Re: [Qemu-devel] [V12 0/4] AMD IOMMU

[Jan Kiszka <jan.kiszka@web.de>]

[-- Attachment #1: Type: text/plain, Size: 1849 bytes --]

On 2016-06-15 19:07, David Kiarie wrote:
> On Wed, Jun 15, 2016 at 5:26 PM, Eduardo Habkost <ehabkost@redhat.com> wrote:
>> On Wed, Jun 15, 2016 at 03:21:48PM +0300, David Kiarie wrote:
>>> Hi all,
>>>
>>> This patchset adds basic AMD IOMMU emulation support to Qemu.
>>>
>>> Changes since V11
>>>    -AMD IOMMU is not started with -device amd-iommu (with a dependency on Marcel's patches).
>>>    -IOMMU commands are represented using bitfields which is less error prone and more readable[Peter]
>>>    -Changed from debug fprintfs to tracing[Jan]
>>
>> What were the issues that required the sysbus+PCI code you sent
>> previously? How did you address them in this series?
> 
> Short answer: Those issues are not present in this patch.
> 
> Long answer: The sysbus + PCI code is necessary for interrupt
> remapping to be implemented(it could also be done without sysbus + PCI
> code but that wouldn't be in line with the Intel IOMMU interrupt
> remapping code, which is already on the list). The idea is that X86
> IOMMUs should have a base class that implements common code. It was
> decided that this class should be a SysBusDev(which works perfectly
> with Intel IOMMU) but not with AMD IOMMU which has PCI properties. I
> had to find a way to provide for the PCI properties once I inherit
> from Intel IOMMU.
> 
> In this patchset, I have not inherited from the base class(it's not
> merged yet) and even if it was merged I'd prefer to only use the
> SysBus + PCI code when necessary(when I work on interrupt remapping)
> so as to avoid delaying this patchset further.

We specifically came to the conclusion that it is better to evolve the
commodities between both IOMMU implementations (will be mostly about
interrupt remapping) in-tree rather than in separate patch sets outside
the mainline.

Jan



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 181 bytes --]