Ocfs2 devel - Apr 2008 - [PATCH 1/1] ocfs2/net: Add debug interface to o2net

This patch exposes o2net information via debugfs. The information includes
the list of sockets (sock_containers) as well as the list of outstanding
messages (send_tracking). Useful for o2dlm debugging.

(This patch is derived from an earlier one written by Zach Brown that
exposed the same information via /proc.)

Signed-off-by: Sunil Mushran <sunil.mushran at oracle.com>
---
 fs/ocfs2/cluster/Makefile       |    2 +-
 fs/ocfs2/cluster/netdebug.c     |  437 +++++++++++++++++++++++++++++++++++++++
 fs/ocfs2/cluster/nodemanager.c  |    5 +-
 fs/ocfs2/cluster/tcp.c          |   20 ++
 fs/ocfs2/cluster/tcp.h          |   32 +++
 fs/ocfs2/cluster/tcp_internal.h |   83 ++++++++-
 6 files changed, 576 insertions(+), 3 deletions(-)
 create mode 100644 fs/ocfs2/cluster/netdebug.c

diff --git a/fs/ocfs2/cluster/Makefile b/fs/ocfs2/cluster/Makefile
index cdd162f..bc8c5e7 100644
--- a/fs/ocfs2/cluster/Makefile
+++ b/fs/ocfs2/cluster/Makefile
@@ -1,4 +1,4 @@
 obj-$(CONFIG_OCFS2_FS) += ocfs2_nodemanager.o
 
 ocfs2_nodemanager-objs := heartbeat.o masklog.o sys.o nodemanager.o \
-	quorum.o tcp.o ver.o
+	quorum.o tcp.o netdebug.o ver.o
diff --git a/fs/ocfs2/cluster/netdebug.c b/fs/ocfs2/cluster/netdebug.c
new file mode 100644
index 0000000..b80e6e9
--- /dev/null
+++ b/fs/ocfs2/cluster/netdebug.c
@@ -0,0 +1,437 @@
+/* -*- mode: c; c-basic-offset: 8; -*-
+ * vim: noexpandtab sw=8 ts=8 sts=0:
+ *
+ * netdebug.c
+ *
+ * debug functionality for o2net
+ *
+ * Copyright (C) 2005, 2008 Oracle.  All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public
+ * License as published by the Free Software Foundation; either
+ * version 2 of the License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public
+ * License along with this program; if not, write to the
+ * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+ * Boston, MA 021110-1307, USA.
+ *
+ */
+
+#ifdef CONFIG_DEBUG_FS
+
+#include <linux/module.h>
+#include <linux/types.h>
+#include <linux/slab.h>
+#include <linux/idr.h>
+#include <linux/kref.h>
+#include <linux/seq_file.h>
+#include <linux/debugfs.h>
+
+#include <asm/uaccess.h>
+
+#include "tcp.h"
+#include "nodemanager.h"
+#define MLOG_MASK_PREFIX ML_TCP
+#include "masklog.h"
+
+#include "tcp_internal.h"
+
+#define O2NET_DEBUG_DIR		"o2net"
+#define SC_DEBUG_NAME		"sock_containers"
+#define NST_DEBUG_NAME		"send_tracking"
+
+static struct dentry *o2net_dentry = NULL;
+static struct dentry *sc_dentry = NULL;
+static struct dentry *nst_dentry = NULL;
+
+static spinlock_t o2net_debug_lock = SPIN_LOCK_UNLOCKED;
+
+static LIST_HEAD(sock_containers);
+static LIST_HEAD(send_tracking);
+
+void o2net_debug_add_nst(struct o2net_send_tracking *nst)
+{
+	spin_lock(&o2net_debug_lock);
+	list_add(&nst->st_net_debug_item, &send_tracking);
+	spin_unlock(&o2net_debug_lock);
+}
+
+void o2net_debug_del_nst(struct o2net_send_tracking *nst)
+{
+	spin_lock(&o2net_debug_lock);
+	if (!list_empty(&nst->st_net_debug_item))
+		list_del_init(&nst->st_net_debug_item);
+	spin_unlock(&o2net_debug_lock);
+}
+
+static struct o2net_send_tracking *next_nst(struct o2net_send_tracking
*nst_start)
+{
+	struct o2net_send_tracking *nst, *ret = NULL;
+
+	assert_spin_locked(&o2net_debug_lock);
+
+	list_for_each_entry(nst, &nst_start->st_net_debug_item,
+			    st_net_debug_item) {
+		/* discover the head of the list */
+		if (&nst->st_net_debug_item == &send_tracking)
+			break;
+
+		/* use st_task to detect real nsts in the list */
+		if (nst->st_task != NULL) {
+			ret = nst;
+			break;
+		}
+	}
+
+	return ret;
+}
+
+static void *nst_seq_start(struct seq_file *seq, loff_t *pos)
+{
+	struct o2net_send_tracking *nst, *dummy_nst = seq->private;
+
+	spin_lock(&o2net_debug_lock);
+	nst = next_nst(dummy_nst);
+	spin_unlock(&o2net_debug_lock);
+
+	return nst;
+}
+
+static void *nst_seq_next(struct seq_file *seq, void *v, loff_t *pos)
+{
+	struct o2net_send_tracking *nst, *dummy_nst = seq->private;
+
+	spin_lock(&o2net_debug_lock);
+	nst = next_nst(dummy_nst);
+	list_del_init(&dummy_nst->st_net_debug_item);
+	if (nst)
+		list_add(&dummy_nst->st_net_debug_item,
+			 &nst->st_net_debug_item);
+	spin_unlock(&o2net_debug_lock);
+
+	return nst; /* unused, just needs to be null when done */
+}
+
+static int nst_seq_show(struct seq_file *seq, void *v)
+{
+	struct o2net_send_tracking *nst, *dummy_nst = seq->private;
+
+	spin_lock(&o2net_debug_lock);
+	nst = next_nst(dummy_nst);
+
+	if (nst != NULL) {
+		/* get_task_comm isn't exported.  oh well. */
+		seq_printf(seq, "%p:\n"
+			   "  pid:          %lu\n"
+			   "  tgid:         %lu\n"
+			   "  process name: %s\n"
+			   "  node:         %u\n"
+			   "  sc:           %p\n"
+			   "  message id:   %d\n"
+			   "  message type: %u\n"
+			   "  message key:  0x%08x\n"
+			   "  sock acquiry: %lu.%lu\n"
+			   "  send start:   %lu.%lu\n"
+			   "  wait start:   %lu.%lu\n",
+			   nst, (unsigned long)nst->st_task->pid,
+			   (unsigned long)nst->st_task->tgid,
+			   nst->st_task->comm, nst->st_node,
+			   nst->st_sc, nst->st_id, nst->st_msg_type,
+			   nst->st_msg_key,
+			   nst->st_sock_time.tv_sec, nst->st_sock_time.tv_usec,
+			   nst->st_send_time.tv_sec, nst->st_send_time.tv_usec,
+			   nst->st_status_time.tv_sec,
+			   nst->st_status_time.tv_usec);
+	}
+
+	spin_unlock(&o2net_debug_lock);
+
+	return 0;
+}
+
+static void nst_seq_stop(struct seq_file *seq, void *v)
+{
+}
+
+static struct seq_operations nst_seq_ops = {
+	.start = nst_seq_start,
+	.next = nst_seq_next,
+	.stop = nst_seq_stop,
+	.show = nst_seq_show,
+};
+
+static int nst_fop_open(struct inode *inode, struct file *file)
+{
+	struct o2net_send_tracking *dummy_nst;
+	struct seq_file *seq;
+	int ret;
+
+	dummy_nst = kmalloc(sizeof(struct o2net_send_tracking), GFP_KERNEL);
+	if (dummy_nst == NULL) {
+		ret = -ENOMEM;
+		goto out;
+	}
+	dummy_nst->st_task = NULL;
+
+	ret = seq_open(file, &nst_seq_ops);
+	if (ret)
+		goto out;
+
+	seq = file->private_data;
+	seq->private = dummy_nst;
+	o2net_debug_add_nst(dummy_nst);
+
+	dummy_nst = NULL;
+
+out:
+	kfree(dummy_nst);
+	return ret;
+}
+
+static int nst_fop_release(struct inode *inode, struct file *file)
+{
+	struct seq_file *seq = file->private_data;
+	struct o2net_send_tracking *dummy_nst = seq->private;
+
+	o2net_debug_del_nst(dummy_nst);
+	return seq_release_private(inode, file);
+}
+
+static struct file_operations nst_seq_fops = {
+	.open = nst_fop_open,
+	.read = seq_read,
+	.llseek = seq_lseek,
+	.release = nst_fop_release,
+};
+
+void o2net_debug_add_sc(struct o2net_sock_container *sc)
+{
+	spin_lock(&o2net_debug_lock);
+	list_add(&sc->sc_net_debug_item, &sock_containers);
+	spin_unlock(&o2net_debug_lock);
+}
+
+void o2net_debug_del_sc(struct o2net_sock_container *sc)
+{
+	spin_lock(&o2net_debug_lock);
+	list_del_init(&sc->sc_net_debug_item);
+	spin_unlock(&o2net_debug_lock);
+}
+
+static struct o2net_sock_container *next_sc(struct o2net_sock_container
*sc_start)
+{
+	struct o2net_sock_container *sc, *ret = NULL;
+
+	assert_spin_locked(&o2net_debug_lock);
+
+	list_for_each_entry(sc, &sc_start->sc_net_debug_item,
sc_net_debug_item) {
+		/* discover the head of the list miscast as a sc */
+		if (&sc->sc_net_debug_item == &sock_containers)
+			break;
+
+		/* use sc_page to detect real scs in the list */
+		if (sc->sc_page != NULL) {
+			ret = sc;
+			break;
+		}
+	}
+
+	return ret;
+}
+
+static void *sc_seq_start(struct seq_file *seq, loff_t *pos)
+{
+	struct o2net_sock_container *sc, *dummy_sc = seq->private;
+
+	spin_lock(&o2net_debug_lock);
+	sc = next_sc(dummy_sc);
+	spin_unlock(&o2net_debug_lock);
+
+	return sc;
+}
+
+static void *sc_seq_next(struct seq_file *seq, void *v, loff_t *pos)
+{
+	struct o2net_sock_container *sc, *dummy_sc = seq->private;
+
+	spin_lock(&o2net_debug_lock);
+	sc = next_sc(dummy_sc);
+	list_del_init(&dummy_sc->sc_net_debug_item);
+	if (sc)
+		list_add(&dummy_sc->sc_net_debug_item, &sc->sc_net_debug_item);
+	spin_unlock(&o2net_debug_lock);
+
+	return sc; /* unused, just needs to be null when done */
+}
+
+#define TV_SEC_USEC(TV) TV.tv_sec, TV.tv_usec
+
+static int sc_seq_show(struct seq_file *seq, void *v)
+{
+	struct o2net_sock_container *sc, *dummy_sc = seq->private;
+
+	spin_lock(&o2net_debug_lock);
+	sc = next_sc(dummy_sc);
+
+	if (sc != NULL) {
+		struct inet_sock *inet = NULL;
+
+		__be32 saddr = 0, daddr = 0;
+		__be16 sport = 0, dport = 0;
+
+		if (sc->sc_sock) {
+			inet = inet_sk(sc->sc_sock->sk);
+			/* the stack's structs aren't sparse endian clean */
+			saddr = (__force __be32)inet->saddr;
+			daddr = (__force __be32)inet->daddr;
+			sport = (__force __be16)inet->sport;
+			dport = (__force __be16)inet->dport;
+		}
+
+		/* XXX sigh, inet-> doesn't have sparse annotation so any
+		 * use of it here generates a warning with -Wbitwise */
+		seq_printf(seq, "%p:\n"
+			   "  krefs:           %d\n"
+			   "  sock:            %u.%u.%u.%u:%u -> %u.%u.%u.%u:%u\n"
+			   "  remote node:     %s\n"
+			   "  page off:        %zu\n"
+			   "  handshake ok:    %u\n"
+			   "  timer:           %lu.%lu\n"
+			   "  data ready:      %lu.%lu\n"
+			   "  advance start:   %lu.%lu\n"
+			   "  advance stop:    %lu.%lu\n"
+			   "  func start:      %lu.%lu\n"
+			   "  func stop:       %lu.%lu\n"
+			   "  func key:        %u\n"
+			   "  func type:       %u\n",
+			   sc,
+			   atomic_read(&sc->sc_kref.refcount),
+			   NIPQUAD(saddr), inet ? ntohs(sport) : 0,
+			   NIPQUAD(daddr), inet ? ntohs(dport) : 0,
+			   sc->sc_node->nd_name,
+			   sc->sc_page_off,
+			   sc->sc_handshake_ok,
+			   TV_SEC_USEC(sc->sc_tv_timer),
+			   TV_SEC_USEC(sc->sc_tv_data_ready),
+			   TV_SEC_USEC(sc->sc_tv_advance_start),
+			   TV_SEC_USEC(sc->sc_tv_advance_stop),
+			   TV_SEC_USEC(sc->sc_tv_func_start),
+			   TV_SEC_USEC(sc->sc_tv_func_stop),
+			   sc->sc_msg_key,
+			   sc->sc_msg_type);
+	}
+
+
+	spin_unlock(&o2net_debug_lock);
+
+	return 0;
+}
+
+static void sc_seq_stop(struct seq_file *seq, void *v)
+{
+}
+
+static struct seq_operations sc_seq_ops = {
+	.start = sc_seq_start,
+	.next = sc_seq_next,
+	.stop = sc_seq_stop,
+	.show = sc_seq_show,
+};
+
+static int sc_fop_open(struct inode *inode, struct file *file)
+{
+	struct o2net_sock_container *dummy_sc;
+	struct seq_file *seq;
+	int ret;
+
+	dummy_sc = kmalloc(sizeof(struct o2net_sock_container), GFP_KERNEL);
+	if (dummy_sc == NULL) {
+		ret = -ENOMEM;
+		goto out;
+	}
+	dummy_sc->sc_page = NULL;
+
+	ret = seq_open(file, &sc_seq_ops);
+	if (ret)
+		goto out;
+
+	seq = file->private_data;
+	seq->private = dummy_sc;
+	o2net_debug_add_sc(dummy_sc);
+
+	dummy_sc = NULL;
+
+out:
+	kfree(dummy_sc);
+	return ret;
+}
+
+static int sc_fop_release(struct inode *inode, struct file *file)
+{
+	struct seq_file *seq = file->private_data;
+	struct o2net_sock_container *dummy_sc = seq->private;
+
+	o2net_debug_del_sc(dummy_sc);
+	return seq_release_private(inode, file);
+}
+
+static struct file_operations sc_seq_fops = {
+	.open = sc_fop_open,
+	.read = seq_read,
+	.llseek = seq_lseek,
+	.release = sc_fop_release,
+};
+
+int o2net_debugfs_init(void)
+{
+	o2net_dentry = debugfs_create_dir(O2NET_DEBUG_DIR, NULL);
+	if (!o2net_dentry) {
+		mlog_errno(-ENOMEM);
+		goto bail;
+	}
+
+	nst_dentry = debugfs_create_file(NST_DEBUG_NAME, S_IFREG|S_IRUSR,
+					 o2net_dentry, NULL,
+					 &nst_seq_fops);
+	if (!nst_dentry) {
+		mlog_errno(-ENOMEM);
+		goto bail;
+	}
+
+	sc_dentry = debugfs_create_file(SC_DEBUG_NAME, S_IFREG|S_IRUSR,
+					o2net_dentry, NULL,
+					&sc_seq_fops);
+	if (!sc_dentry) {
+		mlog_errno(-ENOMEM);
+		goto bail;
+	}
+
+	return 0;
+bail:
+	if (sc_dentry)
+		debugfs_remove(sc_dentry);
+	if (nst_dentry)
+		debugfs_remove(nst_dentry);
+	if (o2net_dentry)
+		debugfs_remove(o2net_dentry);
+	return -ENOMEM;
+}
+
+void o2net_debugfs_exit(void)
+{
+	if (sc_dentry)
+		debugfs_remove(sc_dentry);
+	if (nst_dentry)
+		debugfs_remove(nst_dentry);
+	if (o2net_dentry)
+		debugfs_remove(o2net_dentry);
+}
+
+#endif	/* CONFIG_DEBUG_FS */
diff --git a/fs/ocfs2/cluster/nodemanager.c b/fs/ocfs2/cluster/nodemanager.c
index 709fba2..cf9401e 100644
--- a/fs/ocfs2/cluster/nodemanager.c
+++ b/fs/ocfs2/cluster/nodemanager.c
@@ -959,7 +959,10 @@ static int __init init_o2nm(void)
 	cluster_print_version();
 
