Consult only the basic exit reason, i.e. bits 15:0 of vmcs.EXIT_REASON,
when determining whether a nested VM-Exit should be reflected into L1 or
handled by KVM in L0.
For better or worse, the switch statements nested_vmx_l0_wants_exit()
and nested_vmx_l1_wants_exit() default to reflecting the VM-Exit into L1
for any nested VM-Exit without dedicated logic. Because the case
statements only contain the basic exit reason, any VM-Exit with modifier
bits set will be reflected to L1, even if KVM intended to handle it in
L0.
Practically speaking, this only affects EXIT_REASON_MCE_DURING_VMENTRY,
i.e. a #MC that occurs on nested VM-Enter would be incorrectly routed to
L1, as "failed VM-Entry" is the only modifier that KVM can currently
encounter. The SMM modifiers will never be generated as KVM doesn't
support/employ a SMI Transfer Monitor. Ditto for "exit from enclave",
as KVM doesn't yet support virtualizing SGX, i.e. it's impossible to
enter an enclave in a KVM guest (L1 or L2).
Note, the original version of this fix[*] is functionally equivalent and
far more suited to backporting as the affected code was refactored since
the original patch was posted.
[*] https://lkml.kernel.org/r/[email protected]
Fixes: 644d711aa0e1 ("KVM: nVMX: Deciding if L0 or L1 should handle an L2 exit")
Cc: Jim Mattson <[email protected]>
Cc: Xiaoyao Li <[email protected]>
Cc: [email protected]
Cc: Oliver Upton <[email protected]>
Cc: Krish Sadhukhan <[email protected]>
Cc: Miaohe Lin <[email protected]>
Signed-off-by: Sean Christopherson <[email protected]>
---
Another wounded soldier.
Oliver, Krish, and Miaohe all provided reviews for v1, but I didn't feel
comfortable adding the tags to v2 because this is far from a straight
rebase.
v2: Rebased to kvm/queue, commit fb7333dfd812 ("KVM: SVM: fix calls ...").
arch/x86/kvm/vmx/nested.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
index bcb50724be38..adb11b504d5c 100644
--- a/arch/x86/kvm/vmx/nested.c
+++ b/arch/x86/kvm/vmx/nested.c
@@ -5672,7 +5672,7 @@ static bool nested_vmx_l0_wants_exit(struct kvm_vcpu *vcpu, u32 exit_reason)
{
u32 intr_info;
- switch (exit_reason) {
+ switch ((u16)exit_reason) {
case EXIT_REASON_EXCEPTION_NMI:
intr_info = vmx_get_intr_info(vcpu);
if (is_nmi(intr_info))
@@ -5733,7 +5733,7 @@ static bool nested_vmx_l1_wants_exit(struct kvm_vcpu *vcpu, u32 exit_reason)
struct vmcs12 *vmcs12 = get_vmcs12(vcpu);
u32 intr_info;
- switch (exit_reason) {
+ switch ((u16)exit_reason) {
case EXIT_REASON_EXCEPTION_NMI:
intr_info = vmx_get_intr_info(vcpu);
if (is_nmi(intr_info))
--
2.26.0
On Mon, Jun 8, 2020 at 12:19 PM Sean Christopherson
<[email protected]> wrote:
>
> Consult only the basic exit reason, i.e. bits 15:0 of vmcs.EXIT_REASON,
> when determining whether a nested VM-Exit should be reflected into L1 or
> handled by KVM in L0.
>
> For better or worse, the switch statements nested_vmx_l0_wants_exit()
> and nested_vmx_l1_wants_exit() default to reflecting the VM-Exit into L1
> for any nested VM-Exit without dedicated logic. Because the case
> statements only contain the basic exit reason, any VM-Exit with modifier
> bits set will be reflected to L1, even if KVM intended to handle it in
> L0.
>
> Practically speaking, this only affects EXIT_REASON_MCE_DURING_VMENTRY,
> i.e. a #MC that occurs on nested VM-Enter would be incorrectly routed to
> L1, as "failed VM-Entry" is the only modifier that KVM can currently
> encounter. The SMM modifiers will never be generated as KVM doesn't
> support/employ a SMI Transfer Monitor. Ditto for "exit from enclave",
> as KVM doesn't yet support virtualizing SGX, i.e. it's impossible to
> enter an enclave in a KVM guest (L1 or L2).
>
> Note, the original version of this fix[*] is functionally equivalent and
> far more suited to backporting as the affected code was refactored since
> the original patch was posted.
>
> [*] https://lkml.kernel.org/r/[email protected]
>
> Fixes: 644d711aa0e1 ("KVM: nVMX: Deciding if L0 or L1 should handle an L2 exit")
> Cc: Jim Mattson <[email protected]>
> Cc: Xiaoyao Li <[email protected]>
> Cc: [email protected]
> Cc: Oliver Upton <[email protected]>
> Cc: Krish Sadhukhan <[email protected]>
> Cc: Miaohe Lin <[email protected]>
> Signed-off-by: Sean Christopherson <[email protected]>
Reviewed-by: Oliver Upton <[email protected]>
> ---
>
> Another wounded soldier.
