Xen: integer overflow in xen-netback xenvif_set_hash_mapping 




The xen-netback linux kernel module is the default backend for Xen's virtual network devices. Since commit <a href="https://crrev.com/40d8abdee806d496a60ee607a6d01b1cd7fabaf0" title="" class="" rel="nofollow">40d8abdee806d496a60ee607a6d01b1cd7fabaf0</a> the backend supports an additional control protocol described in (include/xen/interface/io/netif.h) to enable 
receive side scaling (RSS) for Windows guests. 

When a guest sends a control request with the type XEN_NETIF_CTRL_TYPE_SET_HASH_MAPPING to the backend, the function xenvif_set_hash_mapping is called with guest controlled values in gref, off and len:

u32 xenvif_set_hash_mapping(struct xenvif *vif, u32 gref, u32 len,
			    u32 off)
{
	u32 *mapping = &vif->hash.mapping[off];
	struct gnttab_copy copy_op = {
		.source.u.ref = gref,
		.source.domid = vif->domid,
		.dest.u.gmfn = virt_to_gfn(mapping),
		.dest.domid = DOMID_SELF,
		.dest.offset = xen_offset_in_page(mapping),
		.len = len * sizeof(u32),
		.flags = GNTCOPY_source_gref
	};

(A)	if ((off + len > vif->hash.size) || copy_op.len > XEN_PAGE_SIZE)
		return XEN_NETIF_CTRL_STATUS_INVALID_PARAMETER;

(B)	while (len-- != 0)
		if (mapping[off++] >= vif->num_queues)
			return XEN_NETIF_CTRL_STATUS_INVALID_PARAMETER;

	if (copy_op.len != 0) {
(C)		gnttab_batch_copy(&copy_op, 1);

		if (copy_op.status != GNTST_okay)
			return XEN_NETIF_CTRL_STATUS_INVALID_PARAMETER;
	}

	return XEN_NETIF_CTRL_STATUS_SUCCESS;
}

A malicious guest can trigger an integer overflow for off + len and bypass the
check at (A). This leads to OOB read access in (B) and might be turned into a controlled
OOB write in dom0 when reaching (C)








Found by: fwilhelm