[Qemu-devel] [PATCH][RESEND] qcow2: Fix cluster allocation

[Qemu-devel] [PATCH][RESEND] qcow2: Fix cluster allocation

[Kevin Wolf]

[-- Attachment #1: Type: text/plain, Size: 959 bytes --]

When allocating multiple clusters at once, the qcow2 implementation
tries to find as many physically contiguous clusters as possible to
allow larger writes. This search includes allocated clusters which are
in the right place and still free clusters. If the range to allocate
spans clusters in patterns like "10 allocated, then 10 free, then again
10 allocated" it is only checked that the chunks of allocated clusters
are contiguous for themselves.

However, what is actually needed is to have _all_ allocated clusters
contiguous, starting at the first cluster of the allocation and spanning
multiple such chunks. This patch changes the check so that each offset
is not compared to the offset of the first cluster in its own chunk but
to the first cluster in the whole allocation.

I haven't seen it happen, but without this fix data corruption on qcow2
images is possible.

Signed-off-by: Kevin Wolf <kwolf@suse.de>
Acked-by: Gleb Natapov <gleb@redhat.com>

[-- Attachment #2: qcow2-fix-contiguous-alloc.patch --]
[-- Type: text/x-patch, Size: 1939 bytes --]

Index: qemu-svn/block-qcow2.c
===================================================================
--- qemu-svn.orig/block-qcow2.c
+++ qemu-svn/block-qcow2.c
@@ -615,7 +615,7 @@ static int size_to_clusters(BDRVQcowStat
 }
 
 static int count_contiguous_clusters(uint64_t nb_clusters, int cluster_size,
-        uint64_t *l2_table, uint64_t mask)
+        uint64_t *l2_table, uint64_t start, uint64_t mask)
 {
     int i;
     uint64_t offset = be64_to_cpu(l2_table[0]) & ~mask;
@@ -623,11 +623,11 @@ static int count_contiguous_clusters(uin
     if (!offset)
         return 0;
 
-    for (i = 0; i < nb_clusters; i++)
+    for (i = start; i < start + nb_clusters; i++)
         if (offset + i * cluster_size != (be64_to_cpu(l2_table[i]) & ~mask))
             break;
 
-	return i;
+	return (i - start);
 }
 
 static int count_contiguous_free_clusters(uint64_t nb_clusters, uint64_t *l2_table)
@@ -714,7 +714,7 @@ static uint64_t get_cluster_offset(Block
     } else {
         /* how many allocated clusters ? */
         c = count_contiguous_clusters(nb_clusters, s->cluster_size,
-                &l2_table[l2_index], QCOW_OFLAG_COPIED);
+                &l2_table[l2_index], 0, QCOW_OFLAG_COPIED);
     }
 
    nb_available = (c * s->cluster_sectors);
@@ -968,7 +968,7 @@ static uint64_t alloc_cluster_offset(Blo
 
     if (cluster_offset & QCOW_OFLAG_COPIED) {
         nb_clusters = count_contiguous_clusters(nb_clusters, s->cluster_size,
-                &l2_table[l2_index], 0);
+                &l2_table[l2_index], 0, 0);
 
         cluster_offset &= ~QCOW_OFLAG_COPIED;
         m->nb_clusters = 0;
@@ -985,7 +985,7 @@ static uint64_t alloc_cluster_offset(Blo
 
     while (i < nb_clusters) {
         i += count_contiguous_clusters(nb_clusters - i, s->cluster_size,
-                &l2_table[l2_index + i], 0);
+                &l2_table[l2_index], i, 0);
 
         if(be64_to_cpu(l2_table[l2_index + i]))
             break;

[Qemu-devel] Re: [PATCH][RESEND] qcow2: Fix cluster allocation

[Anthony Liguori]

Kevin Wolf wrote:
> When allocating multiple clusters at once, the qcow2 implementation
> tries to find as many physically contiguous clusters as possible to
> allow larger writes. This search includes allocated clusters which are
> in the right place and still free clusters. If the range to allocate
> spans clusters in patterns like "10 allocated, then 10 free, then again
> 10 allocated" it is only checked that the chunks of allocated clusters
> are contiguous for themselves.
>
> However, what is actually needed is to have _all_ allocated clusters
> contiguous, starting at the first cluster of the allocation and spanning
> multiple such chunks. This patch changes the check so that each offset
> is not compared to the offset of the first cluster in its own chunk but
> to the first cluster in the whole allocation.
>
> I haven't seen it happen, but without this fix data corruption on qcow2
> images is possible.
>
> Signed-off-by: Kevin Wolf <kwolf@suse.de>
> Acked-by: Gleb Natapov <gleb@redhat.com>
>   
Applied.  Thanks.

Regards,

Anthony Liguori

[Qemu-devel] Re: [PATCH][RESEND] qcow2: Fix cluster allocation

[Anthony Liguori]

Kevin Wolf wrote:
> When allocating multiple clusters at once, the qcow2 implementation
> tries to find as many physically contiguous clusters as possible to
> allow larger writes. This search includes allocated clusters which are
> in the right place and still free clusters. If the range to allocate
> spans clusters in patterns like "10 allocated, then 10 free, then again
> 10 allocated" it is only checked that the chunks of allocated clusters
> are contiguous for themselves.
>
> However, what is actually needed is to have _all_ allocated clusters
> contiguous, starting at the first cluster of the allocation and spanning
> multiple such chunks. This patch changes the check so that each offset
> is not compared to the offset of the first cluster in its own chunk but
> to the first cluster in the whole allocation.
>
> I haven't seen it happen, but without this fix data corruption on qcow2
> images is possible.
>
> Signed-off-by: Kevin Wolf <kwolf@suse.de>
> Acked-by: Gleb Natapov <gleb@redhat.com>
>   

Applied.  Thanks.

Regards,

Anthony Liguori