Message-ID: <20090305192706.GB12835@hmsreliant.think-freely.org>
Date:	Thu, 5 Mar 2009 14:27:06 -0500
From:	Neil Horman <nhorman@...driver.com>
To:	netdev@...r.kernel.org
Cc:	davem@...emloft.net, kuznet@....inr.ac.ru, pekkas@...core.fi,
	jmorris@...ei.org, yoshfuji@...ux-ipv6.org, kaber@...sh.net
Subject: Re: [Patch 4/5] Network Drop Monitor: Adding drop monitor
	implementation & Netlink protocol

Ok, as requested new version of the patch to implement the drop monitor trace
hook and netlink service

Modification notes:

1) As requested, modified the code to use generic netlink sockets rather than
create a new netlink protocol family.  This implies that patch 1/5 can be
dropped from this series entirely, since all it did was add that protocol family
identifier

2) General checkpatch cleanup.

3) Fix some alignment issues in net_dropmon.h.  Note that I used an explicit
__attribute__, rather than aligned_u64, since the later doesn't appear to be
exposed to user space.


Let me know what you all think.  Thanks!

Neil

Signed-off-by: Neil Horman <nhorman@...driver.com>


 include/linux/net_dropmon.h |   56 +++++++++
 net/core/drop_monitor.c     |  265 ++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 321 insertions(+)


