From: mchristi@redhat.com <mchristi@redhat.com>
Date: Wed, 27 May 2009 16:34:08 -0500
Subject: [scsi] add netlink msg to iscsi IF to support offload
Message-id: 1243460050-16253-4-git-send-email-mchristi@redhat.com
O-Subject: [PATCH 4/6] RHEL 5.4: add netlink msg to iscsi interface to support offload
Bugzilla: 441979

From: Mike Christie <mchristi@redhat.com>

This is for BZ 441979.

bnx2i is not using the normal network stack/tools, so for operations
like DHCP and ARP it uses UIO and a userspace helper. This adds
some netlink messages that bnx2i will use to config the network
for the card after userpace/uio has done whatever setup is necessary.

I have tested this with userspace tools from upstream and RHEL.
This patch should be upstream for 2.6.31 like with the other patches.
Its latest version was posted here:
http://marc.info/?l=linux-scsi&m=124311620603619&w=2
and there do not seem to be any objections now.

This also fixes a issue where userspace had been setting the nl group
expecting multicast to be used, but the kernel was using the unicast
so only one listener was getting messages.

diff --git a/drivers/scsi/scsi_transport_iscsi2.c b/drivers/scsi/scsi_transport_iscsi2.c
index 3f19803..f32d594 100644
--- a/drivers/scsi/scsi_transport_iscsi2.c
+++ b/drivers/scsi/scsi_transport_iscsi2.c
@@ -40,7 +40,6 @@
 #define ISCSI_TRANSPORT_VERSION "2.0-871"
 
 struct iscsi_internal {
-	int daemon_pid;
 	struct scsi_transport_template t;
 	struct iscsi_transport *iscsi_transport;
 	struct list_head list;
@@ -899,23 +898,17 @@ iscsi_if_transport_lookup(struct iscsi_transport *tt)
 }
 
 static int
-iscsi_broadcast_skb(struct sk_buff *skb, gfp_t gfp)
+iscsi_multicast_skb(struct sk_buff *skb, u32 group, gfp_t gfp)
 {
-	return netlink_broadcast(nls, skb, 0, 1, gfp);
-}
+	int err;
 
-static int
-iscsi_unicast_skb(struct sk_buff *skb, int pid)
-{
-	int rc;
+	NETLINK_CB(skb).dst_group = group;
 
-	rc = netlink_unicast(nls, skb, pid, MSG_DONTWAIT);
-	if (rc < 0) {
-		printk(KERN_ERR "iscsi: can not unicast skb (%d)\n", rc);
-		return rc;
-	}
+	err = netlink_broadcast(nls, skb, 0, group, gfp);
+	if (err > 0)
+		err = 0;
 
-	return 0;
+	return err;
 }
 
 int iscsi2_recv_pdu(struct iscsi_cls_conn *conn, struct iscsi_hdr *hdr,
@@ -941,7 +934,7 @@ int iscsi2_recv_pdu(struct iscsi_cls_conn *conn, struct iscsi_hdr *hdr,
 		return -ENOMEM;
 	}
 
-	nlh = __nlmsg_put(skb, priv->daemon_pid, 0, 0, (len - sizeof(*nlh)), 0);
+	nlh = __nlmsg_put(skb, 0, 0, 0, (len - sizeof(*nlh)), 0);
 	ev = NLMSG_DATA(nlh);
 	memset(ev, 0, sizeof(*ev));
 	ev->transport_handle = iscsi_handle(conn->transport);
@@ -952,10 +945,45 @@ int iscsi2_recv_pdu(struct iscsi_cls_conn *conn, struct iscsi_hdr *hdr,
 	memcpy(pdu, hdr, sizeof(struct iscsi_hdr));
 	memcpy(pdu + sizeof(struct iscsi_hdr), data, data_size);
 
-	return iscsi_unicast_skb(skb, priv->daemon_pid);
+	return iscsi_multicast_skb(skb, ISCSI_NL_GRP_ISCSID, GFP_ATOMIC);
 }
 EXPORT_SYMBOL_GPL(iscsi2_recv_pdu);
 
