Merge tag 'nvme-6.14-2025-03-05' of git://git.infradead.org/nvme into block-6.14

Pull NVMe fixe from Keith:

"nvme fixes for Linux 6.14

 - TCP use after free fix on polling (Sagi)
 - Controller memory buffer cleanup fixes (Icenowy)
 - Free leaking requests on bad user passthrough commands (Keith)
 - TCP error message fix (Maurizio)
 - TCP corruption fix on partial PDU (Maurizio)
 - TCP memory ordering fix for weakly ordered archs (Meir)
 - Type coercion fix on message error for TCP (Dan)"

* tag 'nvme-6.14-2025-03-05' of git://git.infradead.org/nvme:
  nvme-tcp: fix signedness bug in nvme_tcp_init_connection()
  nvmet-tcp: Fix a possible sporadic response drops in weakly ordered arch
  nvme-tcp: fix potential memory corruption in nvme_tcp_recv_pdu()
  nvme-tcp: Fix a C2HTermReq error message
  nvmet: remove old function prototype
  nvme-ioctl: fix leaked requests on mapping error
  nvme-pci: skip CMB blocks incompatible with PCI P2P DMA
  nvme-pci: clean up CMBMSC when registering CMB fails
  nvme-tcp: fix possible UAF in nvme_tcp_poll

Author: axboe

drivers/nvme/host/ioctl.c

@@ -128,8 +128,10 @@ static int nvme_map_user_request(struct request *req, u64 ubuffer,
 	if (!nvme_ctrl_sgl_supported(ctrl))
 		dev_warn_once(ctrl->device, "using unchecked data buffer\n");
 	if (has_metadata) {
-		if (!supports_metadata)
-			return -EINVAL;
+		if (!supports_metadata) {
+			ret = -EINVAL;
+			goto out;
+		}
 		if (!nvme_ctrl_meta_sgl_supported(ctrl))
 			dev_warn_once(ctrl->device,
 				      "using unchecked metadata buffer\n");
@@ -139,8 +141,10 @@ static int nvme_map_user_request(struct request *req, u64 ubuffer,
 		struct iov_iter iter;
 
 		/* fixedbufs is only for non-vectored io */
-		if (WARN_ON_ONCE(flags & NVME_IOCTL_VEC))
-			return -EINVAL;
+		if (WARN_ON_ONCE(flags & NVME_IOCTL_VEC)) {
+			ret = -EINVAL;
+			goto out;
+		}
 		ret = io_uring_cmd_import_fixed(ubuffer, bufflen,
 				rq_data_dir(req), &iter, ioucmd);
 		if (ret < 0)

drivers/nvme/host/pci.c

@@ -1982,6 +1982,18 @@ static void nvme_map_cmb(struct nvme_dev *dev)
 	if (offset > bar_size)
 		return;
 
+	/*
+	 * Controllers may support a CMB size larger than their BAR, for
+	 * example, due to being behind a bridge. Reduce the CMB to the
+	 * reported size of the BAR
+	 */
+	size = min(size, bar_size - offset);
+
+	if (!IS_ALIGNED(size, memremap_compat_align()) ||
+	    !IS_ALIGNED(pci_resource_start(pdev, bar),
+			memremap_compat_align()))
+		return;
+
 	/*
 	 * Tell the controller about the host side address mapping the CMB,
 	 * and enable CMB decoding for the NVMe 1.4+ scheme:
@@ -1992,17 +2004,10 @@ static void nvme_map_cmb(struct nvme_dev *dev)
 			     dev->bar + NVME_REG_CMBMSC);
 	}
 
