On Thu, May 20, 2021 at 1:43 PM Michael S. Tsirkin [off-list ref] wrote:

On Thu, May 20, 2021 at 01:25:16PM +0800, Yongji Xie wrote:

quoted

On Wed, May 19, 2021 at 10:42 PM Dan Carpenter [off-list ref] wrote:

quoted

On Wed, May 19, 2021 at 09:39:20PM +0800, Yongji Xie wrote:

quoted

On Mon, May 17, 2021 at 5:56 PM Xie Yongji [off-list ref] wrote:

quoted

This ensures that we will not use an invalid block size
in config space (might come from an untrusted device).

I looked at if I should add this as an untrusted function so that Smatch
could find these sorts of bugs but this is reading data from the host so
there has to be some level of trust...

It would be great if Smatch could detect this case if possible. The
data might be trusted in traditional VM cases. But now the data can be
read from a userspace daemon when VDUSE is enabled.

quoted

I should add some more untrusted data kvm functions to Smatch.  Right
now I only have kvm_register_read() and I've added kvm_read_guest_virt()
just now.

quoted

quoted

Signed-off-by: Xie Yongji <redacted>
---
 drivers/block/virtio_blk.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/block/virtio_blk.c b/drivers/block/virtio_blk.c
index ebb4d3fe803f..c848aa36d49b 100644
--- a/drivers/block/virtio_blk.c
+++ b/drivers/block/virtio_blk.c

@@ -826,7 +826,7 @@ static int virtblk_probe(struct virtio_device *vdev)
        err = virtio_cread_feature(vdev, VIRTIO_BLK_F_BLK_SIZE,
                                   struct virtio_blk_config, blk_size,
                                   &blk_size);
-       if (!err)
+       if (!err && blk_size > 0 && blk_size <= max_size)

The check here is incorrect. I will use PAGE_SIZE as the maximum
boundary in the new version.

What does this bug look like to the user?

The kernel will panic if the block size is larger than PAGE_SIZE.

Kernel panic at this point is par for the course IMHO.

But it seems better if we can avoid this kind of panic. Because this
might also be triggered by a buggy VDUSE daemon.

Let's focus on eliminating data corruption for starters.

OK, now the incorrect used length might cause data corruption in
virtio-net and virtio-console drivers as I mentioned in another mail.
I will send a fix ASAP.

Thanks,
Yongji