[PATCH] KVM: PPC: Book3S HV P9: Use kvm_arch_vcpu_get_wait() to get rcuwait object

Use kvm_arch_vcpu_get_wait() to get a vCPU's rcuwait object instead of
using vcpu->wait directly in kvmhv_run_single_vcpu().  Functionally, this
is a nop as vcpu->arch.waitp is guaranteed to point at vcpu->wait.  But
that is not obvious at first glance, and a future change coming in via
the KVM tree, commit 510958e99721 ("KVM: Force PPC to define its own
rcuwait object"), will hide vcpu->wait from architectures that define
__KVM_HAVE_ARCH_WQP to prevent generic KVM from attepting to wake a vCPU
with the wrong rcuwait object.

Reported-by: Sachin Sant <redacted>
Cc: Nicholas Piggin <npiggin@gmail.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
---

Compile tested only.  Carrying this in the PPC tree will avoid a build
failure due to a silent merge conflict with a change in the KVM tree.

https://lore.kernel.org/all/496ECBB3-36F3-4F07-83B2-875F683BC446@linux.vnet.ibm.com/

 arch/powerpc/kvm/book3s_hv.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c
index 7986911b873c..f64e45d6c0f4 100644
--- a/arch/powerpc/kvm/book3s_hv.c
+++ b/arch/powerpc/kvm/book3s_hv.c

@@ -4442,6 +4442,7 @@ static int kvmppc_run_vcpu(struct kvm_vcpu *vcpu)
 int kvmhv_run_single_vcpu(struct kvm_vcpu *vcpu, u64 time_limit,
 			  unsigned long lpcr)
 {
+	struct rcuwait *wait = kvm_arch_vcpu_get_wait(vcpu);
 	struct kvm_run *run = vcpu->run;
 	int trap, r, pcpu;
 	int srcu_idx;

@@ -4588,7 +4589,7 @@ int kvmhv_run_single_vcpu(struct kvm_vcpu *vcpu, u64 time_limit,
 	if (is_kvmppc_resume_guest(r) && !kvmppc_vcpu_check_block(vcpu)) {
 		kvmppc_set_timer(vcpu);
 
-		prepare_to_rcuwait(&vcpu->wait);
+		prepare_to_rcuwait(wait);
 		for (;;) {
 			set_current_state(TASK_INTERRUPTIBLE);
 			if (signal_pending(current)) {

@@ -4605,7 +4606,7 @@ int kvmhv_run_single_vcpu(struct kvm_vcpu *vcpu, u64 time_limit,
 			schedule();
 			trace_kvmppc_vcore_blocked(vc, 1);
 		}
-		finish_rcuwait(&vcpu->wait);
+		finish_rcuwait(wait);
 	}
 	vcpu->arch.ceded = 0;
 

-- 
2.34.1.173.g76aa8bc2d0-goog

On 13-Dec-2021, at 11:15 PM, Sean Christopherson [off-list ref] wrote:

Use kvm_arch_vcpu_get_wait() to get a vCPU's rcuwait object instead of
using vcpu->wait directly in kvmhv_run_single_vcpu().  Functionally, this
is a nop as vcpu->arch.waitp is guaranteed to point at vcpu->wait.  But
that is not obvious at first glance, and a future change coming in via
the KVM tree, commit 510958e99721 ("KVM: Force PPC to define its own
rcuwait object"), will hide vcpu->wait from architectures that define
__KVM_HAVE_ARCH_WQP to prevent generic KVM from attepting to wake a vCPU
with the wrong rcuwait object.

Reported-by: Sachin Sant <redacted>
Cc: Nicholas Piggin <npiggin@gmail.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
—

Thanks for the patch.

Tested-by: Sachin Sant <redacted>

On Mon, 13 Dec 2021 17:45:56 +0000, Sean Christopherson wrote:

Use kvm_arch_vcpu_get_wait() to get a vCPU's rcuwait object instead of
using vcpu->wait directly in kvmhv_run_single_vcpu().  Functionally, this
is a nop as vcpu->arch.waitp is guaranteed to point at vcpu->wait.  But
that is not obvious at first glance, and a future change coming in via
the KVM tree, commit 510958e99721 ("KVM: Force PPC to define its own
rcuwait object"), will hide vcpu->wait from architectures that define
__KVM_HAVE_ARCH_WQP to prevent generic KVM from attepting to wake a vCPU
with the wrong rcuwait object.

[...]

Applied to powerpc/next.

[1/1] KVM: PPC: Book3S HV P9: Use kvm_arch_vcpu_get_wait() to get rcuwait object
      https://git.kernel.org/powerpc/c/63fa47ba886b86cbd58f03b3b01b04bd57a1f233

cheers