Merge tag 'dlm-5.14' of git://git.kernel.org/pub/scm/linux/kernel/git/teigland/linux-dlm

Pull dlm updates from David Teigland: "This is a major dlm networking enhancement that adds message retransmission so that the dlm can reliably continue operating when network connections fail and nodes reconnect. Previously, this would result in lost messages which could only be handled as a node failure" * tag 'dlm-5.14' of git://git.kernel.org/pub/scm/linux/kernel/git/teigland/linux-dlm: (26 commits) fs: dlm: invalid buffer access in lookup error fs: dlm: fix race in mhandle deletion fs: dlm: rename socket and app buffer defines fs: dlm: introduce proto values fs: dlm: move dlm allow conn fs: dlm: use alloc_ordered_workqueue fs: dlm: fix memory leak when fenced fs: dlm: fix lowcomms_start error case fs: dlm: Fix spelling mistake "stucked" -> "stuck" fs: dlm: Fix memory leak of object mh fs: dlm: don't allow half transmitted messages fs: dlm: add midcomms debugfs functionality fs: dlm: add reliable connection if reconnect fs: dlm: add union in dlm header for lockspace id fs: dlm: move out some hash functionality fs: dlm: add functionality to re-transmit a message fs: dlm: make buffer handling per msg fs: dlm: add more midcomms hooks fs: dlm: public header in out utility fs: dlm: fix connection tcp EOF handling ...
author: Linus Torvalds <torvalds@linux-foundation.org> 2021-06-29 20:30:00 -0700
committer: Linus Torvalds <torvalds@linux-foundation.org> 2021-06-29 20:30:00 -0700
commit: 007b350a58754a93ca9fe50c498cc27780171153 (patch)
tree: c9e1c689bc30662faf4c591fefec2d9d10afddff /fs
parent: 8418dabd97ce4b0713c36dd8226978f737c342b0 (diff)
parent: 957adb68b3f7df8421a05f1647d3027f2acad310 (diff)
14 files changed, 1924 insertions, 191 deletions
diff --git a/fs/dlm/config.c b/fs/dlm/config.c
index 88d95d96e36c..42eee2783756 100644
--- a/fs/dlm/config.c
+++ b/fs/dlm/config.c
@@ -20,6 +20,7 @@
 #include <net/sock.h>
 
 #include "config.h"
+#include "midcomms.h"
 #include "lowcomms.h"
 
 /*
@@ -79,6 +80,9 @@ struct dlm_cluster {
 	unsigned int cl_new_rsb_count;
 	unsigned int cl_recover_callbacks;
 	char cl_cluster_name[DLM_LOCKSPACE_LEN];
+
+	struct dlm_spaces *sps;
+	struct dlm_comms *cms;
 };
 
 static struct dlm_cluster *config_item_to_cluster(struct config_item *i)
@@ -204,7 +208,7 @@ static int dlm_check_zero(unsigned int x)
 
 static int dlm_check_buffer_size(unsigned int x)
 {
-	if (x < DEFAULT_BUFFER_SIZE)
+	if (x < DLM_MAX_SOCKET_BUFSIZE)
 		return -EINVAL;
 
 	return 0;
@@ -409,6 +413,9 @@ static struct config_group *make_cluster(struct config_group *g,
 	if (!cl || !sps || !cms)
 		goto fail;
 
+	cl->sps = sps;
+	cl->cms = cms;
+
 	config_group_init_type_name(&cl->group, name, &cluster_type);
 	config_group_init_type_name(&sps->ss_group, "spaces", &spaces_type);
 	config_group_init_type_name(&cms->cs_group, "comms", &comms_type);
@@ -458,6 +465,9 @@ static void drop_cluster(struct config_group *g, struct config_item *i)
 static void release_cluster(struct config_item *i)
 {
 	struct dlm_cluster *cl = config_item_to_cluster(i);
+
+	kfree(cl->sps);
+	kfree(cl->cms);
 	kfree(cl);
 }
 
@@ -532,7 +542,7 @@ static void drop_comm(struct config_group *g, struct config_item *i)
 	struct dlm_comm *cm = config_item_to_comm(i);
 	if (local_comm == cm)
 		local_comm = NULL;
-	dlm_lowcomms_close(cm->nodeid);
+	dlm_midcomms_close(cm->nodeid);
 	while (cm->addr_count--)
 		kfree(cm->addr[cm->addr_count]);
 	config_item_put(i);
@@ -942,7 +952,7 @@ int dlm_our_addr(struct sockaddr_storage *addr, int num)
 #define DEFAULT_SCAN_SECS          5
 #define DEFAULT_LOG_DEBUG          0
 #define DEFAULT_LOG_INFO           1
-#define DEFAULT_PROTOCOL           0
+#define DEFAULT_PROTOCOL           DLM_PROTO_TCP
 #define DEFAULT_MARK               0
 #define DEFAULT_TIMEWARN_CS      500 /* 5 sec = 500 centiseconds */
 #define DEFAULT_WAITWARN_US	   0
@@ -952,7 +962,7 @@ int dlm_our_addr(struct sockaddr_storage *addr, int num)
 
 struct dlm_config_info dlm_config = {
 	.ci_tcp_port = DEFAULT_TCP_PORT,
-	.ci_buffer_size = DEFAULT_BUFFER_SIZE,
+	.ci_buffer_size = DLM_MAX_SOCKET_BUFSIZE,
 	.ci_rsbtbl_size = DEFAULT_RSBTBL_SIZE,
 	.ci_recover_timer = DEFAULT_RECOVER_TIMER,
 	.ci_toss_secs = DEFAULT_TOSS_SECS,
diff --git a/fs/dlm/config.h b/fs/dlm/config.h
index d2cd4bd20313..df92b0a07fc6 100644
--- a/fs/dlm/config.h
+++ b/fs/dlm/config.h
@@ -12,7 +12,7 @@
 #ifndef __CONFIG_DOT_H__
 #define __CONFIG_DOT_H__
 
-#define DEFAULT_BUFFER_SIZE     4096
+#define DLM_MAX_SOCKET_BUFSIZE	4096
 
 struct dlm_config_node {
 	int nodeid;
@@ -23,6 +23,9 @@ struct dlm_config_node {
 
 #define DLM_MAX_ADDR_COUNT 3
 
+#define DLM_PROTO_TCP	0
+#define DLM_PROTO_SCTP	1
+
 struct dlm_config_info {
 	int ci_tcp_port;
 	int ci_buffer_size;
diff --git a/fs/dlm/debug_fs.c b/fs/dlm/debug_fs.c
index d5bd990bcab8..47e9d57e4cae 100644
--- a/fs/dlm/debug_fs.c
+++ b/fs/dlm/debug_fs.c
@@ -16,6 +16,7 @@
 #include <linux/slab.h>
 
 #include "dlm_internal.h"
+#include "midcomms.h"
 #include "lock.h"
 
 #define DLM_DEBUG_BUF_LEN 4096
@@ -23,6 +24,7 @@ static char debug_buf[DLM_DEBUG_BUF_LEN];
 static struct mutex debug_buf_lock;
 
 static struct dentry *dlm_root;
+static struct dentry *dlm_comms;
 
 static char *print_lockmode(int mode)
 {
@@ -738,6 +740,57 @@ void dlm_delete_debug_file(struct dlm_ls *ls)
 	debugfs_remove(ls->ls_debug_toss_dentry);
 }
 
+static int dlm_state_show(struct seq_file *file, void *offset)
+{
+	seq_printf(file, "%s\n", dlm_midcomms_state(file->private));
+	return 0;
+}
+DEFINE_SHOW_ATTRIBUTE(dlm_state);
+
+static int dlm_flags_show(struct seq_file *file, void *offset)
+{
+	seq_printf(file, "%lu\n", dlm_midcomms_flags(file->private));
+	return 0;
+}
+DEFINE_SHOW_ATTRIBUTE(dlm_flags);
+
+static int dlm_send_queue_cnt_show(struct seq_file *file, void *offset)
+{
+	seq_printf(file, "%d\n", dlm_midcomms_send_queue_cnt(file->private));
+	return 0;
+}
+DEFINE_SHOW_ATTRIBUTE(dlm_send_queue_cnt);
+
+static int dlm_version_show(struct seq_file *file, void *offset)
+{
+	seq_printf(file, "0x%08x\n", dlm_midcomms_version(file->private));
+	return 0;
+}
+DEFINE_SHOW_ATTRIBUTE(dlm_version);
+
+void *dlm_create_debug_comms_file(int nodeid, void *data)
+{
+	struct dentry *d_node;
+	char name[256];
+
+	memset(name, 0, sizeof(name));
+	snprintf(name, 256, "%d", nodeid);
+
+	d_node = debugfs_create_dir(name, dlm_comms);
+	debugfs_create_file("state", 0444, d_node, data, &dlm_state_fops);
+	debugfs_create_file("flags", 0444, d_node, data, &dlm_flags_fops);
+	debugfs_create_file("send_queue_count", 0444, d_node, data,
+			    &dlm_send_queue_cnt_fops);
+	debugfs_create_file("version", 0444, d_node, data, &dlm_version_fops);
+
+	return d_node;
+}
+
+void dlm_delete_debug_comms_file(void *ctx)
+{
+	debugfs_remove(ctx);
+}
+
 void dlm_create_debug_file(struct dlm_ls *ls)
 {
 	char name[DLM_LOCKSPACE_LEN + 8];
@@ -797,6 +850,7 @@ void __init dlm_register_debugfs(void)
 {
 	mutex_init(&debug_buf_lock);
 	dlm_root = debugfs_create_dir("dlm", NULL);
+	dlm_comms = debugfs_create_dir("comms", dlm_root);
 }
 
 void dlm_unregister_debugfs(void)
diff --git a/fs/dlm/dlm_internal.h b/fs/dlm/dlm_internal.h
index 04fe9f525ac7..91d1ca3a121a 100644
--- a/fs/dlm/dlm_internal.h
+++ b/fs/dlm/dlm_internal.h
@@ -57,9 +57,12 @@ struct dlm_header;
 struct dlm_message;
 struct dlm_rcom;
 struct dlm_mhandle;
+struct dlm_msg;
 
 #define log_print(fmt, args...) \
 	printk(KERN_ERR "dlm: "fmt"\n" , ##args)
+#define log_print_ratelimited(fmt, args...) \
+	printk_ratelimited(KERN_ERR "dlm: "fmt"\n", ##args)
 #define log_error(ls, fmt, args...) \
 	printk(KERN_ERR "dlm: %s: " fmt "\n", (ls)->ls_name , ##args)
 
@@ -368,23 +371,33 @@ static inline int rsb_flag(struct dlm_rsb *r, enum rsb_flags flag)
 /* dlm_header is first element of all structs sent between nodes */
 
 #define DLM_HEADER_MAJOR	0x00030000
-#define DLM_HEADER_MINOR	0x00000001
+#define DLM_HEADER_MINOR	0x00000002
+
+#define DLM_VERSION_3_1		0x00030001
+#define DLM_VERSION_3_2		0x00030002
 
 #define DLM_HEADER_SLOTS	0x00000001
 
 #define DLM_MSG			1
 #define DLM_RCOM		2
+#define DLM_OPTS		3
+#define DLM_ACK			4
+#define DLM_FIN			5
 
 struct dlm_header {
 	uint32_t		h_version;
-	uint32_t		h_lockspace;
+	union {
+		/* for DLM_MSG and DLM_RCOM */
+		uint32_t	h_lockspace;
+		/* for DLM_ACK and DLM_OPTS */
+		uint32_t	h_seq;
+	} u;
 	uint32_t		h_nodeid;	/* nodeid of sender */
 	uint16_t		h_length;
 	uint8_t			h_cmd;		/* DLM_MSG, DLM_RCOM */
 	uint8_t			h_pad;
 };
 
-
 #define DLM_MSG_REQUEST		1
 #define DLM_MSG_CONVERT		2
 #define DLM_MSG_UNLOCK		3
@@ -452,10 +465,29 @@ struct dlm_rcom {
 	char			rc_buf[];
 };
 
+struct dlm_opt_header {
+	uint16_t	t_type;
+	uint16_t	t_length;
+	uint32_t	o_pad;
+	/* need to be 8 byte aligned */
+	char		t_value[];
+};
+
+/* encapsulation header */
+struct dlm_opts {
+	struct dlm_header	o_header;
+	uint8_t			o_nextcmd;
+	uint8_t			o_pad;
+	uint16_t		o_optlen;
+	uint32_t		o_pad2;
+	char			o_opts[];
+};
+
 union dlm_packet {
 	struct dlm_header	header;		/* common to other two */
 	struct dlm_message	message;
 	struct dlm_rcom		rcom;
+	struct dlm_opts		opts;
 };
 
 #define DLM_RSF_NEED_SLOTS	0x00000001
@@ -722,11 +754,15 @@ void dlm_register_debugfs(void);
 void dlm_unregister_debugfs(void);
 void dlm_create_debug_file(struct dlm_ls *ls);
 void dlm_delete_debug_file(struct dlm_ls *ls);
+void *dlm_create_debug_comms_file(int nodeid, void *data);
+void dlm_delete_debug_comms_file(void *ctx);
 #else
 static inline void dlm_register_debugfs(void) { }
 static inline void dlm_unregister_debugfs(void) { }
 static inline void dlm_create_debug_file(struct dlm_ls *ls) { }
 static inline void dlm_delete_debug_file(struct dlm_ls *ls) { }
+static inline void *dlm_create_debug_comms_file(int nodeid, void *data) { return NULL; }
+static inline void dlm_delete_debug_comms_file(void *ctx) { }
 #endif
 
 #endif				/* __DLM_INTERNAL_DOT_H__ */
diff --git a/fs/dlm/lock.c b/fs/dlm/lock.c
index b93df39d0915..c502c065d007 100644
--- a/fs/dlm/lock.c
+++ b/fs/dlm/lock.c
@@ -59,7 +59,7 @@
 #include "dlm_internal.h"
 #include <linux/dlm_device.h>
 #include "memory.h"
-#include "lowcomms.h"
+#include "midcomms.h"
 #include "requestqueue.h"
 #include "util.h"
 #include "dir.h"
@@ -3534,17 +3534,17 @@ static int _create_message(struct dlm_ls *ls, int mb_len,
 	char *mb;
 
 	/* get_buffer gives us a message handle (mh) that we need to
-	   pass into lowcomms_commit and a message buffer (mb) that we
+	   pass into midcomms_commit and a message buffer (mb) that we
 	   write our data into */
 
