> From: Nicolin Chen <nicolinc@xxxxxxxxxx> > Sent: Wednesday, February 15, 2023 3:15 PM > > But things will be out of control, if user space continues mapping > something onto domain1's iopt for idev2, even after it is attached > covertly to domain2's iopt by the replace routine. I wonder how > kernel should handle this and keep the consistency between IOMMUFD > objects and iommu_group. > this is where I don't understand. domain mapping just reflects what an address space has. Take Qemu for example. w/o vIOMMU domain mappings is added/ removed according to the change in the GPA address space. w/ vIOMMU then it is synced with guest page table. why would userspace construct mappings for a specific device? when device is moved from one domain to another domain, it just bears with whatever the new domain allows it to access.
