summary refs log tree commit diff
path: root/fs/dlm
diff options
context:
space:
mode:
Diffstat (limited to 'fs/dlm')
-rw-r--r--fs/dlm/Kconfig20
-rw-r--r--fs/dlm/Makefile4
-rw-r--r--fs/dlm/dlm_internal.h4
-rw-r--r--fs/dlm/lock.c16
-rw-r--r--fs/dlm/lockspace.c4
-rw-r--r--fs/dlm/lowcomms-sctp.c (renamed from fs/dlm/lowcomms.c)264
-rw-r--r--fs/dlm/lowcomms-tcp.c1189
-rw-r--r--fs/dlm/lowcomms.h2
-rw-r--r--fs/dlm/main.c10
-rw-r--r--fs/dlm/member.c8
-rw-r--r--fs/dlm/memory.c2
-rw-r--r--fs/dlm/rcom.c58
-rw-r--r--fs/dlm/recover.c1
-rw-r--r--fs/dlm/recoverd.c44
-rw-r--r--fs/dlm/requestqueue.c26
-rw-r--r--fs/dlm/requestqueue.h2
16 files changed, 1464 insertions, 190 deletions
diff --git a/fs/dlm/Kconfig b/fs/dlm/Kconfig
index 81b2c6465eeb..b5654a284fef 100644
--- a/fs/dlm/Kconfig
+++ b/fs/dlm/Kconfig
@@ -1,14 +1,32 @@
 menu "Distributed Lock Manager"
-	depends on INET && IP_SCTP && EXPERIMENTAL
+	depends on EXPERIMENTAL && INET
 
 config DLM
 	tristate "Distributed Lock Manager (DLM)"
 	depends on IPV6 || IPV6=n
 	select CONFIGFS_FS
+	select IP_SCTP if DLM_SCTP
 	help
 	A general purpose distributed lock manager for kernel or userspace
 	applications.
 
+choice
+	prompt "Select DLM communications protocol"
+	depends on DLM
+	default DLM_TCP
+	help
+	The DLM Can use TCP or SCTP for it's network communications.
+	SCTP supports multi-homed operations whereas TCP doesn't.
+	However, SCTP seems to have stability problems at the moment.
+
+config DLM_TCP
+	bool "TCP/IP"
+
+config DLM_SCTP
+	bool "SCTP"
+
+endchoice
+
 config DLM_DEBUG
 	bool "DLM debugging"
 	depends on DLM
diff --git a/fs/dlm/Makefile b/fs/dlm/Makefile
index 1832e0297f7d..65388944eba0 100644
--- a/fs/dlm/Makefile
+++ b/fs/dlm/Makefile
@@ -4,7 +4,6 @@ dlm-y :=			ast.o \
 				dir.o \
 				lock.o \
 				lockspace.o \
-				lowcomms.o \
 				main.o \
 				member.o \
 				memory.o \
@@ -17,3 +16,6 @@ dlm-y :=			ast.o \
 				util.o
 dlm-$(CONFIG_DLM_DEBUG) +=	debug_fs.o
 
+dlm-$(CONFIG_DLM_TCP)   += lowcomms-tcp.o
+
+dlm-$(CONFIG_DLM_SCTP)  += lowcomms-sctp.o
\ No newline at end of file
diff --git a/fs/dlm/dlm_internal.h b/fs/dlm/dlm_internal.h
index 1e5cd67e1b7a..1ee8195e6fc0 100644
--- a/fs/dlm/dlm_internal.h
+++ b/fs/dlm/dlm_internal.h
@@ -471,6 +471,7 @@ struct dlm_ls {
 	char			*ls_recover_buf;
 	int			ls_recover_nodeid; /* for debugging */
 	uint64_t		ls_rcom_seq;
+	spinlock_t		ls_rcom_spin;
 	struct list_head	ls_recover_list;
 	spinlock_t		ls_recover_list_lock;
 	int			ls_recover_list_count;
@@ -488,7 +489,8 @@ struct dlm_ls {
 #define LSFL_RUNNING		1
 #define LSFL_RECOVERY_STOP	2
 #define LSFL_RCOM_READY		3
-#define LSFL_UEVENT_WAIT	4
+#define LSFL_RCOM_WAIT		4
+#define LSFL_UEVENT_WAIT	5
 
 /* much of this is just saving user space pointers associated with the
    lock that we pass back to the user lib with an ast */
diff --git a/fs/dlm/lock.c b/fs/dlm/lock.c
index 3f2befa4797b..30878defaeb6 100644
--- a/fs/dlm/lock.c
+++ b/fs/dlm/lock.c
@@ -2372,6 +2372,7 @@ static int send_lookup_reply(struct dlm_ls *ls, struct dlm_message *ms_in,
 static void receive_flags(struct dlm_lkb *lkb, struct dlm_message *ms)
 {
 	lkb->lkb_exflags = ms->m_exflags;
+	lkb->lkb_sbflags = ms->m_sbflags;
 	lkb->lkb_flags = (lkb->lkb_flags & 0xFFFF0000) |
 		         (ms->m_flags & 0x0000FFFF);
 }
@@ -3028,10 +3029,17 @@ int dlm_receive_message(struct dlm_header *hd, int nodeid, int recovery)
 
 	while (1) {
 		if (dlm_locking_stopped(ls)) {
-			if (!recovery)
-				dlm_add_requestqueue(ls, nodeid, hd);
-			error = -EINTR;
-			goto out;
+			if (recovery) {
+				error = -EINTR;
+				goto out;
+			}
+			error = dlm_add_requestqueue(ls, nodeid, hd);
+			if (error == -EAGAIN)
+				continue;
+			else {
+				error = -EINTR;
+				goto out;
+			}
 		}
 
 		if (lock_recovery_try(ls))
diff --git a/fs/dlm/lockspace.c b/fs/dlm/lockspace.c
index f8842ca443c2..59012b089e8d 100644
--- a/fs/dlm/lockspace.c
+++ b/fs/dlm/lockspace.c
@@ -22,6 +22,7 @@
 #include "memory.h"
 #include "lock.h"
 #include "recover.h"
+#include "requestqueue.h"
 
 #ifdef CONFIG_DLM_DEBUG
 int dlm_create_debug_file(struct dlm_ls *ls);
@@ -478,6 +479,8 @@ static int new_lockspace(char *name, int namelen, void **lockspace,
 	ls->ls_recoverd_task = NULL;
 	mutex_init(&ls->ls_recoverd_active);
 	spin_lock_init(&ls->ls_recover_lock);
+	spin_lock_init(&ls->ls_rcom_spin);
+	get_random_bytes(&ls->ls_rcom_seq, sizeof(uint64_t));
 	ls->ls_recover_status = 0;
 	ls->ls_recover_seq = 0;
 	ls->ls_recover_args = NULL;
@@ -684,6 +687,7 @@ static int release_lockspace(struct dlm_ls *ls, int force)
 	 * Free structures on any other lists
 	 */
 
+	dlm_purge_requestqueue(ls);
 	kfree(ls->ls_recover_args);
 	dlm_clear_free_entries(ls);
 	dlm_clear_members(ls);
diff --git a/fs/dlm/lowcomms.c b/fs/dlm/lowcomms-sctp.c
index 6da6b14d5a61..fe158d7a9285 100644
--- a/fs/dlm/lowcomms.c
+++ b/fs/dlm/lowcomms-sctp.c
@@ -2,7 +2,7 @@
 *******************************************************************************
 **
 **  Copyright (C) Sistina Software, Inc.  1997-2003  All rights reserved.
-**  Copyright (C) 2004-2005 Red Hat, Inc.  All rights reserved.
+**  Copyright (C) 2004-2006 Red Hat, Inc.  All rights reserved.
 **
 **  This copyrighted material is made available to anyone wishing to use,
 **  modify, copy, or redistribute it subject to the terms and conditions
@@ -75,13 +75,13 @@ struct nodeinfo {
 };
 
 static DEFINE_IDR(nodeinfo_idr);
-static struct rw_semaphore	nodeinfo_lock;
-static int			max_nodeid;
+static DECLARE_RWSEM(nodeinfo_lock);
+static int max_nodeid;
 
 struct cbuf {
-	unsigned		base;
-	unsigned		len;
-	unsigned		mask;
+	unsigned int base;
+	unsigned int len;
+	unsigned int mask;
 };
 
 /* Just the one of these, now. But this struct keeps
@@ -90,9 +90,9 @@ struct cbuf {
 #define CF_READ_PENDING 1
 
 struct connection {
-	struct socket          *sock;
+	struct socket           *sock;
 	unsigned long		flags;
-	struct page            *rx_page;
+	struct page             *rx_page;
 	atomic_t		waiting_requests;
 	struct cbuf		cb;
 	int                     eagain_flag;
@@ -102,36 +102,40 @@ struct connection {
 
 struct writequeue_entry {
 	struct list_head	list;
-	struct page            *page;
+	struct page             *page;
 	int			offset;
 	int			len;
 	int			end;
 	int			users;
-	struct nodeinfo        *ni;
+	struct nodeinfo         *ni;
 };
 
-#define CBUF_ADD(cb, n) do { (cb)->len += n; } while(0)
-#define CBUF_EMPTY(cb) ((cb)->len == 0)
-#define CBUF_MAY_ADD(cb, n) (((cb)->len + (n)) < ((cb)->mask + 1))
-#define CBUF_DATA(cb) (((cb)->base + (cb)->len) & (cb)->mask)
+static void cbuf_add(struct cbuf *cb, int n)
+{
+	cb->len += n;
+}
 
-#define CBUF_INIT(cb, size) \
-do { \
-	(cb)->base = (cb)->len = 0; \
-	(cb)->mask = ((size)-1); \
-} while(0)
+static int cbuf_data(struct cbuf *cb)
+{
+	return ((cb->base + cb->len) & cb->mask);
+}
 
-#define CBUF_EAT(cb, n) \
-do { \
-	(cb)->len  -= (n); \
-	(cb)->base += (n); \
-	(cb)->base &= (cb)->mask; \
-} while(0)
+static void cbuf_init(struct cbuf *cb, int size)
+{
+	cb->base = cb->len = 0;
+	cb->mask = size-1;
+}
 
+static void cbuf_eat(struct cbuf *cb, int n)
+{
+	cb->len  -= n;
+	cb->base += n;
+	cb->base &= cb->mask;
+}
 
 /* List of nodes which have writes pending */
-static struct list_head write_nodes;
-static spinlock_t write_nodes_lock;
+static LIST_HEAD(write_nodes);
+static DEFINE_SPINLOCK(write_nodes_lock);
 
 /* Maximum number of incoming messages to process before
  * doing a schedule()
@@ -141,8 +145,7 @@ static spinlock_t write_nodes_lock;
 /* Manage daemons */
 static struct task_struct *recv_task;
 static struct task_struct *send_task;
-static wait_queue_head_t lowcomms_recv_wait;
-static atomic_t accepting;
+static DECLARE_WAIT_QUEUE_HEAD(lowcomms_recv_wait);
 
 /* The SCTP connection */
 static struct connection sctp_con;
@@ -161,11 +164,11 @@ static int nodeid_to_addr(int nodeid, struct sockaddr *retaddr)
 		return error;
 
 	if (dlm_local_addr[0]->ss_family == AF_INET) {
-	        struct sockaddr_in *in4  = (struct sockaddr_in *) &addr;
+		struct sockaddr_in *in4  = (struct sockaddr_in *) &addr;
 		struct sockaddr_in *ret4 = (struct sockaddr_in *) retaddr;
 		ret4->sin_addr.s_addr = in4->sin_addr.s_addr;
 	} else {
-	        struct sockaddr_in6 *in6  = (struct sockaddr_in6 *) &addr;
+		struct sockaddr_in6 *in6  = (struct sockaddr_in6 *) &addr;
 		struct sockaddr_in6 *ret6 = (struct sockaddr_in6 *) retaddr;
 		memcpy(&ret6->sin6_addr, &in6->sin6_addr,
 		       sizeof(in6->sin6_addr));
@@ -174,6 +177,8 @@ static int nodeid_to_addr(int nodeid, struct sockaddr *retaddr)
 	return 0;
 }
 
+/* If alloc is 0 here we will not attempt to allocate a new
+   nodeinfo struct */
 static struct nodeinfo *nodeid2nodeinfo(int nodeid, gfp_t alloc)
 {
 	struct nodeinfo *ni;
@@ -184,44 +189,45 @@ static struct nodeinfo *nodeid2nodeinfo(int nodeid, gfp_t alloc)
 	ni = idr_find(&nodeinfo_idr, nodeid);
 	up_read(&nodeinfo_lock);
 
-	if (!ni && alloc) {
-		down_write(&nodeinfo_lock);
+	if (ni || !alloc)
+		return ni;
 
-		ni = idr_find(&nodeinfo_idr, nodeid);
-		if (ni)
-			goto out_up;
+	down_write(&nodeinfo_lock);
 
-		r = idr_pre_get(&nodeinfo_idr, alloc);
-		if (!r)
-			goto out_up;
+	ni = idr_find(&nodeinfo_idr, nodeid);
+	if (ni)
+		goto out_up;
 
-		ni = kmalloc(sizeof(struct nodeinfo), alloc);
-		if (!ni)
-			goto out_up;
+	r = idr_pre_get(&nodeinfo_idr, alloc);
+	if (!r)
+		goto out_up;
 
-		r = idr_get_new_above(&nodeinfo_idr, ni, nodeid, &n);
-		if (r) {
-			kfree(ni);
-			ni = NULL;
-			goto out_up;
-		}
-		if (n != nodeid) {
-			idr_remove(&nodeinfo_idr, n);
-			kfree(ni);
-			ni = NULL;
-			goto out_up;
-		}
-		memset(ni, 0, sizeof(struct nodeinfo));
-		spin_lock_init(&ni->lock);
-		INIT_LIST_HEAD(&ni->writequeue);
-		spin_lock_init(&ni->writequeue_lock);
-		ni->nodeid = nodeid;
-
-		if (nodeid > max_nodeid)
-			max_nodeid = nodeid;
-	out_up:
-		up_write(&nodeinfo_lock);
+	ni = kmalloc(sizeof(struct nodeinfo), alloc);
+	if (!ni)
+		goto out_up;
+
+	r = idr_get_new_above(&nodeinfo_idr, ni, nodeid, &n);
+	if (r) {
+		kfree(ni);
+		ni = NULL;
+		goto out_up;
 	}
+	if (n != nodeid) {
+		idr_remove(&nodeinfo_idr, n);
+		kfree(ni);
+		ni = NULL;
+		goto out_up;
+	}
+	memset(ni, 0, sizeof(struct nodeinfo));
+	spin_lock_init(&ni->lock);
+	INIT_LIST_HEAD(&ni->writequeue);
+	spin_lock_init(&ni->writequeue_lock);
+	ni->nodeid = nodeid;
+
+	if (nodeid > max_nodeid)
+		max_nodeid = nodeid;
+out_up:
+	up_write(&nodeinfo_lock);
 
 	return ni;
 }
@@ -279,13 +285,13 @@ static void make_sockaddr(struct sockaddr_storage *saddr, uint16_t port,
 		in4_addr->sin_port = cpu_to_be16(port);
 		memset(&in4_addr->sin_zero, 0, sizeof(in4_addr->sin_zero));
 		memset(in4_addr+1, 0, sizeof(struct sockaddr_storage) -
-				      sizeof(struct sockaddr_in));
+		       sizeof(struct sockaddr_in));
 		*addr_len = sizeof(struct sockaddr_in);
 	} else {
 		struct sockaddr_in6 *in6_addr = (struct sockaddr_in6 *)saddr;
 		in6_addr->sin6_port = cpu_to_be16(port);
 		memset(in6_addr+1, 0, sizeof(struct sockaddr_storage) -
-				      sizeof(struct sockaddr_in6));
+		       sizeof(struct sockaddr_in6));
 		*addr_len = sizeof(struct sockaddr_in6);
 	}
 }
@@ -324,7 +330,7 @@ static void send_shutdown(sctp_assoc_t associd)
 	cmsg->cmsg_type = SCTP_SNDRCV;
 	cmsg->cmsg_len = CMSG_LEN(sizeof(struct sctp_sndrcvinfo));
 	outmessage.msg_controllen = cmsg->cmsg_len;
-	sinfo = (struct sctp_sndrcvinfo *)CMSG_DATA(cmsg);
+	sinfo = CMSG_DATA(cmsg);
 	memset(sinfo, 0x00, sizeof(struct sctp_sndrcvinfo));
 
 	sinfo->sinfo_flags |= MSG_EOF;
@@ -387,7 +393,7 @@ static void process_sctp_notification(struct msghdr *msg, char *buf)
 
 			if ((int)sn->sn_assoc_change.sac_assoc_id <= 0) {
 				log_print("COMM_UP for invalid assoc ID %d",
-					 (int)sn->sn_assoc_change.sac_assoc_id);
+					  (int)sn->sn_assoc_change.sac_assoc_id);
 				init_failed();
 				return;
 			}
@@ -398,15 +404,18 @@ static void process_sctp_notification(struct msghdr *msg, char *buf)
 			fs = get_fs();
 			set_fs(get_ds());
 			ret = sctp_con.sock->ops->getsockopt(sctp_con.sock,
-						IPPROTO_SCTP, SCTP_PRIMARY_ADDR,
-						(char*)&prim, &prim_len);
+							     IPPROTO_SCTP,
+							     SCTP_PRIMARY_ADDR,
+							     (char*)&prim,
+							     &prim_len);
 			set_fs(fs);
 			if (ret < 0) {
 				struct nodeinfo *ni;
 
 				log_print("getsockopt/sctp_primary_addr on "
 					  "new assoc %d failed : %d",
-				    (int)sn->sn_assoc_change.sac_assoc_id, ret);
+					  (int)sn->sn_assoc_change.sac_assoc_id,
+					  ret);
 
 				/* Retry INIT later */
 				ni = assoc2nodeinfo(sn->sn_assoc_change.sac_assoc_id);
@@ -426,12 +435,10 @@ static void process_sctp_notification(struct msghdr *msg, char *buf)
 				return;
 
 			/* Save the assoc ID */
-			spin_lock(&ni->lock);
 			ni->assoc_id = sn->sn_assoc_change.sac_assoc_id;
-			spin_unlock(&ni->lock);
 
 			log_print("got new/restarted association %d nodeid %d",
-			       (int)sn->sn_assoc_change.sac_assoc_id, nodeid);
+				  (int)sn->sn_assoc_change.sac_assoc_id, nodeid);
 
 			/* Send any pending writes */
 			clear_bit(NI_INIT_PENDING, &ni->flags);
@@ -507,13 +514,12 @@ static int receive_from_sock(void)
 		sctp_con.rx_page = alloc_page(GFP_ATOMIC);
 		if (sctp_con.rx_page == NULL)
 			goto out_resched;
-		CBUF_INIT(&sctp_con.cb, PAGE_CACHE_SIZE);
+		cbuf_init(&sctp_con.cb, PAGE_CACHE_SIZE);
 	}
 
 	memset(&incmsg, 0, sizeof(incmsg));
 	memset(&msgname, 0, sizeof(msgname));
 
-	memset(incmsg, 0, sizeof(incmsg));
 	msg.msg_name = &msgname;
 	msg.msg_namelen = sizeof(msgname);
 	msg.msg_flags = 0;
@@ -532,17 +538,17 @@ static int receive_from_sock(void)
 	 * iov[0] is the bit of the circular buffer between the current end
 	 * point (cb.base + cb.len) and the end of the buffer.
 	 */
-	iov[0].iov_len = sctp_con.cb.base - CBUF_DATA(&sctp_con.cb);
+	iov[0].iov_len = sctp_con.cb.base - cbuf_data(&sctp_con.cb);
 	iov[0].iov_base = page_address(sctp_con.rx_page) +
-			  CBUF_DATA(&sctp_con.cb);
+		cbuf_data(&sctp_con.cb);
 	iov[1].iov_len = 0;
 
 	/*
 	 * iov[1] is the bit of the circular buffer between the start of the
 	 * buffer and the start of the currently used section (cb.base)
 	 */
-	if (CBUF_DATA(&sctp_con.cb) >= sctp_con.cb.base) {
-		iov[0].iov_len = PAGE_CACHE_SIZE - CBUF_DATA(&sctp_con.cb);
+	if (cbuf_data(&sctp_con.cb) >= sctp_con.cb.base) {
+		iov[0].iov_len = PAGE_CACHE_SIZE - cbuf_data(&sctp_con.cb);
 		iov[1].iov_len = sctp_con.cb.base;
 		iov[1].iov_base = page_address(sctp_con.rx_page);
 		msg.msg_iovlen = 2;
@@ -557,7 +563,7 @@ static int receive_from_sock(void)
 	msg.msg_control = incmsg;
 	msg.msg_controllen = sizeof(incmsg);
 	cmsg = CMSG_FIRSTHDR(&msg);
-	sinfo = (struct sctp_sndrcvinfo *)CMSG_DATA(cmsg);
+	sinfo = CMSG_DATA(cmsg);
 
 	if (msg.msg_flags & MSG_NOTIFICATION) {
 		process_sctp_notification(&msg, page_address(sctp_con.rx_page));
@@ -583,29 +589,29 @@ static int receive_from_sock(void)
 	if (r == 1)
 		return 0;
 
