From mboxrd@z Thu Jan 1 00:00:00 1970 From: Rik van Riel Subject: Re: [PATCH 4/4] kvm,rcu: use RCU extended quiescent state when running KVM guest Date: Thu, 05 Feb 2015 13:09:19 -0500 Message-ID: <54D3B1CF.1000301@redhat.com> References: <1423154134-17391-1-git-send-email-riel@redhat.com> <1423154134-17391-5-git-send-email-riel@redhat.com> <54D39DF0.3020109@de.ibm.com> <54D39FD5.2040402@redhat.com> <20150205175014.GG5370@linux.vnet.ibm.com> Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 7bit Cc: Christian Borntraeger , kvm@vger.kernel.org, linux-kernel@vger.kernel.org, mtosatti@redhat.com, mingo@kernel.org, ak@linux.intel.com, oleg@redhat.com, masami.hiramatsu.pt@hitachi.com, fweisbec@gmail.com, lcapitulino@redhat.com, pbonzini@redhat.com To: paulmck@linux.vnet.ibm.com Return-path: In-Reply-To: <20150205175014.GG5370@linux.vnet.ibm.com> Sender: linux-kernel-owner@vger.kernel.org List-Id: kvm.vger.kernel.org On 02/05/2015 12:50 PM, Paul E. McKenney wrote: > On Thu, Feb 05, 2015 at 11:52:37AM -0500, Rik van Riel wrote: >> On 02/05/2015 11:44 AM, Christian Borntraeger wrote: >>> Am 05.02.2015 um 17:35 schrieb riel@redhat.com: >>>> From: Rik van Riel >>>> >>>> The host kernel is not doing anything while the CPU is executing >>>> a KVM guest VCPU, so it can be marked as being in an extended >>>> quiescent state, identical to that used when running user space >>>> code. >>>> >>>> The only exception to that rule is when the host handles an >>>> interrupt, which is already handled by the irq code, which >>>> calls rcu_irq_enter and rcu_irq_exit. >>>> >>>> The guest_enter and guest_exit functions already switch vtime >>>> accounting independent of context tracking, so leave those calls >>>> where they are, instead of moving them into the context tracking >>>> code. >>>> >>>> Signed-off-by: Rik van Riel >>>> --- >>>> include/linux/context_tracking.h | 8 +++++++- >>>> include/linux/context_tracking_state.h | 1 + >>>> 2 files changed, 8 insertions(+), 1 deletion(-) >>>> >>>> diff --git a/include/linux/context_tracking.h b/include/linux/context_tracking.h >>>> index bd9f000fc98d..a5d3bb44b897 100644 >>>> --- a/include/linux/context_tracking.h >>>> +++ b/include/linux/context_tracking.h >>>> @@ -43,7 +43,7 @@ static inline enum ctx_state exception_enter(void) >>>> static inline void exception_exit(enum ctx_state prev_ctx) >>>> { >>>> if (context_tracking_is_enabled()) { >>>> - if (prev_ctx == IN_USER) >>>> + if (prev_ctx == IN_USER || prev_ctx == IN_GUEST) >>>> context_tracking_user_enter(prev_ctx); >>>> } >>>> } >>>> @@ -78,6 +78,9 @@ static inline void guest_enter(void) >>>> vtime_guest_enter(current); >>>> else >>>> current->flags |= PF_VCPU; >>>> + >>>> + if (context_tracking_is_enabled()) >>>> + context_tracking_user_enter(IN_GUEST); >>>> } >>> >>> >>> Couldnt we make >>> rcu_virt_note_context_switch(smp_processor_id()); >>> conditional in include/linux/kvm_host.h (kvm_guest_enter) >>> >>> e.g. something like >>> if (!context_tracking_is_enabled()) >>> rcu_virt_note_context_switch(smp_processor_id()); >> >> Possibly. I considered the same, but I do not know whether >> or not just rcu_user_enter / rcu_user_exit is enough. >> >> I could certainly try it out and see whether anything >> explodes, but I am not convinced that is careful enough >> when it comes to handling RCU code... >> >> Paul? :) > > That can fail for some odd combinations of Kconfig and boot parameters. > As I understand it, you instead want the following: > > if (!tick_nohz_full_cpu(smp_processor_id())) > rcu_virt_note_context_switch(smp_processor_id()); > > Frederic might know of a better approach. Interesting, in what cases would that happen? If context_tracking_is_enabled() we end up eventually calling into rcu_user_enter & rcu_user_exit. If it is not enabled, we call rcu_virt_note_context_switch. In what cases would we need to call both? I don't see exit-to-userspace do the things that rcu_virt_note_context_switch does, and do not understand why userspace is fine with that...