The standard cluster descriptor in L2 table entries has a field to
store the host cluster offset. When we need to get that offset from an
entry we use L2E_OFFSET_MASK to ensure that we only use the bits that
belong to that field.

But while that mask is used every time we read from an L2 entry, it
is never used when we write to it. Due to the QCOW_MAX_CLUSTER_OFFSET
limit set in the cluster allocation code QEMU can never produce
offsets that don't fit in that field so any such offset would indicate
a bug in QEMU.

Compressed cluster descriptors contain two fields (host cluster offset
and size of the compressed data) and the situation with them is
similar. In this case the masks are not constant but are stored in the
csize_mask and cluster_offset_mask fields of BDRVQcow2State.

Signed-off-by: Alberto Garcia <berto@igalia.com>
---
 block/qcow2-cluster.c | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index 974a4e8656..4a96966ab9 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -799,6 +799,10 @@ int qcow2_alloc_compressed_cluster_offset(BlockDriverState *bs,
     nb_csectors = ((cluster_offset + compressed_size - 1) >> 9) -
                   (cluster_offset >> 9);
 
+    /* The offset and size must fit in their fields of the L2 table entry */
+    assert((cluster_offset & s->cluster_offset_mask) == cluster_offset);
+    assert((nb_csectors & s->csize_mask) == nb_csectors);
+
     cluster_offset |= QCOW_OFLAG_COMPRESSED |
                       ((uint64_t)nb_csectors << s->csize_shift);
 
@@ -984,6 +988,7 @@ int qcow2_alloc_cluster_link_l2(BlockDriverState *bs, QCowL2Meta *m)
 
     assert(l2_index + m->nb_clusters <= s->l2_slice_size);
     for (i = 0; i < m->nb_clusters; i++) {
+        uint64_t offset = cluster_offset + (i << s->cluster_bits);
         /* if two concurrent writes happen to the same unallocated cluster
          * each write allocates separate cluster and writes data concurrently.
          * The first one to complete updates l2 table with pointer to its
@@ -994,8 +999,10 @@ int qcow2_alloc_cluster_link_l2(BlockDriverState *bs, QCowL2Meta *m)
             old_cluster[j++] = l2_slice[l2_index + i];
         }
 
-        l2_slice[l2_index + i] = cpu_to_be64((cluster_offset +
-                    (i << s->cluster_bits)) | QCOW_OFLAG_COPIED);
+        /* The offset must fit in the offset field of the L2 table entry */
+        assert((offset & L2E_OFFSET_MASK) == offset);
+
+        l2_slice[l2_index + i] = cpu_to_be64(offset | QCOW_OFLAG_COPIED);
      }
 
 
@@ -1928,6 +1935,9 @@ static int expand_zero_clusters_in_l1(BlockDriverState *bs, uint64_t *l1_table,
                         goto fail;
                     }
 
+                    /* The offset must fit in the offset field */
+                    assert((offset & L2E_OFFSET_MASK) == offset);
+
                     if (l2_refcount > 1) {
                         /* For shared L2 tables, set the refcount accordingly
                          * (it is already 1 and needs to be l2_refcount) */
-- 
2.11.0

On 5/3/19 4:38 AM, Alberto Garcia wrote:
> The standard cluster descriptor in L2 table entries has a field to
> store the host cluster offset. When we need to get that offset from an
> entry we use L2E_OFFSET_MASK to ensure that we only use the bits that
> belong to that field.
> 
> But while that mask is used every time we read from an L2 entry, it
> is never used when we write to it. Due to the QCOW_MAX_CLUSTER_OFFSET
> limit set in the cluster allocation code QEMU can never produce
> offsets that don't fit in that field so any such offset would indicate
> a bug in QEMU.

Yeah, I'm not seeing where this one could ever overflow.

> 
> Compressed cluster descriptors contain two fields (host cluster offset
> and size of the compressed data) and the situation with them is
> similar. In this case the masks are not constant but are stored in the
> csize_mask and cluster_offset_mask fields of BDRVQcow2State.

For this one, we did have a bug in the past where we were overflowing,
as evidenced by iotest 220 shortly after we patched the bug (77d6a215).

> 
> Signed-off-by: Alberto Garcia <berto@igalia.com>
> ---
>  block/qcow2-cluster.c | 14 ++++++++++++--
>  1 file changed, 12 insertions(+), 2 deletions(-)
> 

Adding more assertions shouldn't hurt.
Reviewed-by: Eric Blake <eblake@redhat.com>

-- 
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3226
Virtualization:  qemu.org | libvirt.org