 	o2hb_init();
-	o2net_init();
+
+	ret = o2net_init();
+	if (ret)
+		goto out;
 
 	ocfs2_table_header = register_sysctl_table(ocfs2_root_table);
 	if (!ocfs2_table_header) {
diff --git a/fs/ocfs2/cluster/tcp.c b/fs/ocfs2/cluster/tcp.c
index b8057c5..c2af3c4 100644
--- a/fs/ocfs2/cluster/tcp.c
+++ b/fs/ocfs2/cluster/tcp.c
@@ -296,6 +296,7 @@ static void sc_kref_release(struct kref *kref)
 	o2nm_node_put(sc->sc_node);
 	sc->sc_node = NULL;
 
+	o2net_debug_del_sc(sc);
 	kfree(sc);
 }
 
@@ -336,6 +337,7 @@ static struct o2net_sock_container *sc_alloc(struct
o2nm_node *node)
 
 	ret = sc;
 	sc->sc_page = page;
+	o2net_debug_add_sc(sc);
 	sc = NULL;
 	page = NULL;
 
@@ -914,6 +916,9 @@ int o2net_send_message_vec(u32 msg_type, u32 key, struct
kvec *caller_vec,
 	struct o2net_status_wait nsw = {
 		.ns_node_item = LIST_HEAD_INIT(nsw.ns_node_item),
 	};
+	struct o2net_send_tracking nst;
+
+	o2net_init_nst(&nst, msg_type, key, current, target_node);
 
 	if (o2net_wq == NULL) {
 		mlog(0, "attempt to tx without o2netd running\n");
@@ -939,6 +944,10 @@ int o2net_send_message_vec(u32 msg_type, u32 key, struct
kvec *caller_vec,
 		goto out;
 	}
 
+	o2net_debug_add_nst(&nst);
+
+	o2net_set_nst_sock_time(&nst);
+
 	ret = wait_event_interruptible(nn->nn_sc_wq,
 				       o2net_tx_can_proceed(nn, &sc, &error));
 	if (!ret && error)
@@ -946,6 +955,8 @@ int o2net_send_message_vec(u32 msg_type, u32 key, struct
kvec *caller_vec,
 	if (ret)
 		goto out;
 
+	o2net_set_nst_sock_container(&nst, sc);
+
 	veclen = caller_veclen + 1;
 	vec = kmalloc(sizeof(struct kvec) * veclen, GFP_ATOMIC);
 	if (vec == NULL) {
@@ -972,6 +983,9 @@ int o2net_send_message_vec(u32 msg_type, u32 key, struct
kvec *caller_vec,
 		goto out;
 
 	msg->msg_num = cpu_to_be32(nsw.ns_id);
+	o2net_set_nst_msg_id(&nst, nsw.ns_id);
+
+	o2net_set_nst_send_time(&nst);
 
 	/* finally, convert the message header to network byte-order
 	 * and send */
@@ -986,6 +1000,7 @@ int o2net_send_message_vec(u32 msg_type, u32 key, struct
kvec *caller_vec,
 	}
 
 	/* wait on other node's handler */
+	o2net_set_nst_status_time(&nst);
 	wait_event(nsw.ns_wq, o2net_nsw_completed(nn, &nsw));
 
 	/* Note that we avoid overwriting the callers status return
@@ -998,6 +1013,7 @@ int o2net_send_message_vec(u32 msg_type, u32 key, struct
kvec *caller_vec,
 	mlog(0, "woken, returning system status %d, user status %d\n",
 	     ret, nsw.ns_status);
 out:
+	o2net_debug_del_nst(&nst); /* must be before dropping sc and node */
 	if (sc)
 		sc_put(sc);
 	if (vec)
@@ -1922,6 +1938,9 @@ int o2net_init(void)
 
 	o2quo_init();
 
+	if (o2net_debugfs_init())
+		return -ENOMEM;
+
 	o2net_hand = kzalloc(sizeof(struct o2net_handshake), GFP_KERNEL);
 	o2net_keep_req = kzalloc(sizeof(struct o2net_msg), GFP_KERNEL);
 	o2net_keep_resp = kzalloc(sizeof(struct o2net_msg), GFP_KERNEL);
@@ -1962,4 +1981,5 @@ void o2net_exit(void)
 	kfree(o2net_hand);
 	kfree(o2net_keep_req);
 	kfree(o2net_keep_resp);
+	o2net_debugfs_exit();
 }
diff --git a/fs/ocfs2/cluster/tcp.h b/fs/ocfs2/cluster/tcp.h
index f36f66a..a705d5d 100644
--- a/fs/ocfs2/cluster/tcp.h
+++ b/fs/ocfs2/cluster/tcp.h
@@ -117,4 +117,36 @@ int o2net_num_connected_peers(void);
 int o2net_init(void);
 void o2net_exit(void);
 
+struct o2net_send_tracking;
+struct o2net_sock_container;
+
+#ifdef CONFIG_DEBUG_FS
+int o2net_debugfs_init(void);
+void o2net_debugfs_exit(void);
+void o2net_debug_add_nst(struct o2net_send_tracking *nst);
+void o2net_debug_del_nst(struct o2net_send_tracking *nst);
+void o2net_debug_add_sc(struct o2net_sock_container *sc);
+void o2net_debug_del_sc(struct o2net_sock_container *sc);
+#else
+static int o2net_debugfs_init(void)
+{
+	return 0;
+}
+static void o2net_debugfs_exit(void)
+{
+}
+static void o2net_debug_add_nst(struct o2net_send_tracking *nst)
+{
+}
+static void o2net_debug_del_nst(struct o2net_send_tracking *nst)
+{
+}
+static void o2net_debug_add_sc(struct o2net_sock_container *sc)
+{
+}
+static void o2net_debug_del_sc(struct o2net_sock_container *sc)
+{
+}
+#endif	/* CONFIG_DEBUG_FS */
+
 #endif /* O2CLUSTER_TCP_H */
diff --git a/fs/ocfs2/cluster/tcp_internal.h b/fs/ocfs2/cluster/tcp_internal.h
index d25b9af..68f2448 100644
--- a/fs/ocfs2/cluster/tcp_internal.h
+++ b/fs/ocfs2/cluster/tcp_internal.h
@@ -164,7 +164,9 @@ struct o2net_sock_container {
 	/* original handlers for the sockets */
 	void			(*sc_state_change)(struct sock *sk);
 	void			(*sc_data_ready)(struct sock *sk, int bytes);
-
+#ifdef CONFIG_DEBUG_FS
+	struct list_head        sc_net_debug_item;
+#endif
 	struct timeval 		sc_tv_timer;
 	struct timeval 		sc_tv_data_ready;
 	struct timeval 		sc_tv_advance_start;
@@ -206,4 +208,83 @@ struct o2net_status_wait {
 	struct list_head	ns_node_item;
 };
 
+#ifdef CONFIG_DEBUG_FS
+/* just for state dumps */
+struct o2net_send_tracking {
+	struct list_head		st_net_debug_item;
+	struct task_struct		*st_task;
+	struct o2net_sock_container	*st_sc;
+	u32				st_id;
+	u32				st_msg_type;
+	u32				st_msg_key;
+	u8				st_node;
+	struct timeval			st_sock_time;
+	struct timeval			st_send_time;
+	struct timeval			st_status_time;
+};
+
+static void o2net_init_nst(struct o2net_send_tracking *nst, u32 msgtype,
+			   u32 msgkey, struct task_struct *task, u8 node)
+{
+	INIT_LIST_HEAD(&nst->st_net_debug_item);
+	nst->st_task = task;
+	nst->st_msg_type = msgtype;
+	nst->st_msg_key = msgkey;
+	nst->st_node = node;
+}
+
+static void o2net_set_nst_sock_time(struct o2net_send_tracking *nst)
+{
+	do_gettimeofday(&nst->st_sock_time);
+}
+
+static void o2net_set_nst_send_time(struct o2net_send_tracking *nst)
+{
+	do_gettimeofday(&nst->st_send_time);
+}
+
+static void o2net_set_nst_status_time(struct o2net_send_tracking *nst)
+{
+	do_gettimeofday(&nst->st_status_time);
+}
+
+static void o2net_set_nst_sock_container(struct o2net_send_tracking *nst,
+					 struct o2net_sock_container *sc)
+{
+	nst->st_sc = sc;
+}
+
+static void o2net_set_nst_msg_id(struct o2net_send_tracking *nst, u32 msg_id)
+{
+	nst->st_id = msg_id;
+}
+
+#else	/* CONFIG_DEBUG_FS */
+
+struct o2net_send_tracking {
+	u32	dummy;
+};
+
+static void o2net_init_nst(struct o2net_send_tracking *nst, u32 msgtype,
+			   u32 msgkey, struct task_struct *task, u8 node)
+{
+}
+static void o2net_set_nst_sock_time(struct o2net_send_tracking *nst)
+{
+}
+static void o2net_set_nst_send_time(struct o2net_send_tracking *nst)
+{
+}
+static void o2net_set_nst_status_time(struct o2net_send_tracking *nst)
+{
+}
+static void o2net_set_nst_sock_container(struct o2net_send_tracking *nst,
+					 struct o2net_sock_container *sc)
+{
+}
+static void o2net_set_nst_msg_id(struct o2net_send_tracking *nst, u32 msg_id)
+{
+}
+#endif	/* CONFIG_DEBUG_FS */
+
 #endif /* O2CLUSTER_TCP_INTERNAL_H */
-- 
1.5.3.4

