On Wed, Oct 09, 2024 at 09:38:07AM -0700, Nicolin Chen wrote:
> For an iommu_dev that can unplug (so far only this selftest does so), the
> viommu->iommu_dev pointer has no guarantee of its life cycle after it is
> copied from the idev->dev->iommu->iommu_dev.
>
> Track the user count of the iommu_dev. Delay the exit routine if refcount
> is unbalanced. The refcount inc/dec will be added in the following patch.
>
> Signed-off-by: Nicolin Chen <nicol...@nvidia.com>
> ---
> drivers/iommu/iommufd/selftest.c | 33 ++++++++++++++++++++++++--------
> 1 file changed, 25 insertions(+), 8 deletions(-)
>
> diff --git a/drivers/iommu/iommufd/selftest.c
> b/drivers/iommu/iommufd/selftest.c
> index f4be87b49447..a89a865617db 100644
> --- a/drivers/iommu/iommufd/selftest.c
> +++ b/drivers/iommu/iommufd/selftest.c
> @@ -508,14 +508,17 @@ static bool mock_domain_capable(struct device *dev,
> enum iommu_cap cap)
>
> static struct iopf_queue *mock_iommu_iopf_queue;
>
> -static struct iommu_device mock_iommu_device = {
> -};
> +static struct mock_iommu_device {
> + struct iommu_device iommu_dev;
> + wait_queue_head_t wait;
Just use a completion instead of a wait_queue, a few more bytes but it
is easier to code. This has some subtle issue where the device memory
could be freed while a concurrent thread is going to trigger the wait.
Jason
