Save and reload guest CET supervisor states, i.e.,PL{0,1,2}_SSP,
when vCPU context is being swapped before and after userspace
<->kernel entry, also do the same operation when vCPU is sched-in
or sched-out.
Enabling CET supervisor state management in KVM due to:
-Introducing unnecessary XSAVE operation when switch to non-vCPU
userspace within current FPU framework.
-Forcing allocating additional space for CET supervisor states in
each thread context regardless whether it's vCPU thread or not.
Add a new helper kvm_arch_sched_out() for that purpose. Adding
the support in kvm_arch_vcpu_put/load() without the new helper
looks possible, but the put/load functions are also called in
vcpu_put()/load(), the latter are heavily used in KVM, so adding
new helper makes the implementation clearer.
Signed-off-by: Yang Weijiang <weijiang.yang@intel.com>
---
arch/arm64/include/asm/kvm_host.h | 1 +
arch/mips/include/asm/kvm_host.h | 1 +
arch/powerpc/include/asm/kvm_host.h | 1 +
arch/riscv/include/asm/kvm_host.h | 1 +
arch/s390/include/asm/kvm_host.h | 1 +
arch/x86/kvm/x86.c | 37 +++++++++++++++++++++++++++++
include/linux/kvm_host.h | 1 +
virt/kvm/kvm_main.c | 1 +
8 files changed, 44 insertions(+)
diff --git a/arch/arm64/include/asm/kvm_host.h b/arch/arm64/include/asm/kvm_host.h
index 7e7e19ef6993..98235cb3d258 100644
--- a/arch/arm64/include/asm/kvm_host.h
+++ b/arch/arm64/include/asm/kvm_host.h
@@ -1023,6 +1023,7 @@ void kvm_arm_vcpu_ptrauth_trap(struct kvm_vcpu *vcpu);
static inline void kvm_arch_sync_events(struct kvm *kvm) {}
static inline void kvm_arch_sched_in(struct kvm_vcpu *vcpu, int cpu) {}
+static inline void kvm_arch_sched_out(struct kvm_vcpu *vcpu, int cpu) {}
void kvm_arm_init_debug(void);
void kvm_arm_vcpu_init_debug(struct kvm_vcpu *vcpu);
diff --git a/arch/mips/include/asm/kvm_host.h b/arch/mips/include/asm/kvm_host.h
index 957121a495f0..56c5e85ba5a3 100644
--- a/arch/mips/include/asm/kvm_host.h
+++ b/arch/mips/include/asm/kvm_host.h
@@ -893,6 +893,7 @@ static inline void kvm_arch_free_memslot(struct kvm *kvm,
struct kvm_memory_slot *slot) {}
static inline void kvm_arch_memslots_updated(struct kvm *kvm, u64 gen) {}
static inline void kvm_arch_sched_in(struct kvm_vcpu *vcpu, int cpu) {}
+static inline void kvm_arch_sched_out(struct kvm_vcpu *vcpu, int cpu) {}
static inline void kvm_arch_vcpu_blocking(struct kvm_vcpu *vcpu) {}
static inline void kvm_arch_vcpu_unblocking(struct kvm_vcpu *vcpu) {}
diff --git a/arch/powerpc/include/asm/kvm_host.h b/arch/powerpc/include/asm/kvm_host.h
index 14ee0dece853..11587d953bf6 100644
--- a/arch/powerpc/include/asm/kvm_host.h
+++ b/arch/powerpc/include/asm/kvm_host.h
@@ -880,6 +880,7 @@ static inline void kvm_arch_sync_events(struct kvm *kvm) {}
static inline void kvm_arch_memslots_updated(struct kvm *kvm, u64 gen) {}
static inline void kvm_arch_flush_shadow_all(struct kvm *kvm) {}
static inline void kvm_arch_sched_in(struct kvm_vcpu *vcpu, int cpu) {}
