From mboxrd@z Thu Jan  1 00:00:00 1970
From: Wen Congyang <wency@cn.fujitsu.com>
Subject: Re: [RFC Patch 25/25] sync mmu before resuming
 secondary vm
Date: Thu, 7 Aug 2014 15:46:40 +0800
Message-ID: <53E32EE0.4050705@cn.fujitsu.com>
References: <1405683551-12579-1-git-send-email-wency@cn.fujitsu.com>
	<1405683551-12579-26-git-send-email-wency@cn.fujitsu.com>
	<20140724105910.GE1817@deinos.phlegethon.org>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
Return-path: <xen-devel-bounces@lists.xen.org>
In-Reply-To: <20140724105910.GE1817@deinos.phlegethon.org>
List-Unsubscribe: <http://lists.xen.org/cgi-bin/mailman/options/xen-devel>,
	<mailto:xen-devel-request@lists.xen.org?subject=unsubscribe>
List-Post: <mailto:xen-devel@lists.xen.org>
List-Help: <mailto:xen-devel-request@lists.xen.org?subject=help>
List-Subscribe: <http://lists.xen.org/cgi-bin/mailman/listinfo/xen-devel>,
	<mailto:xen-devel-request@lists.xen.org?subject=subscribe>
Sender: xen-devel-bounces@lists.xen.org
Errors-To: xen-devel-bounces@lists.xen.org
To: Tim Deegan <tim@xen.org>
Cc: Ian Campbell <Ian.Campbell@citrix.com>, Ian Jackson <Ian.Jackson@eu.citrix.com>, Jiang Yunhong <yunhong.jiang@intel.com>, Dong Eddie <eddie.dong@intel.com>, xen devel <xen-devel@lists.xen.org>, Yang Hongyang <yanghy@cn.fujitsu.com>, Lai Jiangshan <laijs@cn.fujitsu.com>
List-Id: xen-devel@lists.xenproject.org

At 07/24/2014 06:59 PM, Tim Deegan Write:
> At 19:39 +0800 on 18 Jul (1405708750), Wen Congyang wrote:
>> In out test, we find secondary vm will be bluescreen due to memory
>> related problem. If we sync mmu, the problem will disappear.
> 
> Erk.  Do you understand _why_ this happens?  Do you need both the TLB
> flush and the EPT flush to fix it?  

I cannot reproduce this problem now. But Some user processes will exited
due to memory problem.

Secondary vm is running like this:
1. running
2. stop
3. update the the state(both memory and device)
4. continue running

Before step4, I think the tlb for the guest is out-of-data.
I don't have any knowledge about EPT. So I don't know if the
ept is out-of-data too.

Anyway, I will remove this patch in next version.

Thanks
Wen Congyang

> 
> The TLB flush sounds plausible because the migration may have changed
> some pagetables and you've elided any TLB flushes that happened on the
> source.  For that case I think I'd prefer to make the TLB flush
> implicit in the HVM load operation, e.g., by putting a call to
> hvm_asid_flush_vcpu(v) into hvm_load_cpu_ctxt() (on the grounds that the
> TLB is part of the vcpu state).
> 
> Cheers,
> 
> Tim.
>