Commit 678e315e authored by Sean Christopherson's avatar Sean Christopherson Committed by Paolo Bonzini
Browse files

KVM: vmx: add dedicated utility to access guest's kernel_gs_base 



When lazy save/restore of MSR_KERNEL_GS_BASE was introduced[1], the
MSR was intercepted in all modes and was only restored for the host
when the guest is in 64-bit mode.  So at the time, going through the
full host restore prior to accessing MSR_KERNEL_GS_BASE was necessary
to load host state and was not a significant waste of cycles.

Later, MSR_KERNEL_GS_BASE interception was disabled for a 64-bit
guest[2], and then unconditionally saved/restored for the host[3].
As a result, loading full host state is overkill for accesses to
MSR_KERNEL_GS_BASE, and completely unnecessary when the guest is
not in 64-bit mode.

Add a dedicated utility to read/write the guest's MSR_KERNEL_GS_BASE
(outside of the save/restore flow) to minimize the overhead incurred
when accessing the MSR.  When setting EFER, only decache the MSR if
the new EFER will disable long mode.

Removing out-of-band usage of vmx_load_host_state() also eliminates,
or at least reduces, potential corner cases in its usage, which in
turn will (hopefuly) make it easier to reason about future changes
to the save/restore flow, e.g. optimization of saving host state.

[1] commit 44ea2b17 ("KVM: VMX: Move MSR_KERNEL_GS_BASE out of the vmx
                                    autoload msr area")
[2] commit 5897297b ("KVM: VMX: Don't intercept MSR_KERNEL_GS_BASE")
[3] commit c8770e7b ("KVM: VMX: Fix host userspace gsbase corruption")

Signed-off-by: default avatarSean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: default avatarPaolo Bonzini <pbonzini@redhat.com>
parent bd9966de

arch/x86/kvm/vmx.c

+35 −11
Original line number Diff line number Diff line
@@ -2772,12 +2772,30 @@ static void __vmx_load_host_state(struct vcpu_vmx *vmx) 
	load_fixmap_gdt(raw_smp_processor_id());

}



static void vmx_load_host_state(struct vcpu_vmx *vmx)

#ifdef CONFIG_X86_64

static u64 vmx_read_guest_kernel_gs_base(struct vcpu_vmx *vmx)

{

	if (is_long_mode(&vmx->vcpu)) {

		preempt_disable();

	__vmx_load_host_state(vmx);

		if (vmx->loaded_cpu_state)

			rdmsrl(MSR_KERNEL_GS_BASE,

			       vmx->msr_guest_kernel_gs_base);

		preempt_enable();

	}

	return vmx->msr_guest_kernel_gs_base;

}



static void vmx_write_guest_kernel_gs_base(struct vcpu_vmx *vmx, u64 data)

{

	if (is_long_mode(&vmx->vcpu)) {

		preempt_disable();

		if (vmx->loaded_cpu_state)

			wrmsrl(MSR_KERNEL_GS_BASE, data);

		preempt_enable();

	}

	vmx->msr_guest_kernel_gs_base = data;

}

#endif



static void vmx_vcpu_pi_load(struct kvm_vcpu *vcpu, int cpu)

{
@@ -3857,8 +3875,7 @@ static int vmx_get_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info) 
		msr_info->data = vmcs_readl(GUEST_GS_BASE);

		break;

	case MSR_KERNEL_GS_BASE:

		vmx_load_host_state(vmx);

		msr_info->data = vmx->msr_guest_kernel_gs_base;

		msr_info->data = vmx_read_guest_kernel_gs_base(vmx);

		break;

#endif

	case MSR_EFER:
@@ -3958,8 +3975,7 @@ static int vmx_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info) 
		vmcs_writel(GUEST_GS_BASE, data);

		break;

	case MSR_KERNEL_GS_BASE:

		vmx_load_host_state(vmx);

		vmx->msr_guest_kernel_gs_base = data;

		vmx_write_guest_kernel_gs_base(vmx, data);

		break;

#endif

	case MSR_IA32_SYSENTER_CS:
@@ -4849,10 +4865,18 @@ static void vmx_set_efer(struct kvm_vcpu *vcpu, u64 efer) 
		return;



	/*

	 * Force kernel_gs_base reloading before EFER changes, as control

	 * of this msr depends on is_long_mode().

	 * MSR_KERNEL_GS_BASE is not intercepted when the guest is in

	 * 64-bit mode as a 64-bit kernel may frequently access the

	 * MSR.  This means we need to manually save/restore the MSR

	 * when switching between guest and host state, but only if

	 * the guest is in 64-bit mode.  Sync our cached value if the

	 * guest is transitioning to 32-bit mode and the CPU contains

	 * guest state, i.e. the cache is stale.

	 */

	vmx_load_host_state(to_vmx(vcpu));

#ifdef CONFIG_X86_64

	if (!(efer & EFER_LMA))

		(void)vmx_read_guest_kernel_gs_base(vmx);

#endif

	vcpu->arch.efer = efer;

	if (efer & EFER_LMA) {

		vm_entry_controls_setbit(to_vmx(vcpu), VM_ENTRY_IA32E_MODE);

CRA Git | Maintained and supported by SUSTech CRA and CCSE