+static inline void kvm_arch_sched_out(struct kvm_vcpu *vcpu, int cpu) {}
static inline void kvm_arch_vcpu_blocking(struct kvm_vcpu *vcpu) {}
static inline void kvm_arch_vcpu_unblocking(struct kvm_vcpu *vcpu) {}
diff --git a/arch/riscv/include/asm/kvm_host.h b/arch/riscv/include/asm/kvm_host.h
index ee0acccb1d3b..6ff4a04fe0f2 100644
--- a/arch/riscv/include/asm/kvm_host.h
+++ b/arch/riscv/include/asm/kvm_host.h
@@ -244,6 +244,7 @@ struct kvm_vcpu_arch {
static inline void kvm_arch_sync_events(struct kvm *kvm) {}
static inline void kvm_arch_sched_in(struct kvm_vcpu *vcpu, int cpu) {}
+static inline void kvm_arch_sched_out(struct kvm_vcpu *vcpu, int cpu) {}
#define KVM_ARCH_WANT_MMU_NOTIFIER
diff --git a/arch/s390/include/asm/kvm_host.h b/arch/s390/include/asm/kvm_host.h
index 2bbc3d54959d..d1750a6a86cf 100644
--- a/arch/s390/include/asm/kvm_host.h
+++ b/arch/s390/include/asm/kvm_host.h
@@ -1033,6 +1033,7 @@ extern int kvm_s390_gisc_unregister(struct kvm *kvm, u32 gisc);
static inline void kvm_arch_sync_events(struct kvm *kvm) {}
static inline void kvm_arch_sched_in(struct kvm_vcpu *vcpu, int cpu) {}
+static inline void kvm_arch_sched_out(struct kvm_vcpu *vcpu, int cpu) {}
static inline void kvm_arch_free_memslot(struct kvm *kvm,
struct kvm_memory_slot *slot) {}
static inline void kvm_arch_memslots_updated(struct kvm *kvm, u64 gen) {}
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 7a3753c05c09..f7558f0f6fc0 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -11212,6 +11212,33 @@ static void kvm_put_guest_fpu(struct kvm_vcpu *vcpu)
trace_kvm_fpu(0);
}
+static void kvm_save_cet_supervisor_ssp(struct kvm_vcpu *vcpu)
+{
+ preempt_disable();
+ if (unlikely(guest_can_use(vcpu, X86_FEATURE_SHSTK))) {
+ rdmsrl(MSR_IA32_PL0_SSP, vcpu->arch.cet_s_ssp[0]);
+ rdmsrl(MSR_IA32_PL1_SSP, vcpu->arch.cet_s_ssp[1]);
+ rdmsrl(MSR_IA32_PL2_SSP, vcpu->arch.cet_s_ssp[2]);
+ /*
+ * Omit reset to host PL{1,2}_SSP because Linux will never use
+ * these MSRs.
+ */
+ wrmsrl(MSR_IA32_PL0_SSP, 0);
+ }
+ preempt_enable();
+}
+
+static void kvm_reload_cet_supervisor_ssp(struct kvm_vcpu *vcpu)
+{
+ preempt_disable();
+ if (unlikely(guest_can_use(vcpu, X86_FEATURE_SHSTK))) {
+ wrmsrl(MSR_IA32_PL0_SSP, vcpu->arch.cet_s_ssp[0]);
+ wrmsrl(MSR_IA32_PL1_SSP, vcpu->arch.cet_s_ssp[1]);
+ wrmsrl(MSR_IA32_PL2_SSP, vcpu->arch.cet_s_ssp[2]);
+ }
+ preempt_enable();
+}
+
int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
{
struct kvm_queued_exception *ex = &vcpu->arch.exception;
@@ -11222,6 +11249,7 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
kvm_sigset_activate(vcpu);
kvm_run->flags = 0;
kvm_load_guest_fpu(vcpu);
+ kvm_reload_cet_supervisor_ssp(vcpu);
kvm_vcpu_srcu_read_lock(vcpu);
if (unlikely(vcpu->arch.mp_state == KVM_MP_STATE_UNINITIALIZED)) {
@@ -11310,6 +11338,7 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
r = vcpu_run(vcpu);