+int iscsi_offload_mesg(struct Scsi_Host *shost,
+		       struct iscsi_transport *transport, uint32_t type,
+		       char *data, uint16_t data_size)
+{
+	struct nlmsghdr	*nlh;
+	struct sk_buff *skb;
+	struct iscsi_uevent *ev;
+	int len = NLMSG_SPACE(sizeof(*ev) + data_size);
+
+	skb = alloc_skb(len, GFP_NOIO);
+	if (!skb) {
+		printk(KERN_ERR "can not deliver iscsi offload message:OOM\n");
+		return -ENOMEM;
+	}
+
+	nlh = __nlmsg_put(skb, 0, 0, 0, (len - sizeof(*nlh)), 0);
+	ev = NLMSG_DATA(nlh);
+	memset(ev, 0, sizeof(*ev));
+	ev->type = type;
+	ev->transport_handle = iscsi_handle(transport);
+	switch (type) {
+	case ISCSI_KEVENT_PATH_REQ:
+		ev->r.req_path.host_no = shost->host_no;
+		break;
+	case ISCSI_KEVENT_IF_DOWN:
+		ev->r.notify_if_down.host_no = shost->host_no;
+		break;
+	}
+
+	memcpy((char*)ev + sizeof(*ev), data, data_size);
+
+	return iscsi_multicast_skb(skb, ISCSI_NL_GRP_UIP, GFP_NOIO);
+}
+EXPORT_SYMBOL_GPL(iscsi_offload_mesg);
+
 void iscsi2_conn_error_event(struct iscsi_cls_conn *conn, enum iscsi_err error)
 {
 	struct nlmsghdr	*nlh;
@@ -975,7 +1003,7 @@ void iscsi2_conn_error_event(struct iscsi_cls_conn *conn, enum iscsi_err error)
 		return;
 	}
 
-	nlh = __nlmsg_put(skb, priv->daemon_pid, 0, 0, (len - sizeof(*nlh)), 0);
+	nlh = __nlmsg_put(skb, 0, 0, 0, (len - sizeof(*nlh)), 0);
 	ev = NLMSG_DATA(nlh);
 	ev->transport_handle = iscsi_handle(conn->transport);
 	ev->type = ISCSI_KEVENT_CONN_ERROR;
@@ -983,7 +1011,7 @@ void iscsi2_conn_error_event(struct iscsi_cls_conn *conn, enum iscsi_err error)
 	ev->r.connerror.cid = conn->cid;
 	ev->r.connerror.sid = iscsi_conn_get_sid(conn);
 
-	iscsi_broadcast_skb(skb, GFP_ATOMIC);
+	iscsi_multicast_skb(skb, ISCSI_NL_GRP_ISCSID, GFP_ATOMIC);
 
 	iscsi_cls_conn_printk(KERN_INFO, conn, "detected conn error (%d)\n",
 			      error);
@@ -991,8 +1019,8 @@ void iscsi2_conn_error_event(struct iscsi_cls_conn *conn, enum iscsi_err error)
 EXPORT_SYMBOL_GPL(iscsi2_conn_error_event);
 
 static int
-iscsi_if_send_reply(int pid, int seq, int type, int done, int multi,
-		      void *payload, int size)
+iscsi_if_send_reply(u32 group, int seq, int type, int done, int multi,
+		    void *payload, int size)
 {
 	struct sk_buff	*skb;
 	struct nlmsghdr	*nlh;
@@ -1006,10 +1034,10 @@ iscsi_if_send_reply(int pid, int seq, int type, int done, int multi,
 		return -ENOMEM;
 	}
 
