On Thu, Aug 15, 2024 at 12:53:04PM -0700, Nicolin Chen wrote: > > Maybe the iommufd_viommu_invalidate ioctl handler should hold that > > xa_lock around the viommu->ops->cache_invalidate, and then add lock > > assert in iommufd_viommu_find_device? > > xa_lock/spinlock might be too heavy. We can have a mutex to wrap > around viommu ioctl handlers.. A rw semaphore might be reasonable. Jason
