Sophie

Sophie

distrib > Scientific%20Linux > 5x > x86_64 > by-pkgid > fc11cd6e1c513a17304da94a5390f3cd > files > 1973

kernel-2.6.18-194.11.1.el5.src.rpm

From: Neil Horman <nhorman@redhat.com>
Date: Mon, 15 Jun 2009 16:27:09 -0400
Subject: [net] backport net_rx_action tracepoint
Message-id: 20090615202709.GA11967@hmsreliant.think-freely.org
O-Subject: [RHEL 5.4 PATCH] backport net_rx_action tracepoint (bz 506138)
Bugzilla: 506138
RH-Acked-by: Jiri Pirko <jpirko@redhat.com>
RH-Acked-by: Andy Gospodarek <gospo@redhat.com>

Hey all-
	backport of a commit 4ea7e38696c7e798c47ebbecadfd392f23f814f9  from
net-next.  This lets the dropmon protocol detect frame loss in hardware.
Satisfies bz 506138

Neil

diff --git a/include/linux/net_dropmon.h b/include/linux/net_dropmon.h
index 0217fb8..2fd0cc4 100644
--- a/include/linux/net_dropmon.h
+++ b/include/linux/net_dropmon.h
@@ -2,6 +2,7 @@
 #define __NET_DROPMON_H
 
 #include <linux/netlink.h>
+#include <linux/types.h>
 
 struct net_dm_drop_point {
 	__u8 pc[8];
diff --git a/include/trace/napi.h b/include/trace/napi.h
new file mode 100644
index 0000000..3510135
--- /dev/null
+++ b/include/trace/napi.h
@@ -0,0 +1,11 @@
+#ifndef _TRACE_NAPI_H_
+#define _TRACE_NAPI_H_
+
+#include <linux/netdevice.h>
+#include <linux/tracepoint.h>
+
+DEFINE_TRACE(napi_poll,
+	TPPROTO(struct net_device *napi),
+	TPARGS(napi));
+
+#endif
diff --git a/net/core/dev.c b/net/core/dev.c
index f57096e..0813be2 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -117,6 +117,7 @@
 #include <linux/dmaengine.h>
 #include <linux/err.h>
 #include <linux/ctype.h>
+#include <trace/napi.h>
 
 #ifdef CONFIG_XEN
 #include <net/ip.h>
@@ -2337,7 +2338,7 @@ static void net_rx_action(struct softirq_action *h)
 	unsigned long start_time = jiffies;
 	int budget = netdev_budget;
 	void *have;
-
+	int called_poll;
 	local_irq_disable();
 
 	while (!list_empty(&queue->poll_list)) {
@@ -2351,7 +2352,7 @@ static void net_rx_action(struct softirq_action *h)
 		dev = list_entry(queue->poll_list.next,
 				 struct net_device, poll_list);
 		have = netpoll_poll_lock(dev);
-
+		called_poll = (dev->quota <= 0) ? 0:1;
 		if (dev->quota <= 0 || dev->poll(dev, &budget)) {
 			netpoll_poll_unlock(have);
 			local_irq_disable();
@@ -2365,6 +2366,8 @@ static void net_rx_action(struct softirq_action *h)
 			dev_put(dev);
 			local_irq_disable();
 		}
+		if (called_poll)
+			trace_napi_poll(dev);
 	}
 out:
 #ifdef CONFIG_NET_DMA
diff --git a/net/core/drop_monitor.c b/net/core/drop_monitor.c
index 9fe224d..8016ac4 100644
--- a/net/core/drop_monitor.c
+++ b/net/core/drop_monitor.c
@@ -22,8 +22,10 @@
 #include <linux/timer.h>
 #include <linux/bitops.h>
 #include <net/genetlink.h>
+#include <net/netevent.h>
 
 #include <trace/skb.h>
+#include <trace/napi.h>
 
 #include <asm/unaligned.h>
 
@@ -38,7 +40,8 @@ static void send_dm_alert(struct work_struct *unused);
  * and the work handle that will send up
  * netlink alerts
  */
-struct sock *dm_sock;
+static int trace_state = TRACE_OFF;
+static spinlock_t trace_state_lock = SPIN_LOCK_UNLOCKED;
 
 struct per_cpu_dm_data {
 	struct work_struct dm_alert_work;
@@ -47,6 +50,13 @@ struct per_cpu_dm_data {
 	struct timer_list send_timer;
 };
 
+struct dm_hw_stat_delta {
+	struct net_device *dev;
+	struct list_head list;
+	struct rcu_head rcu;
+	unsigned long last_drop_val;
+};
+
 static struct genl_family net_drop_monitor_family = {
 	.id             = GENL_ID_GENERATE,
 	.hdrsize        = 0,
@@ -59,7 +69,8 @@ static DEFINE_PER_CPU(struct per_cpu_dm_data, dm_cpu_data);
 
 static int dm_hit_limit = 64;
 static int dm_delay = 1;
-
+static unsigned long dm_hw_check_delta = 2*HZ;
+static LIST_HEAD(hw_stats_list);
 
 static void reset_per_cpu_data(struct per_cpu_dm_data *data)
 {
@@ -111,7 +122,7 @@ static void sched_send_work(unsigned long unused)
 	schedule_work(&data->dm_alert_work);
 }
 