out:
+ kvm_save_cet_supervisor_ssp(vcpu);
kvm_put_guest_fpu(vcpu);
if (kvm_run->kvm_valid_regs)
store_regs(vcpu);
@@ -12398,9 +12427,17 @@ void kvm_arch_sched_in(struct kvm_vcpu *vcpu, int cpu)
pmu->need_cleanup = true;
kvm_make_request(KVM_REQ_PMU, vcpu);
}
+
+ kvm_reload_cet_supervisor_ssp(vcpu);
+
static_call(kvm_x86_sched_in)(vcpu, cpu);
}
+void kvm_arch_sched_out(struct kvm_vcpu *vcpu, int cpu)
+{
+ kvm_save_cet_supervisor_ssp(vcpu);
+}
+
void kvm_arch_free_vm(struct kvm *kvm)
{
kfree(to_kvm_hv(kvm)->hv_pa_pg);
diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
index d90331f16db1..b3032a5f0641 100644
--- a/include/linux/kvm_host.h
+++ b/include/linux/kvm_host.h
@@ -1423,6 +1423,7 @@ int kvm_arch_vcpu_ioctl_set_guest_debug(struct kvm_vcpu *vcpu,
int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu);
void kvm_arch_sched_in(struct kvm_vcpu *vcpu, int cpu);
+void kvm_arch_sched_out(struct kvm_vcpu *vcpu, int cpu);
void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu);
void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu);
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 66c1447d3c7f..42f28e8905e1 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -5885,6 +5885,7 @@ static void kvm_sched_out(struct preempt_notifier *pn,
{
struct kvm_vcpu *vcpu = preempt_notifier_to_vcpu(pn);
+ kvm_arch_sched_out(vcpu, 0);
if (current->on_rq) {
WRITE_ONCE(vcpu->preempted, true);
WRITE_ONCE(vcpu->ready, true);
--
2.27.0On Thu, Jul 20, 2023 at 11:03:42PM -0400, Yang Weijiang wrote:
>+static void kvm_save_cet_supervisor_ssp(struct kvm_vcpu *vcpu)
>+{
>+ preempt_disable();
what's the purpose of disabling preemption?
>+ if (unlikely(guest_can_use(vcpu, X86_FEATURE_SHSTK))) {
>+ rdmsrl(MSR_IA32_PL0_SSP, vcpu->arch.cet_s_ssp[0]);
>+ rdmsrl(MSR_IA32_PL1_SSP, vcpu->arch.cet_s_ssp[1]);
>+ rdmsrl(MSR_IA32_PL2_SSP, vcpu->arch.cet_s_ssp[2]);
>+ /*
>+ * Omit reset to host PL{1,2}_SSP because Linux will never use
>+ * these MSRs.
>+ */
>+ wrmsrl(MSR_IA32_PL0_SSP, 0);
You don't need to reset the MSR because current host doesn't enable SSS
and leaving guest value in the MSR won't affect host behavior.
>+ }
>+ preempt_enable();
>+}
>+
>+static void kvm_reload_cet_supervisor_ssp(struct kvm_vcpu *vcpu)
>+{
>+ preempt_disable();
>+ if (unlikely(guest_can_use(vcpu, X86_FEATURE_SHSTK))) {
>+ wrmsrl(MSR_IA32_PL0_SSP, vcpu->arch.cet_s_ssp[0]);
>+ wrmsrl(MSR_IA32_PL1_SSP, vcpu->arch.cet_s_ssp[1]);
>+ wrmsrl(MSR_IA32_PL2_SSP, vcpu->arch.cet_s_ssp[2]);
>+ }
>+ preempt_enable();
>+}
save/load PLx_SSP in kvm_sched_in/out() and in VCPU_RUN ioctl is sub-optimal.
How about:
1. expose kvm_save/reload_cet_supervisor_ssp()
2. reload guest PLx_SSP in {vmx,svm}_prepare_switch_to_guest()
3. save guest PLx_SSP in vmx_prepare_switch_to_host() and
svm_prepare_host_switch()?
this way existing svm/vmx->guest_state_loaded can help to reduce a lot of
unnecessary PLx_SSP MSR accesses.