>
> Oliver, Krish, and Miaohe all provided reviews for v1, but I didn't feel
> comfortable adding the tags to v2 because this is far from a straight
> rebase.
>
> v2: Rebased to kvm/queue, commit fb7333dfd812 ("KVM: SVM: fix calls ...").
>
> arch/x86/kvm/vmx/nested.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
> index bcb50724be38..adb11b504d5c 100644
> --- a/arch/x86/kvm/vmx/nested.c
> +++ b/arch/x86/kvm/vmx/nested.c
> @@ -5672,7 +5672,7 @@ static bool nested_vmx_l0_wants_exit(struct kvm_vcpu *vcpu, u32 exit_reason)
> {
> u32 intr_info;
>
> - switch (exit_reason) {
> + switch ((u16)exit_reason) {
> case EXIT_REASON_EXCEPTION_NMI:
> intr_info = vmx_get_intr_info(vcpu);
> if (is_nmi(intr_info))
> @@ -5733,7 +5733,7 @@ static bool nested_vmx_l1_wants_exit(struct kvm_vcpu *vcpu, u32 exit_reason)
> struct vmcs12 *vmcs12 = get_vmcs12(vcpu);
> u32 intr_info;
>
> - switch (exit_reason) {
> + switch ((u16)exit_reason) {
> case EXIT_REASON_EXCEPTION_NMI:
> intr_info = vmx_get_intr_info(vcpu);
> if (is_nmi(intr_info))
> --
> 2.26.0
>
On 08/06/20 21:18, Sean Christopherson wrote:
> Consult only the basic exit reason, i.e. bits 15:0 of vmcs.EXIT_REASON,
> when determining whether a nested VM-Exit should be reflected into L1 or
> handled by KVM in L0.
>
> For better or worse, the switch statements nested_vmx_l0_wants_exit()
> and nested_vmx_l1_wants_exit() default to reflecting the VM-Exit into L1
> for any nested VM-Exit without dedicated logic. Because the case
> statements only contain the basic exit reason, any VM-Exit with modifier
> bits set will be reflected to L1, even if KVM intended to handle it in
> L0.
>
> Practically speaking, this only affects EXIT_REASON_MCE_DURING_VMENTRY,
> i.e. a #MC that occurs on nested VM-Enter would be incorrectly routed to
> L1, as "failed VM-Entry" is the only modifier that KVM can currently
> encounter. The SMM modifiers will never be generated as KVM doesn't
> support/employ a SMI Transfer Monitor. Ditto for "exit from enclave",
> as KVM doesn't yet support virtualizing SGX, i.e. it's impossible to
> enter an enclave in a KVM guest (L1 or L2).
>
> Note, the original version of this fix[*] is functionally equivalent and
> far more suited to backporting as the affected code was refactored since
> the original patch was posted.
>
> [*] https://lkml.kernel.org/r/[email protected]
>
> Fixes: 644d711aa0e1 ("KVM: nVMX: Deciding if L0 or L1 should handle an L2 exit")
> Cc: Jim Mattson <[email protected]>
> Cc: Xiaoyao Li <[email protected]>
> Cc: [email protected]
> Cc: Oliver Upton <[email protected]>
> Cc: Krish Sadhukhan <[email protected]>
> Cc: Miaohe Lin <[email protected]>
> Signed-off-by: Sean Christopherson <[email protected]>
> ---
>
> Another wounded soldier.
>
> Oliver, Krish, and Miaohe all provided reviews for v1, but I didn't feel
> comfortable adding the tags to v2 because this is far from a straight
> rebase.
>
> v2: Rebased to kvm/queue, commit fb7333dfd812 ("KVM: SVM: fix calls ...").
>
> arch/x86/kvm/vmx/nested.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
> index bcb50724be38..adb11b504d5c 100644
> --- a/arch/x86/kvm/vmx/nested.c
> +++ b/arch/x86/kvm/vmx/nested.c
> @@ -5672,7 +5672,7 @@ static bool nested_vmx_l0_wants_exit(struct kvm_vcpu *vcpu, u32 exit_reason)
> {
> u32 intr_info;
>
> - switch (exit_reason) {
> + switch ((u16)exit_reason) {
> case EXIT_REASON_EXCEPTION_NMI:
> intr_info = vmx_get_intr_info(vcpu);
> if (is_nmi(intr_info))
> @@ -5733,7 +5733,7 @@ static bool nested_vmx_l1_wants_exit(struct kvm_vcpu *vcpu, u32 exit_reason)
> struct vmcs12 *vmcs12 = get_vmcs12(vcpu);
> u32 intr_info;
>
> - switch (exit_reason) {
> + switch ((u16)exit_reason) {
> case EXIT_REASON_EXCEPTION_NMI:
> intr_info = vmx_get_intr_info(vcpu);
> if (is_nmi(intr_info))
>
Thanks, queued with git magic (committed v1 based on v5.7, merged into
kvm/queue, added the delta to v2 in the merge commit) so that stable
branches can cherry-pick cleanly.
Paolo