On Wed, Apr 02, 2008 at 12:40:10PM -0700, Sunil Mushran
wrote:
> This patch exposes o2net information via debugfs. The information includes
> the list of sockets (sock_containers) as well as the list of outstanding
> messages (send_tracking). Useful for o2dlm debugging.
> 
> (This patch is derived from an earlier one written by Zach Brown that
> exposed the same information via /proc.)
> 
> Signed-off-by: Sunil Mushran <sunil.mushran at oracle.com>
Signed-off-by: Joel Becker <joel.becker at oracle.com>
> ---
>  fs/ocfs2/cluster/Makefile       |    2 +-
>  fs/ocfs2/cluster/netdebug.c     |  437
+++++++++++++++++++++++++++++++++++++++
>  fs/ocfs2/cluster/nodemanager.c  |    5 +-
>  fs/ocfs2/cluster/tcp.c          |   20 ++
>  fs/ocfs2/cluster/tcp.h          |   32 +++
>  fs/ocfs2/cluster/tcp_internal.h |   83 ++++++++-
>  6 files changed, 576 insertions(+), 3 deletions(-)
>  create mode 100644 fs/ocfs2/cluster/netdebug.c
> 
> diff --git a/fs/ocfs2/cluster/Makefile b/fs/ocfs2/cluster/Makefile
> index cdd162f..bc8c5e7 100644
> --- a/fs/ocfs2/cluster/Makefile
> +++ b/fs/ocfs2/cluster/Makefile
> @@ -1,4 +1,4 @@
>  obj-$(CONFIG_OCFS2_FS) += ocfs2_nodemanager.o
>  
>  ocfs2_nodemanager-objs := heartbeat.o masklog.o sys.o nodemanager.o \
> -	quorum.o tcp.o ver.o
> +	quorum.o tcp.o netdebug.o ver.o
> diff --git a/fs/ocfs2/cluster/netdebug.c b/fs/ocfs2/cluster/netdebug.c
> new file mode 100644
> index 0000000..b80e6e9
> --- /dev/null
> +++ b/fs/ocfs2/cluster/netdebug.c
> @@ -0,0 +1,437 @@
> +/* -*- mode: c; c-basic-offset: 8; -*-
> + * vim: noexpandtab sw=8 ts=8 sts=0:
> + *
> + * netdebug.c
> + *
> + * debug functionality for o2net
> + *
> + * Copyright (C) 2005, 2008 Oracle.  All rights reserved.
> + *
> + * This program is free software; you can redistribute it and/or
> + * modify it under the terms of the GNU General Public
> + * License as published by the Free Software Foundation; either
> + * version 2 of the License, or (at your option) any later version.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> + * General Public License for more details.
> + *
> + * You should have received a copy of the GNU General Public
> + * License along with this program; if not, write to the
> + * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
> + * Boston, MA 021110-1307, USA.
> + *
> + */
> +
> +#ifdef CONFIG_DEBUG_FS
> +
> +#include <linux/module.h>
> +#include <linux/types.h>
> +#include <linux/slab.h>
> +#include <linux/idr.h>
> +#include <linux/kref.h>
> +#include <linux/seq_file.h>
> +#include <linux/debugfs.h>
> +
> +#include <asm/uaccess.h>
> +
> +#include "tcp.h"
> +#include "nodemanager.h"
> +#define MLOG_MASK_PREFIX ML_TCP
> +#include "masklog.h"
> +
> +#include "tcp_internal.h"
> +
> +#define O2NET_DEBUG_DIR		"o2net"
> +#define SC_DEBUG_NAME		"sock_containers"
> +#define NST_DEBUG_NAME		"send_tracking"
> +
> +static struct dentry *o2net_dentry = NULL;
> +static struct dentry *sc_dentry = NULL;
> +static struct dentry *nst_dentry = NULL;
> +
> +static spinlock_t o2net_debug_lock = SPIN_LOCK_UNLOCKED;
> +
> +static LIST_HEAD(sock_containers);
> +static LIST_HEAD(send_tracking);
> +
> +void o2net_debug_add_nst(struct o2net_send_tracking *nst)
> +{
> +	spin_lock(&o2net_debug_lock);
> +	list_add(&nst->st_net_debug_item, &send_tracking);
> +	spin_unlock(&o2net_debug_lock);
> +}
> +
> +void o2net_debug_del_nst(struct o2net_send_tracking *nst)
> +{
> +	spin_lock(&o2net_debug_lock);
> +	if (!list_empty(&nst->st_net_debug_item))
> +		list_del_init(&nst->st_net_debug_item);
> +	spin_unlock(&o2net_debug_lock);
> +}
> +
> +static struct o2net_send_tracking *next_nst(struct o2net_send_tracking
*nst_start)
> +{
> +	struct o2net_send_tracking *nst, *ret = NULL;
> +
> +	assert_spin_locked(&o2net_debug_lock);
> +
> +	list_for_each_entry(nst, &nst_start->st_net_debug_item,
> +			    st_net_debug_item) {
> +		/* discover the head of the list */
> +		if (&nst->st_net_debug_item == &send_tracking)
> +			break;
> +
> +		/* use st_task to detect real nsts in the list */
> +		if (nst->st_task != NULL) {
> +			ret = nst;
> +			break;
> +		}
> +	}
> +
> +	return ret;
> +}
> +
> +static void *nst_seq_start(struct seq_file *seq, loff_t *pos)
> +{
> +	struct o2net_send_tracking *nst, *dummy_nst = seq->private;
> +
> +	spin_lock(&o2net_debug_lock);
> +	nst = next_nst(dummy_nst);
> +	spin_unlock(&o2net_debug_lock);
> +
> +	return nst;
> +}
> +
> +static void *nst_seq_next(struct seq_file *seq, void *v, loff_t *pos)
> +{
> +	struct o2net_send_tracking *nst, *dummy_nst = seq->private;
> +
> +	spin_lock(&o2net_debug_lock);
> +	nst = next_nst(dummy_nst);
> +	list_del_init(&dummy_nst->st_net_debug_item);
> +	if (nst)
> +		list_add(&dummy_nst->st_net_debug_item,
> +			 &nst->st_net_debug_item);
> +	spin_unlock(&o2net_debug_lock);
> +
> +	return nst; /* unused, just needs to be null when done */
> +}
> +
> +static int nst_seq_show(struct seq_file *seq, void *v)
> +{
> +	struct o2net_send_tracking *nst, *dummy_nst = seq->private;
> +
> +	spin_lock(&o2net_debug_lock);
> +	nst = next_nst(dummy_nst);
> +
> +	if (nst != NULL) {
> +		/* get_task_comm isn't exported.  oh well. */
> +		seq_printf(seq, "%p:\n"
> +			   "  pid:          %lu\n"
> +			   "  tgid:         %lu\n"
> +			   "  process name: %s\n"
> +			   "  node:         %u\n"
> +			   "  sc:           %p\n"
> +			   "  message id:   %d\n"
> +			   "  message type: %u\n"
> +			   "  message key:  0x%08x\n"
> +			   "  sock acquiry: %lu.%lu\n"
> +			   "  send start:   %lu.%lu\n"
> +			   "  wait start:   %lu.%lu\n",
> +			   nst, (unsigned long)nst->st_task->pid,
> +			   (unsigned long)nst->st_task->tgid,
> +			   nst->st_task->comm, nst->st_node,
> +			   nst->st_sc, nst->st_id, nst->st_msg_type,
> +			   nst->st_msg_key,
> +			   nst->st_sock_time.tv_sec, nst->st_sock_time.tv_usec,
> +			   nst->st_send_time.tv_sec, nst->st_send_time.tv_usec,
> +			   nst->st_status_time.tv_sec,
> +			   nst->st_status_time.tv_usec);
> +	}
> +
> +	spin_unlock(&o2net_debug_lock);
> +
> +	return 0;
> +}
> +
> +static void nst_seq_stop(struct seq_file *seq, void *v)
> +{
> +}
> +
> +static struct seq_operations nst_seq_ops = {
> +	.start = nst_seq_start,
> +	.next = nst_seq_next,
> +	.stop = nst_seq_stop,
> +	.show = nst_seq_show,
> +};
> +
> +static int nst_fop_open(struct inode *inode, struct file *file)
> +{
> +	struct o2net_send_tracking *dummy_nst;
> +	struct seq_file *seq;
> +	int ret;
> +
> +	dummy_nst = kmalloc(sizeof(struct o2net_send_tracking), GFP_KERNEL);
> +	if (dummy_nst == NULL) {
> +		ret = -ENOMEM;
> +		goto out;
> +	}
> +	dummy_nst->st_task = NULL;
> +
> +	ret = seq_open(file, &nst_seq_ops);
> +	if (ret)
> +		goto out;
> +
> +	seq = file->private_data;
> +	seq->private = dummy_nst;
> +	o2net_debug_add_nst(dummy_nst);
> +
> +	dummy_nst = NULL;
> +
> +out:
> +	kfree(dummy_nst);
> +	return ret;
> +}
> +
> +static int nst_fop_release(struct inode *inode, struct file *file)
> +{
> +	struct seq_file *seq = file->private_data;
> +	struct o2net_send_tracking *dummy_nst = seq->private;
> +
> +	o2net_debug_del_nst(dummy_nst);
> +	return seq_release_private(inode, file);
> +}
> +
> +static struct file_operations nst_seq_fops = {
> +	.open = nst_fop_open,
> +	.read = seq_read,
> +	.llseek = seq_lseek,
> +	.release = nst_fop_release,
> +};
> +
> +void o2net_debug_add_sc(struct o2net_sock_container *sc)
> +{
> +	spin_lock(&o2net_debug_lock);
> +	list_add(&sc->sc_net_debug_item, &sock_containers);
> +	spin_unlock(&o2net_debug_lock);
> +}
> +
> +void o2net_debug_del_sc(struct o2net_sock_container *sc)
> +{
> +	spin_lock(&o2net_debug_lock);
> +	list_del_init(&sc->sc_net_debug_item);
> +	spin_unlock(&o2net_debug_lock);
> +}
> +
> +static struct o2net_sock_container *next_sc(struct o2net_sock_container
*sc_start)
> +{
> +	struct o2net_sock_container *sc, *ret = NULL;
> +
> +	assert_spin_locked(&o2net_debug_lock);
> +
> +	list_for_each_entry(sc, &sc_start->sc_net_debug_item,
sc_net_debug_item) {
> +		/* discover the head of the list miscast as a sc */
> +		if (&sc->sc_net_debug_item == &sock_containers)
> +			break;
> +
> +		/* use sc_page to detect real scs in the list */
> +		if (sc->sc_page != NULL) {
> +			ret = sc;
> +			break;
> +		}
> +	}
> +
> +	return ret;
> +}
> +
> +static void *sc_seq_start(struct seq_file *seq, loff_t *pos)
> +{
> +	struct o2net_sock_container *sc, *dummy_sc = seq->private;
> +
> +	spin_lock(&o2net_debug_lock);
> +	sc = next_sc(dummy_sc);
> +	spin_unlock(&o2net_debug_lock);
> +
> +	return sc;
> +}
> +
> +static void *sc_seq_next(struct seq_file *seq, void *v, loff_t *pos)
> +{
> +	struct o2net_sock_container *sc, *dummy_sc = seq->private;
> +
> +	spin_lock(&o2net_debug_lock);
> +	sc = next_sc(dummy_sc);
> +	list_del_init(&dummy_sc->sc_net_debug_item);
> +	if (sc)
> +		list_add(&dummy_sc->sc_net_debug_item,
&sc->sc_net_debug_item);
> +	spin_unlock(&o2net_debug_lock);
> +
> +	return sc; /* unused, just needs to be null when done */
> +}
> +
> +#define TV_SEC_USEC(TV) TV.tv_sec, TV.tv_usec
> +
> +static int sc_seq_show(struct seq_file *seq, void *v)
> +{
> +	struct o2net_sock_container *sc, *dummy_sc = seq->private;
> +
> +	spin_lock(&o2net_debug_lock);
> +	sc = next_sc(dummy_sc);
> +
> +	if (sc != NULL) {
> +		struct inet_sock *inet = NULL;
> +
> +		__be32 saddr = 0, daddr = 0;
> +		__be16 sport = 0, dport = 0;
> +
> +		if (sc->sc_sock) {
> +			inet = inet_sk(sc->sc_sock->sk);
> +			/* the stack's structs aren't sparse endian clean */
> +			saddr = (__force __be32)inet->saddr;
> +			daddr = (__force __be32)inet->daddr;
> +			sport = (__force __be16)inet->sport;
> +			dport = (__force __be16)inet->dport;
> +		}
> +
> +		/* XXX sigh, inet-> doesn't have sparse annotation so any
> +		 * use of it here generates a warning with -Wbitwise */
> +		seq_printf(seq, "%p:\n"
> +			   "  krefs:           %d\n"
> +			   "  sock:            %u.%u.%u.%u:%u -> %u.%u.%u.%u:%u\n"
> +			   "  remote node:     %s\n"
> +			   "  page off:        %zu\n"
> +			   "  handshake ok:    %u\n"
> +			   "  timer:           %lu.%lu\n"
> +			   "  data ready:      %lu.%lu\n"
> +			   "  advance start:   %lu.%lu\n"
> +			   "  advance stop:    %lu.%lu\n"
> +			   "  func start:      %lu.%lu\n"
> +			   "  func stop:       %lu.%lu\n"
> +			   "  func key:        %u\n"
> +			   "  func type:       %u\n",
> +			   sc,
> +			   atomic_read(&sc->sc_kref.refcount),
> +			   NIPQUAD(saddr), inet ? ntohs(sport) : 0,
