From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755107AbcHSBlv convert rfc822-to-8bit (ORCPT ); Thu, 18 Aug 2016 21:41:51 -0400 Received: from Chamillionaire.breakpoint.cc ([146.0.238.67]:53972 "EHLO Chamillionaire.breakpoint.cc" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754988AbcHSBlm (ORCPT ); Thu, 18 Aug 2016 21:41:42 -0400 Date: Thu, 18 Aug 2016 19:35:35 +0200 From: Sebastian Andrzej Siewior To: "Paul E. McKenney" Cc: Anna-Maria Gleixner , LKML , Peter Zijlstra , Ingo Molnar , Sebastian Andrzej Siewior Subject: Re: [patch 59/66] rcu: Convert rcutree to hotplug state machine Message-ID: <20160818173535.l5i24wepg26ktqbd@breakpoint.cc> References: <20160711122450.923603742@linutronix.de> <20160711122535.613832499@linutronix.de> <20160711183828.GO4650@linux.vnet.ibm.com> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline Content-Transfer-Encoding: 8BIT In-Reply-To: <20160711183828.GO4650@linux.vnet.ibm.com> User-Agent: Mutt/1.6.2-neo (2016-08-08) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 2016-07-11 11:38:28 [-0700], Paul E. McKenney wrote: > commit da7095f39456dd0f28fa21697f2f976a61bc6d0a > Author: Paul E. McKenney > Date: Thu Jun 30 13:58:26 2016 -0700 > > rcu: Exact CPU-online tracking for RCU … I don't pretend that I know what I going on here. I have just one simple question :) > > Signed-off-by: Paul E. McKenney > > --- a/kernel/cpu.c > +++ b/kernel/cpu.c > @@ -882,6 +882,7 @@ void notify_cpu_starting(unsigned int cpu) > struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu); > enum cpuhp_state target = min((int)st->target, CPUHP_AP_ONLINE); > > + rcu_cpu_starting(cpu); /* All CPU_STARTING notifiers can use RCU. */ > while (st->state < target) { > struct cpuhp_step *step; What happens if something post CPUHP_AP_ONLINE fails and we do a rollback to 0? Do we need to revert / undo rcu_cpu_starting() doing? > diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c > index 3121242b8579..5e7c1d6a6108 100644 > --- a/kernel/rcu/tree.c > +++ b/kernel/rcu/tree.c > @@ -3793,8 +3793,6 @@ rcu_init_percpu_data(int cpu, struct rcu_state *rsp) > rnp = rdp->mynode; > mask = rdp->grpmask; > raw_spin_lock_rcu_node(rnp); /* irqs already disabled. */ > - rnp->qsmaskinitnext |= mask; > - rnp->expmaskinitnext |= mask; > if (!rdp->beenonline) > WRITE_ONCE(rsp->ncpus, READ_ONCE(rsp->ncpus) + 1); > rdp->beenonline = true; /* We have now been online. */ > @@ -4211,8 +4235,10 @@ void __init rcu_init(void) > */ > cpu_notifier(rcu_cpu_notify, 0); > pm_notifier(rcu_pm_notify, 0); > - for_each_online_cpu(cpu) > + for_each_online_cpu(cpu) { > rcu_cpu_notify(NULL, CPU_UP_PREPARE, (void *)(long)cpu); > + rcu_cpu_starting(cpu); > + } and looking at this from x86-64 then at this point I have CPU0 online and all other are down (or not yet up). So this is invoked for one CPU only. And later via hotplug callbacks for the other CPUs. > } > > #include "tree_exp.h" Sebastian