-	CBUF_ADD(&sctp_con.cb, ret);
+	cbuf_add(&sctp_con.cb, ret);
 	ret = dlm_process_incoming_buffer(cpu_to_le32(sinfo->sinfo_ppid),
 					  page_address(sctp_con.rx_page),
 					  sctp_con.cb.base, sctp_con.cb.len,
 					  PAGE_CACHE_SIZE);
 	if (ret < 0)
 		goto out_close;
-	CBUF_EAT(&sctp_con.cb, ret);
+	cbuf_eat(&sctp_con.cb, ret);
 
-      out:
+out:
 	ret = 0;
 	goto out_ret;
 
-      out_resched:
+out_resched:
 	lowcomms_data_ready(sctp_con.sock->sk, 0);
 	ret = 0;
-	schedule();
+	cond_resched();
 	goto out_ret;
 
-      out_close:
+out_close:
 	if (ret != -EAGAIN)
 		log_print("error reading from sctp socket: %d", ret);
-      out_ret:
+out_ret:
 	return ret;
 }
 
@@ -619,10 +625,12 @@ static int add_bind_addr(struct sockaddr_storage *addr, int addr_len, int num)
 	set_fs(get_ds());
 	if (num == 1)
 		result = sctp_con.sock->ops->bind(sctp_con.sock,
-					(struct sockaddr *) addr, addr_len);
+						  (struct sockaddr *) addr,
+						  addr_len);
 	else
 		result = sctp_con.sock->ops->setsockopt(sctp_con.sock, SOL_SCTP,
-				SCTP_SOCKOPT_BINDX_ADD, (char *)addr, addr_len);
+							SCTP_SOCKOPT_BINDX_ADD,
+							(char *)addr, addr_len);
 	set_fs(fs);
 
 	if (result < 0)
@@ -719,10 +727,10 @@ static int init_sock(void)
 
 	return 0;
 
- create_delsock:
+create_delsock:
 	sock_release(sock);
 	sctp_con.sock = NULL;
- out:
+out:
 	return result;
 }
 
@@ -756,16 +764,13 @@ void *dlm_lowcomms_get_buffer(int nodeid, int len, gfp_t allocation, char **ppc)
 	int users = 0;
 	struct nodeinfo *ni;
 
-	if (!atomic_read(&accepting))
-		return NULL;
-
 	ni = nodeid2nodeinfo(nodeid, allocation);
 	if (!ni)
 		return NULL;
 
 	spin_lock(&ni->writequeue_lock);
 	e = list_entry(ni->writequeue.prev, struct writequeue_entry, list);
-	if (((struct list_head *) e == &ni->writequeue) ||
+	if ((&e->list == &ni->writequeue) ||
 	    (PAGE_CACHE_SIZE - e->end < len)) {
 		e = NULL;
 	} else {
@@ -776,7 +781,7 @@ void *dlm_lowcomms_get_buffer(int nodeid, int len, gfp_t allocation, char **ppc)
 	spin_unlock(&ni->writequeue_lock);
 
 	if (e) {
-	      got_one:
+	got_one:
 		if (users == 0)
 			kmap(e->page);
 		*ppc = page_address(e->page) + offset;
@@ -803,9 +808,6 @@ void dlm_lowcomms_commit_buffer(void *arg)
 	int users;
 	struct nodeinfo *ni = e->ni;
 
-	if (!atomic_read(&accepting))
-		return;
-
 	spin_lock(&ni->writequeue_lock);
 	users = --e->users;
 	if (users)
@@ -822,7 +824,7 @@ void dlm_lowcomms_commit_buffer(void *arg)
 	}
 	return;
 
-      out:
+out:
 	spin_unlock(&ni->writequeue_lock);
 	return;
 }
@@ -878,7 +880,7 @@ static void initiate_association(int nodeid)
 	cmsg->cmsg_level = IPPROTO_SCTP;
 	cmsg->cmsg_type = SCTP_SNDRCV;
 	cmsg->cmsg_len = CMSG_LEN(sizeof(struct sctp_sndrcvinfo));
-	sinfo = (struct sctp_sndrcvinfo *)CMSG_DATA(cmsg);
+	sinfo = CMSG_DATA(cmsg);
 	memset(sinfo, 0x00, sizeof(struct sctp_sndrcvinfo));
 	sinfo->sinfo_ppid = cpu_to_le32(dlm_local_nodeid);
 
@@ -892,7 +894,7 @@ static void initiate_association(int nodeid)
 }
 
 /* Send a message */
-static int send_to_sock(struct nodeinfo *ni)
+static void send_to_sock(struct nodeinfo *ni)
 {
 	int ret = 0;
 	struct writequeue_entry *e;
@@ -903,13 +905,13 @@ static int send_to_sock(struct nodeinfo *ni)
 	struct sctp_sndrcvinfo *sinfo;
 	struct kvec iov;
 
-        /* See if we need to init an association before we start
+	/* See if we need to init an association before we start
 	   sending precious messages */
 	spin_lock(&ni->lock);
 	if (!ni->assoc_id && !test_and_set_bit(NI_INIT_PENDING, &ni->flags)) {
 		spin_unlock(&ni->lock);
 		initiate_association(ni->nodeid);
-		return 0;
+		return;
 	}
 	spin_unlock(&ni->lock);
 
@@ -923,7 +925,7 @@ static int send_to_sock(struct nodeinfo *ni)
 	cmsg->cmsg_level = IPPROTO_SCTP;
 	cmsg->cmsg_type = SCTP_SNDRCV;
 	cmsg->cmsg_len = CMSG_LEN(sizeof(struct sctp_sndrcvinfo));
-	sinfo = (struct sctp_sndrcvinfo *)CMSG_DATA(cmsg);
+	sinfo = CMSG_DATA(cmsg);
 	memset(sinfo, 0x00, sizeof(struct sctp_sndrcvinfo));
 	sinfo->sinfo_ppid = cpu_to_le32(dlm_local_nodeid);
 	sinfo->sinfo_assoc_id = ni->assoc_id;
@@ -955,7 +957,7 @@ static int send_to_sock(struct nodeinfo *ni)
 				goto send_error;
 		} else {
 			/* Don't starve people filling buffers */
-			schedule();
+			cond_resched();
 		}
 
 		spin_lock(&ni->writequeue_lock);
@@ -964,15 +966,16 @@ static int send_to_sock(struct nodeinfo *ni)
 
 		if (e->len == 0 && e->users == 0) {
 			list_del(&e->list);
+			kunmap(e->page);
 			free_entry(e);
 			continue;
 		}
 	}
 	spin_unlock(&ni->writequeue_lock);
- out:
-	return ret;
+out:
+	return;
 
- send_error:
+send_error:
 	log_print("Error sending to node %d %d", ni->nodeid, ret);
 	spin_lock(&ni->lock);
 	if (!test_and_set_bit(NI_INIT_PENDING, &ni->flags)) {
@@ -982,7 +985,7 @@ static int send_to_sock(struct nodeinfo *ni)
 	} else
 		spin_unlock(&ni->lock);
 
-	return ret;
+	return;
 }
 
 /* Try to send any messages that are pending */
@@ -994,7 +997,7 @@ static void process_output_queue(void)
 	spin_lock_bh(&write_nodes_lock);
 	list_for_each_safe(list, temp, &write_nodes) {
 		struct nodeinfo *ni =
-		    list_entry(list, struct nodeinfo, write_list);
+			list_entry(list, struct nodeinfo, write_list);
 		clear_bit(NI_WRITE_PENDING, &ni->flags);
 		list_del(&ni->write_list);
 
@@ -1106,7 +1109,7 @@ static int dlm_recvd(void *data)
 		set_current_state(TASK_INTERRUPTIBLE);
 		add_wait_queue(&lowcomms_recv_wait, &wait);
 		if (!test_bit(CF_READ_PENDING, &sctp_con.flags))
-			schedule();
+			cond_resched();
 		remove_wait_queue(&lowcomms_recv_wait, &wait);
 		set_current_state(TASK_RUNNING);
 
@@ -1118,12 +1121,12 @@ static int dlm_recvd(void *data)
 
 				/* Don't starve out everyone else */
 				if (++count >= MAX_RX_MSG_COUNT) {
-					schedule();
+					cond_resched();
 					count = 0;
 				}
 			} while (!kthread_should_stop() && ret >=0);
 		}
-		schedule();
+		cond_resched();
 	}
 
 	return 0;
@@ -1138,7 +1141,7 @@ static int dlm_sendd(void *data)
 	while (!kthread_should_stop()) {
 		set_current_state(TASK_INTERRUPTIBLE);
 		if (write_list_empty())
-			schedule();
+			cond_resched();
 		set_current_state(TASK_RUNNING);
 
 		if (sctp_con.eagain_flag) {
@@ -1166,7 +1169,7 @@ static int daemons_start(void)
 
 	p = kthread_run(dlm_recvd, NULL, "dlm_recvd");
 	error = IS_ERR(p);
-       	if (error) {
+	if (error) {
 		log_print("can't start dlm_recvd %d", error);
 		return error;
 	}
@@ -1174,7 +1177,7 @@ static int daemons_start(void)
 
 	p = kthread_run(dlm_sendd, NULL, "dlm_sendd");
 	error = IS_ERR(p);
-       	if (error) {
+	if (error) {
 		log_print("can't start dlm_sendd %d", error);
 		kthread_stop(recv_task);
 		return error;
@@ -1197,43 +1200,28 @@ int dlm_lowcomms_start(void)
 	error = daemons_start();
 	if (error)
 		goto fail_sock;
-	atomic_set(&accepting, 1);
 	return 0;
 
- fail_sock:
+fail_sock:
 	close_connection();
 	return error;
 }
 
-/* Set all the activity flags to prevent any socket activity. */
-
 void dlm_lowcomms_stop(void)
 {
-	atomic_set(&accepting, 0);
+	int i;
+
 	sctp_con.flags = 0x7;
 	daemons_stop();
 	clean_writequeues();
 	close_connection();
 	dealloc_nodeinfo();
 	max_nodeid = 0;
-}
 
