]> rtime.felk.cvut.cz Git - lisovros/linux_canprio.git/commitdiff
KVM: x86: fix information leak to userland
authorVasiliy Kulikov <segooon@gmail.com>
Sat, 30 Oct 2010 18:54:47 +0000 (22:54 +0400)
committerGreg Kroah-Hartman <gregkh@suse.de>
Thu, 9 Dec 2010 21:32:45 +0000 (13:32 -0800)
commit 97e69aa62f8b5d338d6cff49be09e37cc1262838 upstream.

Structures kvm_vcpu_events, kvm_debugregs, kvm_pit_state2 and
kvm_clock_data are copied to userland with some padding and reserved
fields unitialized.  It leads to leaking of contents of kernel stack
memory.  We have to initialize them to zero.

In patch v1 Jan Kiszka suggested to fill reserved fields with zeros
instead of memset'ting the whole struct.  It makes sense as these
fields are explicitly marked as padding.  No more fields need zeroing.

Signed-off-by: Vasiliy Kulikov <segooon@gmail.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
arch/x86/kvm/x86.c

index acc64063c89d83eb81b7350c8142f8415d6ba187..3f2d2b3b4d852eaa5683431952035901879bdd67 100644 (file)
@@ -2305,6 +2305,7 @@ static void kvm_vcpu_ioctl_x86_get_vcpu_events(struct kvm_vcpu *vcpu,
                !kvm_exception_is_soft(vcpu->arch.exception.nr);
        events->exception.nr = vcpu->arch.exception.nr;
        events->exception.has_error_code = vcpu->arch.exception.has_error_code;
+       events->exception.pad = 0;
        events->exception.error_code = vcpu->arch.exception.error_code;
 
        events->interrupt.injected =
@@ -2318,12 +2319,14 @@ static void kvm_vcpu_ioctl_x86_get_vcpu_events(struct kvm_vcpu *vcpu,
        events->nmi.injected = vcpu->arch.nmi_injected;
        events->nmi.pending = vcpu->arch.nmi_pending;
        events->nmi.masked = kvm_x86_ops->get_nmi_mask(vcpu);
+       events->nmi.pad = 0;
 
        events->sipi_vector = vcpu->arch.sipi_vector;
 
        events->flags = (KVM_VCPUEVENT_VALID_NMI_PENDING
                         | KVM_VCPUEVENT_VALID_SIPI_VECTOR
                         | KVM_VCPUEVENT_VALID_SHADOW);
+       memset(&events->reserved, 0, sizeof(events->reserved));
 }
 
 static int kvm_vcpu_ioctl_x86_set_vcpu_events(struct kvm_vcpu *vcpu,
@@ -2366,6 +2369,7 @@ static void kvm_vcpu_ioctl_x86_get_debugregs(struct kvm_vcpu *vcpu,
        dbgregs->dr6 = vcpu->arch.dr6;
        dbgregs->dr7 = vcpu->arch.dr7;
        dbgregs->flags = 0;
+       memset(&dbgregs->reserved, 0, sizeof(dbgregs->reserved));
 }
 
 static int kvm_vcpu_ioctl_x86_set_debugregs(struct kvm_vcpu *vcpu,
@@ -2849,6 +2853,7 @@ static int kvm_vm_ioctl_get_pit2(struct kvm *kvm, struct kvm_pit_state2 *ps)
                sizeof(ps->channels));
        ps->flags = kvm->arch.vpit->pit_state.flags;
        mutex_unlock(&kvm->arch.vpit->pit_state.lock);
+       memset(&ps->reserved, 0, sizeof(ps->reserved));
        return r;
 }
 
@@ -3229,6 +3234,7 @@ long kvm_arch_vm_ioctl(struct file *filp,
                now_ns = timespec_to_ns(&now);
                user_ns.clock = kvm->arch.kvmclock_offset + now_ns;
                user_ns.flags = 0;
+               memset(&user_ns.pad, 0, sizeof(user_ns.pad));
 
                r = -EFAULT;
                if (copy_to_user(argp, &user_ns, sizeof(user_ns)))