[PATCH 3.10 243/250] KVM: kvm_io_bus_unregister_dev() should never fail

[Willy Tarreau]

From: David Hildenbrand <david@xxxxxxxxxx>

commit 90db10434b163e46da413d34db8d0e77404cc645 upstream.

No caller currently checks the return value of
kvm_io_bus_unregister_dev(). This is evil, as all callers silently go on
freeing their device. A stale reference will remain in the io_bus,
getting at least used again, when the iobus gets teared down on
kvm_destroy_vm() - leading to use after free errors.

There is nothing the callers could do, except retrying over and over
again.

So let's simply remove the bus altogether, print an error and make
sure no one can access this broken bus again (returning -ENOMEM on any
attempt to access it).

Fixes: e93f8a0f821e ("KVM: convert io_bus to SRCU")
Reported-by: Dmitry Vyukov <dvyukov@xxxxxxxxxx>
Reviewed-by: Cornelia Huck <cornelia.huck@xxxxxxxxxx>
Signed-off-by: David Hildenbrand <david@xxxxxxxxxx>
Signed-off-by: Paolo Bonzini <pbonzini@xxxxxxxxxx>
[wt: no kvm_io_bus_read_cookie in 3.10, slightly different constructs]

Signed-off-by: Willy Tarreau <w@xxxxxx>
---
 include/linux/kvm_host.h |  4 ++--
 virt/kvm/eventfd.c       |  3 ++-
 virt/kvm/kvm_main.c      | 38 +++++++++++++++++++++-----------------
 3 files changed, 25 insertions(+), 20 deletions(-)

diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
index cbe9083..71bcaf5 100644
--- a/include/linux/kvm_host.h
+++ b/include/linux/kvm_host.h
@@ -163,8 +163,8 @@ int kvm_io_bus_read(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr, int len,
 		    void *val);
 int kvm_io_bus_register_dev(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
 			    int len, struct kvm_io_device *dev);
-int kvm_io_bus_unregister_dev(struct kvm *kvm, enum kvm_bus bus_idx,
-			      struct kvm_io_device *dev);
+void kvm_io_bus_unregister_dev(struct kvm *kvm, enum kvm_bus bus_idx,
+			       struct kvm_io_device *dev);
 
 #ifdef CONFIG_KVM_ASYNC_PF
 struct kvm_async_pf {
diff --git a/virt/kvm/eventfd.c b/virt/kvm/eventfd.c
index 1550637..362908c 100644
--- a/virt/kvm/eventfd.c
+++ b/virt/kvm/eventfd.c
@@ -799,7 +799,8 @@ kvm_deassign_ioeventfd(struct kvm *kvm, struct kvm_ioeventfd *args)
 			continue;
 
 		kvm_io_bus_unregister_dev(kvm, bus_idx, &p->dev);
-		kvm->buses[bus_idx]->ioeventfd_count--;
+		if (kvm->buses[bus_idx])
+			kvm->buses[bus_idx]->ioeventfd_count--;
 		ioeventfd_release(p);
 		ret = 0;
 		break;
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 36d14e5..0715673 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -608,7 +608,8 @@ static void kvm_destroy_vm(struct kvm *kvm)
 	raw_spin_unlock(&kvm_lock);
 	kvm_free_irq_routing(kvm);
 	for (i = 0; i < KVM_NR_BUSES; i++) {
-		kvm_io_bus_destroy(kvm->buses[i]);
+		if (kvm->buses[i])
+			kvm_io_bus_destroy(kvm->buses[i]);
 		kvm->buses[i] = NULL;
 	}
 	kvm_coalesced_mmio_free(kvm);
@@ -2887,6 +2888,8 @@ int kvm_io_bus_write(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
 	};
 
 	bus = srcu_dereference(kvm->buses[bus_idx], &kvm->srcu);
+	if (!bus)
+		return -ENOMEM;
 	idx = kvm_io_bus_get_first_dev(bus, addr, len);
 	if (idx < 0)
 		return -EOPNOTSUPP;
@@ -2915,6 +2918,8 @@ int kvm_io_bus_read(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
 	};
 
 	bus = srcu_dereference(kvm->buses[bus_idx], &kvm->srcu);
+	if (!bus)
+		return -ENOMEM;
 	idx = kvm_io_bus_get_first_dev(bus, addr, len);
 	if (idx < 0)
 		return -EOPNOTSUPP;
@@ -2936,6 +2941,9 @@ int kvm_io_bus_register_dev(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
 	struct kvm_io_bus *new_bus, *bus;
 
 	bus = kvm->buses[bus_idx];
+	if (!bus)
+		return -ENOMEM;
+
 	/* exclude ioeventfd which is limited by maximum fd */
 	if (bus->dev_count - bus->ioeventfd_count > NR_IOBUS_DEVS - 1)
 		return -ENOSPC;
@@ -2955,45 +2963,41 @@ int kvm_io_bus_register_dev(struct kvm *kvm, enum kvm_bus bus_idx, gpa_t addr,
 }
 
 /* Caller must hold slots_lock. */
-int kvm_io_bus_unregister_dev(struct kvm *kvm, enum kvm_bus bus_idx,
-			      struct kvm_io_device *dev)
+void kvm_io_bus_unregister_dev(struct kvm *kvm, enum kvm_bus bus_idx,
+			       struct kvm_io_device *dev)
 {
-	int i, r;
+	int i;
 	struct kvm_io_bus *new_bus, *bus;
 
 	bus = kvm->buses[bus_idx];
-
-	/*
-	 * It's possible the bus being released before hand. If so,
-	 * we're done here.
-	 */
 	if (!bus)
-		return 0;
+		return;
 
-	r = -ENOENT;
 	for (i = 0; i < bus->dev_count; i++)
 		if (bus->range[i].dev == dev) {
-			r = 0;
 			break;
 		}
 
-	if (r)
-		return r;
+	if (i == bus->dev_count)
+		return;
 
 	new_bus = kzalloc(sizeof(*bus) + ((bus->dev_count - 1) *
 			  sizeof(struct kvm_io_range)), GFP_KERNEL);
-	if (!new_bus)
-		return -ENOMEM;
+	if (!new_bus)  {
+		pr_err("kvm: failed to shrink bus, removing it completely\n");
+		goto broken;
+	}
 
 	memcpy(new_bus, bus, sizeof(*bus) + i * sizeof(struct kvm_io_range));
 	new_bus->dev_count--;
 	memcpy(new_bus->range + i, bus->range + i + 1,
 	       (new_bus->dev_count - i) * sizeof(struct kvm_io_range));
 
+broken:
 	rcu_assign_pointer(kvm->buses[bus_idx], new_bus);
 	synchronize_srcu_expedited(&kvm->srcu);
 	kfree(bus);
-	return r;
+	return;
 }
 
 static struct notifier_block kvm_cpu_notifier = {
-- 
2.8.0.rc2.1.gbe9624a