> +			   NIPQUAD(daddr), inet ? ntohs(dport) : 0,
> +			   sc->sc_node->nd_name,
> +			   sc->sc_page_off,
> +			   sc->sc_handshake_ok,
> +			   TV_SEC_USEC(sc->sc_tv_timer),
> +			   TV_SEC_USEC(sc->sc_tv_data_ready),
> +			   TV_SEC_USEC(sc->sc_tv_advance_start),
> +			   TV_SEC_USEC(sc->sc_tv_advance_stop),
> +			   TV_SEC_USEC(sc->sc_tv_func_start),
> +			   TV_SEC_USEC(sc->sc_tv_func_stop),
> +			   sc->sc_msg_key,
> +			   sc->sc_msg_type);
> +	}
> +
> +
> +	spin_unlock(&o2net_debug_lock);
> +
> +	return 0;
> +}
> +
> +static void sc_seq_stop(struct seq_file *seq, void *v)
> +{
> +}
> +
> +static struct seq_operations sc_seq_ops = {
> +	.start = sc_seq_start,
> +	.next = sc_seq_next,
> +	.stop = sc_seq_stop,
> +	.show = sc_seq_show,
> +};
> +
> +static int sc_fop_open(struct inode *inode, struct file *file)
> +{
> +	struct o2net_sock_container *dummy_sc;
> +	struct seq_file *seq;
> +	int ret;
> +
> +	dummy_sc = kmalloc(sizeof(struct o2net_sock_container), GFP_KERNEL);
> +	if (dummy_sc == NULL) {
> +		ret = -ENOMEM;
> +		goto out;
> +	}
> +	dummy_sc->sc_page = NULL;
> +
> +	ret = seq_open(file, &sc_seq_ops);
> +	if (ret)
> +		goto out;
> +
> +	seq = file->private_data;
> +	seq->private = dummy_sc;
> +	o2net_debug_add_sc(dummy_sc);
> +
> +	dummy_sc = NULL;
> +
> +out:
> +	kfree(dummy_sc);
> +	return ret;
> +}
> +
> +static int sc_fop_release(struct inode *inode, struct file *file)
> +{
> +	struct seq_file *seq = file->private_data;
> +	struct o2net_sock_container *dummy_sc = seq->private;
> +
> +	o2net_debug_del_sc(dummy_sc);
> +	return seq_release_private(inode, file);
> +}
> +
> +static struct file_operations sc_seq_fops = {
> +	.open = sc_fop_open,
> +	.read = seq_read,
> +	.llseek = seq_lseek,
> +	.release = sc_fop_release,
> +};
> +
> +int o2net_debugfs_init(void)
> +{
> +	o2net_dentry = debugfs_create_dir(O2NET_DEBUG_DIR, NULL);
> +	if (!o2net_dentry) {
> +		mlog_errno(-ENOMEM);
> +		goto bail;
> +	}
> +
> +	nst_dentry = debugfs_create_file(NST_DEBUG_NAME, S_IFREG|S_IRUSR,
> +					 o2net_dentry, NULL,
> +					 &nst_seq_fops);
> +	if (!nst_dentry) {
> +		mlog_errno(-ENOMEM);
> +		goto bail;
> +	}
> +
> +	sc_dentry = debugfs_create_file(SC_DEBUG_NAME, S_IFREG|S_IRUSR,
> +					o2net_dentry, NULL,
> +					&sc_seq_fops);
> +	if (!sc_dentry) {
> +		mlog_errno(-ENOMEM);
> +		goto bail;
> +	}
> +
> +	return 0;
> +bail:
> +	if (sc_dentry)
> +		debugfs_remove(sc_dentry);
> +	if (nst_dentry)
> +		debugfs_remove(nst_dentry);
> +	if (o2net_dentry)
> +		debugfs_remove(o2net_dentry);
> +	return -ENOMEM;
> +}
> +
> +void o2net_debugfs_exit(void)
> +{
> +	if (sc_dentry)
> +		debugfs_remove(sc_dentry);
> +	if (nst_dentry)
> +		debugfs_remove(nst_dentry);
> +	if (o2net_dentry)
> +		debugfs_remove(o2net_dentry);
> +}
> +
> +#endif	/* CONFIG_DEBUG_FS */
> diff --git a/fs/ocfs2/cluster/nodemanager.c
b/fs/ocfs2/cluster/nodemanager.c
> index 709fba2..cf9401e 100644
> --- a/fs/ocfs2/cluster/nodemanager.c
> +++ b/fs/ocfs2/cluster/nodemanager.c
> @@ -959,7 +959,10 @@ static int __init init_o2nm(void)
>  	cluster_print_version();
>  
>  	o2hb_init();
> -	o2net_init();
> +
> +	ret = o2net_init();
> +	if (ret)
> +		goto out;
>  
>  	ocfs2_table_header = register_sysctl_table(ocfs2_root_table);
>  	if (!ocfs2_table_header) {
> diff --git a/fs/ocfs2/cluster/tcp.c b/fs/ocfs2/cluster/tcp.c
> index b8057c5..c2af3c4 100644
> --- a/fs/ocfs2/cluster/tcp.c
> +++ b/fs/ocfs2/cluster/tcp.c
> @@ -296,6 +296,7 @@ static void sc_kref_release(struct kref *kref)
>  	o2nm_node_put(sc->sc_node);
>  	sc->sc_node = NULL;
>  
> +	o2net_debug_del_sc(sc);
>  	kfree(sc);
>  }
>  
> @@ -336,6 +337,7 @@ static struct o2net_sock_container *sc_alloc(struct
o2nm_node *node)
>  
>  	ret = sc;
>  	sc->sc_page = page;
> +	o2net_debug_add_sc(sc);
>  	sc = NULL;
>  	page = NULL;
>  
> @@ -914,6 +916,9 @@ int o2net_send_message_vec(u32 msg_type, u32 key,
struct kvec *caller_vec,
>  	struct o2net_status_wait nsw = {
>  		.ns_node_item = LIST_HEAD_INIT(nsw.ns_node_item),
>  	};
> +	struct o2net_send_tracking nst;
> +
> +	o2net_init_nst(&nst, msg_type, key, current, target_node);
>  
>  	if (o2net_wq == NULL) {
>  		mlog(0, "attempt to tx without o2netd running\n");
> @@ -939,6 +944,10 @@ int o2net_send_message_vec(u32 msg_type, u32 key,
struct kvec *caller_vec,
>  		goto out;
>  	}
>  
> +	o2net_debug_add_nst(&nst);
> +
> +	o2net_set_nst_sock_time(&nst);
> +
>  	ret = wait_event_interruptible(nn->nn_sc_wq,
>  				       o2net_tx_can_proceed(nn, &sc, &error));
>  	if (!ret && error)
> @@ -946,6 +955,8 @@ int o2net_send_message_vec(u32 msg_type, u32 key,
struct kvec *caller_vec,
>  	if (ret)
>  		goto out;
>  
> +	o2net_set_nst_sock_container(&nst, sc);
> +
>  	veclen = caller_veclen + 1;
>  	vec = kmalloc(sizeof(struct kvec) * veclen, GFP_ATOMIC);
>  	if (vec == NULL) {
> @@ -972,6 +983,9 @@ int o2net_send_message_vec(u32 msg_type, u32 key,
struct kvec *caller_vec,
>  		goto out;
>  
>  	msg->msg_num = cpu_to_be32(nsw.ns_id);
> +	o2net_set_nst_msg_id(&nst, nsw.ns_id);
> +
> +	o2net_set_nst_send_time(&nst);
>  
>  	/* finally, convert the message header to network byte-order
>  	 * and send */
> @@ -986,6 +1000,7 @@ int o2net_send_message_vec(u32 msg_type, u32 key,
struct kvec *caller_vec,
>  	}
>  
>  	/* wait on other node's handler */
> +	o2net_set_nst_status_time(&nst);
>  	wait_event(nsw.ns_wq, o2net_nsw_completed(nn, &nsw));
>  
>  	/* Note that we avoid overwriting the callers status return
> @@ -998,6 +1013,7 @@ int o2net_send_message_vec(u32 msg_type, u32 key,
struct kvec *caller_vec,
>  	mlog(0, "woken, returning system status %d, user status %d\n",
>  	     ret, nsw.ns_status);
>  out:
> +	o2net_debug_del_nst(&nst); /* must be before dropping sc and node */
>  	if (sc)
>  		sc_put(sc);
>  	if (vec)
> @@ -1922,6 +1938,9 @@ int o2net_init(void)
>  
>  	o2quo_init();
>  
> +	if (o2net_debugfs_init())
> +		return -ENOMEM;
> +
>  	o2net_hand = kzalloc(sizeof(struct o2net_handshake), GFP_KERNEL);
>  	o2net_keep_req = kzalloc(sizeof(struct o2net_msg), GFP_KERNEL);
>  	o2net_keep_resp = kzalloc(sizeof(struct o2net_msg), GFP_KERNEL);
> @@ -1962,4 +1981,5 @@ void o2net_exit(void)
>  	kfree(o2net_hand);
>  	kfree(o2net_keep_req);
>  	kfree(o2net_keep_resp);
> +	o2net_debugfs_exit();
>  }
> diff --git a/fs/ocfs2/cluster/tcp.h b/fs/ocfs2/cluster/tcp.h
> index f36f66a..a705d5d 100644
> --- a/fs/ocfs2/cluster/tcp.h
> +++ b/fs/ocfs2/cluster/tcp.h
> @@ -117,4 +117,36 @@ int o2net_num_connected_peers(void);
>  int o2net_init(void);
>  void o2net_exit(void);
>  
> +struct o2net_send_tracking;
> +struct o2net_sock_container;
> +
> +#ifdef CONFIG_DEBUG_FS
> +int o2net_debugfs_init(void);
> +void o2net_debugfs_exit(void);
> +void o2net_debug_add_nst(struct o2net_send_tracking *nst);
> +void o2net_debug_del_nst(struct o2net_send_tracking *nst);
> +void o2net_debug_add_sc(struct o2net_sock_container *sc);
> +void o2net_debug_del_sc(struct o2net_sock_container *sc);
> +#else
> +static int o2net_debugfs_init(void)
> +{
> +	return 0;
> +}
> +static void o2net_debugfs_exit(void)
> +{
> +}
> +static void o2net_debug_add_nst(struct o2net_send_tracking *nst)
> +{
> +}
> +static void o2net_debug_del_nst(struct o2net_send_tracking *nst)
> +{
> +}
> +static void o2net_debug_add_sc(struct o2net_sock_container *sc)
> +{
> +}
> +static void o2net_debug_del_sc(struct o2net_sock_container *sc)
> +{
> +}
> +#endif	/* CONFIG_DEBUG_FS */
> +
>  #endif /* O2CLUSTER_TCP_H */
> diff --git a/fs/ocfs2/cluster/tcp_internal.h
b/fs/ocfs2/cluster/tcp_internal.h
> index d25b9af..68f2448 100644
> --- a/fs/ocfs2/cluster/tcp_internal.h
> +++ b/fs/ocfs2/cluster/tcp_internal.h
> @@ -164,7 +164,9 @@ struct o2net_sock_container {
>  	/* original handlers for the sockets */
>  	void			(*sc_state_change)(struct sock *sk);
>  	void			(*sc_data_ready)(struct sock *sk, int bytes);
> -
> +#ifdef CONFIG_DEBUG_FS
> +	struct list_head        sc_net_debug_item;
> +#endif
>  	struct timeval 		sc_tv_timer;
>  	struct timeval 		sc_tv_data_ready;
>  	struct timeval 		sc_tv_advance_start;
> @@ -206,4 +208,83 @@ struct o2net_status_wait {
>  	struct list_head	ns_node_item;
>  };
>  
> +#ifdef CONFIG_DEBUG_FS
> +/* just for state dumps */
> +struct o2net_send_tracking {
> +	struct list_head		st_net_debug_item;
> +	struct task_struct		*st_task;
> +	struct o2net_sock_container	*st_sc;
> +	u32				st_id;
> +	u32				st_msg_type;
> +	u32				st_msg_key;
> +	u8				st_node;
> +	struct timeval			st_sock_time;
> +	struct timeval			st_send_time;
> +	struct timeval			st_status_time;
> +};
> +
> +static void o2net_init_nst(struct o2net_send_tracking *nst, u32 msgtype,
> +			   u32 msgkey, struct task_struct *task, u8 node)
> +{
> +	INIT_LIST_HEAD(&nst->st_net_debug_item);
> +	nst->st_task = task;
> +	nst->st_msg_type = msgtype;
> +	nst->st_msg_key = msgkey;
> +	nst->st_node = node;
> +}
> +
> +static void o2net_set_nst_sock_time(struct o2net_send_tracking *nst)
> +{
> +	do_gettimeofday(&nst->st_sock_time);
> +}
> +
> +static void o2net_set_nst_send_time(struct o2net_send_tracking *nst)
> +{
> +	do_gettimeofday(&nst->st_send_time);
> +}
> +
> +static void o2net_set_nst_status_time(struct o2net_send_tracking *nst)
> +{
> +	do_gettimeofday(&nst->st_status_time);
> +}
> +
> +static void o2net_set_nst_sock_container(struct o2net_send_tracking *nst,
> +					 struct o2net_sock_container *sc)
> +{
> +	nst->st_sc = sc;
> +}
> +
> +static void o2net_set_nst_msg_id(struct o2net_send_tracking *nst, u32
msg_id)
> +{
> +	nst->st_id = msg_id;
> +}
> +
> +#else	/* CONFIG_DEBUG_FS */
> +
> +struct o2net_send_tracking {
> +	u32	dummy;
> +};
> +
> +static void o2net_init_nst(struct o2net_send_tracking *nst, u32 msgtype,
> +			   u32 msgkey, struct task_struct *task, u8 node)
> +{
> +}
> +static void o2net_set_nst_sock_time(struct o2net_send_tracking *nst)
> +{
> +}
> +static void o2net_set_nst_send_time(struct o2net_send_tracking *nst)
> +{
> +}
> +static void o2net_set_nst_status_time(struct o2net_send_tracking *nst)
> +{
> +}
> +static void o2net_set_nst_sock_container(struct o2net_send_tracking *nst,
> +					 struct o2net_sock_container *sc)
> +{
> +}
> +static void o2net_set_nst_msg_id(struct o2net_send_tracking *nst, u32
msg_id)
> +{
> +}
> +#endif	/* CONFIG_DEBUG_FS */
> +
>  #endif /* O2CLUSTER_TCP_INTERNAL_H */
> -- 
> 1.5.3.4
> 
> 
> _______________________________________________
> Ocfs2-devel mailing list
> Ocfs2-devel at oss.oracle.com
> http://oss.oracle.com/mailman/listinfo/ocfs2-devel
-- 

