Patchwork qed: Fix consistency check on 32-bit hosts

login
register
mail settings
Submitter Stefan Hajnoczi
Date April 24, 2011, 5:38 p.m.
Message ID <1303666738-26558-1-git-send-email-stefanha@linux.vnet.ibm.com>
Download mbox | patch
Permalink /patch/92671/
State New
Headers show

Comments

Stefan Hajnoczi - April 24, 2011, 5:38 p.m.
The qed_bytes_to_clusters() function is normally used with size_t
lengths.  Consistency check used it with file size length and therefore
failed on 32-bit hosts when the image file is 4 GB or more.

Make qed_bytes_to_clusters() explicitly 64-bit and update consistency
check to keep 64-bit cluster counts.

Reported-by: Michael Tokarev <mjt@tls.msk.ru>
Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
---
 block/qed-check.c |    4 ++--
 block/qed.h       |    2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)
Kevin Wolf - April 26, 2011, 8:44 a.m.
Am 24.04.2011 19:38, schrieb Stefan Hajnoczi:
> The qed_bytes_to_clusters() function is normally used with size_t
> lengths.  Consistency check used it with file size length and therefore
> failed on 32-bit hosts when the image file is 4 GB or more.
> 
> Make qed_bytes_to_clusters() explicitly 64-bit and update consistency
> check to keep 64-bit cluster counts.
> 
> Reported-by: Michael Tokarev <mjt@tls.msk.ru>
> Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>

Thanks, applied to the block branch.

Kevin
Stefan Hajnoczi - April 26, 2011, 9:48 a.m.
On Tue, Apr 26, 2011 at 9:44 AM, Kevin Wolf <kwolf@redhat.com> wrote:
> Am 24.04.2011 19:38, schrieb Stefan Hajnoczi:
>> The qed_bytes_to_clusters() function is normally used with size_t
>> lengths.  Consistency check used it with file size length and therefore
>> failed on 32-bit hosts when the image file is 4 GB or more.
>>
>> Make qed_bytes_to_clusters() explicitly 64-bit and update consistency
>> check to keep 64-bit cluster counts.
>>
>> Reported-by: Michael Tokarev <mjt@tls.msk.ru>
>> Signed-off-by: Stefan Hajnoczi <stefanha@linux.vnet.ibm.com>
>
> Thanks, applied to the block branch.

Please apply to stable.  This fixes a segfault when checking an image
on 32-bit hosts.

Stefan

Patch

diff --git a/block/qed-check.c b/block/qed-check.c
index ea4ebc8..22cd07f 100644
--- a/block/qed-check.c
+++ b/block/qed-check.c
@@ -18,7 +18,7 @@  typedef struct {
     BdrvCheckResult *result;
     bool fix;                           /* whether to fix invalid offsets */
 
-    size_t nclusters;
+    uint64_t nclusters;
     uint32_t *used_clusters;            /* referenced cluster bitmap */
 
     QEDRequest request;
@@ -177,7 +177,7 @@  static int qed_check_l1_table(QEDCheck *check, QEDTable *table)
 static void qed_check_for_leaks(QEDCheck *check)
 {
     BDRVQEDState *s = check->s;
-    size_t i;
+    uint64_t i;
 
     for (i = s->header.header_size; i < check->nclusters; i++) {
         if (!qed_test_bit(check->used_clusters, i)) {
diff --git a/block/qed.h b/block/qed.h
index 3e1ab84..1d1421f 100644
--- a/block/qed.h
+++ b/block/qed.h
@@ -252,7 +252,7 @@  static inline uint64_t qed_offset_into_cluster(BDRVQEDState *s, uint64_t offset)
     return offset & (s->header.cluster_size - 1);
 }
 
-static inline unsigned int qed_bytes_to_clusters(BDRVQEDState *s, size_t bytes)
+static inline uint64_t qed_bytes_to_clusters(BDRVQEDState *s, uint64_t bytes)
 {
     return qed_start_of_cluster(s, bytes + (s->header.cluster_size - 1)) /
            (s->header.cluster_size - 1);