Minor fixes about compressed clusters with OFLAG_COPIED

Hi,

while reviewing one previous patch about data corruption and compressed clusters we discussed that the documentation doesn't clarify that L2 entries for compressed clusters are not supposed to have the OFLAG_COPIED bit set.

Here's a patch to update the documentation and another one to fix the error message that "qemu-img check" produces.

Regards,

Berto

Alberto Garcia (2):
Fix error message about compressed clusters with OFLAG_COPIED specs/qcow2: Clarify that compressed clusters have the COPIED bit
reset

 block/qcow2-refcount.c | 4 ++--
 docs/interop/qcow2.txt | 8 ++++----
 2 files changed, 6 insertions(+), 6 deletions(-)

2.11.0

Compressed clusters are not supposed to have the COPIED bit set. "qemu-img check" detects that and prints an error message reporting the number of the affected host cluster. This doesn't make much sense because compressed clusters are not aligned to host clusters, so it would be better to report the offset instead. Plus, the calculation is wrong and it uses the raw L2 entry as if it was simply an offset.

This patch fixes the error message and reports the offset of the compressed cluster.

---
 block/qcow2-refcount.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

index 6b8b63514a..2dc23005b7 100644
@@ -1577,9 +1577,9 @@ static int check_refcounts_l2(BlockDriverState *bs, BdrvCheckResult *res,
         case QCOW2_CLUSTER_COMPRESSED:
             /* Compressed clusters don't have QCOW_OFLAG_COPIED */
             if (l2_entry & QCOW_OFLAG_COPIED) {
-                fprintf(stderr, "ERROR: cluster %" PRId64 ": "
+                fprintf(stderr, "ERROR: coffset=0x%" PRIx64 ": "
                     "copied flag must never be set for compressed "
-                    "clusters\n", l2_entry >> s->cluster_bits);
+                    "clusters\n", l2_entry & s->cluster_offset_mask);
                 l2_entry &= ~QCOW_OFLAG_COPIED;
                 res->corruptions++;
             }

2.11.0

Compressed clusters are not supposed to have the COPIED bit set, but this is not made explicit in the specs, so let's document it.

---
 docs/interop/qcow2.txt | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

index feb711fb6a..8e1547ded2 100644
@@ -400,10 +400,10 @@ L2 table entry:
               62:   0 for standard clusters
                     1 for compressed clusters
 
-              63:   0 for a cluster that is unused or requires COW, 1 if its
-                    refcount is exactly one. This information is only accurate
-                    in L2 tables that are reachable from the active L1
-                    table.
+              63:   0 for clusters that are unused, compressed or require COW.
+                    1 for standard clusters whose refcount is exactly one.
+                    This information is only accurate in L2 tables
+                    that are reachable from the active L1 table.
 
 Standard Cluster Descriptor:
 

2.11.0

Do we have iotests coverage of this?

Should the qcow2 spec be explicit that OFLAG_COPIED should never be set on a compressed cluster? The current documentation for L2 table entry mentions that bit 63 is '1' for a cluster with a refcount of exactly 1 if it is an L2 table reachable from the active L1 table, with no mention of a restriction that it must not be set when bit 62 is set. Or is it feasible that although qemu itself (should) never set OFLAG_COPIED on a compressed cluster, that some third-party tool could still validly set the bit for a compressed cluster that happens to occupy a host cluster with a refcount of exactly 1 (and where writing to that cluster could be done by replacing the cluster in-place with uncompressed data, or by again writing compressed data - compression is rather wasteful in that sense as the tail of the host cluster is unused, and the only way to use the tail of the cluster is with another compressed cluster at which point the refcount is no longer 1).

Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3266
Virtualization:  qemu.org | libvirt.org

I have one half-written, but I was thinking to put it in 214, so I'll wait until Max's patch is merged.

Berto

Thanks, applied to my block-next tree:

https://github.com/XanClic/qemu/commits/block-next

Max