Re: [PATCH v3] iommu: iommu_group_claim_dma_owner() must always assign a domain

[Jason Gunthorpe via iommu <iommu@lists.linux-foundation.org>]

On Wed, May 18, 2022 at 02:50:36PM -0400, Eric Farman wrote:

> I got a heads up from Matt about the s390 KVM vfio- variants failing on
> linux-next.
> 
> For vfio-ap and vfio-ccw, they fail on the above error. Both calls to
> __iommu_domain_alloc fail because while dev->dev->bus is non-NULL (it
> points to the mdev bus_type registered in mdev_init()), the bus-
> >iommu_ops pointer is NULL. Which makes sense; the iommu_group is vfio-
> noiommu, via vfio_register_emulated_iommu_dev(), and mdev didn't
> establish an iommu_ops for its bus.

Oh, I think this is a VFIO problem, the iommu layer should not have to
deal with these fake non-iommu groups.

From 9884850a5ceac957e6715beab0888294d4088877 Mon Sep 17 00:00:00 2001
From: Jason Gunthorpe <jgg@nvidia.com>
Date: Wed, 18 May 2022 16:03:34 -0300
Subject: [PATCH] vfio: Do not manipulate iommu dma_owner for fake iommu groups

Since asserting dma ownership now causes the group to have its DMA blocked
the iommu layer requires a working iommu. This means the dma_owner APIs
cannot be used on the fake groups that VFIO creates. Test for this and
avoid calling them.

Otherwise asserting dma ownership will fail for VFIO mdev devices as a
BLOCKING iommu_domain cannot be allocated due to the NULL iommu ops.

Fixes: 0286300e6045 ("iommu: iommu_group_claim_dma_owner() must always assign a domain")
Reported-by: Eric Farman <farman@linux.ibm.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
---
 drivers/vfio/vfio.c | 15 ++++++++++-----
 1 file changed, 10 insertions(+), 5 deletions(-)

I think this will have to go through Alex's tree due to all the other rework
in this area.

diff --git a/drivers/vfio/vfio.c b/drivers/vfio/vfio.c
index cfcff7764403fc..f5ed03897210c3 100644
--- a/drivers/vfio/vfio.c
+++ b/drivers/vfio/vfio.c
@@ -927,7 +927,8 @@ static void __vfio_group_unset_container(struct vfio_group *group)
 		driver->ops->detach_group(container->iommu_data,
 					  group->iommu_group);
 
-	iommu_group_release_dma_owner(group->iommu_group);
+	if (group->type == VFIO_IOMMU)
+		iommu_group_release_dma_owner(group->iommu_group);
 
 	group->container = NULL;
 	group->container_users = 0;
@@ -1001,9 +1002,11 @@ static int vfio_group_set_container(struct vfio_group *group, int container_fd)
 		goto unlock_out;
 	}
 
-	ret = iommu_group_claim_dma_owner(group->iommu_group, f.file);
-	if (ret)
-		goto unlock_out;
+	if (group->type == VFIO_IOMMU) {
+		ret = iommu_group_claim_dma_owner(group->iommu_group, f.file);
+		if (ret)
+			goto unlock_out;
+	}
 
 	driver = container->iommu_driver;
 	if (driver) {
@@ -1011,7 +1014,9 @@ static int vfio_group_set_container(struct vfio_group *group, int container_fd)
 						group->iommu_group,
 						group->type);
 		if (ret) {
-			iommu_group_release_dma_owner(group->iommu_group);
+			if (group->type == VFIO_IOMMU)
+				iommu_group_release_dma_owner(
+					group->iommu_group);
 			goto unlock_out;
 		}
 	}
-- 
2.36.0

_______________________________________________
iommu mailing list
iommu@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/iommu