Network Drop Monitor: Adding drop monitor implementation & Netlink protocol

diff --git a/include/linux/net_dropmon.h b/include/linux/net_dropmon.h
new file mode 100644
index 0000000..7ef35cb
--- /dev/null
+++ b/include/linux/net_dropmon.h
@@ -0,0 +1,56 @@
+#ifndef __NET_DROPMON_H
+#define __NET_DROPMON_H
+
+#include <linux/netlink.h>
+
+struct net_dm_drop_point {
+	uint8_t pc[8];
+	uint32_t count;
+};
+
+#define NET_DM_CFG_VERSION  0
+#define NET_DM_CFG_ALERT_COUNT  1
+#define NET_DM_CFG_ALERT_DELAY 2
+#define NET_DM_CFG_MAX 3
+
+struct net_dm_config_entry {
+	uint32_t type;
+	uint64_t data __attribute__((aligned(8)));
+};
+
+struct net_dm_config_msg {
+	uint32_t entries;
+	struct net_dm_config_entry options[0];
+};
+
+struct net_dm_alert_msg {
+	uint32_t entries;
+	struct net_dm_drop_point points[0];
+};
+
+struct net_dm_user_msg {
+	union {
+		struct net_dm_config_msg user;
+		struct net_dm_alert_msg alert;
+	} u;
+};
+
+
+/* These are the netlink message types for this protocol */
+
+enum {
+	NET_DM_CMD_UNSPEC = 0,
+	NET_DM_CMD_ALERT,
+	NET_DM_CMD_CONFIG,
+	NET_DM_CMD_START,
+	NET_DM_CMD_STOP,
+	_NET_DM_CMD_MAX,
+};
+
+#define NET_DM_CMD_MAX (_NET_DM_CMD_MAX - 1)
+
+/*
+ * Our group identifiers
+ */
+#define NET_DM_GRP_ALERT 1
+#endif
diff --git a/net/core/drop_monitor.c b/net/core/drop_monitor.c
new file mode 100644
index 0000000..a59ba6f
--- /dev/null
+++ b/net/core/drop_monitor.c
@@ -0,0 +1,265 @@
+/*
+ * Monitoring code for network dropped packet alerts
+ *
+ * Copyright (C) 2009 Neil Horman <nhorman@...driver.com>
+ */
+
+#include <linux/netdevice.h>
+#include <linux/etherdevice.h>
+#include <linux/string.h>
+#include <linux/if_arp.h>
+#include <linux/inetdevice.h>
+#include <linux/inet.h>
+#include <linux/interrupt.h>
+#include <linux/netpoll.h>
+#include <linux/sched.h>
+#include <linux/delay.h>
+#include <linux/types.h>
+#include <linux/workqueue.h>
+#include <linux/netlink.h>
+#include <linux/net_dropmon.h>
+#include <linux/percpu.h>
+#include <linux/timer.h>
+#include <linux/bitops.h>
+#include <net/genetlink.h>
+
+#include <trace/skb.h>
+
+#include <asm/unaligned.h>
+
+#define RCV_SKB_FAIL(err) do { netlink_ack(skb, nlh, (err)); return; } while (0)
+
+#define TRACE_ON 1
+#define TRACE_OFF 0
+
+static void send_dm_alert(struct work_struct *unused);
+
+
+/*
+ * Globals, our netlink socket pointer
+ * and the work handle that will send up
+ * netlink alerts
+ */
+struct sock *dm_sock;
+
+struct per_cpu_dm_data {
+	struct work_struct dm_alert_work;
+	struct sk_buff *skb;
+	atomic_t dm_hit_count;
+	struct timer_list send_timer;
+};
+
+static struct genl_family net_drop_monitor_family = {
+	.id             = GENL_ID_GENERATE,
+	.hdrsize        = 0,
+	.name           = "NET_DM",
+	.version        = 1,
+	.maxattr        = NET_DM_CMD_MAX,
+};
+
+static DEFINE_PER_CPU(struct per_cpu_dm_data, dm_cpu_data);
+
+static int dm_hit_limit = 64;
+static int dm_delay = 1;
+
+
+static void reset_per_cpu_data(struct per_cpu_dm_data *data)
+{
+	size_t al;
+	struct net_dm_alert_msg *msg;
+
+	al = sizeof(struct net_dm_alert_msg);
+	al += dm_hit_limit * sizeof(struct net_dm_drop_point);
+	data->skb = genlmsg_new(al, GFP_KERNEL);
+	genlmsg_put(data->skb, 0, 0, &net_drop_monitor_family,
+			0, NET_DM_CMD_ALERT);
+	msg = __nla_reserve_nohdr(data->skb, sizeof(struct net_dm_alert_msg));
+	memset(msg, 0, al);
+	atomic_set(&data->dm_hit_count, dm_hit_limit);
+}
+
+static void send_dm_alert(struct work_struct *unused)
+{
+	struct sk_buff *skb;
+	struct per_cpu_dm_data *data = &__get_cpu_var(dm_cpu_data);
+
+	/*
+	 * Grab the skb we're about to send
+	 */
+	skb = data->skb;
+
+	/*
+	 * Replace it with a new one
+	 */
+	reset_per_cpu_data(data);
+
+	/*
+	 * Ship it!
+	 */
+	genlmsg_multicast(skb, 0, NET_DM_GRP_ALERT, GFP_KERNEL);