-	/*
-	 * Controllers may support a CMB size larger than their BAR,
-	 * for example, due to being behind a bridge. Reduce the CMB to
-	 * the reported size of the BAR
-	 */
-	if (size > bar_size - offset)
-		size = bar_size - offset;
-
 	if (pci_p2pdma_add_resource(pdev, bar, size, offset)) {
 		dev_warn(dev->ctrl.device,
 			 "failed to register the CMB\n");
+		hi_lo_writeq(0, dev->bar + NVME_REG_CMBMSC);
 		return;
 	}
 

drivers/nvme/host/tcp.c

@@ -217,6 +217,19 @@ static inline int nvme_tcp_queue_id(struct nvme_tcp_queue *queue)
 	return queue - queue->ctrl->queues;
 }
 
+static inline bool nvme_tcp_recv_pdu_supported(enum nvme_tcp_pdu_type type)
+{
+	switch (type) {
+	case nvme_tcp_c2h_term:
+	case nvme_tcp_c2h_data:
+	case nvme_tcp_r2t:
+	case nvme_tcp_rsp:
+		return true;
+	default:
+		return false;
+	}
+}
+
 /*
  * Check if the queue is TLS encrypted
  */
@@ -775,7 +788,7 @@ static void nvme_tcp_handle_c2h_term(struct nvme_tcp_queue *queue,
 		[NVME_TCP_FES_PDU_SEQ_ERR] = "PDU Sequence Error",
 		[NVME_TCP_FES_HDR_DIGEST_ERR] = "Header Digest Error",
 		[NVME_TCP_FES_DATA_OUT_OF_RANGE] = "Data Transfer Out Of Range",
-		[NVME_TCP_FES_R2T_LIMIT_EXCEEDED] = "R2T Limit Exceeded",
+		[NVME_TCP_FES_DATA_LIMIT_EXCEEDED] = "Data Transfer Limit Exceeded",
 		[NVME_TCP_FES_UNSUPPORTED_PARAM] = "Unsupported Parameter",
 	};
 
@@ -818,6 +831,16 @@ static int nvme_tcp_recv_pdu(struct nvme_tcp_queue *queue, struct sk_buff *skb,
 		return 0;
 
 	hdr = queue->pdu;
+	if (unlikely(hdr->hlen != sizeof(struct nvme_tcp_rsp_pdu))) {
+		if (!nvme_tcp_recv_pdu_supported(hdr->type))
+			goto unsupported_pdu;
+
+		dev_err(queue->ctrl->ctrl.device,
+			"pdu type %d has unexpected header length (%d)\n",
+			hdr->type, hdr->hlen);
+		return -EPROTO;
+	}
+
 	if (unlikely(hdr->type == nvme_tcp_c2h_term)) {
 		/*
 		 * C2HTermReq never includes Header or Data digests.
@@ -850,10 +873,13 @@ static int nvme_tcp_recv_pdu(struct nvme_tcp_queue *queue, struct sk_buff *skb,
 		nvme_tcp_init_recv_ctx(queue);
 		return nvme_tcp_handle_r2t(queue, (void *)queue->pdu);
 	default:
-		dev_err(queue->ctrl->ctrl.device,
-			"unsupported pdu type (%d)\n", hdr->type);
-		return -EINVAL;
+		goto unsupported_pdu;
 	}
+
+unsupported_pdu:
+	dev_err(queue->ctrl->ctrl.device,
+		"unsupported pdu type (%d)\n", hdr->type);
+	return -EINVAL;
 }
 
 static inline void nvme_tcp_end_request(struct request *rq, u16 status)
@@ -1495,11 +1521,11 @@ static int nvme_tcp_init_connection(struct nvme_tcp_queue *queue)
 	msg.msg_flags = MSG_WAITALL;
 	ret = kernel_recvmsg(queue->sock, &msg, &iov, 1,
 			iov.iov_len, msg.msg_flags);
-	if (ret < sizeof(*icresp)) {
+	if (ret >= 0 && ret < sizeof(*icresp))
+		ret = -ECONNRESET;
+	if (ret < 0) {
 		pr_warn("queue %d: failed to receive icresp, error %d\n",
 			nvme_tcp_queue_id(queue), ret);
-		if (ret >= 0)
-			ret = -ECONNRESET;
 		goto free_icresp;
 	}
 	ret = -ENOTCONN;
@@ -2699,16 +2725,17 @@ static int nvme_tcp_poll(struct blk_mq_hw_ctx *hctx, struct io_comp_batch *iob)
 {
 	struct nvme_tcp_queue *queue = hctx->driver_data;
 	struct sock *sk = queue->sock->sk;
+	int ret;
 
