From mboxrd@z Thu Jan 1 00:00:00 1970 From: Konrad Rzeszutek Wilk Subject: Re: Re: PCI BAR register space written with garbage in HVM guest. Date: Thu, 18 Mar 2010 14:10:08 -0400 Message-ID: <20100318181008.GA32657@phenom.dumpdata.com> References: <4779de451003151809n6cec813dp32d77fee34b1bda2@mail.gmail.com> <4779de451003161714x45dbca6dh80a9eed56e4fb0c2@mail.gmail.com> <4779de451003171914u2f8da58etc698f07dc91c853f@mail.gmail.com> <20100318143207.GF14445@phenom.dumpdata.com> <4779de451003181127m210171e9xc5fc4ff1772f4782@mail.gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable Return-path: Content-Disposition: inline In-Reply-To: <4779de451003181127m210171e9xc5fc4ff1772f4782@mail.gmail.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: Dan Gora Cc: xen-devel@lists.xensource.com List-Id: xen-devel@lists.xenproject.org On Thu, Mar 18, 2010 at 03:27:30PM -0300, Dan Gora wrote: > On Thu, Mar 18, 2010 at 11:32 AM, Konrad Rzeszutek Wilk > wrote: > > On Wed, Mar 17, 2010 at 11:14:13PM -0300, Dan Gora wrote: > >> Hi all, > >> > >> I apologize for the length of this message, but I figured that it wa= s > >> probably better to supply too much info rather than too little... > >> I've made some good progress debugging this today. =A0The problem lo= oks > >> pretty bizarre... > > > > I just skimmed through your e-mail and didn't dive in the r8169 code = to > > see if there is something there, but I thought you might want to try > > using e1000 instead of r8169 driver and see if that makes it work? >=20 > Sure enough that fixed the problem! No more writes on fd 6, no more > PCI BAR regsiter corruption! > That's good enough for me as a workaround. I guess someone else can > have a look at this rtl8139 driver. That is not what I expected, but I am glad to hear that it makes it possible for you to continue with your project. It is good to know that the failure path is contained in the r8169 implementation. If it wouldn't be too much trouble, could you send an e-mail to the QEMU mailing list this thread and your solution in case there is somebody there itching to work on this.