The non-public logs in https://bugzilla.redhat.com/show_bug.cgi?id=1583346 (sorry...) reveal this problem:
$ (Create a qcow2 file "foo.qcow2" with a corrupted first L1 entry) $ echo 'qemu-io none0 "read 0 512"' \ | x86_64-softmmu/qemu-system-x86_64 -drive if=none,file=foo.qcow2 \ -monitor stdio \ -incoming exec:'cat /dev/null' QEMU 2.12.50 monitor - type 'help' for more information (qemu) qemu-io none0 "read 0 512" qcow2: Marking image as corrupt: L2 table offset 0x44200 unaligned (L1 index: 0); further corruption events will be suppressed qemu-system-x86_64: block/io.c:1691: bdrv_co_pwritev: Assertion `!(bs->open_flags & BDRV_O_INACTIVE)' failed. [1] 18444 done echo 'qemu-io none0 "read 0 512"' | 18445 abort (core dumped) x86_64-softmmu/qemu-system-x86_64 -drive if=none,file=foo.qcow2 -monitor stdi Oops. The first patch in this series fixes this by treating inactive images like read-only images in this regard (which most importantly means not trying to set the corrupt flag on them), the second one adds an iotest case. Max Reitz (2): qcow2: Do not mark inactive images corrupt iotests: Add case for a corrupted inactive image block/qcow2.c | 4 +++- tests/qemu-iotests/060 | 30 ++++++++++++++++++++++++++++++ tests/qemu-iotests/060.out | 14 ++++++++++++++ 3 files changed, 47 insertions(+), 1 deletion(-) -- 2.17.0