-	mh = dlm_lowcomms_get_buffer(to_nodeid, mb_len, GFP_NOFS, &mb);
+	mh = dlm_midcomms_get_mhandle(to_nodeid, mb_len, GFP_NOFS, &mb);
 	if (!mh)
 		return -ENOBUFS;
 
 	ms = (struct dlm_message *) mb;
 
 	ms->m_header.h_version = (DLM_HEADER_MAJOR | DLM_HEADER_MINOR);
-	ms->m_header.h_lockspace = ls->ls_global_id;
+	ms->m_header.u.h_lockspace = ls->ls_global_id;
 	ms->m_header.h_nodeid = dlm_our_nodeid();
 	ms->m_header.h_length = mb_len;
 	ms->m_header.h_cmd = DLM_MSG;
@@ -3589,7 +3589,7 @@ static int create_message(struct dlm_rsb *r, struct dlm_lkb *lkb,
 static int send_message(struct dlm_mhandle *mh, struct dlm_message *ms)
 {
 	dlm_message_out(ms);
-	dlm_lowcomms_commit_buffer(mh);
+	dlm_midcomms_commit_mhandle(mh);
 	return 0;
 }
 
@@ -5038,16 +5038,16 @@ void dlm_receive_buffer(union dlm_packet *p, int nodeid)
 
 	if (hd->h_nodeid != nodeid) {
 		log_print("invalid h_nodeid %d from %d lockspace %x",
-			  hd->h_nodeid, nodeid, hd->h_lockspace);
+			  hd->h_nodeid, nodeid, hd->u.h_lockspace);
 		return;
 	}
 
-	ls = dlm_find_lockspace_global(hd->h_lockspace);
+	ls = dlm_find_lockspace_global(hd->u.h_lockspace);
 	if (!ls) {
 		if (dlm_config.ci_log_debug) {
 			printk_ratelimited(KERN_DEBUG "dlm: invalid lockspace "
 				"%u from %d cmd %d type %d\n",
-				hd->h_lockspace, nodeid, hd->h_cmd, type);
+				hd->u.h_lockspace, nodeid, hd->h_cmd, type);
 		}
 
 		if (hd->h_cmd == DLM_RCOM && type == DLM_RCOM_STATUS)
diff --git a/fs/dlm/lockspace.c b/fs/dlm/lockspace.c
index c14cf2b7faab..d71aba8c3e64 100644
--- a/fs/dlm/lockspace.c
+++ b/fs/dlm/lockspace.c
@@ -16,6 +16,7 @@
 #include "member.h"
 #include "recoverd.h"
 #include "dir.h"
+#include "midcomms.h"
 #include "lowcomms.h"
 #include "config.h"
 #include "memory.h"
@@ -390,7 +391,7 @@ static int threads_start(void)
 	}
 
 	/* Thread for sending/receiving messages for all lockspace's */
-	error = dlm_lowcomms_start();
+	error = dlm_midcomms_start();
 	if (error) {
 		log_print("cannot start dlm lowcomms %d", error);
 		goto scand_fail;
@@ -566,7 +567,12 @@ static int new_lockspace(const char *name, const char *cluster,
 	mutex_init(&ls->ls_requestqueue_mutex);
 	mutex_init(&ls->ls_clear_proc_locks);
 
-	ls->ls_recover_buf = kmalloc(LOWCOMMS_MAX_TX_BUFFER_LEN, GFP_NOFS);
+	/* Due backwards compatibility with 3.1 we need to use maximum
+	 * possible dlm message size to be sure the message will fit and
+	 * not having out of bounds issues. However on sending side 3.2
+	 * might send less.
+	 */
+	ls->ls_recover_buf = kmalloc(DLM_MAX_SOCKET_BUFSIZE, GFP_NOFS);
 	if (!ls->ls_recover_buf)
 		goto out_lkbidr;
 
@@ -698,7 +704,7 @@ int dlm_new_lockspace(const char *name, const char *cluster,
 		error = 0;
 	if (!ls_count) {
 		dlm_scand_stop();
-		dlm_lowcomms_shutdown();
+		dlm_midcomms_shutdown();
 		dlm_lowcomms_stop();
 	}
  out:
@@ -787,7 +793,7 @@ static int release_lockspace(struct dlm_ls *ls, int force)
 
 	if (ls_count == 1) {
 		dlm_scand_stop();
-		dlm_lowcomms_shutdown();
+		dlm_midcomms_shutdown();
 	}
 
 	dlm_callback_stop(ls);
diff --git a/fs/dlm/lowcomms.c b/fs/dlm/lowcomms.c
index 166e36fcf3e4..0ea9ae35da0b 100644
--- a/fs/dlm/lowcomms.c
+++ b/fs/dlm/lowcomms.c
@@ -59,7 +59,6 @@
 #include "config.h"
 
 #define NEEDED_RMEM (4*1024*1024)
-#define CONN_HASH_SIZE 32
 
 /* Number of messages to send before rescheduling */
 #define MAX_SEND_MSG_COUNT 25
@@ -79,14 +78,20 @@ struct connection {
 #define CF_CLOSING 8
 #define CF_SHUTDOWN 9
 #define CF_CONNECTED 10
+#define CF_RECONNECT 11
+#define CF_DELAY_CONNECT 12
+#define CF_EOF 13
 	struct list_head writequeue;  /* List of outgoing writequeue_entries */
 	spinlock_t writequeue_lock;
+	atomic_t writequeue_cnt;
 	void (*connect_action) (struct connection *);	/* What to do to connect */
 	void (*shutdown_action)(struct connection *con); /* What to do to shutdown */
+	bool (*eof_condition)(struct connection *con); /* What to do to eof check */
 	int retries;
 #define MAX_CONNECT_RETRIES 3
 	struct hlist_node list;
 	struct connection *othercon;
+	struct connection *sendcon;
 	struct work_struct rwork; /* Receive workqueue */
 	struct work_struct swork; /* Send workqueue */
 	wait_queue_head_t shutdown_wait; /* wait for graceful shutdown */
@@ -113,7 +118,22 @@ struct writequeue_entry {
 	int len;
 	int end;
 	int users;
+	bool dirty;
 	struct connection *con;
+	struct list_head msgs;
+	struct kref ref;
+};
+
+struct dlm_msg {
+	struct writequeue_entry *entry;
+	struct dlm_msg *orig_msg;
+	bool retransmit;
+	void *ppc;
+	int len;
+	int idx; /* new()/commit() idx exchange */
+
+	struct list_head list;
+	struct kref ref;
 };
 
 struct dlm_node_addr {
@@ -155,33 +175,23 @@ static void sctp_connect_to_sock(struct connection *con);
 static void tcp_connect_to_sock(struct connection *con);
 static void dlm_tcp_shutdown(struct connection *con);
 
-/* This is deliberately very simple because most clusters have simple
-   sequential nodeids, so we should be able to go straight to a connection
-   struct in the array */
-static inline int nodeid_hash(int nodeid)
+static struct connection *__find_con(int nodeid, int r)
 {
-	return nodeid & (CONN_HASH_SIZE-1);
-}
-
-static struct connection *__find_con(int nodeid)
-{
-	int r, idx;
 	struct connection *con;
 
-	r = nodeid_hash(nodeid);
-
-	idx = srcu_read_lock(&connections_srcu);
 	hlist_for_each_entry_rcu(con, &connection_hash[r], list) {
-		if (con->nodeid == nodeid) {
-			srcu_read_unlock(&connections_srcu, idx);
+		if (con->nodeid == nodeid)
 			return con;
-		}
 	}
-	srcu_read_unlock(&connections_srcu, idx);
 
 	return NULL;
 }
 
+static bool tcp_eof_condition(struct connection *con)
+{
+	return atomic_read(&con->writequeue_cnt);
+}
+
 static int dlm_con_init(struct connection *con, int nodeid)
 {
 	con->rx_buflen = dlm_config.ci_buffer_size;
@@ -193,15 +203,23 @@ static int dlm_con_init(struct connection *con, int nodeid)
 	mutex_init(&con->sock_mutex);
 	INIT_LIST_HEAD(&con->writequeue);
 	spin_lock_init(&con->writequeue_lock);
+	atomic_set(&con->writequeue_cnt, 0);
 	INIT_WORK(&con->swork, process_send_sockets);
 	INIT_WORK(&con->rwork, process_recv_sockets);
 	init_waitqueue_head(&con->shutdown_wait);
 
-	if (dlm_config.ci_protocol == 0) {
+	switch (dlm_config.ci_protocol) {
+	case DLM_PROTO_TCP:
 		con->connect_action = tcp_connect_to_sock;
 		con->shutdown_action = dlm_tcp_shutdown;
-	} else {
+		con->eof_condition = tcp_eof_condition;
+		break;
+	case DLM_PROTO_SCTP:
 		con->connect_action = sctp_connect_to_sock;
+		break;
+	default:
+		kfree(con->rx_buf);
+		return -EINVAL;
 	}
 
 	return 0;
@@ -216,7 +234,8 @@ static struct connection *nodeid2con(int nodeid, gfp_t alloc)
 	struct connection *con, *tmp;
 	int r, ret;
 
-	con = __find_con(nodeid);
+	r = nodeid_hash(nodeid);
+	con = __find_con(nodeid, r);
 	if (con || !alloc)
 		return con;
 
@@ -230,8 +249,6 @@ static struct connection *nodeid2con(int nodeid, gfp_t alloc)
 		return NULL;
 	}
 
-	r = nodeid_hash(nodeid);
-
 	spin_lock(&connections_lock);
 	/* Because multiple workqueues/threads calls this function it can
 	 * race on multiple cpu's. Instead of locking hot path __find_con()
@@ -239,7 +256,7 @@ static struct connection *nodeid2con(int nodeid, gfp_t alloc)
 	 * under protection of connections_lock. If this is the case we
 	 * abort our connection creation and return the existing connection.
 	 */
-	tmp = __find_con(nodeid);
+	tmp = __find_con(nodeid, r);
 	if (tmp) {
 		spin_unlock(&connections_lock);
 		kfree(con->rx_buf);
@@ -256,15 +273,13 @@ static struct connection *nodeid2con(int nodeid, gfp_t alloc)
 /* Loop round all connections */
 static void foreach_conn(void (*conn_func)(struct connection *c))
 {
-	int i, idx;
+	int i;
 	struct connection *con;
 
-	idx = srcu_read_lock(&connections_srcu);
 	for (i = 0; i < CONN_HASH_SIZE; i++) {
 		hlist_for_each_entry_rcu(con, &connection_hash[i], list)
 			conn_func(con);
 	}
-	srcu_read_unlock(&connections_srcu, idx);
 }
 
 static struct dlm_node_addr *find_node_addr(int nodeid)
@@ -462,6 +477,9 @@ static void lowcomms_data_ready(struct sock *sk)
 
 static void lowcomms_listen_data_ready(struct sock *sk)
 {
+	if (!dlm_allow_conn)
+		return;
+
 	queue_work(recv_workqueue, &listen_con.rwork);
 }
 
@@ -518,14 +536,21 @@ static void lowcomms_state_change(struct sock *sk)
 int dlm_lowcomms_connect_node(int nodeid)
 {
 	struct connection *con;
+	int idx;
 
 	if (nodeid == dlm_our_nodeid())
 		return 0;
 
+	idx = srcu_read_lock(&connections_srcu);
 	con = nodeid2con(nodeid, GFP_NOFS);
-	if (!con)
+	if (!con) {
+		srcu_read_unlock(&connections_srcu, idx);
 		return -ENOMEM;
+	}
+
 	lowcomms_connect_sock(con);
+	srcu_read_unlock(&connections_srcu, idx);
+
 	return 0;
 }
 
@@ -587,6 +612,22 @@ static void lowcomms_error_report(struct sock *sk)
 				   dlm_config.ci_tcp_port, sk->sk_err,
 				   sk->sk_err_soft);
 	}
+
+	/* below sendcon only handling */
+	if (test_bit(CF_IS_OTHERCON, &con->flags))
+		con = con->sendcon;
+
+	switch (sk->sk_err) {
+	case ECONNREFUSED:
+		set_bit(CF_DELAY_CONNECT, &con->flags);
+		break;
+	default:
+		break;
+	}
+
+	if (!test_and_set_bit(CF_RECONNECT, &con->flags))
+		queue_work(send_workqueue, &con->swork);
+
 out:
 	read_unlock_bh(&sk->sk_callback_lock);
 	if (orig_report)
@@ -669,6 +710,42 @@ static void make_sockaddr(struct sockaddr_storage *saddr, uint16_t port,
 	memset((char *)saddr + *addr_len, 0, sizeof(struct sockaddr_storage) - *addr_len);
 }
 
+static void dlm_page_release(struct kref *kref)
+{
+	struct writequeue_entry *e = container_of(kref, struct writequeue_entry,
+						  ref);
+
+	__free_page(e->page);
+	kfree(e);
+}
+
+static void dlm_msg_release(struct kref *kref)
+{
+	struct dlm_msg *msg = container_of(kref, struct dlm_msg, ref);
+
+	kref_put(&msg->entry->ref, dlm_page_release);
+	kfree(msg);
+}
+
+static void free_entry(struct writequeue_entry *e)
+{
+	struct dlm_msg *msg, *tmp;
+
+	list_for_each_entry_safe(msg, tmp, &e->msgs, list) {
+		if (msg->orig_msg) {
+			msg->orig_msg->retransmit = false;
+			kref_put(&msg->orig_msg->ref, dlm_msg_release);
+		}
+
+		list_del(&msg->list);
+		kref_put(&msg->ref, dlm_msg_release);
+	}
+
+	list_del(&e->list);
+	atomic_dec(&e->con->writequeue_cnt);
+	kref_put(&e->ref, dlm_page_release);
+}
+
 static void dlm_close_sock(struct socket **sock)
 {
 	if (*sock) {
@@ -683,6 +760,7 @@ static void close_connection(struct connection *con, bool and_other,
 			     bool tx, bool rx)
 {
 	bool closing = test_and_set_bit(CF_CLOSING, &con->flags);
+	struct writequeue_entry *e;
 
 	if (tx && !closing && cancel_work_sync(&con->swork)) {
 		log_print("canceled swork for node %d", con->nodeid);
@@ -698,12 +776,35 @@ static void close_connection(struct connection *con, bool and_other,
 
 	if (con->othercon && and_other) {
 		/* Will only re-enter once. */
-		close_connection(con->othercon, false, true, true);
+		close_connection(con->othercon, false, tx, rx);
+	}
+
+	/* if we send a writequeue entry only a half way, we drop the
+	 * whole entry because reconnection and that we not start of the
+	 * middle of a msg which will confuse the other end.
+	 *
+	 * we can always drop messages because retransmits, but what we
+	 * cannot allow is to transmit half messages which may be processed
+	 * at the other side.
+	 *
+	 * our policy is to start on a clean state when disconnects, we don't
+	 * know what's send/received on transport layer in this case.
+	 */
+	spin_lock(&con->writequeue_lock);
+	if (!list_empty(&con->writequeue)) {
+		e = list_first_entry(&con->writequeue, struct writequeue_entry,
+				     list);
+		if (e->dirty)
+			free_entry(e);
 	}
+	spin_unlock(&con->writequeue_lock);
 
 	con->rx_leftover = 0;
 	con->retries = 0;
 	clear_bit(CF_CONNECTED, &con->flags);
+	clear_bit(CF_DELAY_CONNECT, &con->flags);
+	clear_bit(CF_RECONNECT, &con->flags);
+	clear_bit(CF_EOF, &con->flags);
 	mutex_unlock(&con->sock_mutex);
 	clear_bit(CF_CLOSING, &con->flags);
 }
@@ -841,19 +942,26 @@ out_resched:
 	return -EAGAIN;
 
 out_close:
-	mutex_unlock(&con->sock_mutex);
-	if (ret != -EAGAIN) {
-		/* Reconnect when there is something to send */
-		close_connection(con, false, true, false);
-		if (ret == 0) {
-			log_print("connection %p got EOF from %d",
-				  con, con->nodeid);
+	if (ret == 0) {
+		log_print("connection %p got EOF from %d",
+			  con, con->nodeid);
+
+		if (con->eof_condition && con->eof_condition(con)) {
+			set_bit(CF_EOF, &con->flags);
+			mutex_unlock(&con->sock_mutex);
+		} else {
+			mutex_unlock(&con->sock_mutex);
+			close_connection(con, false, true, false);
+
 			/* handling for tcp shutdown */
 			clear_bit(CF_SHUTDOWN, &con->flags);
 			wake_up(&con->shutdown_wait);
-			/* signal to breaking receive worker */
-			ret = -1;
 		}
+
+		/* signal to breaking receive worker */
+		ret = -1;
+	} else {
+		mutex_unlock(&con->sock_mutex);
 	}
 	return ret;
 }
@@ -864,16 +972,12 @@ static int accept_from_sock(struct listen_connection *con)
 	int result;
 	struct sockaddr_storage peeraddr;
 	struct socket *newsock;
-	int len;
+	int len, idx;
 	int nodeid;
 	struct connection *newcon;
 	struct connection *addcon;
 	unsigned int mark;
 
