KVM: PPC: Book3S HV nestedv2: Do not check msr on hcalls
authorJordan Niethe <jniethe5@gmail.com>
Fri, 1 Dec 2023 13:26:08 +0000 (18:56 +0530)
committerMichael Ellerman <mpe@ellerman.id.au>
Thu, 7 Dec 2023 12:33:07 +0000 (23:33 +1100)
The check for a hcall coming from userspace is done for KVM-PR. This is
not supported for nestedv2 and the L0 will directly inject the necessary
exception to the L2 if userspace performs a hcall. Avoid checking the
MSR and thus avoid a H_GUEST_GET_STATE hcall in the L1.

Signed-off-by: Jordan Niethe <jniethe5@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20231201132618.555031-4-vaibhav@linux.ibm.com
arch/powerpc/kvm/book3s_hv.c

index 5543e8490cd9f00c9e2e1558c39e533348424594..069c336b6f3c2e5fd307c4a09d5657699e7d15f1 100644 (file)
@@ -1688,7 +1688,7 @@ static int kvmppc_handle_exit_hv(struct kvm_vcpu *vcpu,
        {
                int i;
 
-               if (unlikely(__kvmppc_get_msr_hv(vcpu) & MSR_PR)) {
+               if (!kvmhv_is_nestedv2() && unlikely(__kvmppc_get_msr_hv(vcpu) & MSR_PR)) {
                        /*
                         * Guest userspace executed sc 1. This can only be
                         * reached by the P9 path because the old path
@@ -4949,7 +4949,7 @@ static int kvmppc_vcpu_run_hv(struct kvm_vcpu *vcpu)
                if (run->exit_reason == KVM_EXIT_PAPR_HCALL) {
                        accumulate_time(vcpu, &vcpu->arch.hcall);
 
-                       if (WARN_ON_ONCE(__kvmppc_get_msr_hv(vcpu) & MSR_PR)) {
+                       if (!kvmhv_is_nestedv2() && WARN_ON_ONCE(__kvmppc_get_msr_hv(vcpu) & MSR_PR)) {
                                /*
                                 * These should have been caught reflected
                                 * into the guest by now. Final sanity check: