iscsi_tcp.c

	switch(conn->in.opcode) {
	case ISCSI_OP_SCSI_DATA_IN:
		rc = iscsi_scsi_data_in(conn);
		break;
	case ISCSI_OP_SCSI_CMD_RSP: {
		/*
		 * SCSI Sense Data:
		 * copying the entire Data Segment.
		 */
		if (iscsi_tcp_copy(conn, conn->data, conn->in.datalen)) {
			rc = -EAGAIN;
			goto exit;
		}

		/*
		 * check for sense
		 */
		conn->in.hdr = &conn->hdr;
		conn->senselen = (conn->data[0] << 8) | conn->data[1];
		rc = iscsi_cmd_rsp(conn, conn->in.ctask);
		if (!rc && conn->datadgst_en)
			iscsi_recv_digest_update(conn, conn->data,
						 conn->in.datalen);
	}
	break;
	case ISCSI_OP_TEXT_RSP:
	case ISCSI_OP_LOGIN_RSP:
	case ISCSI_OP_NOOP_IN: {
		struct iscsi_mgmt_task *mtask = NULL;

		if (conn->in.itt != ISCSI_RESERVED_TAG)
			mtask = (struct iscsi_mgmt_task *)
				session->mgmt_cmds[conn->in.itt -
					ISCSI_MGMT_ITT_OFFSET];

		/*
		 * Collect data segment to the connection's data
		 * placeholder
		 */
		if (iscsi_tcp_copy(conn, conn->data, conn->in.datalen)) {
			rc = -EAGAIN;
			goto exit;
		}

		rc = iscsi_recv_pdu(conn->cls_conn, conn->in.hdr,
				    conn->data, conn->in.datalen);

		if (!rc && conn->datadgst_en &&
			conn->in.opcode != ISCSI_OP_LOGIN_RSP)
			iscsi_recv_digest_update(conn, conn->data,
			  			conn->in.datalen);

		if (mtask && conn->login_mtask != mtask) {
			spin_lock(&session->lock);
			__kfifo_put(session->mgmtpool.queue, (void*)&mtask,
				    sizeof(void*));
			spin_unlock(&session->lock);
		}
	}
	break;
	case ISCSI_OP_ASYNC_EVENT:
	case ISCSI_OP_REJECT:
	default:
		BUG_ON(1);
	}
exit:
	return rc;
}

/**
 * iscsi_tcp_data_recv - TCP receive in sendfile fashion
 * @rd_desc: read descriptor
 * @skb: socket buffer
 * @offset: offset in skb
 * @len: skb->len - offset
 **/
static int
iscsi_tcp_data_recv(read_descriptor_t *rd_desc, struct sk_buff *skb,
		unsigned int offset, size_t len)
{
	int rc;
	struct iscsi_conn *conn = rd_desc->arg.data;
	int processed;
	char pad[ISCSI_PAD_LEN];
	struct scatterlist sg;

	/*
	 * Save current SKB and its offset in the corresponding
	 * connection context.
	 */
	conn->in.copy = skb->len - offset;
	conn->in.offset = offset;
	conn->in.skb = skb;
	conn->in.len = conn->in.copy;
	BUG_ON(conn->in.copy <= 0);
	debug_tcp("in %d bytes\n", conn->in.copy);

more:
	conn->in.copied = 0;
	rc = 0;

	if (unlikely(conn->suspend_rx)) {
		debug_tcp("conn %d Rx suspended!\n", conn->id);
		return 0;
	}

	if (conn->in_progress == IN_PROGRESS_WAIT_HEADER ||
	    conn->in_progress == IN_PROGRESS_HEADER_GATHER) {
		rc = iscsi_hdr_extract(conn);
		if (rc) {
		       if (rc == -EAGAIN)
				goto nomore;
		       else {
				iscsi_conn_failure(conn, rc);
				return 0;
		       }
		}

		/*
		 * Verify and process incoming PDU header.
		 */
		rc = iscsi_hdr_recv(conn);
		if (!rc && conn->in.datalen) {
			if (conn->datadgst_en) {
				BUG_ON(!conn->data_rx_tfm);
				crypto_digest_init(conn->data_rx_tfm);
			}
			conn->in_progress = IN_PROGRESS_DATA_RECV;
		} else if (rc) {
			iscsi_conn_failure(conn, rc);
			return 0;
		}
	}

	if (conn->in_progress == IN_PROGRESS_DDIGEST_RECV) {
		uint32_t recv_digest;
		debug_tcp("extra data_recv offset %d copy %d\n",
			  conn->in.offset, conn->in.copy);
		skb_copy_bits(conn->in.skb, conn->in.offset,
				&recv_digest, 4);
		conn->in.offset += 4;
		conn->in.copy -= 4;
		if (recv_digest != conn->in.datadgst) {
			debug_tcp("iscsi_tcp: data digest error!"
				  "0x%x != 0x%x\n", recv_digest,
				  conn->in.datadgst);
			iscsi_conn_failure(conn, ISCSI_ERR_DATA_DGST);
			return 0;
		} else {
			debug_tcp("iscsi_tcp: data digest match!"
				  "0x%x == 0x%x\n", recv_digest,
				  conn->in.datadgst);
			conn->in_progress = IN_PROGRESS_WAIT_HEADER;
		}
	}

	if (conn->in_progress == IN_PROGRESS_DATA_RECV && conn->in.copy) {

		debug_tcp("data_recv offset %d copy %d\n",
		       conn->in.offset, conn->in.copy);

		rc = iscsi_data_recv(conn);
		if (rc) {
			if (rc == -EAGAIN) {
				rd_desc->count = conn->in.datalen -
						conn->in.ctask->data_count;
				goto again;
			}
			iscsi_conn_failure(conn, rc);
			return 0;
		}
		conn->in.copy -= conn->in.padding;
		conn->in.offset += conn->in.padding;
		if (conn->datadgst_en) {
			if (conn->in.padding) {
				debug_tcp("padding -> %d\n", conn->in.padding);
				memset(pad, 0, conn->in.padding);
				sg_init_one(&sg, pad, conn->in.padding);
				crypto_digest_update(conn->data_rx_tfm, &sg, 1);
			}
			crypto_digest_final(conn->data_rx_tfm,
					    (u8 *) & conn->in.datadgst);
			debug_tcp("rx digest 0x%x\n", conn->in.datadgst);
			conn->in_progress = IN_PROGRESS_DDIGEST_RECV;
		} else
			conn->in_progress = IN_PROGRESS_WAIT_HEADER;
	}

	debug_tcp("f, processed %d from out of %d padding %d\n",
	       conn->in.offset - offset, (int)len, conn->in.padding);
	BUG_ON(conn->in.offset - offset > len);

	if (conn->in.offset - offset != len) {
		debug_tcp("continue to process %d bytes\n",
		       (int)len - (conn->in.offset - offset));
		goto more;
	}

nomore:
	processed = conn->in.offset - offset;
	BUG_ON(processed == 0);
	return processed;

again:
	processed = conn->in.offset - offset;
	debug_tcp("c, processed %d from out of %d rd_desc_cnt %d\n",
	          processed, (int)len, (int)rd_desc->count);
	BUG_ON(processed == 0);
	BUG_ON(processed > len);

	conn->rxdata_octets += processed;
	return processed;
}

static void
iscsi_tcp_data_ready(struct sock *sk, int flag)
{
	struct iscsi_conn *conn = sk->sk_user_data;
	read_descriptor_t rd_desc;

	read_lock(&sk->sk_callback_lock);

	/* use rd_desc to pass 'conn' to iscsi_tcp_data_recv */
	rd_desc.arg.data = conn;
	rd_desc.count = 0;
	tcp_read_sock(sk, &rd_desc, iscsi_tcp_data_recv);

	read_unlock(&sk->sk_callback_lock);
}

static void
iscsi_tcp_state_change(struct sock *sk)
{
	struct iscsi_conn *conn;
	struct iscsi_session *session;
	void (*old_state_change)(struct sock *);

	read_lock(&sk->sk_callback_lock);

	conn = (struct iscsi_conn*)sk->sk_user_data;
	session = conn->session;

	if ((sk->sk_state == TCP_CLOSE_WAIT ||
	     sk->sk_state == TCP_CLOSE) &&
	    !atomic_read(&sk->sk_rmem_alloc)) {
		debug_tcp("iscsi_tcp_state_change: TCP_CLOSE|TCP_CLOSE_WAIT\n");
		iscsi_conn_failure(conn, ISCSI_ERR_CONN_FAILED);
	}

	old_state_change = conn->old_state_change;

	read_unlock(&sk->sk_callback_lock);

	old_state_change(sk);
}

/**
 * iscsi_write_space - Called when more output buffer space is available
 * @sk: socket space is available for
 **/
static void
iscsi_write_space(struct sock *sk)
{
	struct iscsi_conn *conn = (struct iscsi_conn*)sk->sk_user_data;
	conn->old_write_space(sk);
	debug_tcp("iscsi_write_space: cid %d\n", conn->id);
	clear_bit(SUSPEND_BIT, &conn->suspend_tx);
	scsi_queue_work(conn->session->host, &conn->xmitwork);
}

static void
iscsi_conn_set_callbacks(struct iscsi_conn *conn)
{
	struct sock *sk = conn->sock->sk;

