[PATCH] nvmet-auth: complete a request only after freeing the dhchap pointers

[Christoph Hellwig]

On Fri, Oct 13, 2023 at 11:42:39AM +0200, Maurizio Lombardi wrote:
> -	nvmet_req_complete(req, status);
>  	if (req->sq->dhchap_step != NVME_AUTH_DHCHAP_MESSAGE_SUCCESS2 &&
>  	    req->sq->dhchap_step != NVME_AUTH_DHCHAP_MESSAGE_FAILURE2) {
>  		unsigned long auth_expire_secs = ctrl->kato ? ctrl->kato : 120;
>  
>  		mod_delayed_work(system_wq, &req->sq->auth_expired_work,
>  				 auth_expire_secs * HZ);
> +
> +		nvmet_req_complete(req, status);
>  		return;
>  	}
>  	/* Final states, clear up variables */
>  	nvmet_auth_sq_free(req->sq);
>  	if (req->sq->dhchap_step == NVME_AUTH_DHCHAP_MESSAGE_FAILURE2)
>  		nvmet_ctrl_fatal_error(ctrl);
> +	nvmet_req_complete(req, status);

I'd be tempted to use a goto here to have a single completion path.

>  }
>  
>  static int nvmet_auth_challenge(struct nvmet_req *req, void *d, int al)
> @@ -514,11 +516,14 @@ void nvmet_execute_auth_receive(struct nvmet_req *req)
>  	kfree(d);
>  done:
>  	req->cqe->result.u64 = 0;
> -	nvmet_req_complete(req, status);
> +
>  	if (req->sq->dhchap_step == NVME_AUTH_DHCHAP_MESSAGE_SUCCESS2)
>  		nvmet_auth_sq_free(req->sq);
>  	else if (req->sq->dhchap_step == NVME_AUTH_DHCHAP_MESSAGE_FAILURE1) {
>  		nvmet_auth_sq_free(req->sq);
>  		nvmet_ctrl_fatal_error(ctrl);
> +		nvmet_req_complete(req, status);
> +		return;
>  	}
> +	nvmet_req_complete(req, status);

And this looks really odd to me.  Why do a manual complete and return
in the branch just to skip the complete right after it?