Re: [PATCH 09/17] nvme-tcp: do not quiesce admin queue in nvme_tcp_teardown_io_queues()

[Hannes Reinecke <hare@suse.de>]

On 4/7/24 23:24, Sagi Grimberg wrote:
> 
> 
> On 18/03/2024 17:03, Hannes Reinecke wrote:
>> nvme_tcp_teardown_io_queues() is for I/O queues; the admin queue
>> should be left untouched here.
>>
>> Signed-off-by: Hannes Reinecke <hare@suse.de>
>> ---
>>   drivers/nvme/host/tcp.c | 1 -
>>   1 file changed, 1 deletion(-)
>>
>> diff --git a/drivers/nvme/host/tcp.c b/drivers/nvme/host/tcp.c
>> index 7018dc0dd026..66675b2dc197 100644
>> --- a/drivers/nvme/host/tcp.c
>> +++ b/drivers/nvme/host/tcp.c
>> @@ -2173,7 +2173,6 @@ static void nvme_tcp_teardown_io_queues(struct 
>> nvme_ctrl *ctrl,
>>   {
>>       if (ctrl->queue_count <= 1)
>>           return;
>> -    nvme_quiesce_admin_queue(ctrl);
>>       nvme_quiesce_io_queues(ctrl);
>>       nvme_sync_io_queues(ctrl);
>>       nvme_tcp_stop_io_queues(ctrl);
> 
> 1. can you explain why is this needed?
> 2. I have some vague recollection of this being added to address
> something, but git blame leads to a patch that looks completely unrelated:
> d4d61470ae48 ("nvme-tcp: serialize controller teardown sequences")
> 
> So I do not really object this (primarily because I cannot figure out why
> this exists there in the first place).
> 
You are right; this is unrelated. I'll leave it out for the next round.

Cheers,

Hannes