	/* assign new callbacks */
	write_lock_bh(&sk->sk_callback_lock);
	sk->sk_user_data = conn;
	conn->old_data_ready = sk->sk_data_ready;
	conn->old_state_change = sk->sk_state_change;
	conn->old_write_space = sk->sk_write_space;
	sk->sk_data_ready = iscsi_tcp_data_ready;
	sk->sk_state_change = iscsi_tcp_state_change;
	sk->sk_write_space = iscsi_write_space;
	write_unlock_bh(&sk->sk_callback_lock);
}

static void
iscsi_conn_restore_callbacks(struct iscsi_conn *conn)
{
	struct sock *sk = conn->sock->sk;

	/* restore socket callbacks, see also: iscsi_conn_set_callbacks() */
	write_lock_bh(&sk->sk_callback_lock);
	sk->sk_user_data    = NULL;
	sk->sk_data_ready   = conn->old_data_ready;
	sk->sk_state_change = conn->old_state_change;
	sk->sk_write_space  = conn->old_write_space;
	sk->sk_no_check	 = 0;
	write_unlock_bh(&sk->sk_callback_lock);
}

/**
 * iscsi_send - generic send routine
 * @sk: kernel's socket
 * @buf: buffer to write from
 * @size: actual size to write
 * @flags: socket's flags
 */
static inline int
iscsi_send(struct iscsi_conn *conn, struct iscsi_buf *buf, int size, int flags)
{
	struct socket *sk = conn->sock;
	int offset = buf->sg.offset + buf->sent;

	/*
	 * if we got use_sg=0 or are sending something we kmallocd
	 * then we did not have to do kmap (kmap returns page_address)
	 *
	 * if we got use_sg > 0, but had to drop down, we do not
	 * set clustering so this should only happen for that
	 * slab case.
	 */
	if (buf->use_sendmsg)
		return sock_no_sendpage(sk, buf->sg.page, offset, size, flags);
	else
		return conn->sendpage(sk, buf->sg.page, offset, size, flags);
}

/**
 * iscsi_sendhdr - send PDU Header via tcp_sendpage()
 * @conn: iscsi connection
 * @buf: buffer to write from
 * @datalen: lenght of data to be sent after the header
 *
 * Notes:
 *	(Tx, Fast Path)
 **/
static inline int
iscsi_sendhdr(struct iscsi_conn *conn, struct iscsi_buf *buf, int datalen)
{
	int flags = 0; /* MSG_DONTWAIT; */
	int res, size;

	size = buf->sg.length - buf->sent;
	BUG_ON(buf->sent + size > buf->sg.length);
	if (buf->sent + size != buf->sg.length || datalen)
		flags |= MSG_MORE;

	res = iscsi_send(conn, buf, size, flags);
	debug_tcp("sendhdr %d bytes, sent %d res %d\n", size, buf->sent, res);
	if (res >= 0) {
		conn->txdata_octets += res;
		buf->sent += res;
		if (size != res)
			return -EAGAIN;
		return 0;
	} else if (res == -EAGAIN) {
		conn->sendpage_failures_cnt++;
		set_bit(SUSPEND_BIT, &conn->suspend_tx);
	} else if (res == -EPIPE)
		iscsi_conn_failure(conn, ISCSI_ERR_CONN_FAILED);

	return res;
}

/**
 * iscsi_sendpage - send one page of iSCSI Data-Out.
 * @conn: iscsi connection
 * @buf: buffer to write from
 * @count: remaining data
 * @sent: number of bytes sent
 *
 * Notes:
 *	(Tx, Fast Path)
 **/
static inline int
iscsi_sendpage(struct iscsi_conn *conn, struct iscsi_buf *buf,
	       int *count, int *sent)
{
	int flags = 0; /* MSG_DONTWAIT; */
	int res, size;

	size = buf->sg.length - buf->sent;
	BUG_ON(buf->sent + size > buf->sg.length);
	if (size > *count)
		size = *count;
	if (buf->sent + size != buf->sg.length || *count != size)
		flags |= MSG_MORE;

	res = iscsi_send(conn, buf, size, flags);
	debug_tcp("sendpage: %d bytes, sent %d left %d sent %d res %d\n",
		  size, buf->sent, *count, *sent, res);
	if (res >= 0) {
		conn->txdata_octets += res;
		buf->sent += res;
		*count -= res;
		*sent += res;
		if (size != res)
			return -EAGAIN;
		return 0;
	} else if (res == -EAGAIN) {
		conn->sendpage_failures_cnt++;
		set_bit(SUSPEND_BIT, &conn->suspend_tx);
	} else if (res == -EPIPE)
		iscsi_conn_failure(conn, ISCSI_ERR_CONN_FAILED);

