On 2022/9/23 0:00, Jason Gunthorpe wrote:
On Tue, Sep 06, 2022 at 08:44:54PM +0800, Lu Baolu wrote:
+/**
+ * iommu_sva_bind_device() - Bind a process address space to a device
+ * @dev: the device
+ * @mm: the mm to bind, caller must hold a reference to mm_users
+ *
+ * Create a bond between device and address space, allowing the device to access
+ * the mm using the returned PASID. If a bond already exists between @device and
+ * @mm, it is returned and an additional reference is taken. Caller must call
+ * iommu_sva_unbind_device() to release each reference.
+ *
+ * iommu_dev_enable_feature(dev, IOMMU_DEV_FEAT_SVA) must be called first, to
+ * initialize the required SVA features.
Thsi is something else that needs cleaning up. IOMMU_DEV_FEAT_SVA
shouldn't exist.
We need to figure out an appropriate way to allow PRI. IMHO the domain
attach should do this, domains that require PRI should be distinct
from domains that don't. When a PRI domain is attached the HW should
be enabled to do PRI. The domain itself should carry the fault ops/etc
that the caller supplies to respond to the PRI.
That is something to address in the PRI series though..
From Intel IOMMU driver's point of view, with above done,
IOMMU_DEV_FEAT_SVA could be removed. However, it will take more time to
consider other needs.
Reviewed-by: Jason Gunthorpe<jgg@xxxxxxxxxx>
Best regards,
baolu
