertions(+), 8 deletions(-) diff --git a/include/linux/sched.h b/include/linux/sched.h index f16db27..f5e3ee9 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1019,6 +1019,7 @@ struct sched_class { void (*prio_changed) (struct rq *this_rq, struct task_struct *task, int oldprio, int running); int (*normal_prio) (struct task_struct *task); + int (*effective_prio) (struct task_struct *task); #ifdef CONFIG_FAIR_GROUP_SCHED void (*moved_group) (struct task_struct *p); diff --git a/kernel/sched.c b/kernel/sched.c index d78c84e..d388641 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -1826,15 +1826,10 @@ static inline int normal_prio(struct task_struct *p) */ static int effective_prio(struct task_struct *p) { + /* update normal prio */ p->normal_prio = normal_prio(p); - /* - * If we are RT tasks or we were boosted to RT priority, - * keep the priority unchanged. Otherwise, update priority - * to the normal priority: - */ - if (!rt_prio(p->prio)) - return p->normal_prio; - return p->prio; + + return p->sched_class->effective_prio(p); } /* diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index c94519a..42af406 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -1822,6 +1822,7 @@ static const struct sched_class fair_sched_class = { .prio_changed = prio_changed_fair, .normal_prio = normal_prio_fair, + .effective_prio = normal_prio_fair, .switched_to = switched_to_fair, #ifdef CONFIG_FAIR_GROUP_SCHED diff --git a/kernel/sched_idletask.c b/kernel/sched_idletask.c index e52ae7c..675de4e 100644 --- a/kernel/sched_idletask.c +++ b/kernel/sched_idletask.c @@ -128,6 +128,7 @@ static const struct sched_class idle_sched_class = { .prio_changed = prio_changed_idle, .normal_prio = normal_prio_idle, + .effective_prio = normal_prio_idle, .switched_to = switched_to_idle, /* no .task_new for idle tasks */ diff --git a/kernel/sched_rt.c b/kernel/sched_rt.c index 5c9a8f1..87390a9 100644 --- a/kernel/sched_rt.c +++ b/kernel/sched_rt.c @@ -1669,6 +1669,11 @@ static int normal_prio_rt(struct task_struct *p) return MAX_RT_PRIO-1 - p->rt_priority; } +static int effective_prio_rt(struct task_struct *p) +{ + return p->prio; +} + static void watchdog(struct rq *rq, struct task_struct *p) { unsigned long soft, hard; @@ -1758,6 +1763,7 @@ static const struct sched_class rt_sched_class = { .prio_changed = prio_changed_rt, .normal_prio = normal_prio_rt, + .effective_prio = effective_prio_rt, .switched_to = switched_to_rt, }; -- 1.6.1.2