[Qemu-devel] [PATCH for-1.5 1/3] qcow2: Catch some L1 table index overflows

[Kevin Wolf <kwolf@redhat.com>]

This catches the situation that is described in the bug report at
https://bugs.launchpad.net/qemu/+bug/865518 and goes like this:

    $ qemu-img create -f qcow2 huge.qcow2 $((1024*1024))T
    Formatting 'huge.qcow2', fmt=qcow2 size=1152921504606846976 encryption=off cluster_size=65536 lazy_refcounts=off
    $ qemu-io /tmp/huge.qcow2 -c "write $((1024*1024*1024*1024*1024*1024 - 1024)) 512"
    Segmentation fault

With this patch applied the segfault will be avoided, however the case
will still fail, though gracefully:

    $ qemu-img create -f qcow2 /tmp/huge.qcow2 $((1024*1024))T
    Formatting 'huge.qcow2', fmt=qcow2 size=1152921504606846976 encryption=off cluster_size=65536 lazy_refcounts=off
    qemu-img: The image size is too large for file format 'qcow2'

Note that even long before these overflow checks kick in, you get
insanely high memory usage (up to INT_MAX * sizeof(uint64_t) = 16 GB for
the L1 table), so with somewhat smaller image sizes you'll probably see
qemu aborting for a failed g_malloc().

If you need huge image sizes, you should increase the cluster size to
the maximum of 2 MB in order to get higher limits.

Signed-off-by: Kevin Wolf <kwolf@redhat.com>
---
 block/qcow2-cluster.c | 23 +++++++++++++++--------
 block/qcow2.c         | 13 +++++++++++--
 block/qcow2.h         |  5 +++--
 3 files changed, 29 insertions(+), 12 deletions(-)

diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
index c71470a..76f30e5 100644
--- a/block/qcow2-cluster.c
+++ b/block/qcow2-cluster.c
@@ -29,12 +29,13 @@
 #include "block/qcow2.h"
 #include "trace.h"
 
-int qcow2_grow_l1_table(BlockDriverState *bs, int min_size, bool exact_size)
+int qcow2_grow_l1_table(BlockDriverState *bs, uint64_t min_size,
+                        bool exact_size)
 {
     BDRVQcowState *s = bs->opaque;
-    int new_l1_size, new_l1_size2, ret, i;
+    int new_l1_size2, ret, i;
     uint64_t *new_l1_table;
-    int64_t new_l1_table_offset;
+    int64_t new_l1_table_offset, new_l1_size;
     uint8_t data[12];
 
     if (min_size <= s->l1_size)
@@ -53,8 +54,13 @@ int qcow2_grow_l1_table(BlockDriverState *bs, int min_size, bool exact_size)
         }
     }
 
+    if (new_l1_size > INT_MAX) {
+        return -EFBIG;
+    }
+
 #ifdef DEBUG_ALLOC2
-    fprintf(stderr, "grow l1_table from %d to %d\n", s->l1_size, new_l1_size);
+    fprintf(stderr, "grow l1_table from %d to %" PRId64 "\n",
+            s->l1_size, new_l1_size);
 #endif
 
     new_l1_size2 = sizeof(uint64_t) * new_l1_size;
@@ -391,8 +397,8 @@ int qcow2_get_cluster_offset(BlockDriverState *bs, uint64_t offset,
     int *num, uint64_t *cluster_offset)
 {
     BDRVQcowState *s = bs->opaque;
-    unsigned int l1_index, l2_index;
-    uint64_t l2_offset, *l2_table;
+    unsigned int l2_index;
+    uint64_t l1_index, l2_offset, *l2_table;
     int l1_bits, c;
     unsigned int index_in_cluster, nb_clusters;
     uint64_t nb_available, nb_needed;
@@ -507,8 +513,8 @@ static int get_cluster_table(BlockDriverState *bs, uint64_t offset,
                              int *new_l2_index)
 {
     BDRVQcowState *s = bs->opaque;
-    unsigned int l1_index, l2_index;
-    uint64_t l2_offset;
+    unsigned int l2_index;
+    uint64_t l1_index, l2_offset;
     uint64_t *l2_table = NULL;
     int ret;
 
@@ -522,6 +528,7 @@ static int get_cluster_table(BlockDriverState *bs, uint64_t offset,
         }
     }
 
+    assert(l1_index < s->l1_size);
     l2_offset = s->l1_table[l1_index] & L1E_OFFSET_MASK;
 
     /* seek the l2 table of the given l2 offset */
diff --git a/block/qcow2.c b/block/qcow2.c
index 2e346d8..0fa5cb2 100644
--- a/block/qcow2.c
+++ b/block/qcow2.c
@@ -307,6 +307,7 @@ static int qcow2_open(BlockDriverState *bs, QDict *options, int flags)
     QemuOpts *opts;
     Error *local_err = NULL;
     uint64_t ext_end;
+    uint64_t l1_vm_state_index;
 
     ret = bdrv_pread(bs->file, 0, &header, sizeof(header));
     if (ret < 0) {
@@ -424,7 +425,14 @@ static int qcow2_open(BlockDriverState *bs, QDict *options, int flags)
 
     /* read the level 1 table */
     s->l1_size = header.l1_size;
-    s->l1_vm_state_index = size_to_l1(s, header.size);
+
+    l1_vm_state_index = size_to_l1(s, header.size);
+    if (l1_vm_state_index > INT_MAX) {
+        ret = -EFBIG;
+        goto fail;
+    }
+    s->l1_vm_state_index = l1_vm_state_index;
+
     /* the L1 table must contain at least enough entries to put
        header.size bytes */
     if (s->l1_size < s->l1_vm_state_index) {
@@ -1480,7 +1488,8 @@ static coroutine_fn int qcow2_co_discard(BlockDriverState *bs,
 static int qcow2_truncate(BlockDriverState *bs, int64_t offset)
 {
     BDRVQcowState *s = bs->opaque;
-    int ret, new_l1_size;
+    int64_t new_l1_size;
+    int ret;
 
     if (offset & 511) {
         error_report("The new size must be a multiple of 512");
diff --git a/block/qcow2.h b/block/qcow2.h
index 9421843..6959c6a 100644
--- a/block/qcow2.h
+++ b/block/qcow2.h
@@ -284,7 +284,7 @@ static inline int size_to_clusters(BDRVQcowState *s, int64_t size)
     return (size + (s->cluster_size - 1)) >> s->cluster_bits;
 }
 
-static inline int size_to_l1(BDRVQcowState *s, int64_t size)
+static inline int64_t size_to_l1(BDRVQcowState *s, int64_t size)
 {
     int shift = s->cluster_bits + s->l2_bits;
     return (size + (1ULL << shift) - 1) >> shift;
@@ -360,7 +360,8 @@ int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res,
                           BdrvCheckMode fix);
 
 /* qcow2-cluster.c functions */
-int qcow2_grow_l1_table(BlockDriverState *bs, int min_size, bool exact_size);
+int qcow2_grow_l1_table(BlockDriverState *bs, uint64_t min_size,
+                        bool exact_size);
 void qcow2_l2_cache_reset(BlockDriverState *bs);
 int qcow2_decompress_cluster(BlockDriverState *bs, uint64_t cluster_offset);
 void qcow2_encrypt_sectors(BDRVQcowState *s, int64_t sector_num,
-- 
1.8.1.4