"The nearest approach to immortality on Earth is a government
 bureau."
	- James F. Byrnes

Joel Becker
Principal Software Developer
Oracle
E-mail: joel.becker at oracle.com
Phone: (650) 506-8127

On Wed, Apr 02, 2008 at 12:40:10PM -0700, Sunil Mushran
wrote:
> This patch exposes o2net information via debugfs. The information includes
> the list of sockets (sock_containers) as well as the list of outstanding
> messages (send_tracking). Useful for o2dlm debugging.
> 
> (This patch is derived from an earlier one written by Zach Brown that
> exposed the same information via /proc.)
> 
> Signed-off-by: Sunil Mushran <sunil.mushran at oracle.com>
Checkpatch gave me the following warnings:

ERROR: do not initialise statics to 0 or NULL
#91: FILE: fs/ocfs2/cluster/netdebug.c:50:
+static struct dentry *o2net_dentry = NULL;

ERROR: do not initialise statics to 0 or NULL
#92: FILE: fs/ocfs2/cluster/netdebug.c:51:
+static struct dentry *sc_dentry = NULL;

ERROR: do not initialise statics to 0 or NULL
#93: FILE: fs/ocfs2/cluster/netdebug.c:52:
+static struct dentry *nst_dentry = NULL;

ERROR: Use of SPIN_LOCK_UNLOCKED is deprecated: see Documentation/spinlocks.txt
#95: FILE: fs/ocfs2/cluster/netdebug.c:54:
+static spinlock_t o2net_debug_lock = SPIN_LOCK_UNLOCKED;

WARNING: line over 80 characters
#115: FILE: fs/ocfs2/cluster/netdebug.c:74:
+static struct o2net_send_tracking *next_nst(struct o2net_send_tracking
*nst_start)

WARNING: line over 80 characters
#269: FILE: fs/ocfs2/cluster/netdebug.c:228:
+static struct o2net_sock_container *next_sc(struct o2net_sock_container
*sc_start)

WARNING: line over 80 characters
#275: FILE: fs/ocfs2/cluster/netdebug.c:234:
+	list_for_each_entry(sc, &sc_start->sc_net_debug_item,
sc_net_debug_item) {

WARNING: line over 80 characters
#343: FILE: fs/ocfs2/cluster/netdebug.c:302:
+			   "  sock:            %u.%u.%u.%u:%u -> %u.%u.%u.%u:%u\n"

total: 4 errors, 4 warnings, 660 lines checked

0001-ocfs2-net-Add-debug-interface-to-o2net.patch has style problems, please
review.  If any of these errors
are false positives report them to the maintainer, see
CHECKPATCH in MAINTAINERS.


Attached is a quick re-work of the patch to address those which seemed
reasonable - all the ones about static initialization, the spinlock warning
and one of the over-80-columns warnings. Please test and let me know if it
still works with my changes so I can merge it. And next time, run
scripts/checkpatch.pl please ;)
	--Mark

--
Mark Fasheh


From: Sunil Mushran <sunil.mushran at oracle.com>

ocfs2/net: Add debug interface to o2net

This patch exposes o2net information via debugfs. The information includes
the list of sockets (sock_containers) as well as the list of outstanding
messages (send_tracking). Useful for o2dlm debugging.

(This patch is derived from an earlier one written by Zach Brown that
exposed the same information via /proc.)

[Mark: checkpatch fixes]

Signed-off-by: Sunil Mushran <sunil.mushran at oracle.com>
Reviewed-by: Joel Becker <joel.becker at oracle.com>
Signed-off-by: Mark Fasheh <mfasheh at suse.com>
---
 fs/ocfs2/cluster/Makefile       |    2 +-
 fs/ocfs2/cluster/netdebug.c     |  438 +++++++++++++++++++++++++++++++++++++++
 fs/ocfs2/cluster/nodemanager.c  |    5 +-
 fs/ocfs2/cluster/tcp.c          |   20 ++
 fs/ocfs2/cluster/tcp.h          |   32 +++
 fs/ocfs2/cluster/tcp_internal.h |   83 ++++++++-
 6 files changed, 577 insertions(+), 3 deletions(-)
 create mode 100644 fs/ocfs2/cluster/netdebug.c

diff --git a/fs/ocfs2/cluster/Makefile b/fs/ocfs2/cluster/Makefile
index cdd162f..bc8c5e7 100644
--- a/fs/ocfs2/cluster/Makefile
+++ b/fs/ocfs2/cluster/Makefile
@@ -1,4 +1,4 @@
 obj-$(CONFIG_OCFS2_FS) += ocfs2_nodemanager.o
 
 ocfs2_nodemanager-objs := heartbeat.o masklog.o sys.o nodemanager.o \
-	quorum.o tcp.o ver.o
+	quorum.o tcp.o netdebug.o ver.o
diff --git a/fs/ocfs2/cluster/netdebug.c b/fs/ocfs2/cluster/netdebug.c
new file mode 100644
index 0000000..a7b3827
--- /dev/null
+++ b/fs/ocfs2/cluster/netdebug.c
@@ -0,0 +1,438 @@
+/* -*- mode: c; c-basic-offset: 8; -*-
+ * vim: noexpandtab sw=8 ts=8 sts=0:
+ *
+ * netdebug.c
+ *
+ * debug functionality for o2net
+ *
+ * Copyright (C) 2005, 2008 Oracle.  All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public
+ * License as published by the Free Software Foundation; either
+ * version 2 of the License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public
+ * License along with this program; if not, write to the
+ * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+ * Boston, MA 021110-1307, USA.
+ *
+ */
+
+#ifdef CONFIG_DEBUG_FS
+
+#include <linux/module.h>
+#include <linux/types.h>
+#include <linux/slab.h>
+#include <linux/idr.h>
+#include <linux/kref.h>
+#include <linux/seq_file.h>
+#include <linux/debugfs.h>
+
+#include <asm/uaccess.h>
+
+#include "tcp.h"
+#include "nodemanager.h"
+#define MLOG_MASK_PREFIX ML_TCP
+#include "masklog.h"
+
+#include "tcp_internal.h"
+
+#define O2NET_DEBUG_DIR		"o2net"
+#define SC_DEBUG_NAME		"sock_containers"
+#define NST_DEBUG_NAME		"send_tracking"
+
+static struct dentry *o2net_dentry;
+static struct dentry *sc_dentry;
+static struct dentry *nst_dentry;
+
+static DEFINE_SPINLOCK(o2net_debug_lock);
+
+static LIST_HEAD(sock_containers);
+static LIST_HEAD(send_tracking);
+
+void o2net_debug_add_nst(struct o2net_send_tracking *nst)
+{
+	spin_lock(&o2net_debug_lock);
+	list_add(&nst->st_net_debug_item, &send_tracking);
+	spin_unlock(&o2net_debug_lock);
+}
+
+void o2net_debug_del_nst(struct o2net_send_tracking *nst)
+{
+	spin_lock(&o2net_debug_lock);
+	if (!list_empty(&nst->st_net_debug_item))
+		list_del_init(&nst->st_net_debug_item);
+	spin_unlock(&o2net_debug_lock);
+}
+
+static struct o2net_send_tracking *next_nst(struct o2net_send_tracking
*nst_start)
+{
+	struct o2net_send_tracking *nst, *ret = NULL;
+
+	assert_spin_locked(&o2net_debug_lock);
+
+	list_for_each_entry(nst, &nst_start->st_net_debug_item,
+			    st_net_debug_item) {
+		/* discover the head of the list */
+		if (&nst->st_net_debug_item == &send_tracking)
+			break;
+
+		/* use st_task to detect real nsts in the list */
+		if (nst->st_task != NULL) {
+			ret = nst;
+			break;
+		}
+	}
+
+	return ret;
+}
+
+static void *nst_seq_start(struct seq_file *seq, loff_t *pos)
+{
+	struct o2net_send_tracking *nst, *dummy_nst = seq->private;
+
+	spin_lock(&o2net_debug_lock);
+	nst = next_nst(dummy_nst);
+	spin_unlock(&o2net_debug_lock);
+
+	return nst;
+}
+
+static void *nst_seq_next(struct seq_file *seq, void *v, loff_t *pos)
+{
+	struct o2net_send_tracking *nst, *dummy_nst = seq->private;
+
+	spin_lock(&o2net_debug_lock);
+	nst = next_nst(dummy_nst);
+	list_del_init(&dummy_nst->st_net_debug_item);
+	if (nst)
+		list_add(&dummy_nst->st_net_debug_item,
+			 &nst->st_net_debug_item);
+	spin_unlock(&o2net_debug_lock);
+
+	return nst; /* unused, just needs to be null when done */
+}
+
+static int nst_seq_show(struct seq_file *seq, void *v)
+{
+	struct o2net_send_tracking *nst, *dummy_nst = seq->private;
+
+	spin_lock(&o2net_debug_lock);
+	nst = next_nst(dummy_nst);
+
+	if (nst != NULL) {
+		/* get_task_comm isn't exported.  oh well. */
+		seq_printf(seq, "%p:\n"
+			   "  pid:          %lu\n"
+			   "  tgid:         %lu\n"
+			   "  process name: %s\n"
+			   "  node:         %u\n"
+			   "  sc:           %p\n"
+			   "  message id:   %d\n"
+			   "  message type: %u\n"
+			   "  message key:  0x%08x\n"
+			   "  sock acquiry: %lu.%lu\n"
+			   "  send start:   %lu.%lu\n"
+			   "  wait start:   %lu.%lu\n",
+			   nst, (unsigned long)nst->st_task->pid,
+			   (unsigned long)nst->st_task->tgid,
+			   nst->st_task->comm, nst->st_node,
+			   nst->st_sc, nst->st_id, nst->st_msg_type,
+			   nst->st_msg_key,
+			   nst->st_sock_time.tv_sec, nst->st_sock_time.tv_usec,
+			   nst->st_send_time.tv_sec, nst->st_send_time.tv_usec,
+			   nst->st_status_time.tv_sec,
+			   nst->st_status_time.tv_usec);
+	}
+
+	spin_unlock(&o2net_debug_lock);
+
+	return 0;
+}
+
+static void nst_seq_stop(struct seq_file *seq, void *v)
+{
+}
+
+static struct seq_operations nst_seq_ops = {
+	.start = nst_seq_start,
+	.next = nst_seq_next,
+	.stop = nst_seq_stop,
+	.show = nst_seq_show,
+};
+
+static int nst_fop_open(struct inode *inode, struct file *file)
+{
+	struct o2net_send_tracking *dummy_nst;
+	struct seq_file *seq;
+	int ret;
+
+	dummy_nst = kmalloc(sizeof(struct o2net_send_tracking), GFP_KERNEL);
+	if (dummy_nst == NULL) {
+		ret = -ENOMEM;
+		goto out;
+	}
+	dummy_nst->st_task = NULL;
+
+	ret = seq_open(file, &nst_seq_ops);
+	if (ret)
+		goto out;
+
+	seq = file->private_data;
+	seq->private = dummy_nst;
+	o2net_debug_add_nst(dummy_nst);
+
+	dummy_nst = NULL;
+
+out:
+	kfree(dummy_nst);
+	return ret;
+}
+
+static int nst_fop_release(struct inode *inode, struct file *file)
+{
+	struct seq_file *seq = file->private_data;
+	struct o2net_send_tracking *dummy_nst = seq->private;
+
+	o2net_debug_del_nst(dummy_nst);
+	return seq_release_private(inode, file);
+}
+
+static struct file_operations nst_seq_fops = {
+	.open = nst_fop_open,
+	.read = seq_read,
+	.llseek = seq_lseek,
+	.release = nst_fop_release,
+};
+
+void o2net_debug_add_sc(struct o2net_sock_container *sc)
+{
+	spin_lock(&o2net_debug_lock);
+	list_add(&sc->sc_net_debug_item, &sock_containers);
+	spin_unlock(&o2net_debug_lock);
+}
+
+void o2net_debug_del_sc(struct o2net_sock_container *sc)
+{
+	spin_lock(&o2net_debug_lock);
+	list_del_init(&sc->sc_net_debug_item);
+	spin_unlock(&o2net_debug_lock);
+}
+
+static struct o2net_sock_container *next_sc(struct o2net_sock_container
*sc_start)
+{
+	struct o2net_sock_container *sc, *ret = NULL;
+
+	assert_spin_locked(&o2net_debug_lock);
+
+	list_for_each_entry(sc, &sc_start->sc_net_debug_item,
+			    sc_net_debug_item) {
+		/* discover the head of the list miscast as a sc */
+		if (&sc->sc_net_debug_item == &sock_containers)
+			break;
+
+		/* use sc_page to detect real scs in the list */
+		if (sc->sc_page != NULL) {
+			ret = sc;
+			break;
+		}
+	}
+
+	return ret;
+}
+
+static void *sc_seq_start(struct seq_file *seq, loff_t *pos)
+{
+	struct o2net_sock_container *sc, *dummy_sc = seq->private;
+
+	spin_lock(&o2net_debug_lock);
+	sc = next_sc(dummy_sc);
+	spin_unlock(&o2net_debug_lock);
+
+	return sc;
+}
+
+static void *sc_seq_next(struct seq_file *seq, void *v, loff_t *pos)
+{
+	struct o2net_sock_container *sc, *dummy_sc = seq->private;
+
+	spin_lock(&o2net_debug_lock);
+	sc = next_sc(dummy_sc);
+	list_del_init(&dummy_sc->sc_net_debug_item);
+	if (sc)
+		list_add(&dummy_sc->sc_net_debug_item, &sc->sc_net_debug_item);
+	spin_unlock(&o2net_debug_lock);
+
+	return sc; /* unused, just needs to be null when done */
+}
+
+#define TV_SEC_USEC(TV) TV.tv_sec, TV.tv_usec
+
+static int sc_seq_show(struct seq_file *seq, void *v)
+{
+	struct o2net_sock_container *sc, *dummy_sc = seq->private;
+
+	spin_lock(&o2net_debug_lock);
+	sc = next_sc(dummy_sc);
+
+	if (sc != NULL) {
+		struct inet_sock *inet = NULL;
+
+		__be32 saddr = 0, daddr = 0;
+		__be16 sport = 0, dport = 0;
+
+		if (sc->sc_sock) {
+			inet = inet_sk(sc->sc_sock->sk);
+			/* the stack's structs aren't sparse endian clean */
+			saddr = (__force __be32)inet->saddr;
+			daddr = (__force __be32)inet->daddr;
+			sport = (__force __be16)inet->sport;
+			dport = (__force __be16)inet->dport;
+		}
+
+		/* XXX sigh, inet-> doesn't have sparse annotation so any
+		 * use of it here generates a warning with -Wbitwise */
+		seq_printf(seq, "%p:\n"
+			   "  krefs:           %d\n"
+			   "  sock:            %u.%u.%u.%u:%u -> %u.%u.%u.%u:%u\n"
+			   "  remote node:     %s\n"
+			   "  page off:        %zu\n"
+			   "  handshake ok:    %u\n"
+			   "  timer:           %lu.%lu\n"
+			   "  data ready:      %lu.%lu\n"
+			   "  advance start:   %lu.%lu\n"
+			   "  advance stop:    %lu.%lu\n"
+			   "  func start:      %lu.%lu\n"
+			   "  func stop:       %lu.%lu\n"
+			   "  func key:        %u\n"
+			   "  func type:       %u\n",
+			   sc,
+			   atomic_read(&sc->sc_kref.refcount),
+			   NIPQUAD(saddr), inet ? ntohs(sport) : 0,
+			   NIPQUAD(daddr), inet ? ntohs(dport) : 0,
+			   sc->sc_node->nd_name,
+			   sc->sc_page_off,
+			   sc->sc_handshake_ok,
+			   TV_SEC_USEC(sc->sc_tv_timer),
+			   TV_SEC_USEC(sc->sc_tv_data_ready),
+			   TV_SEC_USEC(sc->sc_tv_advance_start),
+			   TV_SEC_USEC(sc->sc_tv_advance_stop),
+			   TV_SEC_USEC(sc->sc_tv_func_start),
+			   TV_SEC_USEC(sc->sc_tv_func_stop),
+			   sc->sc_msg_key,
+			   sc->sc_msg_type);
+	}
+
+
+	spin_unlock(&o2net_debug_lock);
+
+	return 0;
+}
+
+static void sc_seq_stop(struct seq_file *seq, void *v)
+{
+}
+
+static struct seq_operations sc_seq_ops = {
+	.start = sc_seq_start,
+	.next = sc_seq_next,
+	.stop = sc_seq_stop,
+	.show = sc_seq_show,
+};
+
+static int sc_fop_open(struct inode *inode, struct file *file)
+{
+	struct o2net_sock_container *dummy_sc;
+	struct seq_file *seq;
+	int ret;
+
+	dummy_sc = kmalloc(sizeof(struct o2net_sock_container), GFP_KERNEL);
+	if (dummy_sc == NULL) {
+		ret = -ENOMEM;
+		goto out;
+	}
+	dummy_sc->sc_page = NULL;
+
+	ret = seq_open(file, &sc_seq_ops);
+	if (ret)
+		goto out;
+
+	seq = file->private_data;
+	seq->private = dummy_sc;
+	o2net_debug_add_sc(dummy_sc);
+
+	dummy_sc = NULL;
+
+out:
+	kfree(dummy_sc);
+	return ret;
+}
+
+static int sc_fop_release(struct inode *inode, struct file *file)
+{
+	struct seq_file *seq = file->private_data;
+	struct o2net_sock_container *dummy_sc = seq->private;
+
+	o2net_debug_del_sc(dummy_sc);
+	return seq_release_private(inode, file);
+}
+
+static struct file_operations sc_seq_fops = {
+	.open = sc_fop_open,
+	.read = seq_read,
+	.llseek = seq_lseek,
+	.release = sc_fop_release,
+};
+
+int o2net_debugfs_init(void)
+{
+	o2net_dentry = debugfs_create_dir(O2NET_DEBUG_DIR, NULL);
+	if (!o2net_dentry) {
+		mlog_errno(-ENOMEM);
+		goto bail;
+	}
+
+	nst_dentry = debugfs_create_file(NST_DEBUG_NAME, S_IFREG|S_IRUSR,
+					 o2net_dentry, NULL,
+					 &nst_seq_fops);
+	if (!nst_dentry) {
+		mlog_errno(-ENOMEM);
+		goto bail;
+	}
+
+	sc_dentry = debugfs_create_file(SC_DEBUG_NAME, S_IFREG|S_IRUSR,
+					o2net_dentry, NULL,
+					&sc_seq_fops);
+	if (!sc_dentry) {
+		mlog_errno(-ENOMEM);
+		goto bail;
+	}
+
+	return 0;
+bail:
+	if (sc_dentry)
+		debugfs_remove(sc_dentry);
+	if (nst_dentry)
+		debugfs_remove(nst_dentry);
+	if (o2net_dentry)
+		debugfs_remove(o2net_dentry);
+	return -ENOMEM;
+}
+
+void o2net_debugfs_exit(void)
+{
+	if (sc_dentry)
+		debugfs_remove(sc_dentry);
+	if (nst_dentry)
+		debugfs_remove(nst_dentry);
+	if (o2net_dentry)
+		debugfs_remove(o2net_dentry);
+}
+
+#endif	/* CONFIG_DEBUG_FS */
diff --git a/fs/ocfs2/cluster/nodemanager.c b/fs/ocfs2/cluster/nodemanager.c
index 709fba2..cf9401e 100644
--- a/fs/ocfs2/cluster/nodemanager.c
+++ b/fs/ocfs2/cluster/nodemanager.c
@@ -959,7 +959,10 @@ static int __init init_o2nm(void)
 	cluster_print_version();
 