-	if (!dlm_allow_conn) {
-		return -1;
-	}
-
 	if (!con->sock)
 		return -ENOTCONN;
 
@@ -907,8 +1011,10 @@ static int accept_from_sock(struct listen_connection *con)
 	 *  the same time and the connections cross on the wire.
 	 *  In this case we store the incoming one in "othercon"
 	 */
+	idx = srcu_read_lock(&connections_srcu);
 	newcon = nodeid2con(nodeid, GFP_NOFS);
 	if (!newcon) {
+		srcu_read_unlock(&connections_srcu, idx);
 		result = -ENOMEM;
 		goto accept_err;
 	}
@@ -924,6 +1030,7 @@ static int accept_from_sock(struct listen_connection *con)
 			if (!othercon) {
 				log_print("failed to allocate incoming socket");
 				mutex_unlock(&newcon->sock_mutex);
+				srcu_read_unlock(&connections_srcu, idx);
 				result = -ENOMEM;
 				goto accept_err;
 			}
@@ -932,11 +1039,14 @@ static int accept_from_sock(struct listen_connection *con)
 			if (result < 0) {
 				kfree(othercon);
 				mutex_unlock(&newcon->sock_mutex);
+				srcu_read_unlock(&connections_srcu, idx);
 				goto accept_err;
 			}
 
 			lockdep_set_subclass(&othercon->sock_mutex, 1);
+			set_bit(CF_IS_OTHERCON, &othercon->flags);
 			newcon->othercon = othercon;
+			othercon->sendcon = newcon;
 		} else {
 			/* close other sock con if we have something new */
 			close_connection(othercon, false, true, false);
@@ -966,6 +1076,8 @@ static int accept_from_sock(struct listen_connection *con)
 	if (!test_and_set_bit(CF_READ_PENDING, &addcon->flags))
 		queue_work(recv_workqueue, &addcon->rwork);
 
+	srcu_read_unlock(&connections_srcu, idx);
+
 	return 0;
 
 accept_err:
@@ -977,12 +1089,6 @@ accept_err:
 	return result;
 }
 
-static void free_entry(struct writequeue_entry *e)
-{
-	__free_page(e->page);
-	kfree(e);
-}
-
 /*
  * writequeue_entry_complete - try to delete and free write queue entry
  * @e: write queue entry to try to delete
@@ -994,11 +1100,11 @@ static void writequeue_entry_complete(struct writequeue_entry *e, int completed)
 {
 	e->offset += completed;
 	e->len -= completed;
+	/* signal that page was half way transmitted */
+	e->dirty = true;
 
-	if (e->len == 0 && e->users == 0) {
-		list_del(&e->list);
+	if (e->len == 0 && e->users == 0)
 		free_entry(e);
-	}
 }
 
 /*
@@ -1075,7 +1181,7 @@ static void sctp_connect_to_sock(struct connection *con)
 
 	make_sockaddr(&daddr, dlm_config.ci_tcp_port, &addr_len);
 
-	log_print("connecting to %d", con->nodeid);
+	log_print_ratelimited("connecting to %d", con->nodeid);
 
 	/* Turn off Nagle's algorithm */
 	sctp_sock_set_nodelay(sock->sk);
@@ -1171,7 +1277,7 @@ static void tcp_connect_to_sock(struct connection *con)
 
 	make_sockaddr(&saddr, dlm_config.ci_tcp_port, &addr_len);
 
-	log_print("connecting to %d", con->nodeid);
+	log_print_ratelimited("connecting to %d", con->nodeid);
 
 	/* Turn off Nagle's algorithm */
 	tcp_sock_set_nodelay(sock->sk);
@@ -1364,12 +1470,16 @@ static struct writequeue_entry *new_writequeue_entry(struct connection *con,
 
 	entry->con = con;
 	entry->users = 1;
+	kref_init(&entry->ref);
+	INIT_LIST_HEAD(&entry->msgs);
 
 	return entry;
 }
 
 static struct writequeue_entry *new_wq_entry(struct connection *con, int len,
-					     gfp_t allocation, char **ppc)
+					     gfp_t allocation, char **ppc,
+					     void (*cb)(struct dlm_mhandle *mh),
+					     struct dlm_mhandle *mh)
 {
 	struct writequeue_entry *e;
 
@@ -1377,7 +1487,12 @@ static struct writequeue_entry *new_wq_entry(struct connection *con, int len,
 	if (!list_empty(&con->writequeue)) {
 		e = list_last_entry(&con->writequeue, struct writequeue_entry, list);
 		if (DLM_WQ_REMAIN_BYTES(e) >= len) {
+			kref_get(&e->ref);
+
 			*ppc = page_address(e->page) + e->end;
+			if (cb)
+				cb(mh);
+
 			e->end += len;
 			e->users++;
 			spin_unlock(&con->writequeue_lock);
@@ -1391,42 +1506,92 @@ static struct writequeue_entry *new_wq_entry(struct connection *con, int len,
 	if (!e)
 		return NULL;
 
+	kref_get(&e->ref);
 	*ppc = page_address(e->page);
 	e->end += len;
+	atomic_inc(&con->writequeue_cnt);
 
 	spin_lock(&con->writequeue_lock);
+	if (cb)
+		cb(mh);
+
 	list_add_tail(&e->list, &con->writequeue);
 	spin_unlock(&con->writequeue_lock);
 
 	return e;
 };
 
-void *dlm_lowcomms_get_buffer(int nodeid, int len, gfp_t allocation, char **ppc)
+static struct dlm_msg *dlm_lowcomms_new_msg_con(struct connection *con, int len,
+						gfp_t allocation, char **ppc,
+						void (*cb)(struct dlm_mhandle *mh),
+						struct dlm_mhandle *mh)
+{
+	struct writequeue_entry *e;
+	struct dlm_msg *msg;
+
+	msg = kzalloc(sizeof(*msg), allocation);
+	if (!msg)
+		return NULL;
+
+	kref_init(&msg->ref);
+
+	e = new_wq_entry(con, len, allocation, ppc, cb, mh);
+	if (!e) {
+		kfree(msg);
+		return NULL;
+	}
+
+	msg->ppc = *ppc;
+	msg->len = len;
+	msg->entry = e;
+
+	return msg;
+}
+
+struct dlm_msg *dlm_lowcomms_new_msg(int nodeid, int len, gfp_t allocation,
+				     char **ppc, void (*cb)(struct dlm_mhandle *mh),
+				     struct dlm_mhandle *mh)
 {
 	struct connection *con;
+	struct dlm_msg *msg;
+	int idx;
 
-	if (len > DEFAULT_BUFFER_SIZE ||
+	if (len > DLM_MAX_SOCKET_BUFSIZE ||
 	    len < sizeof(struct dlm_header)) {
-		BUILD_BUG_ON(PAGE_SIZE < DEFAULT_BUFFER_SIZE);
+		BUILD_BUG_ON(PAGE_SIZE < DLM_MAX_SOCKET_BUFSIZE);
 		log_print("failed to allocate a buffer of size %d", len);
 		WARN_ON(1);
 		return NULL;
 	}
 
+	idx = srcu_read_lock(&connections_srcu);
 	con = nodeid2con(nodeid, allocation);
-	if (!con)
+	if (!con) {
+		srcu_read_unlock(&connections_srcu, idx);
 		return NULL;
+	}
 
-	return new_wq_entry(con, len, allocation, ppc);
+	msg = dlm_lowcomms_new_msg_con(con, len, allocation, ppc, cb, mh);
+	if (!msg) {
+		srcu_read_unlock(&connections_srcu, idx);
+		return NULL;
+	}
+
+	/* we assume if successful commit must called */
+	msg->idx = idx;
+	return msg;
 }
 
-void dlm_lowcomms_commit_buffer(void *mh)
+static void _dlm_lowcomms_commit_msg(struct dlm_msg *msg)
 {
-	struct writequeue_entry *e = (struct writequeue_entry *)mh;
+	struct writequeue_entry *e = msg->entry;
 	struct connection *con = e->con;
 	int users;
 
 	spin_lock(&con->writequeue_lock);
+	kref_get(&msg->ref);
+	list_add(&msg->list, &e->msgs);
+
 	users = --e->users;
 	if (users)
 		goto out;
@@ -1442,6 +1607,42 @@ out:
 	return;
 }
 
+void dlm_lowcomms_commit_msg(struct dlm_msg *msg)
+{
+	_dlm_lowcomms_commit_msg(msg);
+	srcu_read_unlock(&connections_srcu, msg->idx);
+}
+
+void dlm_lowcomms_put_msg(struct dlm_msg *msg)
+{
+	kref_put(&msg->ref, dlm_msg_release);
+}
+
+/* does not held connections_srcu, usage workqueue only */
+int dlm_lowcomms_resend_msg(struct dlm_msg *msg)
+{
+	struct dlm_msg *msg_resend;
+	char *ppc;
+
+	if (msg->retransmit)
+		return 1;
+
+	msg_resend = dlm_lowcomms_new_msg_con(msg->entry->con, msg->len,
+					      GFP_ATOMIC, &ppc, NULL, NULL);
+	if (!msg_resend)
+		return -ENOMEM;
+
+	msg->retransmit = true;
+	kref_get(&msg->ref);
+	msg_resend->orig_msg = msg;
+
+	memcpy(ppc, msg->ppc, msg->len);
+	_dlm_lowcomms_commit_msg(msg_resend);
+	dlm_lowcomms_put_msg(msg_resend);
+
+	return 0;
+}
+
 /* Send a message */
 static void send_to_sock(struct connection *con)
 {
@@ -1483,7 +1684,7 @@ static void send_to_sock(struct connection *con)
 				cond_resched();
 				goto out;
 			} else if (ret < 0)
-				goto send_error;
+				goto out;
 		}
 
 		/* Don't starve people filling buffers */
@@ -1496,16 +1697,23 @@ static void send_to_sock(struct connection *con)
 		writequeue_entry_complete(e, ret);
 	}
 	spin_unlock(&con->writequeue_lock);
-out:
-	mutex_unlock(&con->sock_mutex);
+
+	/* close if we got EOF */
+	if (test_and_clear_bit(CF_EOF, &con->flags)) {
+		mutex_unlock(&con->sock_mutex);
+		close_connection(con, false, false, true);
+
+		/* handling for tcp shutdown */
+		clear_bit(CF_SHUTDOWN, &con->flags);
+		wake_up(&con->shutdown_wait);
+	} else {
+		mutex_unlock(&con->sock_mutex);
+	}
+
 	return;
 
-send_error:
+out:
 	mutex_unlock(&con->sock_mutex);
-	close_connection(con, false, false, true);
-	/* Requeue the send work. When the work daemon runs again, it will try
-	   a new connection, then call this function again. */
-	queue_work(send_workqueue, &con->swork);
 	return;
 
 out_connect:
@@ -1520,7 +1728,6 @@ static void clean_one_writequeue(struct connection *con)
 
 	spin_lock(&con->writequeue_lock);
 	list_for_each_entry_safe(e, safe, &con->writequeue, list) {
-		list_del(&e->list);
 		free_entry(e);
 	}
 	spin_unlock(&con->writequeue_lock);
@@ -1532,8 +1739,10 @@ int dlm_lowcomms_close(int nodeid)
 {
 	struct connection *con;
 	struct dlm_node_addr *na;
+	int idx;
 
 	log_print("closing connection to node %d", nodeid);
+	idx = srcu_read_lock(&connections_srcu);
 	con = nodeid2con(nodeid, 0);
 	if (con) {
 		set_bit(CF_CLOSE, &con->flags);
@@ -1542,6 +1751,7 @@ int dlm_lowcomms_close(int nodeid)
 		if (con->othercon)
 			clean_one_writequeue(con->othercon);
 	}
+	srcu_read_unlock(&connections_srcu, idx);
 
 	spin_lock(&dlm_node_addrs_spin);
 	na = find_node_addr(nodeid);
@@ -1578,35 +1788,50 @@ static void process_send_sockets(struct work_struct *work)
 {
 	struct connection *con = container_of(work, struct connection, swork);
 
+	WARN_ON(test_bit(CF_IS_OTHERCON, &con->flags));
+
 	clear_bit(CF_WRITE_PENDING, &con->flags);
-	if (con->sock == NULL) /* not mutex protected so check it inside too */
+
+	if (test_and_clear_bit(CF_RECONNECT, &con->flags)) {
+		close_connection(con, false, false, true);
+		dlm_midcomms_unack_msg_resend(con->nodeid);
+	}
+
+	if (con->sock == NULL) { /* not mutex protected so check it inside too */
+		if (test_and_clear_bit(CF_DELAY_CONNECT, &con->flags))
+			msleep(1000);
 		con->connect_action(con);
+	}
 	if (!list_empty(&con->writequeue))
 		send_to_sock(con);
 }
 
 static void work_stop(void)
 {
-	if (recv_workqueue)
+	if (recv_workqueue) {
 		destroy_workqueue(recv_workqueue);
-	if (send_workqueue)
+		recv_workqueue = NULL;
+	}
+
+	if (send_workqueue) {
 		destroy_workqueue(send_workqueue);
+		send_workqueue = NULL;
+	}
 }
 
 static int work_start(void)
 {
-	recv_workqueue = alloc_workqueue("dlm_recv",
-					 WQ_UNBOUND | WQ_MEM_RECLAIM, 1);
+	recv_workqueue = alloc_ordered_workqueue("dlm_recv", WQ_MEM_RECLAIM);
 	if (!recv_workqueue) {
 		log_print("can't start dlm_recv");
 		return -ENOMEM;
 	}
 
