On 10/08/2018 06:27 AM, Paolo Bonzini wrote: > On 06/10/2018 22:43, Guenter Roeck wrote: >>> >>>>> Maybe this works as well? I haven't tested it yet: >>>>> >>>> I am sure there are many possible solutions. I would personally >>>> prefer one >>>> that enforces KVM_AMD=m with CRYPTO_DEV_CCP_DD=m, but that is just me. >>> >>> Well, KVM_AMD=y is a relatively unusual choice to begin with. The >> >> It is common enough that we are not the only ones affected. Also, even a >> "relatively unusual choice" should, in my opinion, not result in a build >> error. > > Of course not! The question is whether to solve it by disabling > KVM_AMD_SEV (which is what the current code attempts to do, and my patch > should fix that) or KVM_AMD (your patch). IMHO, Paolo's patch make sense; it removes all the SEV specific code when KVM_AMD_SEV=n. It saves ~4K in text section. Paolo, Does it make sense to move all the SEV specific code in svm-sev.c ? I am looking to add SEV migration support very soon, and can see myself adding more SEV command handling which will grow svm.c further. -Brijesh