-	nlh = __nlmsg_put(skb, pid, seq, t, (len - sizeof(*nlh)), 0);
+	nlh = __nlmsg_put(skb, 0, 0, t, (len - sizeof(*nlh)), 0);
 	nlh->nlmsg_flags = flags;
 	memcpy(NLMSG_DATA(nlh), payload, size);
-	return iscsi_unicast_skb(skb, pid);
+	return iscsi_multicast_skb(skb, group, GFP_ATOMIC);
 }
 
 static int
@@ -1046,7 +1074,7 @@ iscsi_if_get_stats(struct iscsi_transport *transport, struct nlmsghdr *nlh)
 			return -ENOMEM;
 		}
 
-		nlhstat = __nlmsg_put(skbstat, priv->daemon_pid, 0, 0,
+		nlhstat = __nlmsg_put(skbstat, 0, 0, 0,
 				      (len - sizeof(*nlhstat)), 0);
 		evstat = NLMSG_DATA(nlhstat);
 		memset(evstat, 0, sizeof(*evstat));
@@ -1070,7 +1098,8 @@ iscsi_if_get_stats(struct iscsi_transport *transport, struct nlmsghdr *nlh)
 		skb_trim(skbstat, NLMSG_ALIGN(actual_size));
 		nlhstat->nlmsg_len = actual_size;
 
-		err = iscsi_unicast_skb(skbstat, priv->daemon_pid);
+		err = iscsi_multicast_skb(skbstat, ISCSI_NL_GRP_ISCSID,
+					  GFP_ATOMIC);
 	} while (err < 0 && err != -ECONNREFUSED);
 
 	return err;
@@ -1104,7 +1133,7 @@ int iscsi2_session_event(struct iscsi_cls_session *session,
 		return -ENOMEM;
 	}
 
-	nlh = __nlmsg_put(skb, priv->daemon_pid, 0, 0, (len - sizeof(*nlh)), 0);
+	nlh = __nlmsg_put(skb, 0, 0, 0, (len - sizeof(*nlh)), 0);
 	ev = NLMSG_DATA(nlh);
 	ev->transport_handle = iscsi_handle(session->transport);
 
@@ -1133,7 +1162,7 @@ int iscsi2_session_event(struct iscsi_cls_session *session,
 	 * this will occur if the daemon is not up, so we just warn
 	 * the user and when the daemon is restarted it will handle it
 	 */
-	rc = iscsi_broadcast_skb(skb, GFP_KERNEL);
+	rc = iscsi_multicast_skb(skb, ISCSI_NL_GRP_ISCSID, GFP_KERNEL);
 	if (rc < 0)
 		iscsi_cls_session_printk(KERN_ERR, session,
 					 "Cannot notify userspace of session "
@@ -1354,7 +1383,31 @@ iscsi_set_host_param(struct iscsi_transport *transport,
 }
 
 static int
-iscsi_if_recv_msg(struct sk_buff *skb, struct nlmsghdr *nlh)
+iscsi_set_path(struct iscsi_transport *transport, struct iscsi_uevent *ev)
+{
+	struct Scsi_Host *shost;
+	struct iscsi_path *params;
+	int err;
+
+	if (!transport->set_path)
+		return -ENOSYS;
+
+	shost = scsi_host_lookup(ev->u.set_path.host_no);
+	if (!shost) {
+		printk(KERN_ERR "set path could not find host no %u\n",
+		       ev->u.set_path.host_no);
+		return -ENODEV;
+	}
+
+	params = (struct iscsi_path *)((char*)ev + sizeof(*ev));
+	err = transport->set_path(shost, params);
+				      
+	scsi_host_put(shost);
+	return err;
+}
+
+static int
+iscsi_if_recv_msg(struct sk_buff *skb, struct nlmsghdr *nlh, u32 *group)
 {
 	int err = 0;
 	struct iscsi_uevent *ev = NLMSG_DATA(nlh);
@@ -1364,6 +1417,11 @@ iscsi_if_recv_msg(struct sk_buff *skb, struct nlmsghdr *nlh)
 	struct iscsi_cls_conn *conn;
 	struct iscsi_endpoint *ep = NULL;
 
