diff mbox series

[2/3] KVM: x86/xen: Only do in-kernel acceleration of hypercalls for guest CPL0

Message ID 20221123002030.92716-2-dwmw2@infradead.org (mailing list archive)
State New, archived
Headers show
Series [1/3] KVM: x86/xen: Validate port number in SCHEDOP_poll | expand

Commit Message

David Woodhouse Nov. 23, 2022, 12:20 a.m. UTC 
From: David Woodhouse <dwmw@amazon.co.uk>

There are almost no hypercalls which are valid from CPL > 0, and definitely
none which are handled by the kernel.

Fixes: 2fd6df2f2b47 ("KVM: x86/xen: intercept EVTCHNOP_send from guests")
Reported-by: Michal Luczaj <mhal@rbox.co>
Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Cc: stable@kernel.org
---
 arch/x86/kvm/xen.c | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

Comments

Sean Christopherson Nov. 23, 2022, 4:26 p.m. UTC | #1 
On Wed, Nov 23, 2022, David Woodhouse wrote:
> From: David Woodhouse <dwmw@amazon.co.uk>
> 
> There are almost no hypercalls which are valid from CPL > 0, and definitely
> none which are handled by the kernel.
> 
> Fixes: 2fd6df2f2b47 ("KVM: x86/xen: intercept EVTCHNOP_send from guests")
> Reported-by: Michal Luczaj <mhal@rbox.co>
> Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
> Cc: stable@kernel.org
> ---

Reviewed-by: Sean Christopherson <seanjc@google.com>
diff mbox series

Patch

diff --git a/arch/x86/kvm/xen.c b/arch/x86/kvm/xen.c
index dc2f304f2e69..f3098c0e386a 100644
--- a/arch/x86/kvm/xen.c
+++ b/arch/x86/kvm/xen.c
@@ -1227,6 +1227,7 @@  int kvm_xen_hypercall(struct kvm_vcpu *vcpu)
 	bool longmode;
 	u64 input, params[6], r = -ENOSYS;
 	bool handled = false;
+	u8 cpl;
 
 	input = (u64)kvm_register_read(vcpu, VCPU_REGS_RAX);
 
@@ -1254,9 +1255,17 @@  int kvm_xen_hypercall(struct kvm_vcpu *vcpu)
 		params[5] = (u64)kvm_r9_read(vcpu);
 	}
 #endif
+	cpl = static_call(kvm_x86_get_cpl)(vcpu);
 	trace_kvm_xen_hypercall(input, params[0], params[1], params[2],
 				params[3], params[4], params[5]);
 
+	/*
+	 * Only allow hypercall acceleration for CPL0. The rare hypercalls that
+	 * are permitted in guest userspace can be handled by the VMM.
+	 */
+	if (unlikely(cpl > 0))
+		goto handle_in_userspace;
+
 	switch (input) {
 	case __HYPERVISOR_xen_version:
 		if (params[0] == XENVER_version && vcpu->kvm->arch.xen.xen_version) {
@@ -1291,10 +1300,11 @@  int kvm_xen_hypercall(struct kvm_vcpu *vcpu)
 	if (handled)
 		return kvm_xen_hypercall_set_result(vcpu, r);
 
+handle_in_userspace:
 	vcpu->run->exit_reason = KVM_EXIT_XEN;
 	vcpu->run->xen.type = KVM_EXIT_XEN_HCALL;
 	vcpu->run->xen.u.hcall.longmode = longmode;
-	vcpu->run->xen.u.hcall.cpl = static_call(kvm_x86_get_cpl)(vcpu);
+	vcpu->run->xen.u.hcall.cpl = cpl;
 	vcpu->run->xen.u.hcall.input = input;
 	vcpu->run->xen.u.hcall.params[0] = params[0];
 	vcpu->run->xen.u.hcall.params[1] = params[1];