[PATCH 15/16] nvmet-tcp: enable TLS handshake upcall

[Sagi Grimberg]

>>> @@ -1621,6 +1642,75 @@ static int nvmet_tcp_set_queue_sock(struct 
>>> nvmet_tcp_queue *queue)
>>>       return ret;
>>>   }
>>> +#ifdef CONFIG_NVME_TARGET_TCP_TLS
>>> +static void nvmet_tcp_tls_handshake_done(void *data, int status,
>>> +                     key_serial_t peerid)
>>> +{
>>> +    struct nvmet_tcp_queue *queue = data;
>>> +
>>> +    pr_debug("queue %d: TLS handshake done, key %x, status %d\n",
>>> +         queue->idx, peerid, status);
>>> +    spin_lock_irq(&queue->state_lock);
>>> +    if (queue->state != NVMET_TCP_Q_TLS_HANDSHAKE) {
>>> +        pr_warn("queue %d: TLS handshake already completed\n",
>>> +            queue->idx);
>>> +        spin_unlock_irq(&queue->state_lock);
>>> +        return;
>>> +    }
>>> +    if (!status)
>>> +        queue->tls_pskid = peerid;
>>> +    queue->state = NVMET_TCP_Q_CONNECTING;
>>> +    spin_unlock_irq(&queue->state_lock);
>>> +
>>> +    cancel_delayed_work_sync(&queue->tls_handshake_work);
>>> +    if (status) {
>>
>> I think that after this call, you cannot reference anything
>> in queue as it may have been released. Or I'm missing something?
>>
> I guess I can, as this code is gated by the state change above.
> Once we are here the state has been updated, so the timeout work will
> short circuit and not delete the queue.

But who guarantees that queue->state_lock is a valid dereference?