On 07/01/2025 10:04, Shinichiro Kawasaki wrote:
> On Jan 03, 2025 / 11:19, Li Zhijian wrote:
>> It attempts to connect and disconnect the rnbd service on localhost.
>> Actually, It also reveals a real kernel issue[0].
>
> Good to find another bug and another test case, thanks :)
> Please find a nit comment below.
>
> [...]
>
>> diff --git a/tests/rnbd/001.out b/tests/rnbd/001.out
>> new file mode 100644
>> index 000000000000..c1f9980d0f7b
>> --- /dev/null
>> +++ b/tests/rnbd/001.out
>> @@ -0,0 +1,2 @@
>> +Running rnbd/001
>> +Test complete
>> diff --git a/tests/rnbd/rc b/tests/rnbd/rc
>> new file mode 100644
>> index 000000000000..1cf98ad5c498
>> --- /dev/null
>> +++ b/tests/rnbd/rc
>> @@ -0,0 +1,51 @@
>> +#!/bin/bash
>> +# SPDX-License-Identifier: GPL-3.0+
>> +# Copyright (c) 2024 FUJITSU LIMITED. All Rights Reserved.
>> +#
>> +# RNBD tests.
>> +
>> +. common/rc
>> +. common/multipath-over-rdma
>> +
>> +_have_rnbd() {
>> + if [[ "$USE_RXE" != 1 ]]; then
>> + SKIP_REASONS+=("Only USE_RXE=1 is supported")
>> + fi
>> + _have_driver rdma_rxe
>> + _have_driver rnbd_server
>> + _have_driver rnbd_client
>> +}
>> +
>> +_setup_rnbd() {
>> + start_soft_rdma
>
> The added test cases check exit status of this _setup_rnbd() function, but
> this function is not likely return non-zero exist status. I think the line
> below instead of the line above will make the exit status checks more
> valuable.
>
> start_soft_rdma || return $?
Yes, you are right.