	return res;
}

static inline void
iscsi_data_digest_init(struct iscsi_conn *conn, struct iscsi_cmd_task *ctask)
{
	BUG_ON(!conn->data_tx_tfm);
	crypto_digest_init(conn->data_tx_tfm);
	ctask->digest_count = 4;
}

static int
iscsi_digest_final_send(struct iscsi_conn *conn, struct iscsi_cmd_task *ctask,
			struct iscsi_buf *buf, uint32_t *digest, int final)
{
	int rc = 0;
	int sent = 0;

	if (final)
		crypto_digest_final(conn->data_tx_tfm, (u8*)digest);

	iscsi_buf_init_virt(buf, (char*)digest, 4);
	rc = iscsi_sendpage(conn, buf, &ctask->digest_count, &sent);
	if (rc) {
		ctask->datadigest = *digest;
		ctask->xmstate |= XMSTATE_DATA_DIGEST;
	} else
		ctask->digest_count = 4;
	return rc;
}

/**
 * iscsi_solicit_data_cont - initialize next Data-Out
 * @conn: iscsi connection
 * @ctask: scsi command task
 * @r2t: R2T info
 * @left: bytes left to transfer
 *
 * Notes:
 *	Initialize next Data-Out within this R2T sequence and continue
 *	to process next Scatter-Gather element(if any) of this SCSI command.
 *
 *	Called under connection lock.
 **/
static void
iscsi_solicit_data_cont(struct iscsi_conn *conn, struct iscsi_cmd_task *ctask,
			struct iscsi_r2t_info *r2t, int left)
{
	struct iscsi_data *hdr;
	struct iscsi_data_task *dtask;
	struct scsi_cmnd *sc = ctask->sc;
	int new_offset;

	dtask = mempool_alloc(ctask->datapool, GFP_ATOMIC);
	BUG_ON(!dtask);
	hdr = &dtask->hdr;
	memset(hdr, 0, sizeof(struct iscsi_data));
	hdr->ttt = r2t->ttt;
	hdr->datasn = cpu_to_be32(r2t->solicit_datasn);
	r2t->solicit_datasn++;
	hdr->opcode = ISCSI_OP_SCSI_DATA_OUT;
	memcpy(hdr->lun, ctask->hdr.lun, sizeof(hdr->lun));
	hdr->itt = ctask->hdr.itt;
	hdr->exp_statsn = r2t->exp_statsn;
	new_offset = r2t->data_offset + r2t->sent;
	hdr->offset = cpu_to_be32(new_offset);
	if (left > conn->max_xmit_dlength) {
		hton24(hdr->dlength, conn->max_xmit_dlength);
		r2t->data_count = conn->max_xmit_dlength;
	} else {
		hton24(hdr->dlength, left);
		r2t->data_count = left;
		hdr->flags = ISCSI_FLAG_CMD_FINAL;
	}
	conn->dataout_pdus_cnt++;

	iscsi_buf_init_virt(&r2t->headbuf, (char*)hdr,
			   sizeof(struct iscsi_hdr));

	r2t->dtask = dtask;

	if (sc->use_sg && !iscsi_buf_left(&r2t->sendbuf)) {
		BUG_ON(ctask->bad_sg == r2t->sg);
		iscsi_buf_init_sg(&r2t->sendbuf, r2t->sg);
		r2t->sg += 1;
	} else
		iscsi_buf_init_iov(&ctask->sendbuf,
			    (char*)sc->request_buffer + new_offset,
			    r2t->data_count);

	list_add(&dtask->item, &ctask->dataqueue);
}

static void
iscsi_unsolicit_data_init(struct iscsi_conn *conn, struct iscsi_cmd_task *ctask)
{
	struct iscsi_data *hdr;
	struct iscsi_data_task *dtask;

	dtask = mempool_alloc(ctask->datapool, GFP_ATOMIC);
	BUG_ON(!dtask);
	hdr = &dtask->hdr;
	memset(hdr, 0, sizeof(struct iscsi_data));
	hdr->ttt = cpu_to_be32(ISCSI_RESERVED_TAG);
	hdr->datasn = cpu_to_be32(ctask->unsol_datasn);
	ctask->unsol_datasn++;
	hdr->opcode = ISCSI_OP_SCSI_DATA_OUT;
	memcpy(hdr->lun, ctask->hdr.lun, sizeof(hdr->lun));
	hdr->itt = ctask->hdr.itt;
	hdr->exp_statsn = cpu_to_be32(conn->exp_statsn);
	hdr->offset = cpu_to_be32(ctask->total_length -
				  ctask->r2t_data_count -
				  ctask->unsol_count);
	if (ctask->unsol_count > conn->max_xmit_dlength) {
		hton24(hdr->dlength, conn->max_xmit_dlength);
		ctask->data_count = conn->max_xmit_dlength;
		hdr->flags = 0;
	} else {
		hton24(hdr->dlength, ctask->unsol_count);
		ctask->data_count = ctask->unsol_count;
		hdr->flags = ISCSI_FLAG_CMD_FINAL;
	}

