[PATCH] tests/nvme: Set clear_ids for passthru targets

[PATCH] tests/nvme: Set clear_ids for passthru targets

[Alan Adamson]

This allows to connect to passthru targets when the client and target
are on the same host.

Signed-off-by: Alan Adamson <alan.adamson@oracle.com>
---
 tests/nvme/rc | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/tests/nvme/rc b/tests/nvme/rc
index 4bebbc762cbb..5e50e69fb3f0 100644
--- a/tests/nvme/rc
+++ b/tests/nvme/rc
@@ -303,6 +303,9 @@ _create_nvmet_passthru() {
 
 	_test_dev_nvme_ctrl > "${passthru_path}/device_path"
 	echo 1 > "${passthru_path}/enable"
+	if [ -f "${passthru_path}/clear_ids" ]; then
+		echo 1 > "${passthru_path}/clear_ids"
+	fi
 }
 
 _remove_nvmet_passhtru() {
-- 
2.31.1

Re: [PATCH] tests/nvme: Set clear_ids for passthru targets

[Chaitanya Kulkarni]

On 7/5/22 13:56, Alan Adamson wrote:
> This allows to connect to passthru targets when the client and target
> are on the same host.
> 
> Signed-off-by: Alan Adamson <alan.adamson@oracle.com>
> ---
>   tests/nvme/rc | 3 +++
>   1 file changed, 3 insertions(+)
> 
> diff --git a/tests/nvme/rc b/tests/nvme/rc
> index 4bebbc762cbb..5e50e69fb3f0 100644
> --- a/tests/nvme/rc
> +++ b/tests/nvme/rc
> @@ -303,6 +303,9 @@ _create_nvmet_passthru() {
>   
>   	_test_dev_nvme_ctrl > "${passthru_path}/device_path"
>   	echo 1 > "${passthru_path}/enable"
> +	if [ -f "${passthru_path}/clear_ids" ]; then
> +		echo 1 > "${passthru_path}/clear_ids"
> +	fi
>   }
>   
>   _remove_nvmet_passhtru() {

without looking into the code, just wondering whether we need
an explicit check to ensure that both host and target on the
same machine something like checking nvme_trtype=loop ?

-ck

Re: [PATCH] tests/nvme: Set clear_ids for passthru targets

[Alan Adamson]

> On Jul 5, 2022, at 5:18 PM, Chaitanya Kulkarni <chaitanyak@nvidia.com> wrote:
> 
> On 7/5/22 13:56, Alan Adamson wrote:
>> This allows to connect to passthru targets when the client and target
>> are on the same host.
>> 
>> Signed-off-by: Alan Adamson <alan.adamson@oracle.com>
>> ---
>>  tests/nvme/rc | 3 +++
>>  1 file changed, 3 insertions(+)
>> 
>> diff --git a/tests/nvme/rc b/tests/nvme/rc
>> index 4bebbc762cbb..5e50e69fb3f0 100644
>> --- a/tests/nvme/rc
>> +++ b/tests/nvme/rc
>> @@ -303,6 +303,9 @@ _create_nvmet_passthru() {
>> 
>>  	_test_dev_nvme_ctrl > "${passthru_path}/device_path"
>>  	echo 1 > "${passthru_path}/enable"
>> +	if [ -f "${passthru_path}/clear_ids" ]; then
>> +		echo 1 > "${passthru_path}/clear_ids"
>> +	fi
>>  }
>> 
>>  _remove_nvmet_passhtru() {
> 
> without looking into the code, just wondering whether we need
> an explicit check to ensure that both host and target on the
> same machine something like checking nvme_trtype=loop ?

If nvme_trtype=loop, this code isn’t necessary because loop defaults
to clear_ids, but it is necessary for tcp and rdma.  The check is
for necessary when running a pre 5.19 kernel where clear_ids
isn’t present just to prevent a error message.

Alan

Re: [PATCH] tests/nvme: Set clear_ids for passthru targets

[Alan Adamson]

> On Jul 6, 2022, at 8:58 AM, Alan Adamson <alan.adamson@oracle.com> wrote:
> 
> 
> 
>> On Jul 5, 2022, at 5:18 PM, Chaitanya Kulkarni <chaitanyak@nvidia.com> wrote:
>> 
>> On 7/5/22 13:56, Alan Adamson wrote:
>>> This allows to connect to passthru targets when the client and target
>>> are on the same host.
>>> 
>>> Signed-off-by: Alan Adamson <alan.adamson@oracle.com>
>>> ---
>>> tests/nvme/rc | 3 +++
>>> 1 file changed, 3 insertions(+)
>>> 
>>> diff --git a/tests/nvme/rc b/tests/nvme/rc
>>> index 4bebbc762cbb..5e50e69fb3f0 100644
>>> --- a/tests/nvme/rc
>>> +++ b/tests/nvme/rc
>>> @@ -303,6 +303,9 @@ _create_nvmet_passthru() {
>>> 
>>> 	_test_dev_nvme_ctrl > "${passthru_path}/device_path"
>>> 	echo 1 > "${passthru_path}/enable"
>>> +	if [ -f "${passthru_path}/clear_ids" ]; then
>>> +		echo 1 > "${passthru_path}/clear_ids"
>>> +	fi
>>> }
>>> 
>>> _remove_nvmet_passhtru() {
>> 
>> without looking into the code, just wondering whether we need
>> an explicit check to ensure that both host and target on the
>> same machine something like checking nvme_trtype=loop ?
> 
> If nvme_trtype=loop, this code isn’t necessary because loop defaults
> to clear_ids, but it is necessary for tcp and rdma.  The check is
> for necessary when running a pre 5.19 kernel where clear_ids
> isn’t present just to prevent a error message.
> 
> Alan

Any other comments or feedback?