-	send_workqueue = alloc_workqueue("dlm_send",
-					 WQ_UNBOUND | WQ_MEM_RECLAIM, 1);
+	send_workqueue = alloc_ordered_workqueue("dlm_send", WQ_MEM_RECLAIM);
 	if (!send_workqueue) {
 		log_print("can't start dlm_send");
 		destroy_workqueue(recv_workqueue);
+		recv_workqueue = NULL;
 		return -ENOMEM;
 	}
 
@@ -1621,6 +1846,8 @@ static void shutdown_conn(struct connection *con)
 
 void dlm_lowcomms_shutdown(void)
 {
+	int idx;
+
 	/* Set all the flags to prevent any
 	 * socket activity.
 	 */
@@ -1633,7 +1860,9 @@ void dlm_lowcomms_shutdown(void)
 
 	dlm_close_sock(&listen_con.sock);
 
+	idx = srcu_read_lock(&connections_srcu);
 	foreach_conn(shutdown_conn);
+	srcu_read_unlock(&connections_srcu, idx);
 }
 
 static void _stop_conn(struct connection *con, bool and_other)
@@ -1682,7 +1911,7 @@ static void free_conn(struct connection *con)
 
 static void work_flush(void)
 {
-	int ok, idx;
+	int ok;
 	int i;
 	struct connection *con;
 
@@ -1693,7 +1922,6 @@ static void work_flush(void)
 			flush_workqueue(recv_workqueue);
 		if (send_workqueue)
 			flush_workqueue(send_workqueue);
-		idx = srcu_read_lock(&connections_srcu);
 		for (i = 0; i < CONN_HASH_SIZE && ok; i++) {
 			hlist_for_each_entry_rcu(con, &connection_hash[i],
 						 list) {
@@ -1707,14 +1935,17 @@ static void work_flush(void)
 				}
 			}
 		}
-		srcu_read_unlock(&connections_srcu, idx);
 	} while (!ok);
 }
 
 void dlm_lowcomms_stop(void)
 {
+	int idx;
+
+	idx = srcu_read_lock(&connections_srcu);
 	work_flush();
 	foreach_conn(free_conn);
+	srcu_read_unlock(&connections_srcu, idx);
 	work_stop();
 	deinit_local();
 }
@@ -1738,15 +1969,24 @@ int dlm_lowcomms_start(void)
 
 	error = work_start();
 	if (error)
-		goto fail;
+		goto fail_local;
 
 	dlm_allow_conn = 1;
 
 	/* Start listening */
-	if (dlm_config.ci_protocol == 0)
+	switch (dlm_config.ci_protocol) {
+	case DLM_PROTO_TCP:
 		error = tcp_listen_for_all();
-	else
+		break;
+	case DLM_PROTO_SCTP:
 		error = sctp_listen_for_all(&listen_con);
+		break;
+	default:
+		log_print("Invalid protocol identifier %d set",
+			  dlm_config.ci_protocol);
+		error = -EINVAL;
+		break;
+	}
 	if (error)
 		goto fail_unlisten;
 
@@ -1755,6 +1995,9 @@ int dlm_lowcomms_start(void)
 fail_unlisten:
 	dlm_allow_conn = 0;
 	dlm_close_sock(&listen_con.sock);
+	work_stop();
+fail_local:
+	deinit_local();
 fail:
 	return error;
 }
diff --git a/fs/dlm/lowcomms.h b/fs/dlm/lowcomms.h
index 48bbc4e18761..aaae7115c00d 100644
--- a/fs/dlm/lowcomms.h
+++ b/fs/dlm/lowcomms.h
@@ -12,7 +12,22 @@
 #ifndef __LOWCOMMS_DOT_H__
 #define __LOWCOMMS_DOT_H__
 
-#define LOWCOMMS_MAX_TX_BUFFER_LEN	4096
+#include "dlm_internal.h"
+
+#define DLM_MIDCOMMS_OPT_LEN		sizeof(struct dlm_opts)
+#define DLM_MAX_APP_BUFSIZE		(DLM_MAX_SOCKET_BUFSIZE - \
+					 DLM_MIDCOMMS_OPT_LEN)
+
+#define CONN_HASH_SIZE 32
+
+/* This is deliberately very simple because most clusters have simple
+ * sequential nodeids, so we should be able to go straight to a connection
+ * struct in the array
+ */
+static inline int nodeid_hash(int nodeid)
+{
+	return nodeid & (CONN_HASH_SIZE-1);
+}
 
 /* switch to check if dlm is running */
 extern int dlm_allow_conn;
@@ -22,8 +37,12 @@ void dlm_lowcomms_shutdown(void);
 void dlm_lowcomms_stop(void);
 void dlm_lowcomms_exit(void);
 int dlm_lowcomms_close(int nodeid);
-void *dlm_lowcomms_get_buffer(int nodeid, int len, gfp_t allocation, char **ppc);
-void dlm_lowcomms_commit_buffer(void *mh);
+struct dlm_msg *dlm_lowcomms_new_msg(int nodeid, int len, gfp_t allocation,
+				     char **ppc, void (*cb)(struct dlm_mhandle *mh),
+				     struct dlm_mhandle *mh);
+void dlm_lowcomms_commit_msg(struct dlm_msg *msg);
+void dlm_lowcomms_put_msg(struct dlm_msg *msg);
+int dlm_lowcomms_resend_msg(struct dlm_msg *msg);
 int dlm_lowcomms_connect_node(int nodeid);
 int dlm_lowcomms_nodes_set_mark(int nodeid, unsigned int mark);
 int dlm_lowcomms_addr(int nodeid, struct sockaddr_storage *addr, int len);
diff --git a/fs/dlm/member.c b/fs/dlm/member.c
index ceef3f2074ff..d9e1e4170eb1 100644
--- a/fs/dlm/member.c
+++ b/fs/dlm/member.c
@@ -15,6 +15,7 @@
 #include "recover.h"
 #include "rcom.h"
 #include "config.h"
+#include "midcomms.h"
 #include "lowcomms.h"
 
 int dlm_slots_version(struct dlm_header *h)
@@ -270,7 +271,7 @@ int dlm_slots_assign(struct dlm_ls *ls, int *num_slots, int *slots_size,
 
 	log_slots(ls, gen, num, NULL, array, array_size);
 
-	max_slots = (LOWCOMMS_MAX_TX_BUFFER_LEN - sizeof(struct dlm_rcom) -
+	max_slots = (DLM_MAX_APP_BUFSIZE - sizeof(struct dlm_rcom) -
 		     sizeof(struct rcom_config)) / sizeof(struct rcom_slot);
 
 	if (num > max_slots) {
@@ -329,6 +330,7 @@ static int dlm_add_member(struct dlm_ls *ls, struct dlm_config_node *node)
 	memb->nodeid = node->nodeid;
 	memb->weight = node->weight;
 	memb->comm_seq = node->comm_seq;
+	dlm_midcomms_add_member(node->nodeid);
 	add_ordered_member(ls, memb);
 	ls->ls_num_nodes++;
 	return 0;
@@ -359,26 +361,34 @@ int dlm_is_removed(struct dlm_ls *ls, int nodeid)
 	return 0;
 }
 
-static void clear_memb_list(struct list_head *head)
+static void clear_memb_list(struct list_head *head,
+			    void (*after_del)(int nodeid))
 {
 	struct dlm_member *memb;
 
 	while (!list_empty(head)) {
 		memb = list_entry(head->next, struct dlm_member, list);
 		list_del(&memb->list);
+		if (after_del)
+			after_del(memb->nodeid);
 		kfree(memb);
 	}
 }
 
+static void clear_members_cb(int nodeid)
+{
+	dlm_midcomms_remove_member(nodeid);
+}
+
 void dlm_clear_members(struct dlm_ls *ls)
 {
-	clear_memb_list(&ls->ls_nodes);
+	clear_memb_list(&ls->ls_nodes, clear_members_cb);
 	ls->ls_num_nodes = 0;
 }
 
 void dlm_clear_members_gone(struct dlm_ls *ls)
 {
-	clear_memb_list(&ls->ls_nodes_gone);
+	clear_memb_list(&ls->ls_nodes_gone, NULL);
 }
 
 static void make_member_array(struct dlm_ls *ls)
@@ -552,6 +562,7 @@ int dlm_recover_members(struct dlm_ls *ls, struct dlm_recover *rv, int *neg_out)
 
 		neg++;
 		list_move(&memb->list, &ls->ls_nodes_gone);
+		dlm_midcomms_remove_member(memb->nodeid);
 		ls->ls_num_nodes--;
 		dlm_lsop_recover_slot(ls, memb);
 	}
@@ -576,12 +587,18 @@ int dlm_recover_members(struct dlm_ls *ls, struct dlm_recover *rv, int *neg_out)
 	*neg_out = neg;
 
 	error = ping_members(ls);
-	if (!error || error == -EPROTO) {
-		/* new_lockspace() may be waiting to know if the config
-		   is good or bad */
-		ls->ls_members_result = error;
-		complete(&ls->ls_members_done);
-	}
+	/* error -EINTR means that a new recovery action is triggered.
+	 * We ignore this recovery action and let run the new one which might
+	 * have new member configuration.
+	 */
+	if (error == -EINTR)
+		error = 0;
+
+	/* new_lockspace() may be waiting to know if the config
+	 * is good or bad
+	 */
+	ls->ls_members_result = error;
+	complete(&ls->ls_members_done);
 
 	log_rinfo(ls, "dlm_recover_members %d nodes", ls->ls_num_nodes);
 	return error;
diff --git a/fs/dlm/midcomms.c b/fs/dlm/midcomms.c
index 1c6654a21ec4..e3de268898ed 100644
--- a/fs/dlm/midcomms.c
+++ b/fs/dlm/midcomms.c
@@ -3,7 +3,7 @@
 *******************************************************************************
 **
 **  Copyright (C) Sistina Software, Inc.  1997-2003  All rights reserved.
-**  Copyright (C) 2004-2008 Red Hat, Inc.  All rights reserved.
+**  Copyright (C) 2004-2021 Red Hat, Inc.  All rights reserved.
 **
 **
 *******************************************************************************
@@ -12,22 +12,866 @@
 /*
  * midcomms.c
  *
- * This is the appallingly named "mid-level" comms layer.
+ * This is the appallingly named "mid-level" comms layer. It takes care about
+ * deliver an on application layer "reliable" communication above the used
+ * lowcomms transport layer.
  *
- * Its purpose is to take packets from the "real" comms layer,
- * split them up into packets and pass them to the interested
- * part of the locking mechanism.
+ * How it works:
  *
- * It also takes messages from the locking layer, formats them
- * into packets and sends them to the comms layer.
+ * Each nodes keeps track of all send DLM messages in send_queue with a sequence
+ * number. The receive will send an DLM_ACK message back for every DLM message
+ * received at the other side. If a reconnect happens in lowcomms we will send
+ * all unacknowledged dlm messages again. The receiving side might drop any already
+ * received message by comparing sequence numbers.
+ *
+ * How version detection works:
+ *
+ * Due the fact that dlm has pre-configured node addresses on every side
+ * it is in it's nature that every side connects at starts to transmit
+ * dlm messages which ends in a race. However DLM_RCOM_NAMES, DLM_RCOM_STATUS
+ * and their replies are the first messages which are exchanges. Due backwards
+ * compatibility these messages are not covered by the midcomms re-transmission
+ * layer. These messages have their own re-transmission handling in the dlm
+ * application layer. The version field of every node will be set on these RCOM
+ * messages as soon as they arrived and the node isn't yet part of the nodes
+ * hash. There exists also logic to detect version mismatched if something weird
+ * going on or the first messages isn't an expected one.
+ *
+ * Termination:
+ *
+ * The midcomms layer does a 4 way handshake for termination on DLM protocol
+ * like TCP supports it with half-closed socket support. SCTP doesn't support
+ * half-closed socket, so we do it on DLM layer. Also socket shutdown() can be
+ * interrupted by .e.g. tcp reset itself. Additional there exists the othercon
+ * paradigm in lowcomms which cannot be easily without breaking backwards
+ * compatibility. A node cannot send anything to another node when a DLM_FIN
+ * message was send. There exists additional logic to print a warning if
+ * DLM wants to do it. There exists a state handling like RFC 793 but reduced
+ * to termination only. The event "member removal event" describes the cluster
+ * manager removed the node from internal lists, at this point DLM does not
+ * send any message to the other node. There exists two cases:
+ *
+ * 1. The cluster member was removed and we received a FIN
+ * OR
+ * 2. We received a FIN but the member was not removed yet
+ *
+ * One of these cases will do the CLOSE_WAIT to LAST_ACK change.
+ *
+ *
+ *                              +---------+
+ *                              | CLOSED  |
+ *                              +---------+
+ *                                   | add member/receive RCOM version
+ *                                   |            detection msg
+ *                                   V
+ *                              +---------+
+ *                              |  ESTAB  |
+ *                              +---------+
+ *                       CLOSE    |     |    rcv FIN
+ *                      -------   |     |    -------
+ * +---------+          snd FIN  /       \   snd ACK          +---------+
+ * |  FIN    |<-----------------           ------------------>|  CLOSE  |
+ * | WAIT-1  |------------------                              |   WAIT  |
+ * +---------+          rcv FIN  \                            +---------+
+ * | rcv ACK of FIN   -------   |                            CLOSE  | member
+ * | --------------   snd ACK   |                           ------- | removal
+ * V        x                   V                           snd FIN V event
+ * +---------+                  +---------+                   +---------+
+ * |FINWAIT-2|                  | CLOSING |                   | LAST-ACK|
+ * +---------+                  +---------+                   +---------+
+ * |                rcv ACK of FIN |                 rcv ACK of FIN |
+ * |  rcv FIN       -------------- |                 -------------- |
+ * |  -------              x       V                        x       V
+ *  \ snd ACK                 +---------+                   +---------+
+ *   ------------------------>| CLOSED  |                   | CLOSED  |
+ *                            +---------+                   +---------+
+ *
+ * NOTE: any state can interrupted by midcomms_close() and state will be
+ * switched to CLOSED in case of fencing. There exists also some timeout
+ * handling when we receive the version detection RCOM messages which is
+ * made by observation.
+ *
+ * Future improvements:
+ *
+ * There exists some known issues/improvements of the dlm handling. Some
+ * of them should be done in a next major dlm version bump which makes
+ * it incompatible with previous versions.
+ *
+ * Unaligned memory access:
+ *
+ * There exists cases when the dlm message buffer length is not aligned
+ * to 8 byte. However seems nobody detected any problem with it. This
+ * can be fixed in the next major version bump of dlm.
+ *
+ * Version detection:
+ *
+ * The version detection and how it's done is related to backwards
+ * compatibility. There exists better ways to make a better handling.
+ * However this should be changed in the next major version bump of dlm.
+ *
+ * Ack handling:
+ *
+ * Currently we send an ack message for every dlm message. However we
+ * can ack multiple dlm messages with one ack by just delaying the ack
+ * message. Will reduce some traffic but makes the drop detection slower.
+ *
+ * Tail Size checking:
+ *
+ * There exists a message tail payload in e.g. DLM_MSG however we don't
+ * check it against the message length yet regarding to the receive buffer
+ * length. That need to be validated.
+ *
+ * Fencing bad nodes:
+ *
+ * At timeout places or weird sequence number behaviours we should send
+ * a fencing request to the cluster manager.
+ */
+
+/* Debug switch to enable a 5 seconds sleep waiting of a termination.
+ * This can be useful to test fencing while termination is running.
+ * This requires a setup with only gfs2 as dlm user, so that the
+ * last umount will terminate the connection.
+ *
+ * However it became useful to test, while the 5 seconds block in umount
+ * just press the reset button. In a lot of dropping the termination
+ * process can could take several seconds.
  */
+#define DLM_DEBUG_FENCE_TERMINATION	0
+
+#include <net/tcp.h>
 
