[prev in list] [next in list] [prev in thread] [next in thread] 

List:       linux-iommu
Subject:    AMD IOMMU: Please support Marvell 88SE91xx SATA Controllers
From:       acooks () gmail ! com (Andrew Cooks)
Date:       2013-07-31 11:07:21
Message-ID: CAJtEV7bAVKVW4YZwmFDjMs7F3tab+ifpnVcq3jBqFa9gvtL+8A () mail ! gmail ! com
[Download RAW message or body]

Hi Joerg

On Tue, Jul 30, 2013 at 7:39 PM, Joerg Roedel <joro at 8bytes.org> wrote:
> Hi Andrew,
>
> On Wed, Jul 03, 2013 at 10:53:37AM +0200, Manfred Schwarb wrote:
>> Am 21.05.2013 22:35, schrieb Joerg Roedel:
>> >On Mon, May 20, 2013 at 11:21:52PM +0200, Manfred Schwarb wrote:
>> >>Is anybody working on this issue for the AMD part? I would be very
>> >>grateful if such support could be added to the linux kernel. I would
>> >>be prepared to test patches, if needed.
>
> Somehow the workaround for Intel didn't make it upstream. Do you plan to
> continue working on the workaround?

I think some of the issues that prevented the workaround from being
merged will be addressed by Alex Williamson's PCIe requester ID
interface patch set, which is currently being discussed[1].

That patch should bring the pci_get_dma_source() workaround into the
main dma ops path, after which it can be changed to allow for devices
that use multiple or otherwise incorrect requestor ids.

[1] http://thread.gmane.org/gmane.linux.kernel.pci/23876/focus=2599
>
> I reused a few parts of your VT-d patch to implement the workaround for
> AMD, there is surely potential to have some common code between the
> IOMMU drivers for this.

Thanks for that.

a.

[prev in list] [next in list] [prev in thread] [next in thread]