From: Max Reitz <mreitz@redhat.com>
To: qemu-block@nongnu.org
Cc: qemu-devel@nongnu.org, Max Reitz <mreitz@redhat.com>,
Kevin Wolf <kwolf@redhat.com>, Alberto Garcia <berto@igalia.com>,
John Snow <jsnow@redhat.com>
Subject: [Qemu-devel] [PATCH for-2.11 4/5] qcow2: Add bounds check to get_refblock_offset()
Date: Fri, 10 Nov 2017 21:31:10 +0100 [thread overview]
Message-ID: <20171110203111.7666-5-mreitz@redhat.com> (raw)
In-Reply-To: <20171110203111.7666-1-mreitz@redhat.com>
Reported-by: R. Nageswara Sastry <nasastry@in.ibm.com>
Buglink: https://bugs.launchpad.net/qemu/+bug/1728661
Signed-off-by: Max Reitz <mreitz@redhat.com>
---
block/qcow2.h | 6 ------
block/qcow2-refcount.c | 26 +++++++++++++++++++++++++-
tests/qemu-iotests/060 | 46 ++++++++++++++++++++++++++++++++++++++++++++++
tests/qemu-iotests/060.out | 22 ++++++++++++++++++++++
4 files changed, 93 insertions(+), 7 deletions(-)
diff --git a/block/qcow2.h b/block/qcow2.h
index 782a206ecb..6f0ff15dd0 100644
--- a/block/qcow2.h
+++ b/block/qcow2.h
@@ -527,12 +527,6 @@ uint32_t offset_to_reftable_index(BDRVQcow2State *s, uint64_t offset)
return offset >> (s->refcount_block_bits + s->cluster_bits);
}
-static inline uint64_t get_refblock_offset(BDRVQcow2State *s, uint64_t offset)
-{
- uint32_t index = offset_to_reftable_index(s, offset);
- return s->refcount_table[index] & REFT_OFFSET_MASK;
-}
-
/* qcow2.c functions */
int qcow2_backing_read1(BlockDriverState *bs, QEMUIOVector *qiov,
int64_t sector_num, int nb_sectors);
diff --git a/block/qcow2-refcount.c b/block/qcow2-refcount.c
index 60b8eef3e8..3de1ab51ba 100644
--- a/block/qcow2-refcount.c
+++ b/block/qcow2-refcount.c
@@ -3077,16 +3077,40 @@ done:
return ret;
}
+static int64_t get_refblock_offset(BlockDriverState *bs, uint64_t offset)
+{
+ BDRVQcow2State *s = bs->opaque;
+ uint32_t index = offset_to_reftable_index(s, offset);
+ int64_t covering_refblock_offset = 0;
+
+ if (index < s->refcount_table_size) {
+ covering_refblock_offset = s->refcount_table[index] & REFT_OFFSET_MASK;
+ }
+ if (!covering_refblock_offset) {
+ qcow2_signal_corruption(bs, true, -1, -1, "Refblock at %#" PRIx64 " is "
+ "not covered by the refcount structures",
+ offset);
+ return -EIO;
+ }
+
+ return covering_refblock_offset;
+}
+
static int qcow2_discard_refcount_block(BlockDriverState *bs,
uint64_t discard_block_offs)
{
BDRVQcow2State *s = bs->opaque;
- uint64_t refblock_offs = get_refblock_offset(s, discard_block_offs);
+ int64_t refblock_offs;
uint64_t cluster_index = discard_block_offs >> s->cluster_bits;
uint32_t block_index = cluster_index & (s->refcount_block_size - 1);
void *refblock;
int ret;
+ refblock_offs = get_refblock_offset(bs, discard_block_offs);
+ if (refblock_offs < 0) {
+ return refblock_offs;
+ }
+
assert(discard_block_offs != 0);
ret = qcow2_cache_get(bs, s->refcount_block_cache, refblock_offs,
diff --git a/tests/qemu-iotests/060 b/tests/qemu-iotests/060
index 44141f6243..c230696b3a 100755
--- a/tests/qemu-iotests/060
+++ b/tests/qemu-iotests/060
@@ -359,6 +359,52 @@ echo '--- Repairing ---'
_check_test_img -q -r all
_check_test_img -r all
+echo
+echo "=== Discarding an out-of-bounds refblock ==="
+echo
+
+_make_test_img 64M
+
+# Pretend there's a refblock really up high
+poke_file "$TEST_IMG" "$(($rt_offset+8))" "\x00\xff\xff\xff\x00\x00\x00\x00"
+# Let's try to shrink the qcow2 image so that the block driver tries
+# to discard that refblock (and see what happens!)