 #include "dlm_internal.h"
 #include "lowcomms.h"
 #include "config.h"
 #include "lock.h"
+#include "util.h"
 #include "midcomms.h"
 
+/* init value for sequence numbers for testing purpose only e.g. overflows */
+#define DLM_SEQ_INIT		0
+/* 3 minutes wait to sync ending of dlm */
+#define DLM_SHUTDOWN_TIMEOUT	msecs_to_jiffies(3 * 60 * 1000)
+#define DLM_VERSION_NOT_SET	0
+
+struct midcomms_node {
+	int nodeid;
+	uint32_t version;
+	uint32_t seq_send;
+	uint32_t seq_next;
+	/* These queues are unbound because we cannot drop any message in dlm.
+	 * We could send a fence signal for a specific node to the cluster
+	 * manager if queues hits some maximum value, however this handling
+	 * not supported yet.
+	 */
+	struct list_head send_queue;
+	spinlock_t send_queue_lock;
+	atomic_t send_queue_cnt;
+#define DLM_NODE_FLAG_CLOSE	1
+#define DLM_NODE_FLAG_STOP_TX	2
+#define DLM_NODE_FLAG_STOP_RX	3
+	unsigned long flags;
+	wait_queue_head_t shutdown_wait;
+
+	/* dlm tcp termination state */
+#define DLM_CLOSED	1
+#define DLM_ESTABLISHED	2
+#define DLM_FIN_WAIT1	3
+#define DLM_FIN_WAIT2	4
+#define DLM_CLOSE_WAIT	5
+#define DLM_LAST_ACK	6
+#define DLM_CLOSING	7
+	int state;
+	spinlock_t state_lock;
+
+	/* counts how many lockspaces are using this node
+	 * this refcount is necessary to determine if the
+	 * node wants to disconnect.
+	 */
+	int users;
+
+	/* not protected by srcu, node_hash lifetime */
+	void *debugfs;
+
+	struct hlist_node hlist;
+	struct rcu_head rcu;
+};
+
+struct dlm_mhandle {
+	const struct dlm_header *inner_hd;
+	struct midcomms_node *node;
+	struct dlm_opts *opts;
+	struct dlm_msg *msg;
+	bool committed;
+	uint32_t seq;
+
+	void (*ack_rcv)(struct midcomms_node *node);
+
+	/* get_mhandle/commit srcu idx exchange */
+	int idx;
+
+	struct list_head list;
+	struct rcu_head rcu;
+};
+
+static struct hlist_head node_hash[CONN_HASH_SIZE];
+static DEFINE_SPINLOCK(nodes_lock);
+DEFINE_STATIC_SRCU(nodes_srcu);
+
+/* This mutex prevents that midcomms_close() is running while
+ * stop() or remove(). As I experienced invalid memory access
+ * behaviours when DLM_DEBUG_FENCE_TERMINATION is enabled and
+ * resetting machines. I will end in some double deletion in nodes
+ * datastructure.
+ */
+static DEFINE_MUTEX(close_lock);
+
+static inline const char *dlm_state_str(int state)
+{
+	switch (state) {
+	case DLM_CLOSED:
+		return "CLOSED";
+	case DLM_ESTABLISHED:
+		return "ESTABLISHED";
+	case DLM_FIN_WAIT1:
+		return "FIN_WAIT1";
+	case DLM_FIN_WAIT2:
+		return "FIN_WAIT2";
+	case DLM_CLOSE_WAIT:
+		return "CLOSE_WAIT";
+	case DLM_LAST_ACK:
+		return "LAST_ACK";
+	case DLM_CLOSING:
+		return "CLOSING";
+	default:
+		return "UNKNOWN";
+	}
+}
+
+const char *dlm_midcomms_state(struct midcomms_node *node)
+{
+	return dlm_state_str(node->state);
+}
+
+unsigned long dlm_midcomms_flags(struct midcomms_node *node)
+{
+	return node->flags;
+}
+
+int dlm_midcomms_send_queue_cnt(struct midcomms_node *node)
+{
+	return atomic_read(&node->send_queue_cnt);
+}
+
+uint32_t dlm_midcomms_version(struct midcomms_node *node)
+{
+	return node->version;
+}
+
+static struct midcomms_node *__find_node(int nodeid, int r)
+{
+	struct midcomms_node *node;
+
+	hlist_for_each_entry_rcu(node, &node_hash[r], hlist) {
+		if (node->nodeid == nodeid)
+			return node;
+	}
+
+	return NULL;
+}
+
+static void dlm_mhandle_release(struct rcu_head *rcu)
+{
+	struct dlm_mhandle *mh = container_of(rcu, struct dlm_mhandle, rcu);
+
+	dlm_lowcomms_put_msg(mh->msg);
+	kfree(mh);
+}
+
+static void dlm_mhandle_delete(struct midcomms_node *node,
+			       struct dlm_mhandle *mh)
+{
+	list_del_rcu(&mh->list);
+	atomic_dec(&node->send_queue_cnt);
+	call_rcu(&mh->rcu, dlm_mhandle_release);
+}
+
+static void dlm_send_queue_flush(struct midcomms_node *node)
+{
+	struct dlm_mhandle *mh;
+
+	pr_debug("flush midcomms send queue of node %d\n", node->nodeid);
+
+	rcu_read_lock();
+	spin_lock(&node->send_queue_lock);
+	list_for_each_entry_rcu(mh, &node->send_queue, list) {
+		dlm_mhandle_delete(node, mh);
+	}
+	spin_unlock(&node->send_queue_lock);
+	rcu_read_unlock();
+}
+
+static void midcomms_node_reset(struct midcomms_node *node)
+{
+	pr_debug("reset node %d\n", node->nodeid);
+
+	node->seq_next = DLM_SEQ_INIT;
+	node->seq_send = DLM_SEQ_INIT;
+	node->version = DLM_VERSION_NOT_SET;
+	node->flags = 0;
+
+	dlm_send_queue_flush(node);
+	node->state = DLM_CLOSED;
+	wake_up(&node->shutdown_wait);
+}
+
+static struct midcomms_node *nodeid2node(int nodeid, gfp_t alloc)
+{
+	struct midcomms_node *node, *tmp;
+	int r = nodeid_hash(nodeid);
+
+	node = __find_node(nodeid, r);
+	if (node || !alloc)
+		return node;
+
+	node = kmalloc(sizeof(*node), alloc);
+	if (!node)
+		return NULL;
+
+	node->nodeid = nodeid;
+	spin_lock_init(&node->state_lock);
+	spin_lock_init(&node->send_queue_lock);
+	atomic_set(&node->send_queue_cnt, 0);
+	INIT_LIST_HEAD(&node->send_queue);
+	init_waitqueue_head(&node->shutdown_wait);
+	node->users = 0;
+	midcomms_node_reset(node);
+
+	spin_lock(&nodes_lock);
+	/* check again if there was somebody else
+	 * earlier here to add the node
+	 */
+	tmp = __find_node(nodeid, r);
+	if (tmp) {
+		spin_unlock(&nodes_lock);
+		kfree(node);
+		return tmp;
+	}
+
+	hlist_add_head_rcu(&node->hlist, &node_hash[r]);
+	spin_unlock(&nodes_lock);
+
+	node->debugfs = dlm_create_debug_comms_file(nodeid, node);
+	return node;
+}
+
+static int dlm_send_ack(int nodeid, uint32_t seq)
+{
+	int mb_len = sizeof(struct dlm_header);
+	struct dlm_header *m_header;
+	struct dlm_msg *msg;
+	char *ppc;
+
+	msg = dlm_lowcomms_new_msg(nodeid, mb_len, GFP_NOFS, &ppc,
+				   NULL, NULL);
+	if (!msg)
+		return -ENOMEM;
+
+	m_header = (struct dlm_header *)ppc;
+
+	m_header->h_version = (DLM_HEADER_MAJOR | DLM_HEADER_MINOR);
+	m_header->h_nodeid = dlm_our_nodeid();
+	m_header->h_length = mb_len;
+	m_header->h_cmd = DLM_ACK;
+	m_header->u.h_seq = seq;
+
+	header_out(m_header);
+	dlm_lowcomms_commit_msg(msg);
+	dlm_lowcomms_put_msg(msg);
+
+	return 0;
+}
+
+static int dlm_send_fin(struct midcomms_node *node,
+			void (*ack_rcv)(struct midcomms_node *node))
+{
+	int mb_len = sizeof(struct dlm_header);
+	struct dlm_header *m_header;
+	struct dlm_mhandle *mh;
+	char *ppc;
+
+	mh = dlm_midcomms_get_mhandle(node->nodeid, mb_len, GFP_NOFS, &ppc);
+	if (!mh)
+		return -ENOMEM;
+
+	mh->ack_rcv = ack_rcv;
+
+	m_header = (struct dlm_header *)ppc;
+
+	m_header->h_version = (DLM_HEADER_MAJOR | DLM_HEADER_MINOR);
+	m_header->h_nodeid = dlm_our_nodeid();
+	m_header->h_length = mb_len;
+	m_header->h_cmd = DLM_FIN;
+
+	header_out(m_header);
+
+	pr_debug("sending fin msg to node %d\n", node->nodeid);
+	dlm_midcomms_commit_mhandle(mh);
+	set_bit(DLM_NODE_FLAG_STOP_TX, &node->flags);
+
+	return 0;
+}
+
+static void dlm_receive_ack(struct midcomms_node *node, uint32_t seq)
+{
+	struct dlm_mhandle *mh;
+
+	rcu_read_lock();
+	list_for_each_entry_rcu(mh, &node->send_queue, list) {
+		if (before(mh->seq, seq)) {
+			if (mh->ack_rcv)
+				mh->ack_rcv(node);
+		} else {
+			/* send queue should be ordered */
+			break;
+		}
+	}
+
+	spin_lock(&node->send_queue_lock);
+	list_for_each_entry_rcu(mh, &node->send_queue, list) {
+		if (before(mh->seq, seq)) {
+			dlm_mhandle_delete(node, mh);
+		} else {
+			/* send queue should be ordered */
+			break;
+		}
+	}
+	spin_unlock(&node->send_queue_lock);
+	rcu_read_unlock();
+}
+
+static void dlm_pas_fin_ack_rcv(struct midcomms_node *node)
+{
+	spin_lock(&node->state_lock);
+	pr_debug("receive passive fin ack from node %d with state %s\n",
+		 node->nodeid, dlm_state_str(node->state));
+
+	switch (node->state) {
+	case DLM_LAST_ACK:
+		/* DLM_CLOSED */
+		midcomms_node_reset(node);
+		break;
+	case DLM_CLOSED:
+		/* not valid but somehow we got what we want */
+		wake_up(&node->shutdown_wait);
+		break;
+	default:
+		spin_unlock(&node->state_lock);
+		log_print("%s: unexpected state: %d\n",
+			  __func__, node->state);
+		WARN_ON(1);
+		return;
+	}
+	spin_unlock(&node->state_lock);
+}
+
+static void dlm_midcomms_receive_buffer(union dlm_packet *p,
+					struct midcomms_node *node,
+					uint32_t seq)
+{
+	if (seq == node->seq_next) {
+		node->seq_next++;
+		/* send ack before fin */
+		dlm_send_ack(node->nodeid, node->seq_next);
+
+		switch (p->header.h_cmd) {
+		case DLM_FIN:
+			spin_lock(&node->state_lock);
+			pr_debug("receive fin msg from node %d with state %s\n",
+				 node->nodeid, dlm_state_str(node->state));
+
+			switch (node->state) {
+			case DLM_ESTABLISHED:
+				node->state = DLM_CLOSE_WAIT;
+				pr_debug("switch node %d to state %s\n",
+					 node->nodeid, dlm_state_str(node->state));
+				/* passive shutdown DLM_LAST_ACK case 1
+				 * additional we check if the node is used by
+				 * cluster manager events at all.
+				 */
+				if (node->users == 0) {
+					node->state = DLM_LAST_ACK;
+					pr_debug("switch node %d to state %s case 1\n",
+						 node->nodeid, dlm_state_str(node->state));
+					spin_unlock(&node->state_lock);
+					goto send_fin;
+				}
+				break;
+			case DLM_FIN_WAIT1:
+				node->state = DLM_CLOSING;
+				pr_debug("switch node %d to state %s\n",
+					 node->nodeid, dlm_state_str(node->state));
+				break;
+			case DLM_FIN_WAIT2:
+				midcomms_node_reset(node);
+				pr_debug("switch node %d to state %s\n",
+					 node->nodeid, dlm_state_str(node->state));
+				wake_up(&node->shutdown_wait);
+				break;
+			case DLM_LAST_ACK:
+				/* probably remove_member caught it, do nothing */
+				break;
+			default:
+				spin_unlock(&node->state_lock);
+				log_print("%s: unexpected state: %d\n",
+					  __func__, node->state);
+				WARN_ON(1);
+				return;
+			}
+			spin_unlock(&node->state_lock);
+
+			set_bit(DLM_NODE_FLAG_STOP_RX, &node->flags);
+			break;
+		default:
+			WARN_ON(test_bit(DLM_NODE_FLAG_STOP_RX, &node->flags));
+			dlm_receive_buffer(p, node->nodeid);
+			break;
+		}
+	} else {
+		/* retry to ack message which we already have by sending back
+		 * current node->seq_next number as ack.
+		 */
+		if (seq < node->seq_next)
+			dlm_send_ack(node->nodeid, node->seq_next);
+
+		log_print_ratelimited("ignore dlm msg because seq mismatch, seq: %u, expected: %u, nodeid: %d",
+				      seq, node->seq_next, node->nodeid);
+	}
+
+	return;
+
+send_fin:
+	set_bit(DLM_NODE_FLAG_STOP_RX, &node->flags);
+	dlm_send_fin(node, dlm_pas_fin_ack_rcv);
+}
+
+static struct midcomms_node *
+dlm_midcomms_recv_node_lookup(int nodeid, const union dlm_packet *p,
+			      uint16_t msglen, int (*cb)(struct midcomms_node *node))
+{
+	struct midcomms_node *node = NULL;
+	gfp_t allocation = 0;
+	int ret;
+
+	switch (p->header.h_cmd) {
+	case DLM_RCOM:
+		if (msglen < sizeof(struct dlm_rcom)) {
+			log_print("rcom msg too small: %u, will skip this message from node %d",
+				  msglen, nodeid);
+			return NULL;
+		}
+
+		switch (le32_to_cpu(p->rcom.rc_type)) {
+		case DLM_RCOM_NAMES:
+			fallthrough;
+		case DLM_RCOM_NAMES_REPLY:
+			fallthrough;
+		case DLM_RCOM_STATUS:
+			fallthrough;
+		case DLM_RCOM_STATUS_REPLY:
+			node = nodeid2node(nodeid, 0);
+			if (node) {
+				spin_lock(&node->state_lock);
+				if (node->state != DLM_ESTABLISHED)
+					pr_debug("receive begin RCOM msg from node %d with state %s\n",
+						 node->nodeid, dlm_state_str(node->state));
+
+				switch (node->state) {
+				case DLM_CLOSED:
+					node->state = DLM_ESTABLISHED;
+					pr_debug("switch node %d to state %s\n",
+						 node->nodeid, dlm_state_str(node->state));
+					break;
+				case DLM_ESTABLISHED:
+					break;
+				default:
+					/* some invalid state passive shutdown
+					 * was failed, we try to reset and
+					 * hope it will go on.
+					 */
+					log_print("reset node %d because shutdown stuck",
+						  node->nodeid);
+
+					midcomms_node_reset(node);
+					node->state = DLM_ESTABLISHED;
+					break;
+				}
+				spin_unlock(&node->state_lock);
+			}
+
+			allocation = GFP_NOFS;
+			break;
+		default:
+			break;
+		}
+
+		break;
+	default:
+		break;
+	}
+
+	node = nodeid2node(nodeid, allocation);
+	if (!node) {
+		switch (p->header.h_cmd) {
+		case DLM_OPTS:
+			if (msglen < sizeof(struct dlm_opts)) {
+				log_print("opts msg too small: %u, will skip this message from node %d",
+					  msglen, nodeid);
+				return NULL;
+			}
+
+			log_print_ratelimited("received dlm opts message nextcmd %d from node %d in an invalid sequence",
+					      p->opts.o_nextcmd, nodeid);
+			break;
+		default:
+			log_print_ratelimited("received dlm message cmd %d from node %d in an invalid sequence",
+					      p->header.h_cmd, nodeid);
+			break;
+		}
+
+		return NULL;
+	}
+
+	ret = cb(node);
+	if (ret < 0)
+		return NULL;
+
+	return node;
+}
+
+static int dlm_midcomms_version_check_3_2(struct midcomms_node *node)
+{
+	switch (node->version) {
+	case DLM_VERSION_NOT_SET:
+		node->version = DLM_VERSION_3_2;
+		log_print("version 0x%08x for node %d detected", DLM_VERSION_3_2,
+			  node->nodeid);
+		break;
+	case DLM_VERSION_3_2:
+		break;
+	default:
+		log_print_ratelimited("version mismatch detected, assumed 0x%08x but node %d has 0x%08x",
+				      DLM_VERSION_3_2, node->nodeid, node->version);
+		return -1;
+	}
+
+	return 0;
+}
+
+static int dlm_opts_check_msglen(union dlm_packet *p, uint16_t msglen, int nodeid)
+{
+	int len = msglen;
+
+	/* we only trust outer header msglen because
+	 * it's checked against receive buffer length.
+	 */
+	if (len < sizeof(struct dlm_opts))
+		return -1;
+	len -= sizeof(struct dlm_opts);
+
+	if (len < le16_to_cpu(p->opts.o_optlen))
+		return -1;
+	len -= le16_to_cpu(p->opts.o_optlen);
+
+	switch (p->opts.o_nextcmd) {
+	case DLM_FIN:
+		if (len < sizeof(struct dlm_header)) {
+			log_print("fin too small: %d, will skip this message from node %d",
+				  len, nodeid);
+			return -1;
+		}
+
+		break;
+	case DLM_MSG:
+		if (len < sizeof(struct dlm_message)) {
+			log_print("msg too small: %d, will skip this message from node %d",
+				  msglen, nodeid);
+			return -1;
+		}
+
+		break;
+	case DLM_RCOM:
+		if (len < sizeof(struct dlm_rcom)) {
+			log_print("rcom msg too small: %d, will skip this message from node %d",
+				  len, nodeid);
+			return -1;
+		}
+
+		break;
+	default:
+		log_print("unsupported o_nextcmd received: %u, will skip this message from node %d",
+			  p->opts.o_nextcmd, nodeid);
+		return -1;
+	}
+
+	return 0;
+}
+
+static void dlm_midcomms_receive_buffer_3_2(union dlm_packet *p, int nodeid)
+{
+	uint16_t msglen = le16_to_cpu(p->header.h_length);
+	struct midcomms_node *node;
+	uint32_t seq;
+	int ret, idx;
+
+	idx = srcu_read_lock(&nodes_srcu);
+	node = dlm_midcomms_recv_node_lookup(nodeid, p, msglen,
+					     dlm_midcomms_version_check_3_2);
+	if (!node)
+		goto out;
+
+	switch (p->header.h_cmd) {
+	case DLM_RCOM:
+		/* these rcom message we use to determine version.
+		 * they have their own retransmission handling and
+		 * are the first messages of dlm.
+		 *
+		 * length already checked.
+		 */
+		switch (le32_to_cpu(p->rcom.rc_type)) {
+		case DLM_RCOM_NAMES:
+			fallthrough;
+		case DLM_RCOM_NAMES_REPLY:
+			fallthrough;
+		case DLM_RCOM_STATUS:
+			fallthrough;
+		case DLM_RCOM_STATUS_REPLY:
+			break;
+		default:
+			log_print("unsupported rcom type received: %u, will skip this message from node %d",
+				  le32_to_cpu(p->rcom.rc_type), nodeid);
+			goto out;
+		}
+
+		WARN_ON(test_bit(DLM_NODE_FLAG_STOP_RX, &node->flags));
+		dlm_receive_buffer(p, nodeid);
+		break;
+	case DLM_OPTS:
+		seq = le32_to_cpu(p->header.u.h_seq);
+
+		ret = dlm_opts_check_msglen(p, msglen, nodeid);
+		if (ret < 0) {
+			log_print("opts msg too small: %u, will skip this message from node %d",
+				  msglen, nodeid);
+			goto out;
+		}
+
+		p = (union dlm_packet *)((unsigned char *)p->opts.o_opts +
+					 le16_to_cpu(p->opts.o_optlen));
+
+		/* recheck inner msglen just if it's not garbage */
+		msglen = le16_to_cpu(p->header.h_length);
+		switch (p->header.h_cmd) {
+		case DLM_RCOM:
+			if (msglen < sizeof(struct dlm_rcom)) {
+				log_print("inner rcom msg too small: %u, will skip this message from node %d",
+					  msglen, nodeid);
+				goto out;
+			}
+
+			break;
+		case DLM_MSG:
+			if (msglen < sizeof(struct dlm_message)) {
+				log_print("inner msg too small: %u, will skip this message from node %d",
+					  msglen, nodeid);
+				goto out;
+			}
+
+			break;
+		case DLM_FIN:
+			if (msglen < sizeof(struct dlm_header)) {
+				log_print("inner fin too small: %u, will skip this message from node %d",
+					  msglen, nodeid);
+				goto out;
+			}
+
+			break;
+		default:
+			log_print("unsupported inner h_cmd received: %u, will skip this message from node %d",
+				  msglen, nodeid);
+			goto out;
+		}
+
+		dlm_midcomms_receive_buffer(p, node, seq);
+		break;
+	case DLM_ACK:
+		seq = le32_to_cpu(p->header.u.h_seq);
+		dlm_receive_ack(node, seq);
+		break;
+	default:
+		log_print("unsupported h_cmd received: %u, will skip this message from node %d",
+			  p->header.h_cmd, nodeid);
+		break;
+	}
+
+out:
+	srcu_read_unlock(&nodes_srcu, idx);
+}
+
+static int dlm_midcomms_version_check_3_1(struct midcomms_node *node)
+{
+	switch (node->version) {
+	case DLM_VERSION_NOT_SET:
+		node->version = DLM_VERSION_3_1;
+		log_print("version 0x%08x for node %d detected", DLM_VERSION_3_1,
+			  node->nodeid);
+		break;
+	case DLM_VERSION_3_1:
+		break;
+	default:
+		log_print_ratelimited("version mismatch detected, assumed 0x%08x but node %d has 0x%08x",
+				      DLM_VERSION_3_1, node->nodeid, node->version);
+		return -1;
+	}
+
+	return 0;
+}
+
+static void dlm_midcomms_receive_buffer_3_1(union dlm_packet *p, int nodeid)
+{
+	uint16_t msglen = le16_to_cpu(p->header.h_length);
+	struct midcomms_node *node;
+	int idx;
+
+	idx = srcu_read_lock(&nodes_srcu);
+	node = dlm_midcomms_recv_node_lookup(nodeid, p, msglen,
+					     dlm_midcomms_version_check_3_1);
+	if (!node) {
+		srcu_read_unlock(&nodes_srcu, idx);
+		return;
+	}
+	srcu_read_unlock(&nodes_srcu, idx);
+
+	switch (p->header.h_cmd) {
+	case DLM_RCOM:
+		/* length already checked */
+		break;
+	case DLM_MSG:
+		if (msglen < sizeof(struct dlm_message)) {
+			log_print("msg too small: %u, will skip this message from node %d",
+				  msglen, nodeid);
+			return;
+		}
+
+		break;
+	default:
+		log_print("unsupported h_cmd received: %u, will skip this message from node %d",
+			  p->header.h_cmd, nodeid);
+		return;
+	}
+
+	dlm_receive_buffer(p, nodeid);
+}
+
 /*
  * Called from the low-level comms layer to process a buffer of
  * commands.
@@ -43,7 +887,7 @@ int dlm_process_incoming_buffer(int nodeid, unsigned char *buf, int len)
 	while (len >= sizeof(struct dlm_header)) {
 		hd = (struct dlm_header *)ptr;
 
