From mboxrd@z Thu Jan  1 00:00:00 1970
From: "Jan Beulich" <jbeulich@novell.com>
Subject: RE: [PATCH] Dom0 Kernel - Fixes
	forsaving/restoringMSI/MSI-X across Dom0 S3
Date: Mon, 24 Nov 2008 15:32:35 +0000
Message-ID: <492AD723.76E4.0078.0@novell.com>
References: <492AA0F0.76E4.0078.0@novell.com>
	<C5504899.2950B%keir.fraser@eu.citrix.com>
	<0A882F4D99BBF6449D58E61AAFD7EDD601E23B69@pdsmsx502.ccr.corp.intel.com>
	<492ABBAA.76E4.0078.0@novell.com>
	<0A882F4D99BBF6449D58E61AAFD7EDD601E23B6B@pdsmsx502.ccr.corp.intel.com>
	<61563CE63B4F854986A895DA7AD3C17701FC5837@pdsmsx502.ccr.corp.intel.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: quoted-printable
Return-path: <xen-devel-bounces@lists.xensource.com>
In-Reply-To: <61563CE63B4F854986A895DA7AD3C17701FC5837@pdsmsx502.ccr.corp.intel.com>
Content-Disposition: inline
List-Unsubscribe: <http://lists.xensource.com/mailman/listinfo/xen-devel>,
	<mailto:xen-devel-request@lists.xensource.com?subject=unsubscribe>
List-Post: <mailto:xen-devel@lists.xensource.com>
List-Help: <mailto:xen-devel-request@lists.xensource.com?subject=help>
List-Subscribe: <http://lists.xensource.com/mailman/listinfo/xen-devel>,
	<mailto:xen-devel-request@lists.xensource.com?subject=subscribe>
Sender: xen-devel-bounces@lists.xensource.com
Errors-To: xen-devel-bounces@lists.xensource.com
To: Haitao Shan <haitao.shan@intel.com>, Kevin Tian <kevin.tian@intel.com>
Cc: "'xen-devel@lists.xensource.com'" <xen-devel@lists.xensource.com>, 'Keir Fraser' <keir.fraser@eu.citrix.com>
List-Id: xen-devel@lists.xenproject.org

>>> "Shan, Haitao" <haitao.shan@intel.com> 24.11.08 14:55 >>>
>Yes, clear enough as Kevin said. That is what I saw why S3 failed when =
AHCI is enabled (AHCI uses MSI). I do not know whether it
>is also the reason that Jan sees the need to add force unbind support of =
MSI.
>I have another question for saving/restoring in Xen if we do not use a =
new hypercall. Devices are controller by dom0. At the point
>Xen wants to save MSI during S3, dom0 may already places that device in =
D3hot state, or it may also cease the device's function
>via pci_disable_device. I doubt whether Xen can read device MMIO at that =
time.

But - as asked before - shouldn't the device undergo full re-initialization=
 after coming out of D3? Iirc, the AHCI/S3 bug results in a cannot-rebind-b=
ecause-already-bound error, which would indicate to me that the driver =
tries to create a new binding, and hence doesn't really need the device's =
state saved (admitted, I didn't look closely at the driver's code so far). =
It instead may just need Xen to clean up its internal state properly. But =
really, I'm in no way a suspend/resume expert...

Jan