Eric Blake <ebl...@redhat.com> writes:
> iotests ./check -qcow2 failed for me today:
>
> 191 fail [10:46:04] [10:46:13] 0s (last: 7s) output
> mismatch (see 191.out.bad)
> --- /home/eblake/qemu/tests/qemu-iotests/191.out 2019-05-24
> 09:27:46.449348997 -0500
> +++ /home/eblake/qemu/tests/qemu-iotests/191.out.bad 2019-06-13
> 10:46:13.906004319 -0500
> @@ -8,6 +8,8 @@
> Formatting 'TEST_DIR/t.IMGFMT.ovl2', fmt=IMGFMT size=67108864
> backing_file=TEST_DIR/t.IMGFMT.mid
> wrote 65536/65536 bytes at offset 1048576
> 64 KiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
> +mkfifo: cannot create fifo
> '/home/eblake/qemu/tests/qemu-iotests/scratch/qmp-out-17782_0': File exists
> +mkfifo: cannot create fifo
> '/home/eblake/qemu/tests/qemu-iotests/scratch/qmp-in-17782_0': File exists
> {
> "return": {
> }
>
> If someone wants a good project to take on - it would be nice to revive
> Jeff Cody's initial work on refactoring the iotests to operate with one
> directory per test, allowing things to be run in parallel, and allowing
> cleanup to be much easier (nuke the temp directories, rather than trying
> to track down which test left a possibly-colliding file name behind)
The non-iotests could use the same treatment.
Too big for <https://wiki.qemu.org/BiteSizedTasks>, I'm afraid.