 	if (!test_bit(NVME_TCP_Q_LIVE, &queue->flags))
 		return 0;
 
 	set_bit(NVME_TCP_Q_POLLING, &queue->flags);
 	if (sk_can_busy_loop(sk) && skb_queue_empty_lockless(&sk->sk_receive_queue))
 		sk_busy_loop(sk, true);
-	nvme_tcp_try_recv(queue);
+	ret = nvme_tcp_try_recv(queue);
 	clear_bit(NVME_TCP_Q_POLLING, &queue->flags);
-	return queue->nr_cqe;
+	return ret < 0 ? ret : queue->nr_cqe;
 }
 
 static int nvme_tcp_get_address(struct nvme_ctrl *ctrl, char *buf, int size)

drivers/nvme/target/nvmet.h

@@ -647,7 +647,6 @@ void nvmet_subsys_disc_changed(struct nvmet_subsys *subsys,
 		struct nvmet_host *host);
 void nvmet_add_async_event(struct nvmet_ctrl *ctrl, u8 event_type,
 		u8 event_info, u8 log_page);
-bool nvmet_subsys_nsid_exists(struct nvmet_subsys *subsys, u32 nsid);
 
 #define NVMET_MIN_QUEUE_SIZE	16
 #define NVMET_MAX_QUEUE_SIZE	1024

drivers/nvme/target/tcp.c

@@ -571,10 +571,16 @@ static void nvmet_tcp_queue_response(struct nvmet_req *req)
 	struct nvmet_tcp_cmd *cmd =
 		container_of(req, struct nvmet_tcp_cmd, req);
 	struct nvmet_tcp_queue	*queue = cmd->queue;
+	enum nvmet_tcp_recv_state queue_state;
+	struct nvmet_tcp_cmd *queue_cmd;
 	struct nvme_sgl_desc *sgl;
 	u32 len;
 
-	if (unlikely(cmd == queue->cmd)) {
+	/* Pairs with store_release in nvmet_prepare_receive_pdu() */
+	queue_state = smp_load_acquire(&queue->rcv_state);
+	queue_cmd = READ_ONCE(queue->cmd);
+
+	if (unlikely(cmd == queue_cmd)) {
 		sgl = &cmd->req.cmd->common.dptr.sgl;
 		len = le32_to_cpu(sgl->length);
 
@@ -583,7 +589,7 @@ static void nvmet_tcp_queue_response(struct nvmet_req *req)
 		 * Avoid using helpers, this might happen before
 		 * nvmet_req_init is completed.
 		 */
-		if (queue->rcv_state == NVMET_TCP_RECV_PDU &&
+		if (queue_state == NVMET_TCP_RECV_PDU &&
 		    len && len <= cmd->req.port->inline_data_size &&
 		    nvme_is_write(cmd->req.cmd))
 			return;
@@ -847,8 +853,9 @@ static void nvmet_prepare_receive_pdu(struct nvmet_tcp_queue *queue)
 {
 	queue->offset = 0;
 	queue->left = sizeof(struct nvme_tcp_hdr);
-	queue->cmd = NULL;
-	queue->rcv_state = NVMET_TCP_RECV_PDU;
+	WRITE_ONCE(queue->cmd, NULL);
+	/* Ensure rcv_state is visible only after queue->cmd is set */
+	smp_store_release(&queue->rcv_state, NVMET_TCP_RECV_PDU);
 }
 
 static void nvmet_tcp_free_crypto(struct nvmet_tcp_queue *queue)