[PATCH 1/2] nvmet-tcp: propagate nvmet_tcp_build_pdu_iovec() errors to its callers

[Keith Busch]

On Wed, Mar 11, 2026 at 05:09:19PM +0100, Maurizio Lombardi wrote:
> Currently, when nvmet_tcp_build_pdu_iovec() detects an out-of-bounds
> PDU length or offset, it triggers nvmet_tcp_fatal_error(cmd->queue)
> and returns early. However, because the function returns void, the
> callers are entirely unaware that a fatal error has occurred and
> that the cmd->recv_msg.msg_iter was left uninitialized.
> 
> Callers such as nvmet_tcp_handle_h2c_data_pdu() proceed to blindly
> overwrite the queue state with queue->rcv_state = NVMET_TCP_RECV_DATA
> Consequently, the socket receiving loop may attempt to read incoming
> network data into the uninitialized iterator.
> 
> Fix this by shifting the error handling responsibility to the callers.

This looks good to me. Just some minor comments below.

> +	ret = nvmet_tcp_build_pdu_iovec(cmd);
> +	if (unlikely(ret)) {
> +		pr_err("queue %d: failed to build PDU iovec", queue->idx);

Missing the '\n' in this print.

> @@ -1092,7 +1098,12 @@ static int nvmet_tcp_done_recv_pdu(struct nvmet_tcp_queue *queue)
>  	if (nvmet_tcp_need_data_in(queue->cmd)) {
>  		if (nvmet_tcp_has_inline_data(queue->cmd)) {
>  			queue->rcv_state = NVMET_TCP_RECV_DATA;
> -			nvmet_tcp_build_pdu_iovec(queue->cmd);
> +			ret = nvmet_tcp_build_pdu_iovec(queue->cmd);
> +			if (unlikely(ret)) {
> +				pr_err("queue %d: failed to build PDU iovec\n",
> +					queue->idx);
> +				return ret;
> +			}
>  			return 0;

Minor nit here, you can squash the two rets into a single 'return ret;'.