Skip to contentSkip to conversation

KVM: SVM: SEV{-ES} bug fixes

PATCH Applied Sean Christopherson 2 participants 5 messages Mar 31, 2021

Subsystems (3)
lore
Share & tools
lore

Sean Christopherson

author maintainer
#0 lore raw

cover KVM: SVM: SEV{-ES} bug fixes

Misc bug fixes in SEV/SEV-ES to protect against a malicious userspace. All found by inspection, I didn't actually crash the host to to prove that userspace could hose the kernel in any of these cases. Boot tested an SEV guest, though the SEV-ES side of patch 2 is essentially untested as I don't have an SEV-ES setup at this time.

Sean Christopherson (3):
KVM: SVM: Use online_vcpus, not created_vcpus, to iterate over vCPUs
KVM: SVM: Do not set sev->es_active until KVM_SEV_ES_INIT completes
KVM: SVM: Do not allow SEV/SEV-ES initialization after vCPUs are
created

 arch/x86/kvm/svm/sev.c | 37 ++++++++++++++++++-------------------
 1 file changed, 18 insertions(+), 19 deletions(-)
Signature · 1 line
2.31.0.291.g576ba9dcdaf-goog
author maintainer
#1 lore raw

PATCH 1/3 KVM: SVM: Use online_vcpus, not created_vcpus, to iterate over vCPUs

Use the kvm_for_each_vcpu() helper to iterate over vCPUs when encrypting VMSAs for SEV, which effectively switches to use online_vcpus instead of created_vcpus. This fixes a possible null-pointer dereference as created_vcpus does not guarantee a vCPU exists, since it is updated at the very beginning of KVM_CREATE_VCPU. created_vcpus exists to allow the bulk of vCPU creation to run in parallel, while still correctly restricting the max number of max vCPUs.

Fixes ad73109ae7ec ("KVM: SVM: Provide support to launch and run an SEV-ES guest")
Cc Brijesh Singh <[email protected]>
Cc Tom Lendacky <[email protected]>
---
 arch/x86/kvm/svm/sev.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
arch/x86/kvm/svm/sev.c +3 -2
index 83e00e524513..6481d7165701 100644
@@ -564,6 +564,7 @@ static int sev_launch_update_vmsa(struct kvm *kvm, struct kvm_sev_cmd *argp)
 {
 	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
 	struct sev_data_launch_update_vmsa *vmsa;
+	struct kvm_vcpu *vcpu;
 	int i, ret;
 
 	if (!sev_es_guest(kvm))
@@ -573,8 +574,8 @@ static int sev_launch_update_vmsa(struct kvm *kvm, struct kvm_sev_cmd *argp)
 	if (!vmsa)
 		return -ENOMEM;
 
-	for (i = 0; i < kvm->created_vcpus; i++) {
-		struct vcpu_svm *svm = to_svm(kvm->vcpus[i]);
+	kvm_for_each_vcpu(i, vcpu, kvm) {
+		struct vcpu_svm *svm = to_svm(vcpu);
 
 		/* Perform some pre-encryption checks against the VMSA */
 		ret = sev_es_sync_vmsa(svm);
Signature · 1 line
2.31.0.291.g576ba9dcdaf-goog
author maintainer
#2 lore raw

PATCH 2/3 KVM: SVM: Do not set sev->es_active until KVM_SEV_ES_INIT completes

In reply to Sean Christopherson

Set sev->es_active only after the guts of KVM_SEV_ES_INIT succeeds. If the command fails, e.g. because SEV is already active or there are no available ASIDs, then es_active will be left set even though the VM is not fully SEV-ES capable.

Refactor the code so that "es_active" is passed on the stack instead of being prematurely shoved into sev_info, both to avoid having to unwind sev_info and so that it's more obvious what actually consumes es_active in sev_guest_init() and its helpers.

Fixes ad73109ae7ec ("KVM: SVM: Provide support to launch and run an SEV-ES guest")
Cc Brijesh Singh <[email protected]>
Cc Tom Lendacky <[email protected]>
---
 arch/x86/kvm/svm/sev.c | 29 ++++++++++++-----------------
 1 file changed, 12 insertions(+), 17 deletions(-)
arch/x86/kvm/svm/sev.c +12 -17
index 6481d7165701..97d42a007b86 100644
@@ -87,7 +87,7 @@ static bool __sev_recycle_asids(int min_asid, int max_asid)
 	return true;
 }
 
-static int sev_asid_new(struct kvm_sev_info *sev)
+static int sev_asid_new(bool es_active)
 {
 	int pos, min_asid, max_asid;
 	bool retry = true;
@@ -98,8 +98,8 @@ static int sev_asid_new(struct kvm_sev_info *sev)
 	 * SEV-enabled guests must use asid from min_sev_asid to max_sev_asid.
 	 * SEV-ES-enabled guest can use from 1 to min_sev_asid - 1.
 	 */
-	min_asid = sev->es_active ? 0 : min_sev_asid - 1;
-	max_asid = sev->es_active ? min_sev_asid - 1 : max_sev_asid;
+	min_asid = es_active ? 0 : min_sev_asid - 1;
+	max_asid = es_active ? min_sev_asid - 1 : max_sev_asid;
 again:
 	pos = find_next_zero_bit(sev_asid_bitmap, max_sev_asid, min_asid);
 	if (pos >= max_asid) {
@@ -179,13 +179,14 @@ static void sev_unbind_asid(struct kvm *kvm, unsigned int handle)
 static int sev_guest_init(struct kvm *kvm, struct kvm_sev_cmd *argp)
 {
 	struct kvm_sev_info *sev = &to_kvm_svm(kvm)->sev_info;
+	bool es_active = argp->id == KVM_SEV_ES_INIT;
 	int asid, ret;
 
 	ret = -EBUSY;
 	if (unlikely(sev->active))
 		return ret;
 