	iscsi_buf_init_virt(&ctask->headbuf, (char*)hdr,
			   sizeof(struct iscsi_hdr));

	list_add(&dtask->item, &ctask->dataqueue);

	ctask->dtask = dtask;
}

/**
 * iscsi_cmd_init - Initialize iSCSI SCSI_READ or SCSI_WRITE commands
 * @conn: iscsi connection
 * @ctask: scsi command task
 * @sc: scsi command
 **/
static void
iscsi_cmd_init(struct iscsi_conn *conn, struct iscsi_cmd_task *ctask,
		struct scsi_cmnd *sc)
{
	struct iscsi_session *session = conn->session;

	BUG_ON(__kfifo_len(ctask->r2tqueue));

	ctask->sc = sc;
	ctask->conn = conn;
	ctask->hdr.opcode = ISCSI_OP_SCSI_CMD;
	ctask->hdr.flags = ISCSI_ATTR_SIMPLE;
	int_to_scsilun(sc->device->lun, (struct scsi_lun *)ctask->hdr.lun);
	ctask->hdr.itt = ctask->itt | (conn->id << CID_SHIFT) |
			 (session->age << AGE_SHIFT);
	ctask->hdr.data_length = cpu_to_be32(sc->request_bufflen);
	ctask->hdr.cmdsn = cpu_to_be32(session->cmdsn); session->cmdsn++;
	ctask->hdr.exp_statsn = cpu_to_be32(conn->exp_statsn);
	memcpy(ctask->hdr.cdb, sc->cmnd, sc->cmd_len);
	memset(&ctask->hdr.cdb[sc->cmd_len], 0, MAX_COMMAND_SIZE - sc->cmd_len);

	ctask->mtask = NULL;
	ctask->sent = 0;
	ctask->sg_count = 0;

	ctask->total_length = sc->request_bufflen;

	if (sc->sc_data_direction == DMA_TO_DEVICE) {
		ctask->exp_r2tsn = 0;
		ctask->hdr.flags |= ISCSI_FLAG_CMD_WRITE;
		BUG_ON(ctask->total_length == 0);
		if (sc->use_sg) {
			struct scatterlist *sg = sc->request_buffer;

			iscsi_buf_init_sg(&ctask->sendbuf,
					  &sg[ctask->sg_count++]);
			ctask->sg = sg;
			ctask->bad_sg = sg + sc->use_sg;
		} else {
			iscsi_buf_init_iov(&ctask->sendbuf, sc->request_buffer,
					sc->request_bufflen);
		}

		/*
		 * Write counters:
		 *
		 *	imm_count	bytes to be sent right after
		 *			SCSI PDU Header
		 *
		 *	unsol_count	bytes(as Data-Out) to be sent
		 *			without	R2T ack right after
		 *			immediate data
		 *
		 *	r2t_data_count	bytes to be sent via R2T ack's
		 *
		 *      pad_count       bytes to be sent as zero-padding
		 */
		ctask->imm_count = 0;
		ctask->unsol_count = 0;
		ctask->unsol_datasn = 0;
		ctask->xmstate = XMSTATE_W_HDR;
		/* calculate write padding */
		ctask->pad_count = ctask->total_length & (ISCSI_PAD_LEN-1);
		if (ctask->pad_count) {
			ctask->pad_count = ISCSI_PAD_LEN - ctask->pad_count;
			debug_scsi("write padding %d bytes\n",
				ctask->pad_count);
			ctask->xmstate |= XMSTATE_W_PAD;
		}
		if (session->imm_data_en) {
			if (ctask->total_length >= session->first_burst)
				ctask->imm_count = min(session->first_burst,
							conn->max_xmit_dlength);
			else
				ctask->imm_count = min(ctask->total_length,
							conn->max_xmit_dlength);
			hton24(ctask->hdr.dlength, ctask->imm_count);
			ctask->xmstate |= XMSTATE_IMM_DATA;
		} else
			zero_data(ctask->hdr.dlength);

		if (!session->initial_r2t_en)
			ctask->unsol_count = min(session->first_burst,
				ctask->total_length) - ctask->imm_count;
		if (!ctask->unsol_count)
			/* No unsolicit Data-Out's */
			ctask->hdr.flags |= ISCSI_FLAG_CMD_FINAL;
		else
			ctask->xmstate |= XMSTATE_UNS_HDR | XMSTATE_UNS_INIT;