-int dlm_lowcomms_init(void)
-{
-	init_waitqueue_head(&lowcomms_recv_wait);
-	spin_lock_init(&write_nodes_lock);
-	INIT_LIST_HEAD(&write_nodes);
-	init_rwsem(&nodeinfo_lock);
-	return 0;
-}
-
-void dlm_lowcomms_exit(void)
-{
-	int i;
+	dlm_local_count = 0;
+	dlm_local_nodeid = 0;
 
 	for (i = 0; i < dlm_local_count; i++)
 		kfree(dlm_local_addr[i]);
-	dlm_local_count = 0;
-	dlm_local_nodeid = 0;
 }
 
diff --git a/fs/dlm/lowcomms-tcp.c b/fs/dlm/lowcomms-tcp.c
new file mode 100644
index 000000000000..9be3a440c42a
--- /dev/null
+++ b/fs/dlm/lowcomms-tcp.c
@@ -0,0 +1,1189 @@
+/******************************************************************************
+*******************************************************************************
+**
+**  Copyright (C) Sistina Software, Inc.  1997-2003  All rights reserved.
+**  Copyright (C) 2004-2006 Red Hat, Inc.  All rights reserved.
+**
+**  This copyrighted material is made available to anyone wishing to use,
+**  modify, copy, or redistribute it subject to the terms and conditions
+**  of the GNU General Public License v.2.
+**
+*******************************************************************************
+******************************************************************************/
+
+/*
+ * lowcomms.c
+ *
+ * This is the "low-level" comms layer.
+ *
+ * It is responsible for sending/receiving messages
+ * from other nodes in the cluster.
+ *
+ * Cluster nodes are referred to by their nodeids. nodeids are
+ * simply 32 bit numbers to the locking module - if they need to
+ * be expanded for the cluster infrastructure then that is it's
+ * responsibility. It is this layer's
+ * responsibility to resolve these into IP address or
+ * whatever it needs for inter-node communication.
+ *
+ * The comms level is two kernel threads that deal mainly with
+ * the receiving of messages from other nodes and passing them
+ * up to the mid-level comms layer (which understands the
+ * message format) for execution by the locking core, and
+ * a send thread which does all the setting up of connections
+ * to remote nodes and the sending of data. Threads are not allowed
+ * to send their own data because it may cause them to wait in times
+ * of high load. Also, this way, the sending thread can collect together
+ * messages bound for one node and send them in one block.
+ *
+ * I don't see any problem with the recv thread executing the locking
+ * code on behalf of remote processes as the locking code is
+ * short, efficient and never waits.
+ *
+ */
+
+
+#include <asm/ioctls.h>
+#include <net/sock.h>
+#include <net/tcp.h>
+#include <linux/pagemap.h>
+
+#include "dlm_internal.h"
+#include "lowcomms.h"
+#include "midcomms.h"
+#include "config.h"
+
+struct cbuf {
+	unsigned int base;
+	unsigned int len;
+	unsigned int mask;
+};
+
+#define NODE_INCREMENT 32
+static void cbuf_add(struct cbuf *cb, int n)
+{
+	cb->len += n;
+}
+
+static int cbuf_data(struct cbuf *cb)
+{
+	return ((cb->base + cb->len) & cb->mask);
+}
+
+static void cbuf_init(struct cbuf *cb, int size)
+{
+	cb->base = cb->len = 0;
+	cb->mask = size-1;
+}
+
+static void cbuf_eat(struct cbuf *cb, int n)
+{
+	cb->len  -= n;
+	cb->base += n;
+	cb->base &= cb->mask;
+}
+
+static bool cbuf_empty(struct cbuf *cb)
+{
+	return cb->len == 0;
+}
+
+/* Maximum number of incoming messages to process before
+   doing a cond_resched()
+*/
+#define MAX_RX_MSG_COUNT 25
+
+struct connection {
+	struct socket *sock;	/* NULL if not connected */
+	uint32_t nodeid;	/* So we know who we are in the list */
+	struct rw_semaphore sock_sem; /* Stop connect races */
+	struct list_head read_list;   /* On this list when ready for reading */
+	struct list_head write_list;  /* On this list when ready for writing */
+	struct list_head state_list;  /* On this list when ready to connect */
+	unsigned long flags;	/* bit 1,2 = We are on the read/write lists */
+#define CF_READ_PENDING 1
+#define CF_WRITE_PENDING 2
+#define CF_CONNECT_PENDING 3
+#define CF_IS_OTHERCON 4
+	struct list_head writequeue;  /* List of outgoing writequeue_entries */
+	struct list_head listenlist;  /* List of allocated listening sockets */
+	spinlock_t writequeue_lock;
+	int (*rx_action) (struct connection *);	/* What to do when active */
+	struct page *rx_page;
+	struct cbuf cb;
+	int retries;
+	atomic_t waiting_requests;
+#define MAX_CONNECT_RETRIES 3
+	struct connection *othercon;
+};
+#define sock2con(x) ((struct connection *)(x)->sk_user_data)
+
+/* An entry waiting to be sent */
+struct writequeue_entry {
+	struct list_head list;
+	struct page *page;
+	int offset;
+	int len;
+	int end;
+	int users;
+	struct connection *con;
+};
+
+static struct sockaddr_storage dlm_local_addr;
+
+/* Manage daemons */
+static struct task_struct *recv_task;
+static struct task_struct *send_task;
+
+static wait_queue_t lowcomms_send_waitq_head;
+static DECLARE_WAIT_QUEUE_HEAD(lowcomms_send_waitq);
+static wait_queue_t lowcomms_recv_waitq_head;
+static DECLARE_WAIT_QUEUE_HEAD(lowcomms_recv_waitq);
+
+/* An array of pointers to connections, indexed by NODEID */
+static struct connection **connections;
+static DECLARE_MUTEX(connections_lock);
+static struct kmem_cache *con_cache;
+static int conn_array_size;
+
+/* List of sockets that have reads pending */
+static LIST_HEAD(read_sockets);
+static DEFINE_SPINLOCK(read_sockets_lock);
+
+/* List of sockets which have writes pending */
+static LIST_HEAD(write_sockets);
+static DEFINE_SPINLOCK(write_sockets_lock);
+
+/* List of sockets which have connects pending */
+static LIST_HEAD(state_sockets);
+static DEFINE_SPINLOCK(state_sockets_lock);
+
+static struct connection *nodeid2con(int nodeid, gfp_t allocation)
+{
+	struct connection *con = NULL;
+
+	down(&connections_lock);
+	if (nodeid >= conn_array_size) {
+		int new_size = nodeid + NODE_INCREMENT;
+		struct connection **new_conns;
+
+		new_conns = kzalloc(sizeof(struct connection *) *
+				    new_size, allocation);
+		if (!new_conns)
+			goto finish;
+
+		memcpy(new_conns, connections,  sizeof(struct connection *) * conn_array_size);
+		conn_array_size = new_size;
+		kfree(connections);
+		connections = new_conns;
+
+	}
+
+	con = connections[nodeid];
+	if (con == NULL && allocation) {
+		con = kmem_cache_zalloc(con_cache, allocation);
+		if (!con)
+			goto finish;
+
+		con->nodeid = nodeid;
+		init_rwsem(&con->sock_sem);
+		INIT_LIST_HEAD(&con->writequeue);
+		spin_lock_init(&con->writequeue_lock);
+
+		connections[nodeid] = con;
+	}
+
+finish:
+	up(&connections_lock);
+	return con;
+}
+
+/* Data available on socket or listen socket received a connect */
+static void lowcomms_data_ready(struct sock *sk, int count_unused)
+{
+	struct connection *con = sock2con(sk);
+
+	atomic_inc(&con->waiting_requests);
+	if (test_and_set_bit(CF_READ_PENDING, &con->flags))
+		return;
+
+	spin_lock_bh(&read_sockets_lock);
+	list_add_tail(&con->read_list, &read_sockets);
+	spin_unlock_bh(&read_sockets_lock);
+
+	wake_up_interruptible(&lowcomms_recv_waitq);
+}
+
+static void lowcomms_write_space(struct sock *sk)
+{
+	struct connection *con = sock2con(sk);
+
+	if (test_and_set_bit(CF_WRITE_PENDING, &con->flags))
+		return;
+
+	spin_lock_bh(&write_sockets_lock);
+	list_add_tail(&con->write_list, &write_sockets);
+	spin_unlock_bh(&write_sockets_lock);
+
+	wake_up_interruptible(&lowcomms_send_waitq);
+}
+
+static inline void lowcomms_connect_sock(struct connection *con)
+{
+	if (test_and_set_bit(CF_CONNECT_PENDING, &con->flags))
+		return;
+
+	spin_lock_bh(&state_sockets_lock);
+	list_add_tail(&con->state_list, &state_sockets);
+	spin_unlock_bh(&state_sockets_lock);
+
+	wake_up_interruptible(&lowcomms_send_waitq);
+}
+
+static void lowcomms_state_change(struct sock *sk)
+{
+	if (sk->sk_state == TCP_ESTABLISHED)
+		lowcomms_write_space(sk);
+}
+
+/* Make a socket active */
+static int add_sock(struct socket *sock, struct connection *con)
+{
+	con->sock = sock;
+
+	/* Install a data_ready callback */
+	con->sock->sk->sk_data_ready = lowcomms_data_ready;
+	con->sock->sk->sk_write_space = lowcomms_write_space;
+	con->sock->sk->sk_state_change = lowcomms_state_change;
+
+	return 0;
+}
+
+/* Add the port number to an IP6 or 4 sockaddr and return the address
+   length */
+static void make_sockaddr(struct sockaddr_storage *saddr, uint16_t port,
+			  int *addr_len)
+{
+	saddr->ss_family =  dlm_local_addr.ss_family;
+	if (saddr->ss_family == AF_INET) {
+		struct sockaddr_in *in4_addr = (struct sockaddr_in *)saddr;
+		in4_addr->sin_port = cpu_to_be16(port);
+		*addr_len = sizeof(struct sockaddr_in);
+	} else {
+		struct sockaddr_in6 *in6_addr = (struct sockaddr_in6 *)saddr;
+		in6_addr->sin6_port = cpu_to_be16(port);
+		*addr_len = sizeof(struct sockaddr_in6);
+	}
+}
+
+/* Close a remote connection and tidy up */
+static void close_connection(struct connection *con, bool and_other)
+{
+	down_write(&con->sock_sem);
+
+	if (con->sock) {
+		sock_release(con->sock);
+		con->sock = NULL;
+	}
+	if (con->othercon && and_other) {
+		/* Will only re-enter once. */
+		close_connection(con->othercon, false);
+	}
+	if (con->rx_page) {
+		__free_page(con->rx_page);
+		con->rx_page = NULL;
+	}
+	con->retries = 0;
+	up_write(&con->sock_sem);
+}
+
+/* Data received from remote end */
+static int receive_from_sock(struct connection *con)
+{
+	int ret = 0;
+	struct msghdr msg;
+	struct iovec iov[2];
+	mm_segment_t fs;
+	unsigned len;
+	int r;
+	int call_again_soon = 0;
+
+	down_read(&con->sock_sem);
+
+	if (con->sock == NULL)
+		goto out;
+	if (con->rx_page == NULL) {
+		/*
+		 * This doesn't need to be atomic, but I think it should
+		 * improve performance if it is.
+		 */
+		con->rx_page = alloc_page(GFP_ATOMIC);
+		if (con->rx_page == NULL)
+			goto out_resched;
+		cbuf_init(&con->cb, PAGE_CACHE_SIZE);
+	}
+
+	msg.msg_control = NULL;
+	msg.msg_controllen = 0;
+	msg.msg_iovlen = 1;
+	msg.msg_iov = iov;
+	msg.msg_name = NULL;
+	msg.msg_namelen = 0;
+	msg.msg_flags = 0;
+
+	/*
+	 * iov[0] is the bit of the circular buffer between the current end
