From mboxrd@z Thu Jan  1 00:00:00 1970
From: Paolo Bonzini <pbonzini@redhat.com>
Subject: Re: [PATCH 09/12] KVM: MMU: coalesce zapping page after
 mmu_sync_children
Date: Thu, 25 Feb 2016 10:55:10 +0100
Message-ID: <56CECF7E.3050801@redhat.com>
References: <1456319873-34182-1-git-send-email-pbonzini@redhat.com>
 <1456319873-34182-10-git-send-email-pbonzini@redhat.com>
 <56CE63D1.40009@lab.ntt.co.jp> <56CEAEA7.8080702@linux.intel.com>
 <56CEC003.1070402@redhat.com> <56CEC516.7010306@linux.intel.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 8bit
Cc: mtosatti@redhat.com
To: Xiao Guangrong <guangrong.xiao@linux.intel.com>,
	Takuya Yoshikawa <yoshikawa_takuya_b1@lab.ntt.co.jp>,
	linux-kernel@vger.kernel.org, kvm@vger.kernel.org
Return-path: <linux-kernel-owner@vger.kernel.org>
In-Reply-To: <56CEC516.7010306@linux.intel.com>
Sender: linux-kernel-owner@vger.kernel.org
List-Id: kvm.vger.kernel.org



On 25/02/2016 10:10, Xiao Guangrong wrote:
> 
> 
> On 02/25/2016 04:49 PM, Paolo Bonzini wrote:
>>
>>
>> On 25/02/2016 08:35, Xiao Guangrong wrote:
>>>> This may release the mmu_lock before committing the zapping.
>>>> Is it safe?  If so, we may want to see the reason in the changelog.
>>>
>>> It is unsafe indeed, please do not do it.
>>
>> Can you explain why?  kvm_zap_obsolete_pages does the same.
> 
> It's not the same, please see the comment in 
> kvm_mmu_invalidate_zap_all_pages:
>     /*
>      * Notify all vcpus to reload its shadow page table
>      * and flush TLB. Then all vcpus will switch to new
>      * shadow page table with the new mmu_valid_gen.
>      *
>      * Note: we should do this under the protection of
>      * mmu-lock, otherwise, vcpu would purge shadow page
>      * but miss tlb flush.
>      */
>     kvm_reload_remote_mmus(kvm);
> 
> That means the tlb is flushed before releasing mmu-lock.
> 
> A example is in rmap_write_protect(), when KVM creates a shadow page
> table for
> the the guest, it detects no spte pointing to the gfn, so tlb is not
> flushed so
> that guest can freely updates its pte.

Then I'll do a different patch that checks need_resched||spin_needbreak,
and if so does commit+cond_resched_lock.  I've removed 9/12 from
kvm/queue.  Again, sorry for giving the impression that these patches
were already final.

Paolo