On 09.11.2022 17:11, Andrew Cooper wrote: > On 08/11/2022 11:38, Roger Pau Monne wrote: >> Like on the Arm side, return -EINVAL when attempting to do a p2m >> operation on dying domains. > > Honestly, I'd drop the comment about ARM. "the Arm side" has existed > for of all of a couple of weeks. > > A far better justification is because almost all other DOMCTLs are > rejected with -EINVAL against dying domains.
Would you mind supplying data to prove this statement? When looking at just x86'es arch_do_domctl() I can't see this being the case for ioport_permission, getpageframeinfo3, hypercall_init, set_address_size, get_address_size, sendtrigger, bind_pt_irq, unbind_pt_irq, ioport_mapping, set_ext_vcpucontext, and get_ext_vcpucontext. At which point I stopped checking further because in the order they appear in the file these are _all_ except shadow_op and [gs]ethvmcontext*. Am I overlooking hidden checks of ->is_dying? Jan