+	 * point (cb.base + cb.len) and the end of the buffer.
+	 */
+	iov[0].iov_len = con->cb.base - cbuf_data(&con->cb);
+	iov[0].iov_base = page_address(con->rx_page) + cbuf_data(&con->cb);
+	iov[1].iov_len = 0;
+
+	/*
+	 * iov[1] is the bit of the circular buffer between the start of the
+	 * buffer and the start of the currently used section (cb.base)
+	 */
+	if (cbuf_data(&con->cb) >= con->cb.base) {
+		iov[0].iov_len = PAGE_CACHE_SIZE - cbuf_data(&con->cb);
+		iov[1].iov_len = con->cb.base;
+		iov[1].iov_base = page_address(con->rx_page);
+		msg.msg_iovlen = 2;
+	}
+	len = iov[0].iov_len + iov[1].iov_len;
+
+	fs = get_fs();
+	set_fs(get_ds());
+	r = ret = sock_recvmsg(con->sock, &msg, len,
+			       MSG_DONTWAIT | MSG_NOSIGNAL);
+	set_fs(fs);
+
+	if (ret <= 0)
+		goto out_close;
+	if (ret == len)
+		call_again_soon = 1;
+	cbuf_add(&con->cb, ret);
+	ret = dlm_process_incoming_buffer(con->nodeid,
+					  page_address(con->rx_page),
+					  con->cb.base, con->cb.len,
+					  PAGE_CACHE_SIZE);
+	if (ret == -EBADMSG) {
+		printk(KERN_INFO "dlm: lowcomms: addr=%p, base=%u, len=%u, "
+		       "iov_len=%u, iov_base[0]=%p, read=%d\n",
+		       page_address(con->rx_page), con->cb.base, con->cb.len,
+		       len, iov[0].iov_base, r);
+	}
+	if (ret < 0)
+		goto out_close;
+	cbuf_eat(&con->cb, ret);
+
+	if (cbuf_empty(&con->cb) && !call_again_soon) {
+		__free_page(con->rx_page);
+		con->rx_page = NULL;
+	}
+
+out:
+	if (call_again_soon)
+		goto out_resched;
+	up_read(&con->sock_sem);
+	return 0;
+
+out_resched:
+	lowcomms_data_ready(con->sock->sk, 0);
+	up_read(&con->sock_sem);
+	cond_resched();
+	return 0;
+
+out_close:
+	up_read(&con->sock_sem);
+	if (ret != -EAGAIN && !test_bit(CF_IS_OTHERCON, &con->flags)) {
+		close_connection(con, false);
+		/* Reconnect when there is something to send */
+	}
+
+	return ret;
+}
+
+/* Listening socket is busy, accept a connection */
+static int accept_from_sock(struct connection *con)
+{
+	int result;
+	struct sockaddr_storage peeraddr;
+	struct socket *newsock;
+	int len;
+	int nodeid;
+	struct connection *newcon;
+
+	memset(&peeraddr, 0, sizeof(peeraddr));
+	result = sock_create_kern(dlm_local_addr.ss_family, SOCK_STREAM,
+				  IPPROTO_TCP, &newsock);
+	if (result < 0)
+		return -ENOMEM;
+
+	down_read(&con->sock_sem);
+
+	result = -ENOTCONN;
+	if (con->sock == NULL)
+		goto accept_err;
+
+	newsock->type = con->sock->type;
+	newsock->ops = con->sock->ops;
+
+	result = con->sock->ops->accept(con->sock, newsock, O_NONBLOCK);
+	if (result < 0)
+		goto accept_err;
+
+	/* Get the connected socket's peer */
+	memset(&peeraddr, 0, sizeof(peeraddr));
+	if (newsock->ops->getname(newsock, (struct sockaddr *)&peeraddr,
+				  &len, 2)) {
+		result = -ECONNABORTED;
+		goto accept_err;
+	}
+
+	/* Get the new node's NODEID */
+	make_sockaddr(&peeraddr, 0, &len);
+	if (dlm_addr_to_nodeid(&peeraddr, &nodeid)) {
+		printk("dlm: connect from non cluster node\n");
+		sock_release(newsock);
+		up_read(&con->sock_sem);
+		return -1;
+	}
+
+	log_print("got connection from %d", nodeid);
+
+	/*  Check to see if we already have a connection to this node. This
+	 *  could happen if the two nodes initiate a connection at roughly
+	 *  the same time and the connections cross on the wire.
+	 * TEMPORARY FIX:
+	 *  In this case we store the incoming one in "othercon"
+	 */
+	newcon = nodeid2con(nodeid, GFP_KERNEL);
+	if (!newcon) {
+		result = -ENOMEM;
+		goto accept_err;
+	}
+	down_write(&newcon->sock_sem);
+	if (newcon->sock) {
+		struct connection *othercon = newcon->othercon;
+
+		if (!othercon) {
+			othercon = kmem_cache_zalloc(con_cache, GFP_KERNEL);
+			if (!othercon) {
+				printk("dlm: failed to allocate incoming socket\n");
+				up_write(&newcon->sock_sem);
+				result = -ENOMEM;
+				goto accept_err;
+			}
+			othercon->nodeid = nodeid;
+			othercon->rx_action = receive_from_sock;
+			init_rwsem(&othercon->sock_sem);
+			set_bit(CF_IS_OTHERCON, &othercon->flags);
+			newcon->othercon = othercon;
+		}
+		othercon->sock = newsock;
+		newsock->sk->sk_user_data = othercon;
+		add_sock(newsock, othercon);
+	}
+	else {
+		newsock->sk->sk_user_data = newcon;
+		newcon->rx_action = receive_from_sock;
+		add_sock(newsock, newcon);
+
+	}
+
+	up_write(&newcon->sock_sem);
+
+	/*
+	 * Add it to the active queue in case we got data
+	 * beween processing the accept adding the socket
+	 * to the read_sockets list
+	 */
+	lowcomms_data_ready(newsock->sk, 0);
+	up_read(&con->sock_sem);
+
+	return 0;
+
+accept_err:
+	up_read(&con->sock_sem);
+	sock_release(newsock);
+
+	if (result != -EAGAIN)
+		printk("dlm: error accepting connection from node: %d\n", result);
+	return result;
+}
+
+/* Connect a new socket to its peer */
+static void connect_to_sock(struct connection *con)
+{
+	int result = -EHOSTUNREACH;
+	struct sockaddr_storage saddr;
+	int addr_len;
+	struct socket *sock;
+
+	if (con->nodeid == 0) {
+		log_print("attempt to connect sock 0 foiled");
+		return;
+	}
+
+	down_write(&con->sock_sem);
+	if (con->retries++ > MAX_CONNECT_RETRIES)
+		goto out;
+
+	/* Some odd races can cause double-connects, ignore them */
+	if (con->sock) {
+		result = 0;
+		goto out;
+	}
+
+	/* Create a socket to communicate with */
+	result = sock_create_kern(dlm_local_addr.ss_family, SOCK_STREAM,
+				  IPPROTO_TCP, &sock);
+	if (result < 0)
+		goto out_err;
+
+	memset(&saddr, 0, sizeof(saddr));
+	if (dlm_nodeid_to_addr(con->nodeid, &saddr))
+		goto out_err;
+
+	sock->sk->sk_user_data = con;
+	con->rx_action = receive_from_sock;
+
+	make_sockaddr(&saddr, dlm_config.tcp_port, &addr_len);
+
+	add_sock(sock, con);
+
+	log_print("connecting to %d", con->nodeid);
+	result =
+		sock->ops->connect(sock, (struct sockaddr *)&saddr, addr_len,
+				   O_NONBLOCK);
+	if (result == -EINPROGRESS)
+		result = 0;
+	if (result == 0)
+		goto out;
+
+out_err:
+	if (con->sock) {
+		sock_release(con->sock);
+		con->sock = NULL;
+	}
+	/*
+	 * Some errors are fatal and this list might need adjusting. For other
+	 * errors we try again until the max number of retries is reached.
+	 */
+	if (result != -EHOSTUNREACH && result != -ENETUNREACH &&
+	    result != -ENETDOWN && result != EINVAL
+	    && result != -EPROTONOSUPPORT) {
+		lowcomms_connect_sock(con);
+		result = 0;
+	}
+out:
+	up_write(&con->sock_sem);
+	return;
+}
+
+static struct socket *create_listen_sock(struct connection *con,
+					 struct sockaddr_storage *saddr)
+{
+	struct socket *sock = NULL;
+	mm_segment_t fs;
+	int result = 0;
+	int one = 1;
+	int addr_len;
+
+	if (dlm_local_addr.ss_family == AF_INET)
+		addr_len = sizeof(struct sockaddr_in);
+	else
+		addr_len = sizeof(struct sockaddr_in6);
+
+	/* Create a socket to communicate with */
+	result = sock_create_kern(dlm_local_addr.ss_family, SOCK_STREAM, IPPROTO_TCP, &sock);
+	if (result < 0) {
+		printk("dlm: Can't create listening comms socket\n");
+		goto create_out;
+	}
+
+	fs = get_fs();
+	set_fs(get_ds());
+	result = sock_setsockopt(sock, SOL_SOCKET, SO_REUSEADDR,
+				 (char *)&one, sizeof(one));
+	set_fs(fs);
+	if (result < 0) {
+		printk("dlm: Failed to set SO_REUSEADDR on socket: result=%d\n",
+		       result);
+	}
+	sock->sk->sk_user_data = con;
+	con->rx_action = accept_from_sock;
+	con->sock = sock;
+
+	/* Bind to our port */
+	make_sockaddr(saddr, dlm_config.tcp_port, &addr_len);
+	result = sock->ops->bind(sock, (struct sockaddr *) saddr, addr_len);
+	if (result < 0) {
+		printk("dlm: Can't bind to port %d\n", dlm_config.tcp_port);
+		sock_release(sock);
+		sock = NULL;
+		con->sock = NULL;
+		goto create_out;
+	}
+
+	fs = get_fs();
+	set_fs(get_ds());
+
+	result = sock_setsockopt(sock, SOL_SOCKET, SO_KEEPALIVE,
+				 (char *)&one, sizeof(one));
+	set_fs(fs);
+	if (result < 0) {
+		printk("dlm: Set keepalive failed: %d\n", result);
+	}
+
+	result = sock->ops->listen(sock, 5);
+	if (result < 0) {
+		printk("dlm: Can't listen on port %d\n", dlm_config.tcp_port);
+		sock_release(sock);
+		sock = NULL;
+		goto create_out;
+	}
+
+create_out:
+	return sock;
+}
+
+
+/* Listen on all interfaces */
+static int listen_for_all(void)
+{
+	struct socket *sock = NULL;
+	struct connection *con = nodeid2con(0, GFP_KERNEL);
+	int result = -EINVAL;
+
+	/* We don't support multi-homed hosts */
+	set_bit(CF_IS_OTHERCON, &con->flags);
+
+	sock = create_listen_sock(con, &dlm_local_addr);
+	if (sock) {
+		add_sock(sock, con);
+		result = 0;
+	}
+	else {
+		result = -EADDRINUSE;
+	}
+
+	return result;
+}
+
+
+
+static struct writequeue_entry *new_writequeue_entry(struct connection *con,
+						     gfp_t allocation)
+{
+	struct writequeue_entry *entry;
+
+	entry = kmalloc(sizeof(struct writequeue_entry), allocation);
+	if (!entry)
+		return NULL;
+
+	entry->page = alloc_page(allocation);
+	if (!entry->page) {
+		kfree(entry);
+		return NULL;
+	}
+
+	entry->offset = 0;
+	entry->len = 0;
+	entry->end = 0;
+	entry->users = 0;
+	entry->con = con;
+
+	return entry;
+}
+
+void *dlm_lowcomms_get_buffer(int nodeid, int len,
+			      gfp_t allocation, char **ppc)
+{
+	struct connection *con;
+	struct writequeue_entry *e;
+	int offset = 0;
+	int users = 0;
+
+	con = nodeid2con(nodeid, allocation);
+	if (!con)
+		return NULL;
+
+	e = list_entry(con->writequeue.prev, struct writequeue_entry, list);
+	if ((&e->list == &con->writequeue) ||
+	    (PAGE_CACHE_SIZE - e->end < len)) {
+		e = NULL;
+	} else {
+		offset = e->end;
+		e->end += len;
+		users = e->users++;
+	}
+	spin_unlock(&con->writequeue_lock);
+
+	if (e) {
+	got_one:
+		if (users == 0)
+			kmap(e->page);
+		*ppc = page_address(e->page) + offset;
+		return e;
+	}
+
+	e = new_writequeue_entry(con, allocation);
+	if (e) {
+		spin_lock(&con->writequeue_lock);
+		offset = e->end;
+		e->end += len;
+		users = e->users++;
+		list_add_tail(&e->list, &con->writequeue);
+		spin_unlock(&con->writequeue_lock);
+		goto got_one;
+	}
+	return NULL;
+}
+
+void dlm_lowcomms_commit_buffer(void *mh)
+{
+	struct writequeue_entry *e = (struct writequeue_entry *)mh;