 	o2hb_init();
-	o2net_init();
+
+	ret = o2net_init();
+	if (ret)
+		goto out;
 
 	ocfs2_table_header = register_sysctl_table(ocfs2_root_table);
 	if (!ocfs2_table_header) {
diff --git a/fs/ocfs2/cluster/tcp.c b/fs/ocfs2/cluster/tcp.c
index b8057c5..c2af3c4 100644
--- a/fs/ocfs2/cluster/tcp.c
+++ b/fs/ocfs2/cluster/tcp.c
@@ -296,6 +296,7 @@ static void sc_kref_release(struct kref *kref)
 	o2nm_node_put(sc->sc_node);
 	sc->sc_node = NULL;
 
+	o2net_debug_del_sc(sc);
 	kfree(sc);
 }
 
@@ -336,6 +337,7 @@ static struct o2net_sock_container *sc_alloc(struct
o2nm_node *node)
 
 	ret = sc;
 	sc->sc_page = page;
+	o2net_debug_add_sc(sc);
 	sc = NULL;
 	page = NULL;
 
@@ -914,6 +916,9 @@ int o2net_send_message_vec(u32 msg_type, u32 key, struct
kvec *caller_vec,
 	struct o2net_status_wait nsw = {
 		.ns_node_item = LIST_HEAD_INIT(nsw.ns_node_item),
 	};
+	struct o2net_send_tracking nst;
+
+	o2net_init_nst(&nst, msg_type, key, current, target_node);
 
 	if (o2net_wq == NULL) {
 		mlog(0, "attempt to tx without o2netd running\n");
@@ -939,6 +944,10 @@ int o2net_send_message_vec(u32 msg_type, u32 key, struct
kvec *caller_vec,
 		goto out;
 	}
 
+	o2net_debug_add_nst(&nst);
+
+	o2net_set_nst_sock_time(&nst);
+
 	ret = wait_event_interruptible(nn->nn_sc_wq,
 				       o2net_tx_can_proceed(nn, &sc, &error));
 	if (!ret && error)
@@ -946,6 +955,8 @@ int o2net_send_message_vec(u32 msg_type, u32 key, struct
kvec *caller_vec,
 	if (ret)
 		goto out;
 
+	o2net_set_nst_sock_container(&nst, sc);
+
 	veclen = caller_veclen + 1;
 	vec = kmalloc(sizeof(struct kvec) * veclen, GFP_ATOMIC);
 	if (vec == NULL) {
@@ -972,6 +983,9 @@ int o2net_send_message_vec(u32 msg_type, u32 key, struct
kvec *caller_vec,
 		goto out;
 
 	msg->msg_num = cpu_to_be32(nsw.ns_id);
+	o2net_set_nst_msg_id(&nst, nsw.ns_id);
+
+	o2net_set_nst_send_time(&nst);
 
 	/* finally, convert the message header to network byte-order
 	 * and send */
@@ -986,6 +1000,7 @@ int o2net_send_message_vec(u32 msg_type, u32 key, struct
kvec *caller_vec,
 	}
 
 	/* wait on other node's handler */
+	o2net_set_nst_status_time(&nst);
 	wait_event(nsw.ns_wq, o2net_nsw_completed(nn, &nsw));
 
 	/* Note that we avoid overwriting the callers status return
@@ -998,6 +1013,7 @@ int o2net_send_message_vec(u32 msg_type, u32 key, struct
kvec *caller_vec,
 	mlog(0, "woken, returning system status %d, user status %d\n",
 	     ret, nsw.ns_status);
 out:
+	o2net_debug_del_nst(&nst); /* must be before dropping sc and node */
 	if (sc)
 		sc_put(sc);
 	if (vec)
@@ -1922,6 +1938,9 @@ int o2net_init(void)
 
 	o2quo_init();
 
+	if (o2net_debugfs_init())
+		return -ENOMEM;
+
 	o2net_hand = kzalloc(sizeof(struct o2net_handshake), GFP_KERNEL);
 	o2net_keep_req = kzalloc(sizeof(struct o2net_msg), GFP_KERNEL);
 	o2net_keep_resp = kzalloc(sizeof(struct o2net_msg), GFP_KERNEL);
@@ -1962,4 +1981,5 @@ void o2net_exit(void)
 	kfree(o2net_hand);
 	kfree(o2net_keep_req);
 	kfree(o2net_keep_resp);
+	o2net_debugfs_exit();
 }
diff --git a/fs/ocfs2/cluster/tcp.h b/fs/ocfs2/cluster/tcp.h
index f36f66a..a705d5d 100644
--- a/fs/ocfs2/cluster/tcp.h
+++ b/fs/ocfs2/cluster/tcp.h
@@ -117,4 +117,36 @@ int o2net_num_connected_peers(void);
 int o2net_init(void);
 void o2net_exit(void);
 
+struct o2net_send_tracking;
+struct o2net_sock_container;
+
+#ifdef CONFIG_DEBUG_FS
+int o2net_debugfs_init(void);
+void o2net_debugfs_exit(void);
+void o2net_debug_add_nst(struct o2net_send_tracking *nst);
+void o2net_debug_del_nst(struct o2net_send_tracking *nst);
+void o2net_debug_add_sc(struct o2net_sock_container *sc);
+void o2net_debug_del_sc(struct o2net_sock_container *sc);
+#else
+static int o2net_debugfs_init(void)
+{
+	return 0;
+}
+static void o2net_debugfs_exit(void)
+{
+}
+static void o2net_debug_add_nst(struct o2net_send_tracking *nst)
+{
+}
+static void o2net_debug_del_nst(struct o2net_send_tracking *nst)
+{
+}
+static void o2net_debug_add_sc(struct o2net_sock_container *sc)
+{
+}
+static void o2net_debug_del_sc(struct o2net_sock_container *sc)
+{
+}
+#endif	/* CONFIG_DEBUG_FS */
+
 #endif /* O2CLUSTER_TCP_H */
diff --git a/fs/ocfs2/cluster/tcp_internal.h b/fs/ocfs2/cluster/tcp_internal.h
index d25b9af..68f2448 100644
--- a/fs/ocfs2/cluster/tcp_internal.h
+++ b/fs/ocfs2/cluster/tcp_internal.h
@@ -164,7 +164,9 @@ struct o2net_sock_container {
 	/* original handlers for the sockets */
 	void			(*sc_state_change)(struct sock *sk);
 	void			(*sc_data_ready)(struct sock *sk, int bytes);
-
+#ifdef CONFIG_DEBUG_FS
+	struct list_head        sc_net_debug_item;
+#endif
 	struct timeval 		sc_tv_timer;
 	struct timeval 		sc_tv_data_ready;
 	struct timeval 		sc_tv_advance_start;
@@ -206,4 +208,83 @@ struct o2net_status_wait {
 	struct list_head	ns_node_item;
 };
 
+#ifdef CONFIG_DEBUG_FS
+/* just for state dumps */
+struct o2net_send_tracking {
+	struct list_head		st_net_debug_item;
+	struct task_struct		*st_task;
+	struct o2net_sock_container	*st_sc;
+	u32				st_id;
+	u32				st_msg_type;
+	u32				st_msg_key;
+	u8				st_node;
+	struct timeval			st_sock_time;
+	struct timeval			st_send_time;
+	struct timeval			st_status_time;
+};
+
+static void o2net_init_nst(struct o2net_send_tracking *nst, u32 msgtype,
+			   u32 msgkey, struct task_struct *task, u8 node)
+{
+	INIT_LIST_HEAD(&nst->st_net_debug_item);
+	nst->st_task = task;
+	nst->st_msg_type = msgtype;
+	nst->st_msg_key = msgkey;
+	nst->st_node = node;
+}
+
+static void o2net_set_nst_sock_time(struct o2net_send_tracking *nst)
+{
+	do_gettimeofday(&nst->st_sock_time);
+}
+
+static void o2net_set_nst_send_time(struct o2net_send_tracking *nst)
+{
+	do_gettimeofday(&nst->st_send_time);
+}
+
+static void o2net_set_nst_status_time(struct o2net_send_tracking *nst)
+{
+	do_gettimeofday(&nst->st_status_time);
+}
+
+static void o2net_set_nst_sock_container(struct o2net_send_tracking *nst,
+					 struct o2net_sock_container *sc)
+{
+	nst->st_sc = sc;
+}
+
+static void o2net_set_nst_msg_id(struct o2net_send_tracking *nst, u32 msg_id)
+{
+	nst->st_id = msg_id;
+}
+
+#else	/* CONFIG_DEBUG_FS */
+
+struct o2net_send_tracking {
+	u32	dummy;
+};
+
+static void o2net_init_nst(struct o2net_send_tracking *nst, u32 msgtype,
+			   u32 msgkey, struct task_struct *task, u8 node)
+{
+}
+static void o2net_set_nst_sock_time(struct o2net_send_tracking *nst)
+{
+}
+static void o2net_set_nst_send_time(struct o2net_send_tracking *nst)
+{
+}
+static void o2net_set_nst_status_time(struct o2net_send_tracking *nst)
+{
+}
+static void o2net_set_nst_sock_container(struct o2net_send_tracking *nst,
+					 struct o2net_sock_container *sc)
+{
+}
+static void o2net_set_nst_msg_id(struct o2net_send_tracking *nst, u32 msg_id)
+{
+}
+#endif	/* CONFIG_DEBUG_FS */
+
 #endif /* O2CLUSTER_TCP_INTERNAL_H */
-- 
1.5.4.1

This patch exposes o2net information via debugfs. The information includes
the list of sockets (sock_containers) as well as the list of outstanding
messages (send_tracking). Useful for o2dlm debugging.