+	if (nlh->nlmsg_type == ISCSI_UEVENT_PATH_UPDATE)
+		*group = ISCSI_NL_GRP_UIP;
+	else
+		*group = ISCSI_NL_GRP_ISCSID;
+
 	priv = iscsi_if_transport_lookup(iscsi_ptr(ev->transport_handle));
 	if (!priv)
 		return -EINVAL;
@@ -1372,8 +1430,6 @@ iscsi_if_recv_msg(struct sk_buff *skb, struct nlmsghdr *nlh)
 	if (!try_module_get(transport->owner))
 		return -EINVAL;
 
-	priv->daemon_pid = NETLINK_CREDS(skb)->pid;
-
 	switch (nlh->nlmsg_type) {
 	case ISCSI_UEVENT_CREATE_SESSION:
 		err = iscsi_if_create_session(priv, ep, ev,
@@ -1467,6 +1523,9 @@ iscsi_if_recv_msg(struct sk_buff *skb, struct nlmsghdr *nlh)
 	case ISCSI_UEVENT_SET_HOST_PARAM:
 		err = iscsi_set_host_param(transport, ev);
 		break;
+	case ISCSI_UEVENT_PATH_UPDATE:
+		err = iscsi_set_path(transport, ev);
+		break;
 	default:
 		err = -ENOSYS;
 		break;
@@ -1498,6 +1557,7 @@ iscsi_if_rx(struct sock *sk, int len)
 			uint32_t rlen;
 			struct nlmsghdr	*nlh;
 			struct iscsi_uevent *ev;
+			uint32_t group;
 
 			nlh = nlmsg_hdr(skb);
 			if (nlh->nlmsg_len < sizeof(*nlh) ||
@@ -1510,7 +1570,7 @@ iscsi_if_rx(struct sock *sk, int len)
 			if (rlen > skb->len)
 				rlen = skb->len;
 
-			err = iscsi_if_recv_msg(skb, nlh);
+			err = iscsi_if_recv_msg(skb, nlh, &group);
 			if (err) {
 				ev->type = ISCSI_KEVENT_IF_ERROR;
 				ev->iferror = err;
@@ -1524,8 +1584,7 @@ iscsi_if_rx(struct sock *sk, int len)
 				 */
 				if (ev->type == ISCSI_UEVENT_GET_STATS && !err)
 					break;
-				err = iscsi_if_send_reply(
-					NETLINK_CREDS(skb)->pid, nlh->nlmsg_seq,
+				err = iscsi_if_send_reply(group, nlh->nlmsg_seq,
 					nlh->nlmsg_type, 0, 0, ev, sizeof(*ev));
 			} while (err < 0 && err != -ECONNREFUSED);
 			skb_pull(skb, rlen);
@@ -1775,7 +1834,6 @@ iscsi2_register_transport(struct iscsi_transport *tt)
 	if (!priv)
 		return NULL;
 	INIT_LIST_HEAD(&priv->list);
-	priv->daemon_pid = -1;
 	priv->iscsi_transport = tt;
 	priv->t.user_scan = iscsi_user_scan;
 	priv->t.create_work_queue = 1;
diff --git a/include/scsi/iscsi_if2.h b/include/scsi/iscsi_if2.h
index 8721bc5..9eda614 100644
--- a/include/scsi/iscsi_if2.h
+++ b/include/scsi/iscsi_if2.h
@@ -22,6 +22,11 @@
 #define ISCSI_IF_H
 
 #include <scsi/iscsi_proto2.h>
+#include <linux/in.h>
+#include <linux/in6.h>
+
+#define ISCSI_NL_GRP_ISCSID	1
+#define ISCSI_NL_GRP_UIP	2
 
 #define UEVENT_BASE			10
 #define KEVENT_BASE			100
@@ -53,6 +58,8 @@ enum iscsi_uevent_e {
 	ISCSI_UEVENT_CREATE_BOUND_SESSION		= UEVENT_BASE + 18,
 	ISCSI_UEVENT_TRANSPORT_EP_CONNECT_THROUGH_HOST	= UEVENT_BASE + 19,
 