+	struct connection *con = e->con;
+	int users;
+
+	users = --e->users;
+	if (users)
+		goto out;
+	e->len = e->end - e->offset;
+	kunmap(e->page);
+	spin_unlock(&con->writequeue_lock);
+
+	if (test_and_set_bit(CF_WRITE_PENDING, &con->flags) == 0) {
+		spin_lock_bh(&write_sockets_lock);
+		list_add_tail(&con->write_list, &write_sockets);
+		spin_unlock_bh(&write_sockets_lock);
+
+		wake_up_interruptible(&lowcomms_send_waitq);
+	}
+	return;
+
+out:
+	spin_unlock(&con->writequeue_lock);
+	return;
+}
+
+static void free_entry(struct writequeue_entry *e)
+{
+	__free_page(e->page);
+	kfree(e);
+}
+
+/* Send a message */
+static void send_to_sock(struct connection *con)
+{
+	int ret = 0;
+	ssize_t(*sendpage) (struct socket *, struct page *, int, size_t, int);
+	const int msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL;
+	struct writequeue_entry *e;
+	int len, offset;
+
+	down_read(&con->sock_sem);
+	if (con->sock == NULL)
+		goto out_connect;
+
+	sendpage = con->sock->ops->sendpage;
+
+	spin_lock(&con->writequeue_lock);
+	for (;;) {
+		e = list_entry(con->writequeue.next, struct writequeue_entry,
+			       list);
+		if ((struct list_head *) e == &con->writequeue)
+			break;
+
+		len = e->len;
+		offset = e->offset;
+		BUG_ON(len == 0 && e->users == 0);
+		spin_unlock(&con->writequeue_lock);
+
+		ret = 0;
+		if (len) {
+			ret = sendpage(con->sock, e->page, offset, len,
+				       msg_flags);
+			if (ret == -EAGAIN || ret == 0)
+				goto out;
+			if (ret <= 0)
+				goto send_error;
+		}
+		else {
+			/* Don't starve people filling buffers */
+			cond_resched();
+		}
+
+		spin_lock(&con->writequeue_lock);
+		e->offset += ret;
+		e->len -= ret;
+
+		if (e->len == 0 && e->users == 0) {
+			list_del(&e->list);
+			kunmap(e->page);
+			free_entry(e);
+			continue;
+		}
+	}
+	spin_unlock(&con->writequeue_lock);
+out:
+	up_read(&con->sock_sem);
+	return;
+
+send_error:
+	up_read(&con->sock_sem);
+	close_connection(con, false);
+	lowcomms_connect_sock(con);
+	return;
+
+out_connect:
+	up_read(&con->sock_sem);
+	lowcomms_connect_sock(con);
+	return;
+}
+
+static void clean_one_writequeue(struct connection *con)
+{
+	struct list_head *list;
+	struct list_head *temp;
+
+	spin_lock(&con->writequeue_lock);
+	list_for_each_safe(list, temp, &con->writequeue) {
+		struct writequeue_entry *e =
+			list_entry(list, struct writequeue_entry, list);
+		list_del(&e->list);
+		free_entry(e);
+	}
+	spin_unlock(&con->writequeue_lock);
+}
+
+/* Called from recovery when it knows that a node has
+   left the cluster */
+int dlm_lowcomms_close(int nodeid)
+{
+	struct connection *con;
+
+	if (!connections)
+		goto out;
+
+	log_print("closing connection to node %d", nodeid);
+	con = nodeid2con(nodeid, 0);
+	if (con) {
+		clean_one_writequeue(con);
+		close_connection(con, true);
+		atomic_set(&con->waiting_requests, 0);
+	}
+	return 0;
+
+out:
+	return -1;
+}
+
+/* API send message call, may queue the request */
+/* N.B. This is the old interface - use the new one for new calls */
+int lowcomms_send_message(int nodeid, char *buf, int len, gfp_t allocation)
+{
+	struct writequeue_entry *e;
+	char *b;
+
+	e = dlm_lowcomms_get_buffer(nodeid, len, allocation, &b);
+	if (e) {
+		memcpy(b, buf, len);
+		dlm_lowcomms_commit_buffer(e);
+		return 0;
+	}
+	return -ENOBUFS;
+}
+
+/* Look for activity on active sockets */
+static void process_sockets(void)
+{
+	struct list_head *list;
+	struct list_head *temp;
+	int count = 0;
+
+	spin_lock_bh(&read_sockets_lock);
+	list_for_each_safe(list, temp, &read_sockets) {
+
+		struct connection *con =
+			list_entry(list, struct connection, read_list);
+		list_del(&con->read_list);
+		clear_bit(CF_READ_PENDING, &con->flags);
+
+		spin_unlock_bh(&read_sockets_lock);
+
+		/* This can reach zero if we are processing requests
+		 * as they come in.
+		 */
+		if (atomic_read(&con->waiting_requests) == 0) {
+			spin_lock_bh(&read_sockets_lock);
+			continue;
+		}
+
+		do {
+			con->rx_action(con);
+
+			/* Don't starve out everyone else */
+			if (++count >= MAX_RX_MSG_COUNT) {
+				cond_resched();
+				count = 0;
+			}
+
+		} while (!atomic_dec_and_test(&con->waiting_requests) &&
+			 !kthread_should_stop());
+
+		spin_lock_bh(&read_sockets_lock);
+	}
+	spin_unlock_bh(&read_sockets_lock);
+}
+
+/* Try to send any messages that are pending
+ */
+static void process_output_queue(void)
+{
+	struct list_head *list;
+	struct list_head *temp;
+
+	spin_lock_bh(&write_sockets_lock);
+	list_for_each_safe(list, temp, &write_sockets) {
+		struct connection *con =
+			list_entry(list, struct connection, write_list);
+		clear_bit(CF_WRITE_PENDING, &con->flags);
+		list_del(&con->write_list);
+
+		spin_unlock_bh(&write_sockets_lock);
+		send_to_sock(con);
+		spin_lock_bh(&write_sockets_lock);
+	}
+	spin_unlock_bh(&write_sockets_lock);
+}
+
+static void process_state_queue(void)
+{
+	struct list_head *list;
+	struct list_head *temp;
+
+	spin_lock_bh(&state_sockets_lock);
+	list_for_each_safe(list, temp, &state_sockets) {
+		struct connection *con =
+			list_entry(list, struct connection, state_list);
+		list_del(&con->state_list);
+		clear_bit(CF_CONNECT_PENDING, &con->flags);
+		spin_unlock_bh(&state_sockets_lock);
+
+		connect_to_sock(con);
+		spin_lock_bh(&state_sockets_lock);
+	}
+	spin_unlock_bh(&state_sockets_lock);
+}
+
+
+/* Discard all entries on the write queues */
+static void clean_writequeues(void)
+{
+	int nodeid;
+
+	for (nodeid = 1; nodeid < conn_array_size; nodeid++) {
+		struct connection *con = nodeid2con(nodeid, 0);
+
+		if (con)
+			clean_one_writequeue(con);
+	}
+}
+
+static int read_list_empty(void)
+{
+	int status;
+
+	spin_lock_bh(&read_sockets_lock);
+	status = list_empty(&read_sockets);
+	spin_unlock_bh(&read_sockets_lock);
+
+	return status;
+}
+
+/* DLM Transport comms receive daemon */
+static int dlm_recvd(void *data)
+{
+	init_waitqueue_entry(&lowcomms_recv_waitq_head, current);
+	add_wait_queue(&lowcomms_recv_waitq, &lowcomms_recv_waitq_head);
+
+	while (!kthread_should_stop()) {
+		set_current_state(TASK_INTERRUPTIBLE);
+		if (read_list_empty())
+			cond_resched();
+		set_current_state(TASK_RUNNING);
+
+		process_sockets();
+	}
+
+	return 0;
+}
+
+static int write_and_state_lists_empty(void)
+{
+	int status;
+
+	spin_lock_bh(&write_sockets_lock);
+	status = list_empty(&write_sockets);
+	spin_unlock_bh(&write_sockets_lock);
+
+	spin_lock_bh(&state_sockets_lock);
+	if (list_empty(&state_sockets) == 0)
+		status = 0;
+	spin_unlock_bh(&state_sockets_lock);
+
+	return status;
+}
+
+/* DLM Transport send daemon */
+static int dlm_sendd(void *data)
+{
+	init_waitqueue_entry(&lowcomms_send_waitq_head, current);
+	add_wait_queue(&lowcomms_send_waitq, &lowcomms_send_waitq_head);
+
+	while (!kthread_should_stop()) {
+		set_current_state(TASK_INTERRUPTIBLE);
+		if (write_and_state_lists_empty())
+			cond_resched();
+		set_current_state(TASK_RUNNING);
+
+		process_state_queue();
+		process_output_queue();
+	}
+
+	return 0;
+}
+
+static void daemons_stop(void)
+{
+	kthread_stop(recv_task);
+	kthread_stop(send_task);
+}
+
+static int daemons_start(void)
+{
+	struct task_struct *p;
+	int error;
+
+	p = kthread_run(dlm_recvd, NULL, "dlm_recvd");
+	error = IS_ERR(p);
+	if (error) {
+		log_print("can't start dlm_recvd %d", error);
+		return error;
+	}
+	recv_task = p;
+
+	p = kthread_run(dlm_sendd, NULL, "dlm_sendd");
+	error = IS_ERR(p);
+	if (error) {
+		log_print("can't start dlm_sendd %d", error);
+		kthread_stop(recv_task);
+		return error;
+	}
+	send_task = p;
+
+	return 0;
+}
+
+/*
+ * Return the largest buffer size we can cope with.
+ */
+int lowcomms_max_buffer_size(void)
+{
+	return PAGE_CACHE_SIZE;
+}
+
+void dlm_lowcomms_stop(void)
+{
+	int i;
+
+	/* Set all the flags to prevent any
+	   socket activity.
+	*/
+	for (i = 0; i < conn_array_size; i++) {
+		if (connections[i])
+			connections[i]->flags |= 0xFF;
+	}
+
+	daemons_stop();
+	clean_writequeues();
+
+	for (i = 0; i < conn_array_size; i++) {
+		if (connections[i]) {
+			close_connection(connections[i], true);
+			if (connections[i]->othercon)
+				kmem_cache_free(con_cache, connections[i]->othercon);
+			kmem_cache_free(con_cache, connections[i]);
+		}
+	}
+
+	kfree(connections);
+	connections = NULL;
+
+	kmem_cache_destroy(con_cache);
+}
+
+/* This is quite likely to sleep... */
+int dlm_lowcomms_start(void)
+{
+	int error = 0;
+
+	error = -ENOMEM;
+	connections = kzalloc(sizeof(struct connection *) *
+			      NODE_INCREMENT, GFP_KERNEL);
+	if (!connections)
+		goto out;
+
+	conn_array_size = NODE_INCREMENT;
+
+	if (dlm_our_addr(&dlm_local_addr, 0)) {
+		log_print("no local IP address has been set");
+		goto fail_free_conn;
+	}
+	if (!dlm_our_addr(&dlm_local_addr, 1)) {
+		log_print("This dlm comms module does not support multi-homed clustering");
+		goto fail_free_conn;
+	}
+
+	con_cache = kmem_cache_create("dlm_conn", sizeof(struct connection),
+				      __alignof__(struct connection), 0,
+				      NULL, NULL);
+	if (!con_cache)
+		goto fail_free_conn;
+
+
+	/* Start listening */
+	error = listen_for_all();
+	if (error)
+		goto fail_unlisten;
+
+	error = daemons_start();
+	if (error)
+		goto fail_unlisten;
+
+	return 0;
+
+fail_unlisten:
+	close_connection(connections[0], false);
+	kmem_cache_free(con_cache, connections[0]);
+	kmem_cache_destroy(con_cache);
+
+fail_free_conn:
+	kfree(connections);
+
+out:
+	return error;
+}
+
+/*
+ * Overrides for Emacs so that we follow Linus's tabbing style.
+ * Emacs will notice this stuff at the end of the file and automatically
+ * adjust the settings for this buffer only.  This must remain at the end
+ * of the file.
+ * ---------------------------------------------------------------------------
+ * Local variables:
+ * c-file-style: "linux"
+ * End:
+ */
diff --git a/fs/dlm/lowcomms.h b/fs/dlm/lowcomms.h
index 2d045e0daae1..a9a9618c0d3f 100644
--- a/fs/dlm/lowcomms.h
+++ b/fs/dlm/lowcomms.h
@@ -14,8 +14,6 @@
 #ifndef __LOWCOMMS_DOT_H__
 #define __LOWCOMMS_DOT_H__
 
