On Tue, 2024-03-12 at 21:41 +0000, Huang, Kai wrote: > On Tue, 2024-03-12 at 07:20 -0700, Sean Christopherson wrote: > > On Tue, Mar 12, 2024, Kai Huang wrote: > > > > Wait. KVM doesn't *need* to do PAGE.ADD from deep in the MMU. The only inputs to > > > > PAGE.ADD are the gfn, pfn, tdr (vm), and source. The S-EPT structures need to be > > > > pre-built, but when they are built is irrelevant, so long as they are in place > > > > before PAGE.ADD. > > > > > > > > Crazy idea. For TDX S-EPT, what if KVM_MAP_MEMORY does all of the SEPT.ADD stuff, > > > > which doesn't affect the measurement, and even fills in KVM's copy of the leaf EPTE, > > > > but tdx_sept_set_private_spte() doesn't do anything if the TD isn't finalized? > > > > > > > > Then KVM provides a dedicated TDX ioctl(), i.e. what is/was KVM_TDX_INIT_MEM_REGION, > > > > to do PAGE.ADD. KVM_TDX_INIT_MEM_REGION wouldn't need to map anything, it would > > > > simply need to verify that the pfn from guest_memfd() is the same as what's in > > > > the TDP MMU. > > > > > > One small question: > > > > > > What if the memory region passed to KVM_TDX_INIT_MEM_REGION hasn't been pre- > > > populated? If we want to make KVM_TDX_INIT_MEM_REGION work with these regions, > > > then we still need to do the real map. Or we can make KVM_TDX_INIT_MEM_REGION > > > return error when it finds the region hasn't been pre-populated? > > > > Return an error. I don't love the idea of bleeding so many TDX details into > > userspace, but I'm pretty sure that ship sailed a long, long time ago. > > In this case, IIUC the KVM_MAP_MEMORY ioctl() will be mandatory for TDX > (presumbly also SNP) guests, but _optional_ for other VMs. Not sure whether > this is ideal. > > And just want to make sure I understand the background correctly: > > The KVM_MAP_MEMORY ioctl() is supposed to be generic, and it should be able to > be used by any VM but not just CoCo VMs (including SW_PROTECTED ones)? > > But it is only supposed to be used by the VMs which use guest_memfd()? Because > IIUC for normal VMs using mmap() we already have MAP_POPULATE for this purpose. > > Looking at [*], it doesn't say what kind of VM the sender was trying to use. > > Therefore can we interpret KVM_MAP_MEMORY ioctl() is effectively for CoCo VMs? > SW_PROTECTED VMs can also use guest_memfd(), but I believe nobody is going to > use it seriously. > > [*] https://lore.kernel.org/all/65262e67-7885-971a-896d-ad9c0a760907@xxxxxxxxx/ > > Hmm.. Just after sending I realized the MAP_POPULATE only pre-populate page table at host side, not EPT... So the KVM_MAP_MEMORY is indeed can be used by _ALL_ VMs. My bad :-(