-	asid = sev_asid_new(sev);
+	asid = sev_asid_new(es_active);
 	if (asid < 0)
 		return ret;
 
@@ -194,6 +195,7 @@ static int sev_guest_init(struct kvm *kvm, struct kvm_sev_cmd *argp)
 		goto e_free;
 
 	sev->active = true;
+	sev->es_active = es_active;
 	sev->asid = asid;
 	INIT_LIST_HEAD(&sev->regions_list);
 
@@ -204,16 +206,6 @@ static int sev_guest_init(struct kvm *kvm, struct kvm_sev_cmd *argp)
 	return ret;
 }
 
-static int sev_es_guest_init(struct kvm *kvm, struct kvm_sev_cmd *argp)
-{
-	if (!sev_es)
-		return -ENOTTY;
-
-	to_kvm_svm(kvm)->sev_info.es_active = true;
-
-	return sev_guest_init(kvm, argp);
-}
-
 static int sev_bind_asid(struct kvm *kvm, unsigned int handle, int *error)
 {
 	struct sev_data_activate *data;
@@ -1128,12 +1120,15 @@ int svm_mem_enc_op(struct kvm *kvm, void __user *argp)
 	mutex_lock(&kvm->lock);
 
 	switch (sev_cmd.id) {
+	case KVM_SEV_ES_INIT:
+		if (!sev_es) {
+			r = -ENOTTY;
+			goto out;
+		}
+		fallthrough;
 	case KVM_SEV_INIT:
 		r = sev_guest_init(kvm, &sev_cmd);
 		break;
-	case KVM_SEV_ES_INIT:
-		r = sev_es_guest_init(kvm, &sev_cmd);
-		break;
 	case KVM_SEV_LAUNCH_START:
 		r = sev_launch_start(kvm, &sev_cmd);
 		break;
Signature · 1 line
2.31.0.291.g576ba9dcdaf-goog
author maintainer
#3 lore raw

PATCH 3/3 KVM: SVM: Do not allow SEV/SEV-ES initialization after vCPUs are created

In reply to Sean Christopherson

Reject KVM_SEV_INIT and KVM_SEV_ES_INIT if they are attempted after one or more vCPUs have been created. KVM assumes a VM is tagged SEV/SEV-ES prior to vCPU creation, e.g. init_vmcb() needs to mark the VMCB as SEV enabled, and svm_create_vcpu() needs to allocate the VMSA. At best, creating vCPUs before SEV/SEV-ES init will lead to unexpected errors and/or behavior, and at worst it will crash the host, e.g. sev_launch_update_vmsa() will dereference a null svm->vmsa pointer.

Fixes 1654efcbc431 ("KVM: SVM: Add KVM_SEV_INIT command")
Fixes ad73109ae7ec ("KVM: SVM: Provide support to launch and run an SEV-ES guest")
Cc Brijesh Singh <[email protected]>
Cc Tom Lendacky <[email protected]>
---
 arch/x86/kvm/svm/sev.c | 3 +++
 1 file changed, 3 insertions(+)
arch/x86/kvm/svm/sev.c +3 -0
index 97d42a007b86..824bc7d22e77 100644
@@ -182,6 +182,9 @@ static int sev_guest_init(struct kvm *kvm, struct kvm_sev_cmd *argp)
 	bool es_active = argp->id == KVM_SEV_ES_INIT;
 	int asid, ret;
 
+	if (kvm->created_vcpus)
+		return -EINVAL;
+
 	ret = -EBUSY;
 	if (unlikely(sev->active))
 		return ret;
Signature · 1 line
2.31.0.291.g576ba9dcdaf-goog
maintainer Applied
#4 lore raw
In reply to Sean Christopherson
Quoted 6 lines · Sean Christopherson

Misc bug fixes in SEV/SEV-ES to protect against a malicious userspace. All found by inspection, I didn't actually crash the host to to prove that userspace could hose the kernel in any of these cases. Boot tested an SEV guest, though the SEV-ES side of patch 2 is essentially untested as I don't have an SEV-ES setup at this time.

Quoted 9 lines

Sean Christopherson (3):
KVM: SVM: Use online_vcpus, not created_vcpus, to iterate over vCPUs
KVM: SVM: Do not set sev->es_active until KVM_SEV_ES_INIT completes
KVM: SVM: Do not allow SEV/SEV-ES initialization after vCPUs are
created

  arch/x86/kvm/svm/sev.c | 37 ++++++++++++++++++-------------------
  1 file changed, 18 insertions(+), 19 deletions(-)

Queued, thanks.

Paolo