Re: [PATCH] vfio: platform: use vfio_iommu_group_get/put

[Peng Fan]

Hi,

On Fri, May 20, 2016 at 10:20:32AM +0200, Baptiste Reynal wrote:
>Hi,
>
>On Fri, May 20, 2016 at 9:58 AM, Eric Auger <eric.auger@xxxxxxxxxx> wrote:
>>
>> Hi,
>> On 05/19/2016 11:01 PM, Alex Williamson wrote:
>> > On Tue, 10 May 2016 15:40:28 +0800
>> > Peng Fan <van.freenix@xxxxxxxxx> wrote:
>> >
>> >> Hi Alex,
>> >>
>> >> On Mon, May 09, 2016 at 09:32:38AM -0600, Alex Williamson wrote:
>> >>> On Mon,  9 May 2016 18:01:43 +0800
>> >>> Peng Fan <van.freenix@xxxxxxxxx> wrote:
>> >>>
>> >>>> Use vfio_iommu_group_get and vfio_iommu_group_put, but not
>> >>>> iommu_group_get or iommu_group_put.
>> >>>
>> >>> I assume you're trying to duplicate the vfio_pci changes from commit
>> >>> 03a76b60f8ba to enable no-iommu mode.  That would be really relevant
>> >>> information for the commit log.
>> >>
>> >> This is not to support non-iommu for vfio platform. I just think
>> >> vfio_iommu_group_get/put is vfio core API and should be used by
>> >> vfio-pci and vfio-platform.
>> >
>> > Hi Peng,
>> >
>> > I suppose I would consider this an optional part of the internal vfio
>> > API, it's only real purpose it to provide the ability to create fake
>> > groups which are only used for no-iommu.  It's perfectly legitimate to
>> > use iommu_group_get/put if there is no desire to enable no-iommu.
>> > Baptiste, Eric, do you have an opinion whether enabling no-iommu in
>> > vfio/platform is something we should do?  Thanks,
>> I think it would make sense to introduce that no-iommu feature for
>> vfio-platform because it allows potential users to get familiar with
>> VFIO platform without having the proper HW. I also thought the primary
>> purpose of this patch was to introduce that support.

Yeah. Initially I would like to try vfio on my i.MX6/7 boards which does not
have HW iommu. I am new to this, so just wrote this simple patch when I was
reading the source code.

>
>Since safety issues have already been addressed for the PCI patch
>series, I don't see any reason not to enable it. You might extend the
>commit message to remind those issues.

Eric, Baptiste,

Is vfio-platform no-iommu on your TODO list? I am happy to help test.
If not, I can try work on this if you could guide me -:)

Thanks,
Peng.
>
>Best regards,
>Baptiste
>
>>
>> Best Regards
>>
>> Eric
>> >
>> > Alex
>> >
>> >>>>
>> >>>> Signed-off-by: Peng Fan <van.freenix@xxxxxxxxx>
>> >>>> Cc: Baptiste Reynal <b.reynal@xxxxxxxxxxxxxxxxxxxxxx>
>> >>>> Cc: Alex Williamson <alex.williamson@xxxxxxxxxx>
>> >>>> ---
>> >>>>  drivers/vfio/platform/vfio_platform_common.c | 6 +++---
>> >>>>  1 file changed, 3 insertions(+), 3 deletions(-)
>> >>>>
>> >>>> diff --git a/drivers/vfio/platform/vfio_platform_common.c b/drivers/vfio/platform/vfio_platform_common.c
>> >>>> index e65b142..582885e 100644
>> >>>> --- a/drivers/vfio/platform/vfio_platform_common.c
>> >>>> +++ b/drivers/vfio/platform/vfio_platform_common.c
>> >>>> @@ -561,7 +561,7 @@ int vfio_platform_probe_common(struct vfio_platform_device *vdev,
>> >>>>
>> >>>>    vdev->device = dev;
>> >>>>
>> >>>> -  group = iommu_group_get(dev);
>> >>>> +  group = vfio_iommu_group_get(dev);
>> >>>>    if (!group) {
>> >>>>            pr_err("VFIO: No IOMMU group for device %s\n", vdev->name);
>> >>>>            return -EINVAL;
>> >>>> @@ -569,7 +569,7 @@ int vfio_platform_probe_common(struct vfio_platform_device *vdev,
>> >>>>
>> >>>>    ret = vfio_add_group_dev(dev, &vfio_platform_ops, vdev);
>> >>>>    if (ret) {
>> >>>> -          iommu_group_put(group);
>> >>>> +          vfio_iommu_group_put(group);
>> >>>>            return ret;
>> >>>>    }
>> >>>>
>> >>>> @@ -589,7 +589,7 @@ struct vfio_platform_device *vfio_platform_remove_common(struct device *dev)
>> >>>>
>> >>>>    if (vdev) {
>> >>>>            vfio_platform_put_reset(vdev);
>> >>>> -          iommu_group_put(dev->iommu_group);
>> >>>> +          vfio_iommu_group_put(dev->iommu_group, dev);
>> >>>>    }
>> >>>>
>> >>>>    return vdev;
>> >>>
>> >
>>