+
+}
+
+/*
+ * This is the timer function to delay the sending of an alert
+ * in the event that more drops will arrive during the
+ * hysteresis period.  Note that it operates under the timer interrupt
+ * so we don't need to disable preemption here
+ */
+static void sched_send_work(unsigned long unused)
+{
+	struct per_cpu_dm_data *data =  &__get_cpu_var(dm_cpu_data);
+
+	schedule_work(&data->dm_alert_work);
+}
+
+static void trace_kfree_skb_hit(struct sk_buff *skb, void *location)
+{
+	struct net_dm_alert_msg *msg;
+	struct nlmsghdr *nlh;
+	int i;
+	struct per_cpu_dm_data *data = &__get_cpu_var(dm_cpu_data);
+
+
+	if (!atomic_add_unless(&data->dm_hit_count, -1, 0)) {
+		/*
+		 * we're already at zero, discard this hit
+		 */
+		goto out;
+	}
+
+	nlh = (struct nlmsghdr *)data->skb->data;
+	msg = genlmsg_data(nlmsg_data(nlh));
+	for (i = 0; i < msg->entries; i++) {
+		if (!memcmp(&location, msg->points[i].pc, sizeof(void *))) {
+			msg->points[i].count++;
+			goto out;
+		}
+	}
+
+	/*
+	 * We need to create a new entry
+	 */
+	__nla_reserve_nohdr(data->skb, sizeof(struct net_dm_drop_point));
+	memcpy(msg->points[msg->entries].pc, &location, sizeof(void *));
+	msg->points[msg->entries].count = 1;
+	msg->entries++;
+
+	if (!timer_pending(&data->send_timer)) {
+		data->send_timer.expires = jiffies + dm_delay * HZ;
+		add_timer_on(&data->send_timer, smp_processor_id());
+	}
+
+out:
+	return;
+}
+
+static int set_all_monitor_traces(int state)
+{
+	int rc = 0;
+
+	switch (state) {
+	case TRACE_ON:
+		rc |= register_trace_kfree_skb(trace_kfree_skb_hit);
+		break;
+	case TRACE_OFF:
+		rc |= unregister_trace_kfree_skb(trace_kfree_skb_hit);
+
+		tracepoint_synchronize_unregister();
+		break;
+	default:
+		rc = 1;
+		break;
+	}
+
+	if (rc)
+		return -EINPROGRESS;
+	return rc;
+}
+
+
+static int net_dm_cmd_config(struct sk_buff *skb,
+			struct genl_info *info)
+{
+	return -ENOTSUPP;
+}
+
+static int net_dm_cmd_trace(struct sk_buff *skb,
+			struct genl_info *info)
+{
+	switch (info->genlhdr->cmd) {
+	case NET_DM_CMD_START:
+		return set_all_monitor_traces(TRACE_ON);
+		break;
+	case NET_DM_CMD_STOP:
+		return set_all_monitor_traces(TRACE_OFF);
+		break;
+	}
+
+	return -ENOTSUPP;
+}
+
+
+static struct genl_ops dropmon_ops[] = {
+	{
+		.cmd = NET_DM_CMD_CONFIG,
+		.doit = net_dm_cmd_config,
+	},
+	{
+		.cmd = NET_DM_CMD_START,
+		.doit = net_dm_cmd_trace,
+	},
+	{
+		.cmd = NET_DM_CMD_STOP,
+		.doit = net_dm_cmd_trace,
+	},
+};
+
+static int __init init_net_drop_monitor(void)
+{
+	int cpu;
+	int rc, i, ret;
+	struct per_cpu_dm_data *data;
+	printk(KERN_INFO "Initalizing network drop monitor service\n");
+
+	if (sizeof(void *) > 8) {
+		printk(KERN_ERR "Unable to store program counters on this arch, Drop monitor failed\n");
+		return -ENOSPC;
+	}
+
+	if (genl_register_family(&net_drop_monitor_family) < 0) {
+		printk(KERN_ERR "Could not create drop monitor netlink family\n");
+		return -EFAULT;
+	}
+
+	rc = -EFAULT;
+
+	for (i = 0; i < ARRAY_SIZE(dropmon_ops); i++) {
+		ret = genl_register_ops(&net_drop_monitor_family,
+					&dropmon_ops[i]);
+		if (ret) {
+			printk(KERN_CRIT "failed to register operation %d\n",
+				dropmon_ops[i].cmd);
+			goto out_unreg;
+		}
+	}
+
+	rc = 0;
+
+	for_each_present_cpu(cpu) {
+		data = &per_cpu(dm_cpu_data, cpu);
+		reset_per_cpu_data(data);
+		INIT_WORK(&data->dm_alert_work, send_dm_alert);
+		init_timer(&data->send_timer);
+		data->send_timer.data = cpu;
+		data->send_timer.function = sched_send_work;
+	}
+	goto out;
+
+out_unreg:
+	genl_unregister_family(&net_drop_monitor_family);
+out:
+	return rc;
+}
+
+late_initcall(init_net_drop_monitor);
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Powered by blists - more mailing lists