(This patch is derived from an earlier one written by Zach Brown that
exposed the same information via /proc.)

[Mark: checkpatch fixes]

Signed-off-by: Sunil Mushran <sunil.mushran at oracle.com>
Reviewed-by: Joel Becker <joel.becker at oracle.com>
Signed-off-by: Mark Fasheh <mfasheh at suse.com>
---
 fs/ocfs2/cluster/Makefile       |    2 +-
 fs/ocfs2/cluster/netdebug.c     |  441 +++++++++++++++++++++++++++++++++++++++
 fs/ocfs2/cluster/nodemanager.c  |    5 +-
 fs/ocfs2/cluster/tcp.c          |   68 ++++++
 fs/ocfs2/cluster/tcp.h          |   32 +++
 fs/ocfs2/cluster/tcp_internal.h |   24 ++-
 6 files changed, 569 insertions(+), 3 deletions(-)
 create mode 100644 fs/ocfs2/cluster/netdebug.c

diff --git a/fs/ocfs2/cluster/Makefile b/fs/ocfs2/cluster/Makefile
index cdd162f..bc8c5e7 100644
--- a/fs/ocfs2/cluster/Makefile
+++ b/fs/ocfs2/cluster/Makefile
@@ -1,4 +1,4 @@
 obj-$(CONFIG_OCFS2_FS) += ocfs2_nodemanager.o
 
 ocfs2_nodemanager-objs := heartbeat.o masklog.o sys.o nodemanager.o \
-	quorum.o tcp.o ver.o
+	quorum.o tcp.o netdebug.o ver.o
diff --git a/fs/ocfs2/cluster/netdebug.c b/fs/ocfs2/cluster/netdebug.c
new file mode 100644
index 0000000..7bf3c0e
--- /dev/null
+++ b/fs/ocfs2/cluster/netdebug.c
@@ -0,0 +1,441 @@
+/* -*- mode: c; c-basic-offset: 8; -*-
+ * vim: noexpandtab sw=8 ts=8 sts=0:
+ *
+ * netdebug.c
+ *
+ * debug functionality for o2net
+ *
+ * Copyright (C) 2005, 2008 Oracle.  All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public
+ * License as published by the Free Software Foundation; either
+ * version 2 of the License, or (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public
+ * License along with this program; if not, write to the
+ * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+ * Boston, MA 021110-1307, USA.
+ *
+ */
+
+#ifdef CONFIG_DEBUG_FS
+
+#include <linux/module.h>
+#include <linux/types.h>
+#include <linux/slab.h>
+#include <linux/idr.h>
+#include <linux/kref.h>
+#include <linux/seq_file.h>
+#include <linux/debugfs.h>
+
+#include <linux/uaccess.h>
+
+#include "tcp.h"
+#include "nodemanager.h"
+#define MLOG_MASK_PREFIX ML_TCP
+#include "masklog.h"
+
+#include "tcp_internal.h"
+
+#define O2NET_DEBUG_DIR		"o2net"
+#define SC_DEBUG_NAME		"sock_containers"
+#define NST_DEBUG_NAME		"send_tracking"
+
+static struct dentry *o2net_dentry;
+static struct dentry *sc_dentry;
+static struct dentry *nst_dentry;
+
+static DEFINE_SPINLOCK(o2net_debug_lock);
+
+static LIST_HEAD(sock_containers);
+static LIST_HEAD(send_tracking);
+
+void o2net_debug_add_nst(struct o2net_send_tracking *nst)
+{
+	spin_lock(&o2net_debug_lock);
+	list_add(&nst->st_net_debug_item, &send_tracking);
+	spin_unlock(&o2net_debug_lock);
+}
+
+void o2net_debug_del_nst(struct o2net_send_tracking *nst)
+{
+	spin_lock(&o2net_debug_lock);
+	if (!list_empty(&nst->st_net_debug_item))
+		list_del_init(&nst->st_net_debug_item);
+	spin_unlock(&o2net_debug_lock);
+}
+
+static struct o2net_send_tracking
+			*next_nst(struct o2net_send_tracking *nst_start)
+{
+	struct o2net_send_tracking *nst, *ret = NULL;
+
+	assert_spin_locked(&o2net_debug_lock);
+
+	list_for_each_entry(nst, &nst_start->st_net_debug_item,
+			    st_net_debug_item) {
+		/* discover the head of the list */
+		if (&nst->st_net_debug_item == &send_tracking)
+			break;
+
+		/* use st_task to detect real nsts in the list */
+		if (nst->st_task != NULL) {
+			ret = nst;
+			break;
+		}
+	}
+
+	return ret;
+}
+
+static void *nst_seq_start(struct seq_file *seq, loff_t *pos)
+{
+	struct o2net_send_tracking *nst, *dummy_nst = seq->private;
+
+	spin_lock(&o2net_debug_lock);
+	nst = next_nst(dummy_nst);
+	spin_unlock(&o2net_debug_lock);
+
+	return nst;
+}
+
+static void *nst_seq_next(struct seq_file *seq, void *v, loff_t *pos)
+{
+	struct o2net_send_tracking *nst, *dummy_nst = seq->private;
+
+	spin_lock(&o2net_debug_lock);
+	nst = next_nst(dummy_nst);
+	list_del_init(&dummy_nst->st_net_debug_item);
+	if (nst)
+		list_add(&dummy_nst->st_net_debug_item,
+			 &nst->st_net_debug_item);
+	spin_unlock(&o2net_debug_lock);
+
+	return nst; /* unused, just needs to be null when done */
+}
+
+static int nst_seq_show(struct seq_file *seq, void *v)
+{
+	struct o2net_send_tracking *nst, *dummy_nst = seq->private;
+
+	spin_lock(&o2net_debug_lock);
+	nst = next_nst(dummy_nst);
+
+	if (nst != NULL) {
+		/* get_task_comm isn't exported.  oh well. */
+		seq_printf(seq, "%p:\n"
+			   "  pid:          %lu\n"
+			   "  tgid:         %lu\n"
+			   "  process name: %s\n"
+			   "  node:         %u\n"
+			   "  sc:           %p\n"
+			   "  message id:   %d\n"
+			   "  message type: %u\n"
+			   "  message key:  0x%08x\n"
+			   "  sock acquiry: %lu.%lu\n"
+			   "  send start:   %lu.%lu\n"
+			   "  wait start:   %lu.%lu\n",
+			   nst, (unsigned long)nst->st_task->pid,
+			   (unsigned long)nst->st_task->tgid,
+			   nst->st_task->comm, nst->st_node,
+			   nst->st_sc, nst->st_id, nst->st_msg_type,
+			   nst->st_msg_key,
+			   nst->st_sock_time.tv_sec, nst->st_sock_time.tv_usec,
+			   nst->st_send_time.tv_sec, nst->st_send_time.tv_usec,
+			   nst->st_status_time.tv_sec,
+			   nst->st_status_time.tv_usec);
+	}
+
+	spin_unlock(&o2net_debug_lock);
+
+	return 0;
+}
+
+static void nst_seq_stop(struct seq_file *seq, void *v)
+{
+}
+
+static struct seq_operations nst_seq_ops = {
+	.start = nst_seq_start,
+	.next = nst_seq_next,
+	.stop = nst_seq_stop,
+	.show = nst_seq_show,
+};
+
+static int nst_fop_open(struct inode *inode, struct file *file)
+{
+	struct o2net_send_tracking *dummy_nst;
+	struct seq_file *seq;
+	int ret;
+
+	dummy_nst = kmalloc(sizeof(struct o2net_send_tracking), GFP_KERNEL);
+	if (dummy_nst == NULL) {
+		ret = -ENOMEM;
+		goto out;
+	}
+	dummy_nst->st_task = NULL;
+
+	ret = seq_open(file, &nst_seq_ops);
+	if (ret)
+		goto out;
+
+	seq = file->private_data;
+	seq->private = dummy_nst;
+	o2net_debug_add_nst(dummy_nst);
+
+	dummy_nst = NULL;
+
+out:
+	kfree(dummy_nst);
+	return ret;
+}
+
+static int nst_fop_release(struct inode *inode, struct file *file)
+{
+	struct seq_file *seq = file->private_data;
+	struct o2net_send_tracking *dummy_nst = seq->private;
+
+	o2net_debug_del_nst(dummy_nst);
+	return seq_release_private(inode, file);
+}
+
+static struct file_operations nst_seq_fops = {
+	.open = nst_fop_open,
+	.read = seq_read,
+	.llseek = seq_lseek,
+	.release = nst_fop_release,
+};
+
+void o2net_debug_add_sc(struct o2net_sock_container *sc)
+{
+	spin_lock(&o2net_debug_lock);
+	list_add(&sc->sc_net_debug_item, &sock_containers);
+	spin_unlock(&o2net_debug_lock);
+}
+
+void o2net_debug_del_sc(struct o2net_sock_container *sc)
+{
+	spin_lock(&o2net_debug_lock);
+	list_del_init(&sc->sc_net_debug_item);
+	spin_unlock(&o2net_debug_lock);
+}
+
+static struct o2net_sock_container
+			*next_sc(struct o2net_sock_container *sc_start)
+{
+	struct o2net_sock_container *sc, *ret = NULL;
+
+	assert_spin_locked(&o2net_debug_lock);
+
+	list_for_each_entry(sc, &sc_start->sc_net_debug_item,
+			    sc_net_debug_item) {
+		/* discover the head of the list miscast as a sc */
+		if (&sc->sc_net_debug_item == &sock_containers)
+			break;
+
+		/* use sc_page to detect real scs in the list */
+		if (sc->sc_page != NULL) {
+			ret = sc;
+			break;
+		}
+	}
+
+	return ret;
+}
+
+static void *sc_seq_start(struct seq_file *seq, loff_t *pos)
+{
+	struct o2net_sock_container *sc, *dummy_sc = seq->private;
+
+	spin_lock(&o2net_debug_lock);
+	sc = next_sc(dummy_sc);
+	spin_unlock(&o2net_debug_lock);
+
+	return sc;
+}
+
+static void *sc_seq_next(struct seq_file *seq, void *v, loff_t *pos)
+{
+	struct o2net_sock_container *sc, *dummy_sc = seq->private;
+
+	spin_lock(&o2net_debug_lock);
+	sc = next_sc(dummy_sc);
+	list_del_init(&dummy_sc->sc_net_debug_item);
+	if (sc)
+		list_add(&dummy_sc->sc_net_debug_item, &sc->sc_net_debug_item);
+	spin_unlock(&o2net_debug_lock);
+
+	return sc; /* unused, just needs to be null when done */
+}
+
+#define TV_SEC_USEC(TV) TV.tv_sec, TV.tv_usec
+
+static int sc_seq_show(struct seq_file *seq, void *v)
+{
+	struct o2net_sock_container *sc, *dummy_sc = seq->private;
+
+	spin_lock(&o2net_debug_lock);
+	sc = next_sc(dummy_sc);
+
+	if (sc != NULL) {
+		struct inet_sock *inet = NULL;
+
+		__be32 saddr = 0, daddr = 0;
+		__be16 sport = 0, dport = 0;
+
+		if (sc->sc_sock) {
+			inet = inet_sk(sc->sc_sock->sk);
+			/* the stack's structs aren't sparse endian clean */
+			saddr = (__force __be32)inet->saddr;
+			daddr = (__force __be32)inet->daddr;
+			sport = (__force __be16)inet->sport;
+			dport = (__force __be16)inet->dport;
+		}
+
+		/* XXX sigh, inet-> doesn't have sparse annotation so any
+		 * use of it here generates a warning with -Wbitwise */
+		seq_printf(seq, "%p:\n"
+			   "  krefs:           %d\n"
+			   "  sock:            %u.%u.%u.%u:%u -> "
+					      "%u.%u.%u.%u:%u\n"
+			   "  remote node:     %s\n"
+			   "  page off:        %zu\n"
+			   "  handshake ok:    %u\n"
+			   "  timer:           %lu.%lu\n"
+			   "  data ready:      %lu.%lu\n"
+			   "  advance start:   %lu.%lu\n"
+			   "  advance stop:    %lu.%lu\n"
+			   "  func start:      %lu.%lu\n"
+			   "  func stop:       %lu.%lu\n"
+			   "  func key:        %u\n"
+			   "  func type:       %u\n",
+			   sc,
+			   atomic_read(&sc->sc_kref.refcount),
+			   NIPQUAD(saddr), inet ? ntohs(sport) : 0,
+			   NIPQUAD(daddr), inet ? ntohs(dport) : 0,
+			   sc->sc_node->nd_name,
+			   sc->sc_page_off,
+			   sc->sc_handshake_ok,
+			   TV_SEC_USEC(sc->sc_tv_timer),
+			   TV_SEC_USEC(sc->sc_tv_data_ready),
+			   TV_SEC_USEC(sc->sc_tv_advance_start),
+			   TV_SEC_USEC(sc->sc_tv_advance_stop),
+			   TV_SEC_USEC(sc->sc_tv_func_start),
+			   TV_SEC_USEC(sc->sc_tv_func_stop),
+			   sc->sc_msg_key,
+			   sc->sc_msg_type);
+	}
+
+
+	spin_unlock(&o2net_debug_lock);
+
+	return 0;
+}
+
+static void sc_seq_stop(struct seq_file *seq, void *v)
+{
+}
+
+static struct seq_operations sc_seq_ops = {
+	.start = sc_seq_start,
+	.next = sc_seq_next,
+	.stop = sc_seq_stop,
+	.show = sc_seq_show,
+};
+
+static int sc_fop_open(struct inode *inode, struct file *file)
+{
+	struct o2net_sock_container *dummy_sc;
+	struct seq_file *seq;
+	int ret;
+
+	dummy_sc = kmalloc(sizeof(struct o2net_sock_container), GFP_KERNEL);
+	if (dummy_sc == NULL) {
+		ret = -ENOMEM;
+		goto out;
+	}
+	dummy_sc->sc_page = NULL;
+
+	ret = seq_open(file, &sc_seq_ops);
+	if (ret)
+		goto out;
+
+	seq = file->private_data;
+	seq->private = dummy_sc;
+	o2net_debug_add_sc(dummy_sc);
+
+	dummy_sc = NULL;
+
+out:
+	kfree(dummy_sc);
+	return ret;
+}
+
+static int sc_fop_release(struct inode *inode, struct file *file)
+{
+	struct seq_file *seq = file->private_data;
+	struct o2net_sock_container *dummy_sc = seq->private;
+
+	o2net_debug_del_sc(dummy_sc);
+	return seq_release_private(inode, file);
+}
+
+static struct file_operations sc_seq_fops = {
+	.open = sc_fop_open,
+	.read = seq_read,
+	.llseek = seq_lseek,
+	.release = sc_fop_release,
+};
+
+int o2net_debugfs_init(void)
+{
+	o2net_dentry = debugfs_create_dir(O2NET_DEBUG_DIR, NULL);
+	if (!o2net_dentry) {
+		mlog_errno(-ENOMEM);
+		goto bail;
+	}
+
+	nst_dentry = debugfs_create_file(NST_DEBUG_NAME, S_IFREG|S_IRUSR,
+					 o2net_dentry, NULL,
+					 &nst_seq_fops);
+	if (!nst_dentry) {
+		mlog_errno(-ENOMEM);
+		goto bail;
+	}
+
+	sc_dentry = debugfs_create_file(SC_DEBUG_NAME, S_IFREG|S_IRUSR,
+					o2net_dentry, NULL,
+					&sc_seq_fops);
+	if (!sc_dentry) {
+		mlog_errno(-ENOMEM);
+		goto bail;
+	}
+
+	return 0;
+bail:
+	if (sc_dentry)
+		debugfs_remove(sc_dentry);
+	if (nst_dentry)
+		debugfs_remove(nst_dentry);
+	if (o2net_dentry)
+		debugfs_remove(o2net_dentry);
+	return -ENOMEM;
+}
+
+void o2net_debugfs_exit(void)
+{
+	if (sc_dentry)
+		debugfs_remove(sc_dentry);
+	if (nst_dentry)
+		debugfs_remove(nst_dentry);
+	if (o2net_dentry)
+		debugfs_remove(o2net_dentry);
+}
+
+#endif	/* CONFIG_DEBUG_FS */
diff --git a/fs/ocfs2/cluster/nodemanager.c b/fs/ocfs2/cluster/nodemanager.c
index 709fba2..cf9401e 100644
--- a/fs/ocfs2/cluster/nodemanager.c
+++ b/fs/ocfs2/cluster/nodemanager.c
@@ -959,7 +959,10 @@ static int __init init_o2nm(void)
 	cluster_print_version();
 