		ctask->r2t_data_count = ctask->total_length -
				    ctask->imm_count -
				    ctask->unsol_count;

		debug_scsi("cmd [itt %x total %d imm %d imm_data %d "
			   "r2t_data %d]\n",
			   ctask->itt, ctask->total_length, ctask->imm_count,
			   ctask->unsol_count, ctask->r2t_data_count);
	} else {
		ctask->hdr.flags |= ISCSI_FLAG_CMD_FINAL;
		if (sc->sc_data_direction == DMA_FROM_DEVICE)
			ctask->hdr.flags |= ISCSI_FLAG_CMD_READ;
		ctask->datasn = 0;
		ctask->xmstate = XMSTATE_R_HDR;
		zero_data(ctask->hdr.dlength);
	}

	iscsi_buf_init_virt(&ctask->headbuf, (char*)&ctask->hdr,
			    sizeof(struct iscsi_hdr));
	conn->scsicmd_pdus_cnt++;
}

/**
 * iscsi_mtask_xmit - xmit management(immediate) task
 * @conn: iscsi connection
 * @mtask: task management task
 *
 * Notes:
 *	The function can return -EAGAIN in which case caller must
 *	call it again later, or recover. '0' return code means successful
 *	xmit.
 *
 *	Management xmit state machine consists of two states:
 *		IN_PROGRESS_IMM_HEAD - PDU Header xmit in progress
 *		IN_PROGRESS_IMM_DATA - PDU Data xmit in progress
 **/
static int
iscsi_mtask_xmit(struct iscsi_conn *conn, struct iscsi_mgmt_task *mtask)
{

	debug_scsi("mtask deq [cid %d state %x itt 0x%x]\n",
		conn->id, mtask->xmstate, mtask->itt);

	if (mtask->xmstate & XMSTATE_IMM_HDR) {
		mtask->xmstate &= ~XMSTATE_IMM_HDR;
		if (mtask->data_count)
			mtask->xmstate |= XMSTATE_IMM_DATA;
		if (conn->c_stage != ISCSI_CONN_INITIAL_STAGE &&
	    	    conn->stop_stage != STOP_CONN_RECOVER &&
		    conn->hdrdgst_en)
			iscsi_hdr_digest(conn, &mtask->headbuf,
					(u8*)mtask->hdrext);
		if (iscsi_sendhdr(conn, &mtask->headbuf, mtask->data_count)) {
			mtask->xmstate |= XMSTATE_IMM_HDR;
			if (mtask->data_count)
				mtask->xmstate &= ~XMSTATE_IMM_DATA;
			return -EAGAIN;
		}
	}

	if (mtask->xmstate & XMSTATE_IMM_DATA) {
		BUG_ON(!mtask->data_count);
		mtask->xmstate &= ~XMSTATE_IMM_DATA;
		/* FIXME: implement.
		 * Virtual buffer could be spreaded across multiple pages...
		 */
		do {
			if (iscsi_sendpage(conn, &mtask->sendbuf,
				   &mtask->data_count, &mtask->sent)) {
				mtask->xmstate |= XMSTATE_IMM_DATA;
				return -EAGAIN;
			}
		} while (mtask->data_count);
	}

	BUG_ON(mtask->xmstate != XMSTATE_IDLE);
	return 0;
}

static inline int
handle_xmstate_r_hdr(struct iscsi_conn *conn, struct iscsi_cmd_task *ctask)
{
	ctask->xmstate &= ~XMSTATE_R_HDR;
	if (conn->hdrdgst_en)
		iscsi_hdr_digest(conn, &ctask->headbuf, (u8*)ctask->hdrext);
	if (!iscsi_sendhdr(conn, &ctask->headbuf, 0)) {
		BUG_ON(ctask->xmstate != XMSTATE_IDLE);
		return 0; /* wait for Data-In */
	}
	ctask->xmstate |= XMSTATE_R_HDR;
	return -EAGAIN;
}

static inline int
handle_xmstate_w_hdr(struct iscsi_conn *conn, struct iscsi_cmd_task *ctask)
{
	ctask->xmstate &= ~XMSTATE_W_HDR;
	if (conn->hdrdgst_en)
		iscsi_hdr_digest(conn, &ctask->headbuf, (u8*)ctask->hdrext);
	if (iscsi_sendhdr(conn, &ctask->headbuf, ctask->imm_count)) {
		ctask->xmstate |= XMSTATE_W_HDR;
		return -EAGAIN;
	}
	return 0;
}