+	ISCSI_UEVENT_PATH_UPDATE	= UEVENT_BASE + 20,
+
 	/* up events */
 	ISCSI_KEVENT_RECV_PDU		= KEVENT_BASE + 1,
 	ISCSI_KEVENT_CONN_ERROR		= KEVENT_BASE + 2,
@@ -60,6 +67,9 @@ enum iscsi_uevent_e {
 	ISCSI_KEVENT_DESTROY_SESSION	= KEVENT_BASE + 4,
 	ISCSI_KEVENT_UNBIND_SESSION	= KEVENT_BASE + 5,
 	ISCSI_KEVENT_CREATE_SESSION	= KEVENT_BASE + 6,
+
+	ISCSI_KEVENT_PATH_REQ		= KEVENT_BASE + 7,
+	ISCSI_KEVENT_IF_DOWN		= KEVENT_BASE + 8,
 };
 
 enum iscsi_tgt_dscvr {
@@ -159,6 +169,9 @@ struct iscsi_uevent {
 			uint32_t	param; /* enum iscsi_host_param */
 			uint32_t	len;
 		} set_host_param;
+		struct msg_set_path {
+			uint32_t	host_no;
+		} set_path;
 	} u;
 	union {
 		/* messages k -> u */
@@ -192,10 +205,39 @@ struct iscsi_uevent {
 		struct msg_transport_connect_ret {
 			uint64_t	handle;
 		} ep_connect_ret;
+		struct msg_req_path {
+			uint32_t	host_no;
+		} req_path;
+		struct msg_notify_if_down {
+			uint32_t	host_no;
+		} notify_if_down;
 	} r;
 } __attribute__ ((aligned (sizeof(uint64_t))));
 
 /*
+ * To keep the struct iscsi_uevent size the same for userspace code
+ * compatibility, the main structure for ISCSI_UEVENT_PATH_UPDATE and
+ * ISCSI_KEVENT_PATH_REQ is defined separately and comes after the
+ * struct iscsi_uevent in the NETLINK_ISCSI message.
+ */
+struct iscsi_path {
+	uint64_t	handle;
+	uint8_t		mac_addr[6];
+	uint8_t		mac_addr_old[6];
+	uint32_t	ip_addr_len;	/* 4 or 16 */
+	union {
+		struct in_addr	v4_addr;
+		struct in6_addr	v6_addr;
+	} src;
+	union {
+		struct in_addr	v4_addr;
+		struct in6_addr	v6_addr;
+	} dst;
+	uint16_t	vlan_id;
+	uint16_t	pmtu;
+} __attribute__ ((aligned (sizeof(uint64_t))));
+
+/*
  * Common error codes
  */
 enum iscsi_err {
diff --git a/include/scsi/scsi_transport_iscsi2.h b/include/scsi/scsi_transport_iscsi2.h
index 42b5e64..d337cf4 100644
--- a/include/scsi/scsi_transport_iscsi2.h
+++ b/include/scsi/scsi_transport_iscsi2.h
@@ -134,6 +134,7 @@ struct iscsi_transport {
 	void (*ep_disconnect) (struct iscsi_endpoint *ep);
 	int (*tgt_dscvr) (struct Scsi_Host *shost, enum iscsi_tgt_dscvr type,
 			  uint32_t enable, struct sockaddr *dst_addr);
+	int (*set_path) (struct Scsi_Host *shost, struct iscsi_path *params);
 };
 
 /*
@@ -149,6 +150,9 @@ extern void iscsi2_conn_error_event(struct iscsi_cls_conn *conn,
 				    enum iscsi_err error);
 extern int iscsi2_recv_pdu(struct iscsi_cls_conn *conn, struct iscsi_hdr *hdr,
 			   char *data, uint32_t data_size);
+extern int iscsi_offload_mesg(struct Scsi_Host *shost,
+			      struct iscsi_transport *transport, uint32_t type,
+			      char *data, uint16_t data_size);
 
 struct iscsi_cls_conn {
 	struct list_head conn_list;	/* item in connlist */