>+
> int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
> {
> struct kvm_queued_exception *ex = &vcpu->arch.exception;
>@@ -11222,6 +11249,7 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
> kvm_sigset_activate(vcpu);
> kvm_run->flags = 0;
> kvm_load_guest_fpu(vcpu);
>+ kvm_reload_cet_supervisor_ssp(vcpu);
>
> kvm_vcpu_srcu_read_lock(vcpu);
> if (unlikely(vcpu->arch.mp_state == KVM_MP_STATE_UNINITIALIZED)) {
>@@ -11310,6 +11338,7 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
> r = vcpu_run(vcpu);
>
> out:
>+ kvm_save_cet_supervisor_ssp(vcpu);
> kvm_put_guest_fpu(vcpu);
> if (kvm_run->kvm_valid_regs)
> store_regs(vcpu);
>@@ -12398,9 +12427,17 @@ void kvm_arch_sched_in(struct kvm_vcpu *vcpu, int cpu)
> pmu->need_cleanup = true;
> kvm_make_request(KVM_REQ_PMU, vcpu);
> }
>+
>+ kvm_reload_cet_supervisor_ssp(vcpu);
>+
> static_call(kvm_x86_sched_in)(vcpu, cpu);
> }
>
>+void kvm_arch_sched_out(struct kvm_vcpu *vcpu, int cpu)
>+{
@cpu its meaning isn't clear and isn't used and ...
>+ kvm_save_cet_supervisor_ssp(vcpu);
>+}
>+
> void kvm_arch_free_vm(struct kvm *kvm)
> {
> kfree(to_kvm_hv(kvm)->hv_pa_pg);
>diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
>index d90331f16db1..b3032a5f0641 100644
>--- a/include/linux/kvm_host.h
>+++ b/include/linux/kvm_host.h
>@@ -1423,6 +1423,7 @@ int kvm_arch_vcpu_ioctl_set_guest_debug(struct kvm_vcpu *vcpu,
> int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu);
>
> void kvm_arch_sched_in(struct kvm_vcpu *vcpu, int cpu);
>+void kvm_arch_sched_out(struct kvm_vcpu *vcpu, int cpu);
>
> void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu);
> void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu);
>diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
>index 66c1447d3c7f..42f28e8905e1 100644
>--- a/virt/kvm/kvm_main.c
>+++ b/virt/kvm/kvm_main.c
>@@ -5885,6 +5885,7 @@ static void kvm_sched_out(struct preempt_notifier *pn,
> {
> struct kvm_vcpu *vcpu = preempt_notifier_to_vcpu(pn);
>
>+ kvm_arch_sched_out(vcpu, 0);
passing 0 always looks problematic.
> if (current->on_rq) {
> WRITE_ONCE(vcpu->preempted, true);
> WRITE_ONCE(vcpu->ready, true);
>--
>2.27.0
>
On 7/24/2023 4:26 PM, Chao Gao wrote:
> On Thu, Jul 20, 2023 at 11:03:42PM -0400, Yang Weijiang wrote:
>> +static void kvm_save_cet_supervisor_ssp(struct kvm_vcpu *vcpu)
>> +{
>> + preempt_disable();
> what's the purpose of disabling preemption?
Thanks!
These preempt_disable/enable() becomes unnecessary due to the PLx_SSP
handling
in sched_in/out(). Will remove them.
>
>> + if (unlikely(guest_can_use(vcpu, X86_FEATURE_SHSTK))) {
>> + rdmsrl(MSR_IA32_PL0_SSP, vcpu->arch.cet_s_ssp[0]);
>> + rdmsrl(MSR_IA32_PL1_SSP, vcpu->arch.cet_s_ssp[1]);
>> + rdmsrl(MSR_IA32_PL2_SSP, vcpu->arch.cet_s_ssp[2]);
>> + /*
>> + * Omit reset to host PL{1,2}_SSP because Linux will never use
>> + * these MSRs.
>> + */
>> + wrmsrl(MSR_IA32_PL0_SSP, 0);
> You don't need to reset the MSR because current host doesn't enable SSS
> and leaving guest value in the MSR won't affect host behavior.
Yes, I just want to make the host PLx_SSPs as clean as possible.