static inline int
handle_xmstate_data_digest(struct iscsi_conn *conn,
			   struct iscsi_cmd_task *ctask)
{
	ctask->xmstate &= ~XMSTATE_DATA_DIGEST;
	debug_tcp("resent data digest 0x%x\n", ctask->datadigest);
	if (iscsi_digest_final_send(conn, ctask, &ctask->immbuf,
				    &ctask->datadigest, 0)) {
		ctask->xmstate |= XMSTATE_DATA_DIGEST;
		debug_tcp("resent data digest 0x%x fail!\n",
			  ctask->datadigest);
		return -EAGAIN;
	}
	return 0;
}

static inline int
handle_xmstate_imm_data(struct iscsi_conn *conn, struct iscsi_cmd_task *ctask)
{
	BUG_ON(!ctask->imm_count);
	ctask->xmstate &= ~XMSTATE_IMM_DATA;

	if (conn->datadgst_en) {
		iscsi_data_digest_init(conn, ctask);
		ctask->immdigest = 0;
	}

	for (;;) {
		if (iscsi_sendpage(conn, &ctask->sendbuf, &ctask->imm_count,
				   &ctask->sent)) {
			ctask->xmstate |= XMSTATE_IMM_DATA;
			if (conn->datadgst_en) {
				crypto_digest_final(conn->data_tx_tfm,
						(u8*)&ctask->immdigest);
				debug_tcp("tx imm sendpage fail 0x%x\n",
					  ctask->datadigest);
			}
			return -EAGAIN;
		}
		if (conn->datadgst_en)
			crypto_digest_update(conn->data_tx_tfm,
					     &ctask->sendbuf.sg, 1);

		if (!ctask->imm_count)
			break;
		iscsi_buf_init_sg(&ctask->sendbuf,
				  &ctask->sg[ctask->sg_count++]);
	}

	if (conn->datadgst_en && !(ctask->xmstate & XMSTATE_W_PAD)) {
		if (iscsi_digest_final_send(conn, ctask, &ctask->immbuf,
				            &ctask->immdigest, 1)) {
			debug_tcp("sending imm digest 0x%x fail!\n",
				  ctask->immdigest);
			return -EAGAIN;
		}
		debug_tcp("sending imm digest 0x%x\n", ctask->immdigest);
	}

	return 0;
}

static inline int
handle_xmstate_uns_hdr(struct iscsi_conn *conn, struct iscsi_cmd_task *ctask)
{
	struct iscsi_data_task *dtask;

	ctask->xmstate |= XMSTATE_UNS_DATA;
	if (ctask->xmstate & XMSTATE_UNS_INIT) {
		iscsi_unsolicit_data_init(conn, ctask);
		BUG_ON(!ctask->dtask);
		dtask = ctask->dtask;
		if (conn->hdrdgst_en)
			iscsi_hdr_digest(conn, &ctask->headbuf,
					(u8*)dtask->hdrext);
		ctask->xmstate &= ~XMSTATE_UNS_INIT;
	}
	if (iscsi_sendhdr(conn, &ctask->headbuf, ctask->data_count)) {
		ctask->xmstate &= ~XMSTATE_UNS_DATA;
		ctask->xmstate |= XMSTATE_UNS_HDR;
		return -EAGAIN;
	}

	debug_scsi("uns dout [itt 0x%x dlen %d sent %d]\n",
		   ctask->itt, ctask->unsol_count, ctask->sent);
	return 0;
}

static inline int
handle_xmstate_uns_data(struct iscsi_conn *conn, struct iscsi_cmd_task *ctask)
{
	struct iscsi_data_task *dtask = ctask->dtask;

	BUG_ON(!ctask->data_count);
	ctask->xmstate &= ~XMSTATE_UNS_DATA;

	if (conn->datadgst_en) {
		iscsi_data_digest_init(conn, ctask);
		dtask->digest = 0;
	}

	for (;;) {
		int start = ctask->sent;

		if (iscsi_sendpage(conn, &ctask->sendbuf, &ctask->data_count,
				   &ctask->sent)) {
			ctask->unsol_count -= ctask->sent - start;
			ctask->xmstate |= XMSTATE_UNS_DATA;
			/* will continue with this ctask later.. */
			if (conn->datadgst_en) {
				crypto_digest_final(conn->data_tx_tfm,
						(u8 *)&dtask->digest);
				debug_tcp("tx uns data fail 0x%x\n",
					  dtask->digest);
			}
			return -EAGAIN;
		}

		BUG_ON(ctask->sent > ctask->total_length);
		ctask->unsol_count -= ctask->sent - start;

		/*
		 * XXX:we may run here with un-initial sendbuf.
		 * so pass it
		 */
		if (conn->datadgst_en && ctask->sent - start > 0)
			crypto_digest_update(conn->data_tx_tfm,
					     &ctask->sendbuf.sg, 1);

		if (!ctask->data_count)
			break;
		iscsi_buf_init_sg(&ctask->sendbuf,
				  &ctask->sg[ctask->sg_count++]);
	}
	BUG_ON(ctask->unsol_count < 0);