Alan

Re: [PATCH] tests/nvme: Set clear_ids for passthru targets

[Shinichiro Kawasaki]

On Jul 14, 2022 / 16:25, Alan Adamson wrote:
> 
> 
> > On Jul 6, 2022, at 8:58 AM, Alan Adamson <alan.adamson@oracle.com> wrote:
> > 
> > 
> > 
> >> On Jul 5, 2022, at 5:18 PM, Chaitanya Kulkarni <chaitanyak@nvidia.com> wrote:
> >> 
> >> On 7/5/22 13:56, Alan Adamson wrote:
> >>> This allows to connect to passthru targets when the client and target
> >>> are on the same host.
> >>> 
> >>> Signed-off-by: Alan Adamson <alan.adamson@oracle.com>
> >>> ---
> >>> tests/nvme/rc | 3 +++
> >>> 1 file changed, 3 insertions(+)
> >>> 
> >>> diff --git a/tests/nvme/rc b/tests/nvme/rc
> >>> index 4bebbc762cbb..5e50e69fb3f0 100644
> >>> --- a/tests/nvme/rc
> >>> +++ b/tests/nvme/rc
> >>> @@ -303,6 +303,9 @@ _create_nvmet_passthru() {
> >>> 
> >>> 	_test_dev_nvme_ctrl > "${passthru_path}/device_path"
> >>> 	echo 1 > "${passthru_path}/enable"
> >>> +	if [ -f "${passthru_path}/clear_ids" ]; then
> >>> +		echo 1 > "${passthru_path}/clear_ids"
> >>> +	fi
> >>> }
> >>> 
> >>> _remove_nvmet_passhtru() {
> >> 
> >> without looking into the code, just wondering whether we need
> >> an explicit check to ensure that both host and target on the
> >> same machine something like checking nvme_trtype=loop ?
> > 
> > If nvme_trtype=loop, this code isn’t necessary because loop defaults
> > to clear_ids, but it is necessary for tcp and rdma.  The check is
> > for necessary when running a pre 5.19 kernel where clear_ids
> > isn’t present just to prevent a error message.
> > 
> > Alan
> 
> Any other comments or feedback?

I've applied the patch with an edit to replace single bracket to double
brackets. Wish this edit is ok for you. Thanks!

-- 
Shin'ichiro Kawasaki

Re: [PATCH] tests/nvme: Set clear_ids for passthru targets

[Alan Adamson]

> On Jul 15, 2022, at 1:46 AM, Shinichiro Kawasaki <shinichiro.kawasaki@wdc.com> wrote:
> 
> On Jul 14, 2022 / 16:25, Alan Adamson wrote:
>> 
>> 
>>> On Jul 6, 2022, at 8:58 AM, Alan Adamson <alan.adamson@oracle.com> wrote:
>>> 
>>> 
>>> 
>>>> On Jul 5, 2022, at 5:18 PM, Chaitanya Kulkarni <chaitanyak@nvidia.com> wrote:
>>>> 
>>>> On 7/5/22 13:56, Alan Adamson wrote:
>>>>> This allows to connect to passthru targets when the client and target
>>>>> are on the same host.
>>>>> 
>>>>> Signed-off-by: Alan Adamson <alan.adamson@oracle.com>
>>>>> ---
>>>>> tests/nvme/rc | 3 +++
>>>>> 1 file changed, 3 insertions(+)
>>>>> 
>>>>> diff --git a/tests/nvme/rc b/tests/nvme/rc
>>>>> index 4bebbc762cbb..5e50e69fb3f0 100644
>>>>> --- a/tests/nvme/rc
>>>>> +++ b/tests/nvme/rc
>>>>> @@ -303,6 +303,9 @@ _create_nvmet_passthru() {
>>>>> 
>>>>> 	_test_dev_nvme_ctrl > "${passthru_path}/device_path"
>>>>> 	echo 1 > "${passthru_path}/enable"
>>>>> +	if [ -f "${passthru_path}/clear_ids" ]; then
>>>>> +		echo 1 > "${passthru_path}/clear_ids"
>>>>> +	fi
>>>>> }
>>>>> 
>>>>> _remove_nvmet_passhtru() {
>>>> 
>>>> without looking into the code, just wondering whether we need
>>>> an explicit check to ensure that both host and target on the
>>>> same machine something like checking nvme_trtype=loop ?
>>> 
>>> If nvme_trtype=loop, this code isn’t necessary because loop defaults
>>> to clear_ids, but it is necessary for tcp and rdma.  The check is
>>> for necessary when running a pre 5.19 kernel where clear_ids
>>> isn’t present just to prevent a error message.
>>> 
>>> Alan
>> 
>> Any other comments or feedback?
> 
> I've applied the patch with an edit to replace single bracket to double
> brackets. Wish this edit is ok for you. Thanks!
Looks good.

Thanks!

Alan


> 
> -- 
> Shin'ichiro Kawasaki