 	o2hb_init();
-	o2net_init();
+
+	ret = o2net_init();
+	if (ret)
+		goto out;
 
 	ocfs2_table_header = register_sysctl_table(ocfs2_root_table);
 	if (!ocfs2_table_header) {
diff --git a/fs/ocfs2/cluster/tcp.c b/fs/ocfs2/cluster/tcp.c
index 1170918..1e44ad1 100644
--- a/fs/ocfs2/cluster/tcp.c
+++ b/fs/ocfs2/cluster/tcp.c
@@ -142,6 +142,54 @@ static void o2net_idle_timer(unsigned long data);
 static void o2net_sc_postpone_idle(struct o2net_sock_container *sc);
 static void o2net_sc_reset_idle_timer(struct o2net_sock_container *sc);
 
+static void o2net_init_nst(struct o2net_send_tracking *nst, u32 msgtype,
+			   u32 msgkey, struct task_struct *task, u8 node)
+{
+#ifdef CONFIG_DEBUG_FS
+	INIT_LIST_HEAD(&nst->st_net_debug_item);
+	nst->st_task = task;
+	nst->st_msg_type = msgtype;
+	nst->st_msg_key = msgkey;
+	nst->st_node = node;
+#endif
+}
+
+static void o2net_set_nst_sock_time(struct o2net_send_tracking *nst)
+{
+#ifdef CONFIG_DEBUG_FS
+	do_gettimeofday(&nst->st_sock_time);
+#endif
+}
+
+static void o2net_set_nst_send_time(struct o2net_send_tracking *nst)
+{
+#ifdef CONFIG_DEBUG_FS
+	do_gettimeofday(&nst->st_send_time);
+#endif
+}
+
+static void o2net_set_nst_status_time(struct o2net_send_tracking *nst)
+{
+#ifdef CONFIG_DEBUG_FS
+	do_gettimeofday(&nst->st_status_time);
+#endif
+}
+
+static void o2net_set_nst_sock_container(struct o2net_send_tracking *nst,
+					 struct o2net_sock_container *sc)
+{
+#ifdef CONFIG_DEBUG_FS
+	nst->st_sc = sc;
+#endif
+}
+
+static void o2net_set_nst_msg_id(struct o2net_send_tracking *nst, u32 msg_id)
+{
+#ifdef CONFIG_DEBUG_FS
+	nst->st_id = msg_id;
+#endif
+}
+
 static inline int o2net_reconnect_delay(void)
 {
 	return o2nm_single_cluster->cl_reconnect_delay_ms;
@@ -290,6 +338,7 @@ static void sc_kref_release(struct kref *kref)
 	o2nm_node_put(sc->sc_node);
 	sc->sc_node = NULL;
 
+	o2net_debug_del_sc(sc);
 	kfree(sc);
 }
 
@@ -330,6 +379,7 @@ static struct o2net_sock_container *sc_alloc(struct
o2nm_node *node)
 
 	ret = sc;
 	sc->sc_page = page;
+	o2net_debug_add_sc(sc);
 	sc = NULL;
 	page = NULL;
 
@@ -913,6 +963,9 @@ int o2net_send_message_vec(u32 msg_type, u32 key, struct
kvec *caller_vec,
 	struct o2net_status_wait nsw = {
 		.ns_node_item = LIST_HEAD_INIT(nsw.ns_node_item),
 	};
+	struct o2net_send_tracking nst;
+
+	o2net_init_nst(&nst, msg_type, key, current, target_node);
 
 	if (o2net_wq == NULL) {
 		mlog(0, "attempt to tx without o2netd running\n");
@@ -938,6 +991,10 @@ int o2net_send_message_vec(u32 msg_type, u32 key, struct
kvec *caller_vec,
 		goto out;
 	}
 
+	o2net_debug_add_nst(&nst);
+
+	o2net_set_nst_sock_time(&nst);
+
 	ret = wait_event_interruptible(nn->nn_sc_wq,
 				       o2net_tx_can_proceed(nn, &sc, &error));
 	if (!ret && error)
@@ -945,6 +1002,8 @@ int o2net_send_message_vec(u32 msg_type, u32 key, struct
kvec *caller_vec,
 	if (ret)
 		goto out;
 
+	o2net_set_nst_sock_container(&nst, sc);
+
 	veclen = caller_veclen + 1;
 	vec = kmalloc(sizeof(struct kvec) * veclen, GFP_ATOMIC);
 	if (vec == NULL) {
@@ -971,6 +1030,9 @@ int o2net_send_message_vec(u32 msg_type, u32 key, struct
kvec *caller_vec,
 		goto out;
 
 	msg->msg_num = cpu_to_be32(nsw.ns_id);
+	o2net_set_nst_msg_id(&nst, nsw.ns_id);
+
+	o2net_set_nst_send_time(&nst);
 
 	/* finally, convert the message header to network byte-order
 	 * and send */
@@ -985,6 +1047,7 @@ int o2net_send_message_vec(u32 msg_type, u32 key, struct
kvec *caller_vec,
 	}
 
 	/* wait on other node's handler */
+	o2net_set_nst_status_time(&nst);
 	wait_event(nsw.ns_wq, o2net_nsw_completed(nn, &nsw));
 
 	/* Note that we avoid overwriting the callers status return
@@ -997,6 +1060,7 @@ int o2net_send_message_vec(u32 msg_type, u32 key, struct
kvec *caller_vec,
 	mlog(0, "woken, returning system status %d, user status %d\n",
 	     ret, nsw.ns_status);
 out:
+	o2net_debug_del_nst(&nst); /* must be before dropping sc and node */
 	if (sc)
 		sc_put(sc);
 	if (vec)
@@ -1935,6 +1999,9 @@ int o2net_init(void)
 
 	o2quo_init();
 
+	if (o2net_debugfs_init())
+		return -ENOMEM;
+
 	o2net_hand = kzalloc(sizeof(struct o2net_handshake), GFP_KERNEL);
 	o2net_keep_req = kzalloc(sizeof(struct o2net_msg), GFP_KERNEL);
 	o2net_keep_resp = kzalloc(sizeof(struct o2net_msg), GFP_KERNEL);
@@ -1976,4 +2043,5 @@ void o2net_exit(void)
 	kfree(o2net_hand);
 	kfree(o2net_keep_req);
 	kfree(o2net_keep_resp);
+	o2net_debugfs_exit();
 }
diff --git a/fs/ocfs2/cluster/tcp.h b/fs/ocfs2/cluster/tcp.h
index f36f66a..a705d5d 100644
--- a/fs/ocfs2/cluster/tcp.h
+++ b/fs/ocfs2/cluster/tcp.h
@@ -117,4 +117,36 @@ int o2net_num_connected_peers(void);
 int o2net_init(void);
 void o2net_exit(void);
 
+struct o2net_send_tracking;
+struct o2net_sock_container;
+
+#ifdef CONFIG_DEBUG_FS
+int o2net_debugfs_init(void);
+void o2net_debugfs_exit(void);
+void o2net_debug_add_nst(struct o2net_send_tracking *nst);
+void o2net_debug_del_nst(struct o2net_send_tracking *nst);
+void o2net_debug_add_sc(struct o2net_sock_container *sc);
+void o2net_debug_del_sc(struct o2net_sock_container *sc);
+#else
+static int o2net_debugfs_init(void)
+{
+	return 0;
+}
+static void o2net_debugfs_exit(void)
+{
+}
+static void o2net_debug_add_nst(struct o2net_send_tracking *nst)
+{
+}
+static void o2net_debug_del_nst(struct o2net_send_tracking *nst)
+{
+}
+static void o2net_debug_add_sc(struct o2net_sock_container *sc)
+{
+}
+static void o2net_debug_del_sc(struct o2net_sock_container *sc)
+{
+}
+#endif	/* CONFIG_DEBUG_FS */
+
 #endif /* O2CLUSTER_TCP_H */
diff --git a/fs/ocfs2/cluster/tcp_internal.h b/fs/ocfs2/cluster/tcp_internal.h
index b4c5586..8d58cfe 100644
--- a/fs/ocfs2/cluster/tcp_internal.h
+++ b/fs/ocfs2/cluster/tcp_internal.h
@@ -166,7 +166,9 @@ struct o2net_sock_container {
 	/* original handlers for the sockets */
 	void			(*sc_state_change)(struct sock *sk);
 	void			(*sc_data_ready)(struct sock *sk, int bytes);
-
+#ifdef CONFIG_DEBUG_FS
+	struct list_head        sc_net_debug_item;
+#endif
 	struct timeval 		sc_tv_timer;
 	struct timeval 		sc_tv_data_ready;
 	struct timeval 		sc_tv_advance_start;
@@ -208,4 +210,24 @@ struct o2net_status_wait {
 	struct list_head	ns_node_item;
 };
 
+#ifdef CONFIG_DEBUG_FS
+/* just for state dumps */
+struct o2net_send_tracking {
+	struct list_head		st_net_debug_item;
+	struct task_struct		*st_task;
+	struct o2net_sock_container	*st_sc;
+	u32				st_id;
+	u32				st_msg_type;
+	u32				st_msg_key;
+	u8				st_node;
+	struct timeval			st_sock_time;
+	struct timeval			st_send_time;
+	struct timeval			st_status_time;
+};
+#else
+struct o2net_send_tracking {
+	u32	dummy;
+};
+#endif	/* CONFIG_DEBUG_FS */
+
 #endif /* O2CLUSTER_TCP_INTERNAL_H */
-- 
1.5.3.6