-int dlm_lowcomms_init(void);
-void dlm_lowcomms_exit(void);
 int dlm_lowcomms_start(void);
 void dlm_lowcomms_stop(void);
 int dlm_lowcomms_close(int nodeid);
diff --git a/fs/dlm/main.c b/fs/dlm/main.c
index a8da8dc36b2e..162fbae58fe5 100644
--- a/fs/dlm/main.c
+++ b/fs/dlm/main.c
@@ -16,7 +16,6 @@
 #include "lock.h"
 #include "user.h"
 #include "memory.h"
-#include "lowcomms.h"
 #include "config.h"
 
 #ifdef CONFIG_DLM_DEBUG
@@ -47,20 +46,14 @@ static int __init init_dlm(void)
 	if (error)
 		goto out_config;
 
-	error = dlm_lowcomms_init();
-	if (error)
-		goto out_debug;
-
 	error = dlm_user_init();
 	if (error)
-		goto out_lowcomms;
+		goto out_debug;
 
 	printk("DLM (built %s %s) installed\n", __DATE__, __TIME__);
 
 	return 0;
 
- out_lowcomms:
-	dlm_lowcomms_exit();
  out_debug:
 	dlm_unregister_debugfs();
  out_config:
@@ -76,7 +69,6 @@ static int __init init_dlm(void)
 static void __exit exit_dlm(void)
 {
 	dlm_user_exit();
-	dlm_lowcomms_exit();
 	dlm_config_exit();
 	dlm_memory_exit();
 	dlm_lockspace_exit();
diff --git a/fs/dlm/member.c b/fs/dlm/member.c
index a3f7de7f3a8f..85e2897bd740 100644
--- a/fs/dlm/member.c
+++ b/fs/dlm/member.c
@@ -186,6 +186,14 @@ int dlm_recover_members(struct dlm_ls *ls, struct dlm_recover *rv, int *neg_out)
 	struct dlm_member *memb, *safe;
 	int i, error, found, pos = 0, neg = 0, low = -1;
 
+	/* previously removed members that we've not finished removing need to
+	   count as a negative change so the "neg" recovery steps will happen */
+
+	list_for_each_entry(memb, &ls->ls_nodes_gone, list) {
+		log_debug(ls, "prev removed member %d", memb->nodeid);
+		neg++;
+	}
+
 	/* move departed members from ls_nodes to ls_nodes_gone */
 
 	list_for_each_entry_safe(memb, safe, &ls->ls_nodes, list) {
diff --git a/fs/dlm/memory.c b/fs/dlm/memory.c
index 989b608fd836..5352b03ff5aa 100644
--- a/fs/dlm/memory.c
+++ b/fs/dlm/memory.c
@@ -15,7 +15,7 @@
 #include "config.h"
 #include "memory.h"
 
-static kmem_cache_t *lkb_cache;
+static struct kmem_cache *lkb_cache;
 
 
 int dlm_memory_init(void)
diff --git a/fs/dlm/rcom.c b/fs/dlm/rcom.c
index 518239a8b1e9..4cc31be9cd9d 100644
--- a/fs/dlm/rcom.c
+++ b/fs/dlm/rcom.c
@@ -90,13 +90,28 @@ static int check_config(struct dlm_ls *ls, struct rcom_config *rf, int nodeid)
 	return 0;
 }
 
+static void allow_sync_reply(struct dlm_ls *ls, uint64_t *new_seq)
+{
+	spin_lock(&ls->ls_rcom_spin);
+	*new_seq = ++ls->ls_rcom_seq;
+	set_bit(LSFL_RCOM_WAIT, &ls->ls_flags);
+	spin_unlock(&ls->ls_rcom_spin);
+}
+
+static void disallow_sync_reply(struct dlm_ls *ls)
+{
+	spin_lock(&ls->ls_rcom_spin);
+	clear_bit(LSFL_RCOM_WAIT, &ls->ls_flags);
+	clear_bit(LSFL_RCOM_READY, &ls->ls_flags);
+	spin_unlock(&ls->ls_rcom_spin);
+}
+
 int dlm_rcom_status(struct dlm_ls *ls, int nodeid)
 {
 	struct dlm_rcom *rc;
 	struct dlm_mhandle *mh;
 	int error = 0;
 
-	memset(ls->ls_recover_buf, 0, dlm_config.buffer_size);
 	ls->ls_recover_nodeid = nodeid;
 
 	if (nodeid == dlm_our_nodeid()) {
@@ -108,12 +123,14 @@ int dlm_rcom_status(struct dlm_ls *ls, int nodeid)
 	error = create_rcom(ls, nodeid, DLM_RCOM_STATUS, 0, &rc, &mh);
 	if (error)
 		goto out;
-	rc->rc_id = ++ls->ls_rcom_seq;
+
+	allow_sync_reply(ls, &rc->rc_id);
+	memset(ls->ls_recover_buf, 0, dlm_config.buffer_size);
 
 	send_rcom(ls, mh, rc);
 
 	error = dlm_wait_function(ls, &rcom_response);
-	clear_bit(LSFL_RCOM_READY, &ls->ls_flags);
+	disallow_sync_reply(ls);
 	if (error)
 		goto out;
 
@@ -150,14 +167,21 @@ static void receive_rcom_status(struct dlm_ls *ls, struct dlm_rcom *rc_in)
 
 static void receive_sync_reply(struct dlm_ls *ls, struct dlm_rcom *rc_in)
 {
-	if (rc_in->rc_id != ls->ls_rcom_seq) {
-		log_debug(ls, "reject old reply %d got %llx wanted %llx",
-			  rc_in->rc_type, rc_in->rc_id, ls->ls_rcom_seq);
-		return;
+	spin_lock(&ls->ls_rcom_spin);
+	if (!test_bit(LSFL_RCOM_WAIT, &ls->ls_flags) ||
+	    rc_in->rc_id != ls->ls_rcom_seq) {
+		log_debug(ls, "reject reply %d from %d seq %llx expect %llx",
+			  rc_in->rc_type, rc_in->rc_header.h_nodeid,
+			  (unsigned long long)rc_in->rc_id,
+			  (unsigned long long)ls->ls_rcom_seq);
+		goto out;
 	}
 	memcpy(ls->ls_recover_buf, rc_in, rc_in->rc_header.h_length);
 	set_bit(LSFL_RCOM_READY, &ls->ls_flags);
+	clear_bit(LSFL_RCOM_WAIT, &ls->ls_flags);
 	wake_up(&ls->ls_wait_general);
+ out:
+	spin_unlock(&ls->ls_rcom_spin);
 }
 
 static void receive_rcom_status_reply(struct dlm_ls *ls, struct dlm_rcom *rc_in)
@@ -171,7 +195,6 @@ int dlm_rcom_names(struct dlm_ls *ls, int nodeid, char *last_name, int last_len)
 	struct dlm_mhandle *mh;
 	int error = 0, len = sizeof(struct dlm_rcom);
 
