From mboxrd@z Thu Jan 1 00:00:00 1970 From: Study Xen Subject: Re: unable to capture an expected page fault Date: Tue, 17 Jan 2012 13:11:37 -0500 Message-ID: References: <20120117111222.GC74654@ocelot.phlegethon.org> Mime-Version: 1.0 Content-Type: multipart/mixed; boundary="===============3524151603254786564==" Return-path: In-Reply-To: <20120117111222.GC74654@ocelot.phlegethon.org> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: Tim Deegan Cc: xen-devel@lists.xensource.com List-Id: xen-devel@lists.xenproject.org --===============3524151603254786564== Content-Type: multipart/alternative; boundary=0015175d07b212dcf804b6bd4367 --0015175d07b212dcf804b6bd4367 Content-Type: text/plain; charset=ISO-8859-1 Hello Tim, Thanks for your reply. I am actually learning (and the same time trying to port) the code written by a former student. That is why it is currently hard for me to clearly explain what have been modified in the code. The former member's intention was to make use of Xen's ability of modifying pagetables underneath the guest kernel to achieve data protection of user applications from a compromised kernel. So his modification scattered into several places, mainly in Xen's memory management code and interrupt handling code. Unfortunately I have not been able to pinpoint which code caused the current bug I mentioned in the previous mail. X On Tue, Jan 17, 2012 at 6:12 AM, Tim Deegan wrote: > Hi, > > At 03:08 -0500 on 17 Jan (1326769684), Study Xen wrote: > > I was trying to modify part of Xen and faced a page fault missing issue. > > > > I am testing a PV Linux 64-bit guest (kernel 3.1.1, as dom0, the only > > domain in my setting) atop Xen 4.1.2. In an unmodified Xen, the Linux > > kernel's "native_set_pte" in "arch/x86/include/asm/pgtable_64.h" traps > into > > Xen due to Xen's write-protection of the pagetable page. But in our > > modified version, this expected page fault seems missing. > > Well, I suppose it must be something you changed. :) But since you don't > say what you changed I'm not sure how much we can help you. If the > pagetables are indeed correct then maybe you're missing a TLB flush > somewhere? > > Tim. > --0015175d07b212dcf804b6bd4367 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable Hello Tim,

Thanks for your reply. I am actually learning= (and the same time trying to port) the code written by a former student. T= hat is why it is currently hard for me to clearly explain what have been mo= dified in the code.=A0

The former member's intention was to make use of Xe= n's ability of modifying pagetables underneath the guest kernel to achi= eve data protection of user applications from a compromised kernel. So his = modification scattered into several places, mainly in Xen's memory mana= gement code and interrupt handling code. Unfortunately I have not been able= to pinpoint which code caused the current bug I mentioned in the previous = mail.=A0

X


On Tue, = Jan 17, 2012 at 6:12 AM, Tim Deegan <tim@xen.org> wrote:
Hi,

At 03:08 -0500 on 17 Jan (1326769684), Study Xen wrote:
> I was trying to modify part of Xen and faced a page fault missing issu= e.
>
> I am testing a PV Linux 64-bit guest (kernel 3.1.1, as dom0, the only<= br> > domain in my setting) atop Xen 4.1.2. In an unmodified Xen, the Linux<= br> > kernel's "native_set_pte" in "arch/x86/include/asm/= pgtable_64.h" traps into
> Xen due to Xen's write-protection of the pagetable page. But in ou= r
> modified version, this expected page fault seems missing.

Well, I suppose it must be something you changed. :) =A0But since you= don't
say what you changed I'm not sure how much we can help you. =A0If the pagetables are indeed correct then maybe you're missing a TLB flush
somewhere?

Tim.

--0015175d07b212dcf804b6bd4367-- --===============3524151603254786564== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel --===============3524151603254786564==--