-		/* no message should be more than DEFAULT_BUFFER_SIZE or
+		/* no message should be more than DLM_MAX_SOCKET_BUFSIZE or
 		 * less than dlm_header size.
 		 *
 		 * Some messages does not have a 8 byte length boundary yet
@@ -55,7 +899,7 @@ int dlm_process_incoming_buffer(int nodeid, unsigned char *buf, int len)
 		 * the next major version bump.
 		 */
 		msglen = le16_to_cpu(hd->h_length);
-		if (msglen > DEFAULT_BUFFER_SIZE ||
+		if (msglen > DLM_MAX_SOCKET_BUFSIZE ||
 		    msglen < sizeof(struct dlm_header)) {
 			log_print("received invalid length header: %u from node %d, will abort message parsing",
 				  msglen, nodeid);
@@ -68,32 +912,19 @@ int dlm_process_incoming_buffer(int nodeid, unsigned char *buf, int len)
 		if (msglen > len)
 			break;
 
-		switch (hd->h_cmd) {
-		case DLM_MSG:
-			if (msglen < sizeof(struct dlm_message)) {
-				log_print("dlm msg too small: %u, will skip this message",
-					  msglen);
-				goto skip;
-			}
-
+		switch (le32_to_cpu(hd->h_version)) {
+		case DLM_VERSION_3_1:
+			dlm_midcomms_receive_buffer_3_1((union dlm_packet *)ptr, nodeid);
 			break;
-		case DLM_RCOM:
-			if (msglen < sizeof(struct dlm_rcom)) {
-				log_print("dlm rcom msg too small: %u, will skip this message",
-					  msglen);
-				goto skip;
-			}
-
+		case DLM_VERSION_3_2:
+			dlm_midcomms_receive_buffer_3_2((union dlm_packet *)ptr, nodeid);
 			break;
 		default:
-			log_print("unsupported h_cmd received: %u, will skip this message",
-				  hd->h_cmd);
-			goto skip;
+			log_print("received invalid version header: %u from node %d, will skip this message",
+				  le32_to_cpu(hd->h_version), nodeid);
+			break;
 		}
 
-		dlm_receive_buffer((union dlm_packet *)ptr, nodeid);
-
-skip:
 		ret += msglen;
 		len -= msglen;
 		ptr += msglen;
@@ -102,3 +933,455 @@ skip:
 	return ret;
 }
 