-	memset(ls->ls_recover_buf, 0, dlm_config.buffer_size);
 	ls->ls_recover_nodeid = nodeid;
 
 	if (nodeid == dlm_our_nodeid()) {
@@ -185,12 +208,14 @@ int dlm_rcom_names(struct dlm_ls *ls, int nodeid, char *last_name, int last_len)
 	if (error)
 		goto out;
 	memcpy(rc->rc_buf, last_name, last_len);
-	rc->rc_id = ++ls->ls_rcom_seq;
+
+	allow_sync_reply(ls, &rc->rc_id);
+	memset(ls->ls_recover_buf, 0, dlm_config.buffer_size);
 
 	send_rcom(ls, mh, rc);
 
 	error = dlm_wait_function(ls, &rcom_response);
-	clear_bit(LSFL_RCOM_READY, &ls->ls_flags);
+	disallow_sync_reply(ls);
  out:
 	return error;
 }
@@ -370,9 +395,10 @@ static void receive_rcom_lock_reply(struct dlm_ls *ls, struct dlm_rcom *rc_in)
 static int send_ls_not_ready(int nodeid, struct dlm_rcom *rc_in)
 {
 	struct dlm_rcom *rc;
+	struct rcom_config *rf;
 	struct dlm_mhandle *mh;
 	char *mb;
-	int mb_len = sizeof(struct dlm_rcom);
+	int mb_len = sizeof(struct dlm_rcom) + sizeof(struct rcom_config);
 
 	mh = dlm_lowcomms_get_buffer(nodeid, mb_len, GFP_KERNEL, &mb);
 	if (!mh)
@@ -391,6 +417,9 @@ static int send_ls_not_ready(int nodeid, struct dlm_rcom *rc_in)
 	rc->rc_id = rc_in->rc_id;
 	rc->rc_result = -ESRCH;
 
+	rf = (struct rcom_config *) rc->rc_buf;
+	rf->rf_lvblen = -1;
+
 	dlm_rcom_out(rc);
 	dlm_lowcomms_commit_buffer(mh);
 
@@ -412,9 +441,10 @@ void dlm_receive_rcom(struct dlm_header *hd, int nodeid)
 
 	ls = dlm_find_lockspace_global(hd->h_lockspace);
 	if (!ls) {
-		log_print("lockspace %x from %d not found",
-			  hd->h_lockspace, nodeid);
-		send_ls_not_ready(nodeid, rc);
+		log_print("lockspace %x from %d type %x not found",
+			  hd->h_lockspace, nodeid, rc->rc_type);
+		if (rc->rc_type == DLM_RCOM_STATUS)
+			send_ls_not_ready(nodeid, rc);
 		return;
 	}
 
diff --git a/fs/dlm/recover.c b/fs/dlm/recover.c
index a5e6d184872e..cf9f6831bab5 100644
--- a/fs/dlm/recover.c
+++ b/fs/dlm/recover.c
@@ -252,6 +252,7 @@ static void recover_list_clear(struct dlm_ls *ls)
 	spin_lock(&ls->ls_recover_list_lock);
 	list_for_each_entry_safe(r, s, &ls->ls_recover_list, res_recover_list) {
 		list_del_init(&r->res_recover_list);
+		r->res_recover_locks_count = 0;
 		dlm_put_rsb(r);
 		ls->ls_recover_list_count--;
 	}
diff --git a/fs/dlm/recoverd.c b/fs/dlm/recoverd.c
index 362e3eff4dc9..650536aa5139 100644
--- a/fs/dlm/recoverd.c
+++ b/fs/dlm/recoverd.c
@@ -45,7 +45,7 @@ static int ls_recover(struct dlm_ls *ls, struct dlm_recover *rv)
 	unsigned long start;
 	int error, neg = 0;
 
-	log_debug(ls, "recover %llx", rv->seq);
+	log_debug(ls, "recover %llx", (unsigned long long)rv->seq);
 
 	mutex_lock(&ls->ls_recoverd_active);
 
@@ -94,14 +94,6 @@ static int ls_recover(struct dlm_ls *ls, struct dlm_recover *rv)
 	}
 
 	/*
-	 * Purge directory-related requests that are saved in requestqueue.
-	 * All dir requests from before recovery are invalid now due to the dir
-	 * rebuild and will be resent by the requesting nodes.
-	 */
-
-	dlm_purge_requestqueue(ls);
-
-	/*
 	 * Wait for all nodes to complete directory rebuild.
 	 */
 
@@ -164,10 +156,31 @@ static int ls_recover(struct dlm_ls *ls, struct dlm_recover *rv)
 		 */
 
 		dlm_recover_rsbs(ls);
+	} else {
+		/*
+		 * Other lockspace members may be going through the "neg" steps
+		 * while also adding us to the lockspace, in which case they'll
+		 * be doing the recover_locks (RS_LOCKS) barrier.
+		 */
+		dlm_set_recover_status(ls, DLM_RS_LOCKS);
+
+		error = dlm_recover_locks_wait(ls);
+		if (error) {
+			log_error(ls, "recover_locks_wait failed %d", error);
+			goto fail;
+		}
 	}
 
 	dlm_release_root_list(ls);
 
+	/*
+	 * Purge directory-related requests that are saved in requestqueue.
+	 * All dir requests from before recovery are invalid now due to the dir
+	 * rebuild and will be resent by the requesting nodes.
+	 */
+
+	dlm_purge_requestqueue(ls);
+
 	dlm_set_recover_status(ls, DLM_RS_DONE);
 	error = dlm_recover_done_wait(ls);
 	if (error) {
@@ -199,7 +212,8 @@ static int ls_recover(struct dlm_ls *ls, struct dlm_recover *rv)
 
 	dlm_astd_wake();
 
-	log_debug(ls, "recover %llx done: %u ms", rv->seq,
+	log_debug(ls, "recover %llx done: %u ms",
+		  (unsigned long long)rv->seq,
 		  jiffies_to_msecs(jiffies - start));
 	mutex_unlock(&ls->ls_recoverd_active);
 
@@ -207,11 +221,16 @@ static int ls_recover(struct dlm_ls *ls, struct dlm_recover *rv)
 
  fail:
 	dlm_release_root_list(ls);
-	log_debug(ls, "recover %llx error %d", rv->seq, error);
+	log_debug(ls, "recover %llx error %d",
+		  (unsigned long long)rv->seq, error);
 	mutex_unlock(&ls->ls_recoverd_active);
 	return error;
 }
 
+/* The dlm_ls_start() that created the rv we take here may already have been
+   stopped via dlm_ls_stop(); in that case we need to leave the RECOVERY_STOP
+   flag set. */
+
 static void do_ls_recovery(struct dlm_ls *ls)
 {
 	struct dlm_recover *rv = NULL;
@@ -219,7 +238,8 @@ static void do_ls_recovery(struct dlm_ls *ls)
 	spin_lock(&ls->ls_recover_lock);
 	rv = ls->ls_recover_args;
 	ls->ls_recover_args = NULL;
-	clear_bit(LSFL_RECOVERY_STOP, &ls->ls_flags);
+	if (rv && ls->ls_recover_seq == rv->seq)
+		clear_bit(LSFL_RECOVERY_STOP, &ls->ls_flags);
 	spin_unlock(&ls->ls_recover_lock);
 
 	if (rv) {
diff --git a/fs/dlm/requestqueue.c b/fs/dlm/requestqueue.c
index 7b2b089634a2..65008d79c96d 100644
--- a/fs/dlm/requestqueue.c
+++ b/fs/dlm/requestqueue.c
@@ -30,26 +30,36 @@ struct rq_entry {
  * lockspace is enabled on some while still suspended on others.
  */
 
-void dlm_add_requestqueue(struct dlm_ls *ls, int nodeid, struct dlm_header *hd)
+int dlm_add_requestqueue(struct dlm_ls *ls, int nodeid, struct dlm_header *hd)
 {
 	struct rq_entry *e;
 	int length = hd->h_length;
-
-	if (dlm_is_removed(ls, nodeid))
-		return;
+	int rv = 0;
 
 	e = kmalloc(sizeof(struct rq_entry) + length, GFP_KERNEL);
 	if (!e) {
 		log_print("dlm_add_requestqueue: out of memory\n");
-		return;
+		return 0;
 	}
 
 	e->nodeid = nodeid;
 	memcpy(e->request, hd, length);
 
+	/* We need to check dlm_locking_stopped() after taking the mutex to
+	   avoid a race where dlm_recoverd enables locking and runs
+	   process_requestqueue between our earlier dlm_locking_stopped check
+	   and this addition to the requestqueue. */
+
 	mutex_lock(&ls->ls_requestqueue_mutex);
-	list_add_tail(&e->list, &ls->ls_requestqueue);
+	if (dlm_locking_stopped(ls))
+		list_add_tail(&e->list, &ls->ls_requestqueue);
+	else {
+		log_debug(ls, "dlm_add_requestqueue skip from %d", nodeid);
+		kfree(e);
+		rv = -EAGAIN;
+	}
 	mutex_unlock(&ls->ls_requestqueue_mutex);
+	return rv;
 }
 
 int dlm_process_requestqueue(struct dlm_ls *ls)
@@ -120,6 +130,10 @@ static int purge_request(struct dlm_ls *ls, struct dlm_message *ms, int nodeid)
 {
 	uint32_t type = ms->m_type;
 
+	/* the ls is being cleaned up and freed by release_lockspace */
+	if (!ls->ls_count)
+		return 1;
+
 	if (dlm_is_removed(ls, nodeid))
 		return 1;
 
diff --git a/fs/dlm/requestqueue.h b/fs/dlm/requestqueue.h
index 349f0d292d95..6a53ea03335d 100644
--- a/fs/dlm/requestqueue.h
+++ b/fs/dlm/requestqueue.h
@@ -13,7 +13,7 @@
 #ifndef __REQUESTQUEUE_DOT_H__
 #define __REQUESTQUEUE_DOT_H__
 
-void dlm_add_requestqueue(struct dlm_ls *ls, int nodeid, struct dlm_header *hd);
+int dlm_add_requestqueue(struct dlm_ls *ls, int nodeid, struct dlm_header *hd);
 int dlm_process_requestqueue(struct dlm_ls *ls);
 void dlm_wait_requestqueue(struct dlm_ls *ls);
 void dlm_purge_requestqueue(struct dlm_ls *ls);
generated by cgit-pink 1.4.1 (git 2.36.1) at 2024-09-23 19:31:20 +0000