+$QEMU_IMG resize --shrink "$TEST_IMG" 32M
+
+echo '--- Checking and retrying ---'
+# Image should not be resized
+_img_info | grep 'virtual size'
+# But it should pass this check, because the "partial" resize has
+# already overwritten refblocks past the end
+_check_test_img -r all
+# So let's try again
+$QEMU_IMG resize --shrink "$TEST_IMG" 32M
+_img_info | grep 'virtual size'
+
+echo
+echo "=== Discarding a non-covered in-bounds refblock ==="
+echo
+
+IMGOPTS='refcount_bits=1' _make_test_img 64M
+
+# Pretend there's a refblock somewhere where there is no refblock to
+# cover it (but the covering refblock has a valid index in the
+# reftable)
+# Every refblock covers 65536 * 8 * 65536 = 32 GB, so we have to point
+# to 0x10_0000_0000 (64G) to point to the third refblock
+poke_file "$TEST_IMG" "$(($rt_offset+8))" "\x00\x00\x00\x10\x00\x00\x00\x00"
+$QEMU_IMG resize --shrink "$TEST_IMG" 32M
+
+echo '--- Checking and retrying ---'
+# Image should not be resized
+_img_info | grep 'virtual size'
+# But it should pass this check, because the "partial" resize has
+# already overwritten refblocks past the end
+_check_test_img -r all
+# So let's try again
+$QEMU_IMG resize --shrink "$TEST_IMG" 32M
+_img_info | grep 'virtual size'
+
# success, all done
echo "*** done"
rm -f $seq.full
diff --git a/tests/qemu-iotests/060.out b/tests/qemu-iotests/060.out
index 07dfdcac99..358e54cdc9 100644
--- a/tests/qemu-iotests/060.out
+++ b/tests/qemu-iotests/060.out
@@ -348,4 +348,26 @@ The following inconsistencies were found and repaired:
Double checking the fixed image now...
No errors were found on the image.
+
+=== Discarding an out-of-bounds refblock ===
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=67108864
+qcow2: Marking image as corrupt: Refblock at 0xffffff00000000 is not covered by the refcount structures; further corruption events will be suppressed
+qemu-img: Failed to discard unused refblocks: Input/output error
+--- Checking and retrying ---
+virtual size: 64M (67108864 bytes)
+No errors were found on the image.
+Image resized.
+virtual size: 32M (33554432 bytes)
+
+=== Discarding a non-covered in-bounds refblock ===
+
+Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=67108864
+qcow2: Marking image as corrupt: Refblock at 0x1000000000 is not covered by the refcount structures; further corruption events will be suppressed
+qemu-img: Failed to discard unused refblocks: Input/output error
+--- Checking and retrying ---
+virtual size: 64M (67108864 bytes)
+No errors were found on the image.
+Image resized.
+virtual size: 32M (33554432 bytes)
*** done
--
2.13.6
next prev parent reply other threads:[~2017-11-10 20:31 UTC|newest]
Thread overview: 25+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-11-10 20:31 [Qemu-devel] [PATCH for-2.11 0/5] qcow2: Fixes for corrupted images Max Reitz
2017-11-10 20:31 ` [Qemu-devel] [PATCH for-2.11 1/5] qcow2: check_errors are fatal Max Reitz
2017-11-10 20:55 ` Eric Blake
2017-11-10 20:31 ` [Qemu-devel] [PATCH for-2.11 2/5] qcow2: Unaligned zero cluster in handle_alloc() Max Reitz
2017-11-10 20:58 ` Eric Blake
2017-11-14 14:55 ` Alberto Garcia
2017-11-10 20:31 ` [Qemu-devel] [PATCH for-2.11 3/5] block: Guard against NULL bs->drv Max Reitz
2017-11-10 21:46 ` Eric Blake
2017-11-14 15:36 ` Max Reitz
2017-11-10 20:31 ` Max Reitz [this message]
2017-11-10 21:49 ` [Qemu-devel] [PATCH for-2.11 4/5] qcow2: Add bounds check to get_refblock_offset() Eric Blake
2017-11-14 15:02 ` Alberto Garcia
2017-11-14 15:27 ` Max Reitz
2017-11-14 15:38 ` Alberto Garcia
2017-11-14 15:40 ` Max Reitz
2017-11-10 20:31 ` [Qemu-devel] [PATCH for-2.11 5/5] qcow2: Refuse to get unaligned offsets from cache Max Reitz
2017-11-10 21:54 ` Eric Blake
2017-11-10 22:00 ` Max Reitz
2017-11-10 22:15 ` Eric Blake
2017-11-10 22:16 ` Max Reitz
2017-11-14 15:06 ` Alberto Garcia
2017-11-14 15:09 ` Max Reitz
2017-11-14 15:31 ` Alberto Garcia
2017-11-10 20:41 ` [Qemu-devel] [PATCH for-2.11 0/5] qcow2: Fixes for corrupted images Max Reitz
2017-11-15 20:25 ` Max Reitz
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20171110203111.7666-5-mreitz@redhat.com \
--to=mreitz@redhat.com \
--cc=berto@igalia.com \
--cc=jsnow@redhat.com \
--cc=kwolf@redhat.com \
--cc=qemu-block@nongnu.org \
--cc=qemu-devel@nongnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).