+void dlm_midcomms_unack_msg_resend(int nodeid)
+{
+	struct midcomms_node *node;
+	struct dlm_mhandle *mh;
+	int idx, ret;
+
+	idx = srcu_read_lock(&nodes_srcu);
+	node = nodeid2node(nodeid, 0);
+	if (!node) {
+		srcu_read_unlock(&nodes_srcu, idx);
+		return;
+	}
+
+	/* old protocol, we don't support to retransmit on failure */
+	switch (node->version) {
+	case DLM_VERSION_3_2:
+		break;
+	default:
+		srcu_read_unlock(&nodes_srcu, idx);
+		return;
+	}
+
+	rcu_read_lock();
+	list_for_each_entry_rcu(mh, &node->send_queue, list) {
+		if (!mh->committed)
+			continue;
+
+		ret = dlm_lowcomms_resend_msg(mh->msg);
+		if (!ret)
+			log_print_ratelimited("retransmit dlm msg, seq %u, nodeid %d",
+					      mh->seq, node->nodeid);
+	}
+	rcu_read_unlock();
+	srcu_read_unlock(&nodes_srcu, idx);
+}
+
+static void dlm_fill_opts_header(struct dlm_opts *opts, uint16_t inner_len,
+				 uint32_t seq)
+{
+	opts->o_header.h_cmd = DLM_OPTS;
+	opts->o_header.h_version = (DLM_HEADER_MAJOR | DLM_HEADER_MINOR);
+	opts->o_header.h_nodeid = dlm_our_nodeid();
+	opts->o_header.h_length = DLM_MIDCOMMS_OPT_LEN + inner_len;
+	opts->o_header.u.h_seq = seq;
+	header_out(&opts->o_header);
+}
+
+static void midcomms_new_msg_cb(struct dlm_mhandle *mh)
+{
+	atomic_inc(&mh->node->send_queue_cnt);
+
+	spin_lock(&mh->node->send_queue_lock);
+	list_add_tail_rcu(&mh->list, &mh->node->send_queue);
+	spin_unlock(&mh->node->send_queue_lock);
+
+	mh->seq = mh->node->seq_send++;
+}
+
+static struct dlm_msg *dlm_midcomms_get_msg_3_2(struct dlm_mhandle *mh, int nodeid,
+						int len, gfp_t allocation, char **ppc)
+{
+	struct dlm_opts *opts;
+	struct dlm_msg *msg;
+
+	msg = dlm_lowcomms_new_msg(nodeid, len + DLM_MIDCOMMS_OPT_LEN,
+				   allocation, ppc, midcomms_new_msg_cb, mh);
+	if (!msg)
+		return NULL;
+
+	opts = (struct dlm_opts *)*ppc;
+	mh->opts = opts;
+
+	/* add possible options here */
+	dlm_fill_opts_header(opts, len, mh->seq);
+
+	*ppc += sizeof(*opts);
+	mh->inner_hd = (const struct dlm_header *)*ppc;
+	return msg;
+}
+
+struct dlm_mhandle *dlm_midcomms_get_mhandle(int nodeid, int len,
+					     gfp_t allocation, char **ppc)
+{
+	struct midcomms_node *node;
+	struct dlm_mhandle *mh;
+	struct dlm_msg *msg;
+	int idx;
+
+	idx = srcu_read_lock(&nodes_srcu);
+	node = nodeid2node(nodeid, 0);
+	if (!node) {
+		WARN_ON_ONCE(1);
+		goto err;
+	}
+
+	/* this is a bug, however we going on and hope it will be resolved */
+	WARN_ON(test_bit(DLM_NODE_FLAG_STOP_TX, &node->flags));
+
+	mh = kzalloc(sizeof(*mh), GFP_NOFS);
+	if (!mh)
+		goto err;
+
+	mh->idx = idx;
+	mh->node = node;
+
+	switch (node->version) {
+	case DLM_VERSION_3_1:
+		msg = dlm_lowcomms_new_msg(nodeid, len, allocation, ppc,
+					   NULL, NULL);
+		if (!msg) {
+			kfree(mh);
+			goto err;
+		}
+
+		break;
+	case DLM_VERSION_3_2:
+		msg = dlm_midcomms_get_msg_3_2(mh, nodeid, len, allocation,
+					       ppc);
+		if (!msg) {
+			kfree(mh);
+			goto err;
+		}
+
+		break;
+	default:
+		kfree(mh);
+		WARN_ON(1);
+		goto err;
+	}
+
+	mh->msg = msg;
+
+	/* keep in mind that is a must to call
+	 * dlm_midcomms_commit_msg() which releases
+	 * nodes_srcu using mh->idx which is assumed
+	 * here that the application will call it.
+	 */
+	return mh;
+
+err:
+	srcu_read_unlock(&nodes_srcu, idx);
+	return NULL;
+}
+
+static void dlm_midcomms_commit_msg_3_2(struct dlm_mhandle *mh)
+{
+	/* nexthdr chain for fast lookup */
+	mh->opts->o_nextcmd = mh->inner_hd->h_cmd;
+	mh->committed = true;
+	dlm_lowcomms_commit_msg(mh->msg);
+}
+
+void dlm_midcomms_commit_mhandle(struct dlm_mhandle *mh)
+{
+	switch (mh->node->version) {
+	case DLM_VERSION_3_1:
+		srcu_read_unlock(&nodes_srcu, mh->idx);
+
+		dlm_lowcomms_commit_msg(mh->msg);
+		dlm_lowcomms_put_msg(mh->msg);
+		/* mh is not part of rcu list in this case */
+		kfree(mh);
+		break;
+	case DLM_VERSION_3_2:
+		dlm_midcomms_commit_msg_3_2(mh);
+		srcu_read_unlock(&nodes_srcu, mh->idx);
+		break;
+	default:
+		srcu_read_unlock(&nodes_srcu, mh->idx);
+		WARN_ON(1);
+		break;
+	}
+}
+
+int dlm_midcomms_start(void)
+{
+	int i;
+
+	for (i = 0; i < CONN_HASH_SIZE; i++)
+		INIT_HLIST_HEAD(&node_hash[i]);
+
+	return dlm_lowcomms_start();
+}
+
+static void dlm_act_fin_ack_rcv(struct midcomms_node *node)
+{
+	spin_lock(&node->state_lock);
+	pr_debug("receive active fin ack from node %d with state %s\n",
+		 node->nodeid, dlm_state_str(node->state));
+
+	switch (node->state) {
+	case DLM_FIN_WAIT1:
+		node->state = DLM_FIN_WAIT2;
+		pr_debug("switch node %d to state %s\n",
+			 node->nodeid, dlm_state_str(node->state));
+		break;
+	case DLM_CLOSING:
+		midcomms_node_reset(node);
+		pr_debug("switch node %d to state %s\n",
+			 node->nodeid, dlm_state_str(node->state));
+		wake_up(&node->shutdown_wait);
+		break;
+	case DLM_CLOSED:
+		/* not valid but somehow we got what we want */
+		wake_up(&node->shutdown_wait);
+		break;
+	default:
+		spin_unlock(&node->state_lock);
+		log_print("%s: unexpected state: %d\n",
+			  __func__, node->state);
+		WARN_ON(1);
+		return;
+	}
+	spin_unlock(&node->state_lock);
+}
+
+void dlm_midcomms_add_member(int nodeid)
+{
+	struct midcomms_node *node;
+	int idx;
+
+	if (nodeid == dlm_our_nodeid())
+		return;
+
+	idx = srcu_read_lock(&nodes_srcu);
+	node = nodeid2node(nodeid, GFP_NOFS);
+	if (!node) {
+		srcu_read_unlock(&nodes_srcu, idx);
+		return;
+	}
+
+	spin_lock(&node->state_lock);
+	if (!node->users) {
+		pr_debug("receive add member from node %d with state %s\n",
+			 node->nodeid, dlm_state_str(node->state));
+		switch (node->state) {
+		case DLM_ESTABLISHED:
+			break;
+		case DLM_CLOSED:
+			node->state = DLM_ESTABLISHED;
+			pr_debug("switch node %d to state %s\n",
+				 node->nodeid, dlm_state_str(node->state));
+			break;
+		default:
+			/* some invalid state passive shutdown
+			 * was failed, we try to reset and
+			 * hope it will go on.
+			 */
+			log_print("reset node %d because shutdown stuck",
+				  node->nodeid);
+
+			midcomms_node_reset(node);
+			node->state = DLM_ESTABLISHED;
+			break;
+		}
+	}
+
+	node->users++;
+	pr_debug("users inc count %d\n", node->users);
+	spin_unlock(&node->state_lock);
+
+	srcu_read_unlock(&nodes_srcu, idx);
+}
+
+void dlm_midcomms_remove_member(int nodeid)
+{
+	struct midcomms_node *node;
+	int idx;
+
+	if (nodeid == dlm_our_nodeid())
+		return;
+
+	idx = srcu_read_lock(&nodes_srcu);
+	node = nodeid2node(nodeid, 0);
+	if (!node) {
+		srcu_read_unlock(&nodes_srcu, idx);
+		return;
+	}
+
+	spin_lock(&node->state_lock);
+	node->users--;
+	pr_debug("users dec count %d\n", node->users);
+
+	/* hitting users count to zero means the
+	 * other side is running dlm_midcomms_stop()
+	 * we meet us to have a clean disconnect.
+	 */
+	if (node->users == 0) {
+		pr_debug("receive remove member from node %d with state %s\n",
+			 node->nodeid, dlm_state_str(node->state));
+		switch (node->state) {
+		case DLM_ESTABLISHED:
+			break;
+		case DLM_CLOSE_WAIT:
+			/* passive shutdown DLM_LAST_ACK case 2 */
+			node->state = DLM_LAST_ACK;
+			spin_unlock(&node->state_lock);
+
+			pr_debug("switch node %d to state %s case 2\n",
+				 node->nodeid, dlm_state_str(node->state));
+			goto send_fin;
+		case DLM_LAST_ACK:
+			/* probably receive fin caught it, do nothing */
+			break;
+		case DLM_CLOSED:
+			/* already gone, do nothing */
+			break;
+		default:
+			log_print("%s: unexpected state: %d\n",
+				  __func__, node->state);
+			break;
+		}
+	}
+	spin_unlock(&node->state_lock);
+
+	srcu_read_unlock(&nodes_srcu, idx);
+	return;
+
+send_fin:
+	set_bit(DLM_NODE_FLAG_STOP_RX, &node->flags);
+	dlm_send_fin(node, dlm_pas_fin_ack_rcv);
+	srcu_read_unlock(&nodes_srcu, idx);
+}
+
+static void midcomms_node_release(struct rcu_head *rcu)
+{
+	struct midcomms_node *node = container_of(rcu, struct midcomms_node, rcu);
+
+	WARN_ON(atomic_read(&node->send_queue_cnt));
+	kfree(node);
+}
+
+static void midcomms_shutdown(struct midcomms_node *node)
+{
+	int ret;
+
+	/* old protocol, we don't wait for pending operations */
+	switch (node->version) {
+	case DLM_VERSION_3_2:
+		break;
+	default:
+		return;
+	}
+
+	spin_lock(&node->state_lock);
+	pr_debug("receive active shutdown for node %d with state %s\n",
+		 node->nodeid, dlm_state_str(node->state));
+	switch (node->state) {
+	case DLM_ESTABLISHED:
+		node->state = DLM_FIN_WAIT1;
+		pr_debug("switch node %d to state %s case 2\n",
+			 node->nodeid, dlm_state_str(node->state));
+		break;
+	case DLM_CLOSED:
+		/* we have what we want */
+		spin_unlock(&node->state_lock);
+		return;
+	default:
+		/* busy to enter DLM_FIN_WAIT1, wait until passive
+		 * done in shutdown_wait to enter DLM_CLOSED.
+		 */
+		break;
+	}
+	spin_unlock(&node->state_lock);
+
+	if (node->state == DLM_FIN_WAIT1) {
+		dlm_send_fin(node, dlm_act_fin_ack_rcv);
+
+		if (DLM_DEBUG_FENCE_TERMINATION)
+			msleep(5000);
+	}
+
+	/* wait for other side dlm + fin */
+	ret = wait_event_timeout(node->shutdown_wait,
+				 node->state == DLM_CLOSED ||
+				 test_bit(DLM_NODE_FLAG_CLOSE, &node->flags),
+				 DLM_SHUTDOWN_TIMEOUT);
+	if (!ret || test_bit(DLM_NODE_FLAG_CLOSE, &node->flags)) {
+		pr_debug("active shutdown timed out for node %d with state %s\n",
+			 node->nodeid, dlm_state_str(node->state));
+		midcomms_node_reset(node);
+		return;
+	}
+
+	pr_debug("active shutdown done for node %d with state %s\n",
+		 node->nodeid, dlm_state_str(node->state));
+}
+
+void dlm_midcomms_shutdown(void)
+{
+	struct midcomms_node *node;
+	int i, idx;
+
+	mutex_lock(&close_lock);
+	idx = srcu_read_lock(&nodes_srcu);
+	for (i = 0; i < CONN_HASH_SIZE; i++) {
+		hlist_for_each_entry_rcu(node, &node_hash[i], hlist) {
+			midcomms_shutdown(node);
+
+			dlm_delete_debug_comms_file(node->debugfs);
+
+			spin_lock(&nodes_lock);
+			hlist_del_rcu(&node->hlist);
+			spin_unlock(&nodes_lock);
+
+			call_srcu(&nodes_srcu, &node->rcu, midcomms_node_release);
+		}
+	}
+	srcu_read_unlock(&nodes_srcu, idx);
+	mutex_unlock(&close_lock);
+
+	dlm_lowcomms_shutdown();
+}
+
+int dlm_midcomms_close(int nodeid)
+{
+	struct midcomms_node *node;
+	int idx, ret;
+
+	if (nodeid == dlm_our_nodeid())
+		return 0;
+
+	idx = srcu_read_lock(&nodes_srcu);
+	/* Abort pending close/remove operation */
+	node = nodeid2node(nodeid, 0);
+	if (node) {
+		/* let shutdown waiters leave */
+		set_bit(DLM_NODE_FLAG_CLOSE, &node->flags);
+		wake_up(&node->shutdown_wait);
+	}
+	srcu_read_unlock(&nodes_srcu, idx);
+
+	synchronize_srcu(&nodes_srcu);
+
+	idx = srcu_read_lock(&nodes_srcu);
+	mutex_lock(&close_lock);
+	node = nodeid2node(nodeid, 0);
+	if (!node) {
+		mutex_unlock(&close_lock);
+		srcu_read_unlock(&nodes_srcu, idx);
+		return dlm_lowcomms_close(nodeid);
+	}
+
+	ret = dlm_lowcomms_close(nodeid);
+	spin_lock(&node->state_lock);
+	midcomms_node_reset(node);
+	spin_unlock(&node->state_lock);
+	srcu_read_unlock(&nodes_srcu, idx);
+	mutex_unlock(&close_lock);
+
+	return ret;
+}
diff --git a/fs/dlm/midcomms.h b/fs/dlm/midcomms.h
index 61e90a921849..579abc6929be 100644
--- a/fs/dlm/midcomms.h
+++ b/fs/dlm/midcomms.h
@@ -12,7 +12,22 @@
 #ifndef __MIDCOMMS_DOT_H__
 #define __MIDCOMMS_DOT_H__
 
+struct midcomms_node;
+
 int dlm_process_incoming_buffer(int nodeid, unsigned char *buf, int buflen);
+struct dlm_mhandle *dlm_midcomms_get_mhandle(int nodeid, int len,
+					     gfp_t allocation, char **ppc);
+void dlm_midcomms_commit_mhandle(struct dlm_mhandle *mh);
+int dlm_midcomms_close(int nodeid);
+int dlm_midcomms_start(void);
+void dlm_midcomms_shutdown(void);
+void dlm_midcomms_add_member(int nodeid);
+void dlm_midcomms_remove_member(int nodeid);
+void dlm_midcomms_unack_msg_resend(int nodeid);
+const char *dlm_midcomms_state(struct midcomms_node *node);
+unsigned long dlm_midcomms_flags(struct midcomms_node *node);
+int dlm_midcomms_send_queue_cnt(struct midcomms_node *node);
+uint32_t dlm_midcomms_version(struct midcomms_node *node);
 
 #endif				/* __MIDCOMMS_DOT_H__ */
 
diff --git a/fs/dlm/rcom.c b/fs/dlm/rcom.c
index f5b1bd65728d..5651933f54a4 100644
--- a/fs/dlm/rcom.c
+++ b/fs/dlm/rcom.c
@@ -27,25 +27,15 @@ static int rcom_response(struct dlm_ls *ls)
 	return test_bit(LSFL_RCOM_READY, &ls->ls_flags);
 }
 
