| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-Id: <20220113133703.1976665-4-vkuznets@redhat.com>
Date: Thu, 13 Jan 2022 14:37:01 +0100
From: Vitaly Kuznetsov <vkuznets@...hat.com>
To: kvm@...r.kernel.org, Paolo Bonzini <pbonzini@...hat.com>
Cc: Sean Christopherson <seanjc@...gle.com>,
Wanpeng Li <wanpengli@...cent.com>,
Jim Mattson <jmattson@...gle.com>,
Igor Mammedov <imammedo@...hat.com>,
linux-kernel@...r.kernel.org
Subject: [PATCH 3/5] KVM: x86: Partially allow KVM_SET_CPUID{,2} after KVM_RUN
Commit feb627e8d6f6 ("KVM: x86: Forbid KVM_SET_CPUID{,2} after KVM_RUN")
forbade changing CPUID altogether but unfortunately this is not fully
compatible with existing VMMs. In particular, QEMU reuses vCPU fds for
CPU hotplug after unplug and it calls KVM_SET_CPUID2. Checking for
full equality of the data with what's in 'vcpu->arch.cpuid_entries' is
not feasible for two reasons: the data is tweaked by KVM itself
(__kvm_update_cpuid_runtime()) and QEMU may use vCPU fd for some other
virtual CPU so it may want to change LAPIC/x2APIC ids. Relax the
requirement by implementing an allowlist of entries which are allowed
to change.
Reported-by: Igor Mammedov <imammedo@...hat.com>
Fixes: feb627e8d6f6 ("KVM: x86: Forbid KVM_SET_CPUID{,2} after KVM_RUN")
Signed-off-by: Vitaly Kuznetsov <vkuznets@...hat.com>
---
arch/x86/kvm/cpuid.c | 70 +++++++++++++++++++++++++++++++++++++++++---
arch/x86/kvm/x86.c | 19 ------------
2 files changed, 66 insertions(+), 23 deletions(-)
diff --git a/arch/x86/kvm/cpuid.c b/arch/x86/kvm/cpuid.c
index c4d4c350afbe..4a8e1c8ada6c 100644
--- a/arch/x86/kvm/cpuid.c
+++ b/arch/x86/kvm/cpuid.c
@@ -80,9 +80,11 @@ static inline struct kvm_cpuid_entry2 *cpuid_entry2_find(
return NULL;
}
-static int kvm_check_cpuid(struct kvm_cpuid_entry2 *entries, int nent)
+static int kvm_check_cpuid(struct kvm_vcpu *vcpu, struct kvm_cpuid_entry2 *entries,
+ int nent, bool is_update)
{
- struct kvm_cpuid_entry2 *best;
+ struct kvm_cpuid_entry2 *best, *e;
+ int i;
/*
* The existing code assumes virtual address is 48-bit or 57-bit in the
@@ -96,6 +98,59 @@ static int kvm_check_cpuid(struct kvm_cpuid_entry2 *entries, int nent)
return -EINVAL;
}
+ if (!is_update)
+ return 0;
+
+ /*
+ * KVM does not correctly handle changing guest CPUID after KVM_RUN, as
+ * MAXPHYADDR, GBPAGES support, AMD reserved bit behavior, etc.. aren't
+ * tracked in kvm_mmu_page_role. As a result, KVM may miss guest page
+ * faults due to reusing SPs/SPTEs. In practice no sane VMM mucks with
+ * the core vCPU model on the fly. It would've been better to forbid any
+ * KVM_SET_CPUID{,2} calls after KVM_RUN altogether but unfortunately
+ * some VMMs (e.g. QEMU) reuse vCPU fds for CPU hotplug/unplug and they
+ * need to set CPUID to e.g. change [x2]APIC id. Implement an allowlist
+ * of CPUIDs which are allowed to change.
+ */
+ for (i = 0; i < nent; i++) {
+ e = &entries[i];
+
+ best = kvm_find_cpuid_entry(vcpu, e->function, e->index);
+ if (!best)
+ return -EINVAL;
+
+ switch (e->function) {
+ case 0x1:
+ /* Only initial LAPIC id is allowed to change */
+ if (e->eax != best->eax ||
+ (e->ebx & ~0xff000000u) != (best->ebx & ~0xff000000u) ||
+ e->ecx != best->ecx || e->edx != best->edx)
+ return -EINVAL;
+ break;
+ case 0x3:
+ /* processor serial number */
+ case 0x4:
+ /* cache parameters */
+ case 0xb:
+ case 0x1f:
+ /* x2APIC id and CPU topology */
+ case 0x80000005:
+ /* AMD l1 cache information */
+ case 0x80000006:
+ /* l2 cache information */
+ case 0x8000001d:
+ /* AMD cache topology */
+ case 0x8000001e:
+ /* AMD processor topology */
+ break;
+ default:
+ if (e->eax != best->eax || e->ebx != best->ebx ||
+ e->ecx != best->ecx || e->edx != best->edx)
+ return -EINVAL;
+ break;
+ }
+ }
+
return 0;
}
@@ -290,10 +345,11 @@ static int kvm_set_cpuid(struct kvm_vcpu *vcpu, struct kvm_cpuid_entry2 *e2,
int nent)
{
int r;
+ bool is_update = vcpu->arch.last_vmentry_cpu != -1;
__kvm_update_cpuid_runtime(vcpu, e2, nent);
- r = kvm_check_cpuid(e2, nent);
+ r = kvm_check_cpuid(vcpu, e2, nent, is_update);
if (r)
return r;
@@ -302,7 +358,13 @@ static int kvm_set_cpuid(struct kvm_vcpu *vcpu, struct kvm_cpuid_entry2 *e2,
vcpu->arch.cpuid_nent = nent;
kvm_update_kvm_cpuid_base(vcpu);
- kvm_vcpu_after_set_cpuid(vcpu);
+
+ /*
+ * KVM_SET_CPUID{,2} after KVM_RUN is not allowed to change vCPU features, see
+ * kvm_check_cpuid().
+ */
+ if (!is_update)
+ kvm_vcpu_after_set_cpuid(vcpu);
return 0;
}
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 829d03fcb481..935b60c37c6d 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -5148,17 +5148,6 @@ long kvm_arch_vcpu_ioctl(struct file *filp,
struct kvm_cpuid __user *cpuid_arg = argp;
struct kvm_cpuid cpuid;
- /*
- * KVM does not correctly handle changing guest CPUID after KVM_RUN, as
- * MAXPHYADDR, GBPAGES support, AMD reserved bit behavior, etc.. aren't
- * tracked in kvm_mmu_page_role. As a result, KVM may miss guest page
- * faults due to reusing SPs/SPTEs. In practice no sane VMM mucks with
- * the core vCPU model on the fly, so fail.
- */
- r = -EINVAL;
- if (vcpu->arch.last_vmentry_cpu != -1)
- goto out;
-
r = -EFAULT;
if (copy_from_user(&cpuid, cpuid_arg, sizeof(cpuid)))
goto out;
@@ -5169,14 +5158,6 @@ long kvm_arch_vcpu_ioctl(struct file *filp,
struct kvm_cpuid2 __user *cpuid_arg = argp;
struct kvm_cpuid2 cpuid;
- /*
- * KVM_SET_CPUID{,2} after KVM_RUN is forbidded, see the comment in
- * KVM_SET_CPUID case above.
- */
- r = -EINVAL;
- if (vcpu->arch.last_vmentry_cpu != -1)
- goto out;
-
r = -EFAULT;
if (copy_from_user(&cpuid, cpuid_arg, sizeof(cpuid)))
goto out;
--
2.34.1
Powered by blists - more mailing lists