On Thu, 2022-11-24 at 00:33 +0000, David Woodhouse wrote: > Could do with a test case. It's fairly simple to flip the 'longmode' > config even when the guest is actually in 64-bit mode, so it should be > fairly easy to add this in the xen_shinfo_test. Hm, not so. This doesn't depend on the 'longmode' flag; that's only for the structs in shared memory. For an actual hypercall, the compat behaviour depends more sensibly on the actual CPU mode in which the hypercall was made. I'm not sure I see how to use 32-bit guest mode from the KVM self tests. Even KVM_HC_SEND_IPI doesn't seem to have a 32-bit test... or a positive test at all, in fact. All we seem to do is test that it returns -ENOSYS when *disabled*.
Attachment:
smime.p7s
Description: S/MIME cryptographic signature