-static int create_rcom(struct dlm_ls *ls, int to_nodeid, int type, int len,
-		       struct dlm_rcom **rc_ret, struct dlm_mhandle **mh_ret)
+static void _create_rcom(struct dlm_ls *ls, int to_nodeid, int type, int len,
+			 struct dlm_rcom **rc_ret, char *mb, int mb_len)
 {
 	struct dlm_rcom *rc;
-	struct dlm_mhandle *mh;
-	char *mb;
-	int mb_len = sizeof(struct dlm_rcom) + len;
-
-	mh = dlm_lowcomms_get_buffer(to_nodeid, mb_len, GFP_NOFS, &mb);
-	if (!mh) {
-		log_print("create_rcom to %d type %d len %d ENOBUFS",
-			  to_nodeid, type, len);
-		return -ENOBUFS;
-	}
 
 	rc = (struct dlm_rcom *) mb;
 
 	rc->rc_header.h_version = (DLM_HEADER_MAJOR | DLM_HEADER_MINOR);
-	rc->rc_header.h_lockspace = ls->ls_global_id;
+	rc->rc_header.u.h_lockspace = ls->ls_global_id;
 	rc->rc_header.h_nodeid = dlm_our_nodeid();
 	rc->rc_header.h_length = mb_len;
 	rc->rc_header.h_cmd = DLM_RCOM;
@@ -56,16 +46,67 @@ static int create_rcom(struct dlm_ls *ls, int to_nodeid, int type, int len,
 	rc->rc_seq = ls->ls_recover_seq;
 	spin_unlock(&ls->ls_recover_lock);
 
-	*mh_ret = mh;
 	*rc_ret = rc;
+}
+
+static int create_rcom(struct dlm_ls *ls, int to_nodeid, int type, int len,
+		       struct dlm_rcom **rc_ret, struct dlm_mhandle **mh_ret)
+{
+	int mb_len = sizeof(struct dlm_rcom) + len;
+	struct dlm_mhandle *mh;
+	char *mb;
+
+	mh = dlm_midcomms_get_mhandle(to_nodeid, mb_len, GFP_NOFS, &mb);
+	if (!mh) {
+		log_print("%s to %d type %d len %d ENOBUFS",
+			  __func__, to_nodeid, type, len);
+		return -ENOBUFS;
+	}
+
+	_create_rcom(ls, to_nodeid, type, len, rc_ret, mb, mb_len);
+	*mh_ret = mh;
+	return 0;
+}
+
+static int create_rcom_stateless(struct dlm_ls *ls, int to_nodeid, int type,
+				 int len, struct dlm_rcom **rc_ret,
+				 struct dlm_msg **msg_ret)
+{
+	int mb_len = sizeof(struct dlm_rcom) + len;
+	struct dlm_msg *msg;
+	char *mb;
+
+	msg = dlm_lowcomms_new_msg(to_nodeid, mb_len, GFP_NOFS, &mb,
+				   NULL, NULL);
+	if (!msg) {
+		log_print("create_rcom to %d type %d len %d ENOBUFS",
+			  to_nodeid, type, len);
+		return -ENOBUFS;
+	}
+
+	_create_rcom(ls, to_nodeid, type, len, rc_ret, mb, mb_len);
+	*msg_ret = msg;
 	return 0;
 }
 
+static void _send_rcom(struct dlm_ls *ls, struct dlm_rcom *rc)
+{
+	dlm_rcom_out(rc);
+}
+
 static void send_rcom(struct dlm_ls *ls, struct dlm_mhandle *mh,
 		      struct dlm_rcom *rc)
 {
-	dlm_rcom_out(rc);
-	dlm_lowcomms_commit_buffer(mh);
+	_send_rcom(ls, rc);
+	dlm_midcomms_commit_mhandle(mh);
+}
+
+static void send_rcom_stateless(struct dlm_ls *ls, struct dlm_msg *msg,
+				struct dlm_rcom *rc)
+{
+	_send_rcom(ls, rc);
+	dlm_lowcomms_commit_msg(msg);
+	dlm_lowcomms_put_msg(msg);
 }
 
 static void set_rcom_status(struct dlm_ls *ls, struct rcom_status *rs,
@@ -141,7 +182,7 @@ static void disallow_sync_reply(struct dlm_ls *ls)
 int dlm_rcom_status(struct dlm_ls *ls, int nodeid, uint32_t status_flags)
 {
 	struct dlm_rcom *rc;
-	struct dlm_mhandle *mh;
+	struct dlm_msg *msg;
 	int error = 0;
 
 	ls->ls_recover_nodeid = nodeid;
@@ -153,17 +194,17 @@ int dlm_rcom_status(struct dlm_ls *ls, int nodeid, uint32_t status_flags)
 	}
 
 retry:
-	error = create_rcom(ls, nodeid, DLM_RCOM_STATUS,
-			    sizeof(struct rcom_status), &rc, &mh);
+	error = create_rcom_stateless(ls, nodeid, DLM_RCOM_STATUS,
+				      sizeof(struct rcom_status), &rc, &msg);
 	if (error)
 		goto out;
 
 	set_rcom_status(ls, (struct rcom_status *)rc->rc_buf, status_flags);
 
 	allow_sync_reply(ls, &rc->rc_id);
-	memset(ls->ls_recover_buf, 0, LOWCOMMS_MAX_TX_BUFFER_LEN);
+	memset(ls->ls_recover_buf, 0, DLM_MAX_SOCKET_BUFSIZE);
 
-	send_rcom(ls, mh, rc);
+	send_rcom_stateless(ls, msg, rc);
 
 	error = dlm_wait_function(ls, &rcom_response);
 	disallow_sync_reply(ls);
@@ -191,11 +232,11 @@ retry:
 static void receive_rcom_status(struct dlm_ls *ls, struct dlm_rcom *rc_in)
 {
 	struct dlm_rcom *rc;
-	struct dlm_mhandle *mh;
 	struct rcom_status *rs;
 	uint32_t status;
 	int nodeid = rc_in->rc_header.h_nodeid;
 	int len = sizeof(struct rcom_config);
+	struct dlm_msg *msg;
 	int num_slots = 0;
 	int error;
 
@@ -218,8 +259,8 @@ static void receive_rcom_status(struct dlm_ls *ls, struct dlm_rcom *rc_in)
 	len += num_slots * sizeof(struct rcom_slot);
 
  do_create:
-	error = create_rcom(ls, nodeid, DLM_RCOM_STATUS_REPLY,
-			    len, &rc, &mh);
+	error = create_rcom_stateless(ls, nodeid, DLM_RCOM_STATUS_REPLY,
+				      len, &rc, &msg);
 	if (error)
 		return;
 
@@ -246,7 +287,7 @@ static void receive_rcom_status(struct dlm_ls *ls, struct dlm_rcom *rc_in)
 	spin_unlock(&ls->ls_recover_lock);
 
  do_send:
-	send_rcom(ls, mh, rc);
+	send_rcom_stateless(ls, msg, rc);
 }
 
 static void receive_sync_reply(struct dlm_ls *ls, struct dlm_rcom *rc_in)
@@ -271,21 +312,22 @@ static void receive_sync_reply(struct dlm_ls *ls, struct dlm_rcom *rc_in)
 int dlm_rcom_names(struct dlm_ls *ls, int nodeid, char *last_name, int last_len)
 {
 	struct dlm_rcom *rc;
-	struct dlm_mhandle *mh;
+	struct dlm_msg *msg;
 	int error = 0;
 
 	ls->ls_recover_nodeid = nodeid;
 
 retry:
-	error = create_rcom(ls, nodeid, DLM_RCOM_NAMES, last_len, &rc, &mh);
+	error = create_rcom_stateless(ls, nodeid, DLM_RCOM_NAMES, last_len,
+				      &rc, &msg);
 	if (error)
 		goto out;
 	memcpy(rc->rc_buf, last_name, last_len);
 
 	allow_sync_reply(ls, &rc->rc_id);
-	memset(ls->ls_recover_buf, 0, LOWCOMMS_MAX_TX_BUFFER_LEN);
+	memset(ls->ls_recover_buf, 0, DLM_MAX_SOCKET_BUFSIZE);
 
-	send_rcom(ls, mh, rc);
+	send_rcom_stateless(ls, msg, rc);
 
 	error = dlm_wait_function(ls, &rcom_response);
 	disallow_sync_reply(ls);
@@ -298,14 +340,15 @@ retry:
 static void receive_rcom_names(struct dlm_ls *ls, struct dlm_rcom *rc_in)
 {
 	struct dlm_rcom *rc;
-	struct dlm_mhandle *mh;
 	int error, inlen, outlen, nodeid;
+	struct dlm_msg *msg;
 
 	nodeid = rc_in->rc_header.h_nodeid;
 	inlen = rc_in->rc_header.h_length - sizeof(struct dlm_rcom);
-	outlen = LOWCOMMS_MAX_TX_BUFFER_LEN - sizeof(struct dlm_rcom);
+	outlen = DLM_MAX_APP_BUFSIZE - sizeof(struct dlm_rcom);
 
-	error = create_rcom(ls, nodeid, DLM_RCOM_NAMES_REPLY, outlen, &rc, &mh);
+	error = create_rcom_stateless(ls, nodeid, DLM_RCOM_NAMES_REPLY, outlen,
+				      &rc, &msg);
 	if (error)
 		return;
 	rc->rc_id = rc_in->rc_id;
@@ -313,7 +356,7 @@ static void receive_rcom_names(struct dlm_ls *ls, struct dlm_rcom *rc_in)
 
 	dlm_copy_master_names(ls, rc_in->rc_buf, inlen, rc->rc_buf, outlen,
 			      nodeid);
-	send_rcom(ls, mh, rc);
+	send_rcom_stateless(ls, msg, rc);
 }
 
 int dlm_send_rcom_lookup(struct dlm_rsb *r, int dir_nodeid)
@@ -342,10 +385,6 @@ static void receive_rcom_lookup(struct dlm_ls *ls, struct dlm_rcom *rc_in)
 	int error, ret_nodeid, nodeid = rc_in->rc_header.h_nodeid;
 	int len = rc_in->rc_header.h_length - sizeof(struct dlm_rcom);
 
-	error = create_rcom(ls, nodeid, DLM_RCOM_LOOKUP_REPLY, 0, &rc, &mh);
-	if (error)
-		return;
-
 	/* Old code would send this special id to trigger a debug dump. */
 	if (rc_in->rc_id == 0xFFFFFFFF) {
 		log_error(ls, "receive_rcom_lookup dump from %d", nodeid);
@@ -353,6 +392,10 @@ static void receive_rcom_lookup(struct dlm_ls *ls, struct dlm_rcom *rc_in)
 		return;
 	}
 
+	error = create_rcom(ls, nodeid, DLM_RCOM_LOOKUP_REPLY, 0, &rc, &mh);
+	if (error)
+		return;
+
 	error = dlm_master_lookup(ls, nodeid, rc_in->rc_buf, len,
 				  DLM_LU_RECOVER_MASTER, &ret_nodeid, NULL);
 	if (error)
@@ -458,14 +501,14 @@ int dlm_send_ls_not_ready(int nodeid, struct dlm_rcom *rc_in)
 	char *mb;
 	int mb_len = sizeof(struct dlm_rcom) + sizeof(struct rcom_config);
 
-	mh = dlm_lowcomms_get_buffer(nodeid, mb_len, GFP_NOFS, &mb);
+	mh = dlm_midcomms_get_mhandle(nodeid, mb_len, GFP_NOFS, &mb);
 	if (!mh)
 		return -ENOBUFS;
 
 	rc = (struct dlm_rcom *) mb;
 
 	rc->rc_header.h_version = (DLM_HEADER_MAJOR | DLM_HEADER_MINOR);
-	rc->rc_header.h_lockspace = rc_in->rc_header.h_lockspace;
+	rc->rc_header.u.h_lockspace = rc_in->rc_header.u.h_lockspace;
 	rc->rc_header.h_nodeid = dlm_our_nodeid();
 	rc->rc_header.h_length = mb_len;
 	rc->rc_header.h_cmd = DLM_RCOM;
@@ -479,7 +522,7 @@ int dlm_send_ls_not_ready(int nodeid, struct dlm_rcom *rc_in)
 	rf->rf_lvblen = cpu_to_le32(~0U);
 
 	dlm_rcom_out(rc);
-	dlm_lowcomms_commit_buffer(mh);
+	dlm_midcomms_commit_mhandle(mh);
 
 	return 0;
 }
diff --git a/fs/dlm/util.c b/fs/dlm/util.c
index cfd0d00b19ae..58acbcc2081a 100644
--- a/fs/dlm/util.c
+++ b/fs/dlm/util.c
@@ -20,18 +20,20 @@
 #define DLM_ERRNO_ETIMEDOUT	       110
 #define DLM_ERRNO_EINPROGRESS	       115
 
-static void header_out(struct dlm_header *hd)
+void header_out(struct dlm_header *hd)
 {
 	hd->h_version		= cpu_to_le32(hd->h_version);
-	hd->h_lockspace		= cpu_to_le32(hd->h_lockspace);
+	/* does it for others u32 in union as well */
+	hd->u.h_lockspace	= cpu_to_le32(hd->u.h_lockspace);
 	hd->h_nodeid		= cpu_to_le32(hd->h_nodeid);
 	hd->h_length		= cpu_to_le16(hd->h_length);
 }
 
-static void header_in(struct dlm_header *hd)
+void header_in(struct dlm_header *hd)
 {
 	hd->h_version		= le32_to_cpu(hd->h_version);
-	hd->h_lockspace		= le32_to_cpu(hd->h_lockspace);
+	/* does it for others u32 in union as well */
+	hd->u.h_lockspace	= le32_to_cpu(hd->u.h_lockspace);
 	hd->h_nodeid		= le32_to_cpu(hd->h_nodeid);
 	hd->h_length		= le16_to_cpu(hd->h_length);
 }
diff --git a/fs/dlm/util.h b/fs/dlm/util.h
index cc719ca9397e..d46f23c7a6a0 100644
--- a/fs/dlm/util.h
+++ b/fs/dlm/util.h
@@ -15,6 +15,8 @@ void dlm_message_out(struct dlm_message *ms);
 void dlm_message_in(struct dlm_message *ms);
 void dlm_rcom_out(struct dlm_rcom *rc);
 void dlm_rcom_in(struct dlm_rcom *rc);
+void header_out(struct dlm_header *hd);
+void header_in(struct dlm_header *hd);
 
 #endif
author	Linus Torvalds <torvalds@linux-foundation.org>	2021-06-29 20:30:00 -0700
committer	Linus Torvalds <torvalds@linux-foundation.org>	2021-06-29 20:30:00 -0700
commit	007b350a58754a93ca9fe50c498cc27780171153 (patch)
tree	c9e1c689bc30662faf4c591fefec2d9d10afddff /fs
parent	8418dabd97ce4b0713c36dd8226978f737c342b0 (diff)
parent	957adb68b3f7df8421a05f1647d3027f2acad310 (diff)