	/*
	 * Done with the Data-Out. Next, check if we need
	 * to send another unsolicited Data-Out.
	 */
	if (ctask->unsol_count) {
		if (conn->datadgst_en) {
			if (iscsi_digest_final_send(conn, ctask,
						    &dtask->digestbuf,
						    &dtask->digest, 1)) {
				debug_tcp("send uns digest 0x%x fail\n",
					  dtask->digest);
				return -EAGAIN;
			}
			debug_tcp("sending uns digest 0x%x, more uns\n",
				  dtask->digest);
		}
		ctask->xmstate |= XMSTATE_UNS_INIT;
		return 1;
	}

	if (conn->datadgst_en && !(ctask->xmstate & XMSTATE_W_PAD)) {
		if (iscsi_digest_final_send(conn, ctask,
					    &dtask->digestbuf,
					    &dtask->digest, 1)) {
			debug_tcp("send last uns digest 0x%x fail\n",
				   dtask->digest);
			return -EAGAIN;
		}
		debug_tcp("sending uns digest 0x%x\n",dtask->digest);
	}

	return 0;
}

static inline int
handle_xmstate_sol_data(struct iscsi_conn *conn, struct iscsi_cmd_task *ctask)
{
	struct iscsi_session *session = conn->session;
	struct iscsi_r2t_info *r2t = ctask->r2t;
	struct iscsi_data_task *dtask = r2t->dtask;
	int left;

	ctask->xmstate &= ~XMSTATE_SOL_DATA;
	ctask->dtask = dtask;

	if (conn->datadgst_en) {
		iscsi_data_digest_init(conn, ctask);
		dtask->digest = 0;
	}
solicit_again:
	/*
	 * send Data-Out whitnin this R2T sequence.
	 */
	if (!r2t->data_count)
		goto data_out_done;

	if (iscsi_sendpage(conn, &r2t->sendbuf, &r2t->data_count, &r2t->sent)) {
		ctask->xmstate |= XMSTATE_SOL_DATA;
		/* will continue with this ctask later.. */
		if (conn->datadgst_en) {
			crypto_digest_final(conn->data_tx_tfm,
					  (u8 *)&dtask->digest);
			debug_tcp("r2t data send fail 0x%x\n", dtask->digest);
		}
		return -EAGAIN;
	}

	BUG_ON(r2t->data_count < 0);
	if (conn->datadgst_en)
		crypto_digest_update(conn->data_tx_tfm, &r2t->sendbuf.sg, 1);

	if (r2t->data_count) {
		BUG_ON(ctask->sc->use_sg == 0);
		if (!iscsi_buf_left(&r2t->sendbuf)) {
			BUG_ON(ctask->bad_sg == r2t->sg);
			iscsi_buf_init_sg(&r2t->sendbuf, r2t->sg);
			r2t->sg += 1;
		}
		goto solicit_again;
	}

data_out_done:
	/*
	 * Done with this Data-Out. Next, check if we have
	 * to send another Data-Out for this R2T.
	 */
	BUG_ON(r2t->data_length - r2t->sent < 0);
	left = r2t->data_length - r2t->sent;
	if (left) {
		if (conn->datadgst_en) {
			if (iscsi_digest_final_send(conn, ctask,
						    &dtask->digestbuf,
						    &dtask->digest, 1)) {
				debug_tcp("send r2t data digest 0x%x"
					  "fail\n", dtask->digest);
				return -EAGAIN;
			}
			debug_tcp("r2t data send digest 0x%x\n",
				  dtask->digest);
		}
		iscsi_solicit_data_cont(conn, ctask, r2t, left);
		ctask->xmstate |= XMSTATE_SOL_DATA;
		ctask->xmstate &= ~XMSTATE_SOL_HDR;
		return 1;
	}

	/*
	 * Done with this R2T. Check if there are more
	 * outstanding R2Ts ready to be processed.
	 */
	BUG_ON(ctask->r2t_data_count - r2t->data_length < 0);
	if (conn->datadgst_en) {
		if (iscsi_digest_final_send(conn, ctask, &dtask->digestbuf,
					    &dtask->digest, 1)) {
			debug_tcp("send last r2t data digest 0x%x"
				  "fail\n", dtask->digest);
			return -EAGAIN;
		}
		debug_tcp("r2t done dout digest 0x%x\n", dtask->digest);
	}

	ctask->r2t_data_count -= r2t->data_length;
	ctask->r2t = NULL;
	spin_lock_bh(&session->lock);