>
>> + }
>> + preempt_enable();
>> +}
>> +
>> +static void kvm_reload_cet_supervisor_ssp(struct kvm_vcpu *vcpu)
>> +{
>> + preempt_disable();
>> + if (unlikely(guest_can_use(vcpu, X86_FEATURE_SHSTK))) {
>> + wrmsrl(MSR_IA32_PL0_SSP, vcpu->arch.cet_s_ssp[0]);
>> + wrmsrl(MSR_IA32_PL1_SSP, vcpu->arch.cet_s_ssp[1]);
>> + wrmsrl(MSR_IA32_PL2_SSP, vcpu->arch.cet_s_ssp[2]);
>> + }
>> + preempt_enable();
>> +}
> save/load PLx_SSP in kvm_sched_in/out() and in VCPU_RUN ioctl is sub-optimal.
>
> How about:
> 1. expose kvm_save/reload_cet_supervisor_ssp()
> 2. reload guest PLx_SSP in {vmx,svm}_prepare_switch_to_guest()
> 3. save guest PLx_SSP in vmx_prepare_switch_to_host() and
> svm_prepare_host_switch()?
>
> this way existing svm/vmx->guest_state_loaded can help to reduce a lot of
> unnecessary PLx_SSP MSR accesses.
Nice suggestion! It looks workable. I'll try this, thanks!
>
>> +
>> int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
>> {
>> struct kvm_queued_exception *ex = &vcpu->arch.exception;
>> @@ -11222,6 +11249,7 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
>> kvm_sigset_activate(vcpu);
>> kvm_run->flags = 0;
>> kvm_load_guest_fpu(vcpu);
>> + kvm_reload_cet_supervisor_ssp(vcpu);
>>
>> kvm_vcpu_srcu_read_lock(vcpu);
>> if (unlikely(vcpu->arch.mp_state == KVM_MP_STATE_UNINITIALIZED)) {
>> @@ -11310,6 +11338,7 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
>> r = vcpu_run(vcpu);
>>
>> out:
>> + kvm_save_cet_supervisor_ssp(vcpu);
>> kvm_put_guest_fpu(vcpu);
>> if (kvm_run->kvm_valid_regs)
>> store_regs(vcpu);
>> @@ -12398,9 +12427,17 @@ void kvm_arch_sched_in(struct kvm_vcpu *vcpu, int cpu)
>> pmu->need_cleanup = true;
>> kvm_make_request(KVM_REQ_PMU, vcpu);
>> }
>> +
>> + kvm_reload_cet_supervisor_ssp(vcpu);
>> +
>> static_call(kvm_x86_sched_in)(vcpu, cpu);
>> }
>>
>> +void kvm_arch_sched_out(struct kvm_vcpu *vcpu, int cpu)
>> +{
> @cpu its meaning isn't clear and isn't used and ...
Yes, I should have removed it.
>
>> + kvm_save_cet_supervisor_ssp(vcpu);
>> +}
>> +
>> void kvm_arch_free_vm(struct kvm *kvm)
>> {
>> kfree(to_kvm_hv(kvm)->hv_pa_pg);
>> diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
>> index d90331f16db1..b3032a5f0641 100644
>> --- a/include/linux/kvm_host.h
>> +++ b/include/linux/kvm_host.h
>> @@ -1423,6 +1423,7 @@ int kvm_arch_vcpu_ioctl_set_guest_debug(struct kvm_vcpu *vcpu,
>> int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu);
>>
>> void kvm_arch_sched_in(struct kvm_vcpu *vcpu, int cpu);
>> +void kvm_arch_sched_out(struct kvm_vcpu *vcpu, int cpu);
>>
>> void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu);
>> void kvm_arch_vcpu_put(struct kvm_vcpu *vcpu);
>> diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
>> index 66c1447d3c7f..42f28e8905e1 100644
>> --- a/virt/kvm/kvm_main.c
>> +++ b/virt/kvm/kvm_main.c
>> @@ -5885,6 +5885,7 @@ static void kvm_sched_out(struct preempt_notifier *pn,
>> {
>> struct kvm_vcpu *vcpu = preempt_notifier_to_vcpu(pn);
>>
>> + kvm_arch_sched_out(vcpu, 0);
> passing 0 always looks problematic.
Can you elaborate? I have no intent to use @cpu now.
>> if (current->on_rq) {
>> WRITE_ONCE(vcpu->preempted, true);
>> WRITE_ONCE(vcpu->ready, true);
>> --
>> 2.27.0
>>
© 2016 - 2025 Red Hat, Inc.