On 11/30/20 22:47, Damien Le Moal wrote: > This passes for me: > > ./zonefs-tests.sh -t 0012 /dev/nullb0 > Gathering information on /dev/nullb0... > zonefs-tests on /dev/nullb0: > 4 zones (0 conventional zones, 4 sequential zones) > 524288 512B sectors zone size (256 MiB) > 1 max open zones > Running tests > Test 0012: mkzonefs (invalid device) ... PASS > > 1 / 1 tests passed > > This test create a a regular nullb device and tries mkzonefs on it, which should > fail since the block device is not zoned. In your case, it looks like the test > endup using an existing zoned nullb device, which should not happen (nullb4 and > nullb5). How many nullb devices do you have on that system ? True, my test script creates random number nvmet namespaces and nullblk devices. I think testcase is reading the wrong device, let me finish the V4 with removing bio checks as per your comments and I'll look into this.