-static void trace_kfree_skb_hit(struct sk_buff *skb, void *location)
+static void trace_drop_common(struct sk_buff *skb, void *location)
 {
 	struct net_dm_alert_msg *msg;
 	struct nlmsghdr *nlh;
@@ -152,27 +163,88 @@ out:
 	return;
 }
 
+
+static void trace_kfree_skb_hit(struct sk_buff *skb, void *location)
+{
+	trace_drop_common(skb, location);
+}
+  
+static void trace_napi_poll_hit(struct net_device *napi)
+{
+	struct dm_hw_stat_delta *new_stat;
+ 	struct net_device_stats *stats;
+ 
+	/*
+	 * Ratelimit our check time to dm_hw_check_delta jiffies
+	 */
+	if (!time_after(jiffies, napi->last_rx + dm_hw_check_delta))
+		return;
+  
+	rcu_read_lock();
+	list_for_each_entry_rcu(new_stat, &hw_stats_list, list) {
+		stats = napi->get_stats(napi);
+		if ((new_stat->dev == napi)  &&
+		    (stats->rx_dropped != new_stat->last_drop_val)) {
+			trace_drop_common(NULL, NULL);
+			new_stat->last_drop_val = stats->rx_dropped;
+			break;
+		}
+	}
+	rcu_read_unlock();
+}
+  
+static void free_dm_hw_stat(struct rcu_head *head)
+{
+	struct dm_hw_stat_delta *n;
+	n = container_of(head, struct dm_hw_stat_delta, rcu);
+	kfree(n);
+}
+
 static int set_all_monitor_traces(int state)
 {
+	struct dm_hw_stat_delta *new_stat = NULL;
+	struct dm_hw_stat_delta *temp;
 	int rc = 0;
 
+	spin_lock(&trace_state_lock);
+
 	switch (state) {
 	case TRACE_ON:
 		rc |= register_trace_kfree_skb(trace_kfree_skb_hit);
+		rc |= register_trace_napi_poll(trace_napi_poll_hit);
 		break;
 	case TRACE_OFF:
-		unregister_trace_kfree_skb(trace_kfree_skb_hit);
+		rc = unregister_trace_kfree_skb(trace_kfree_skb_hit);
+		rc |= unregister_trace_napi_poll(trace_napi_poll_hit);
+
+		/*
+		 * Clean the device list
+		 */
+		list_for_each_entry_safe(new_stat, temp, &hw_stats_list, list) {
+			if (new_stat->dev == NULL) {
+				list_del_rcu(&new_stat->list);
+				call_rcu(&new_stat->rcu, free_dm_hw_stat);
+			}
+		}
+
 		break;
 	default:
 		rc = 1;
 		break;
 	}
 
+	if (!rc)
+		trace_state = state;
+
+	spin_unlock(&trace_state_lock);
+
 	if (rc)
 		return -EINPROGRESS;
 	return rc;
 }
 
+  
+
 
 static int net_dm_cmd_config(struct sk_buff *skb,
 			struct genl_info *info)
@@ -195,6 +267,44 @@ static int net_dm_cmd_trace(struct sk_buff *skb,
 	return -ENOTSUPP;
 }
 
+static int dropmon_net_event(struct notifier_block *ev_block,
+			unsigned long event, void *ptr)
+{
+	struct net_device *dev = ptr;
+	struct dm_hw_stat_delta *new_stat = NULL;
+	struct dm_hw_stat_delta *tmp;
+
+	switch (event) {
+	case NETDEV_REGISTER:
+		new_stat = kzalloc(sizeof(struct dm_hw_stat_delta), GFP_KERNEL);
+
+		if (!new_stat)
+			goto out;
+
+		new_stat->dev = dev;
+		INIT_RCU_HEAD(&new_stat->rcu);
+		spin_lock(&trace_state_lock);
+		list_add_rcu(&new_stat->list, &hw_stats_list);
+		spin_unlock(&trace_state_lock);
+		break;
+	case NETDEV_UNREGISTER:
+		spin_lock(&trace_state_lock);
+		list_for_each_entry_safe(new_stat, tmp, &hw_stats_list, list) {
+			if (new_stat->dev == dev) {
+				new_stat->dev = NULL;
+				if (trace_state == TRACE_OFF) {
+					list_del_rcu(&new_stat->list);
+					call_rcu(&new_stat->rcu, free_dm_hw_stat);
+					break;
+				}
+			}
+		}
+		spin_unlock(&trace_state_lock);
+		break;
+	}
+out:
+	return NOTIFY_DONE;
+}
 
 static struct genl_ops dropmon_ops[] = {
 	{
@@ -211,6 +321,10 @@ static struct genl_ops dropmon_ops[] = {
 	},
 };
 
+static struct notifier_block dropmon_net_notifier = {
+	.notifier_call = dropmon_net_event
+};
+
 static int __init init_net_drop_monitor(void)
 {
 	int cpu;
@@ -234,12 +348,18 @@ static int __init init_net_drop_monitor(void)
 		ret = genl_register_ops(&net_drop_monitor_family,
 					&dropmon_ops[i]);
 		if (ret) {
-			printk(KERN_CRIT "failed to register operation %d\n",
+			printk(KERN_CRIT "Failed to register operation %d\n",
 				dropmon_ops[i].cmd);
 			goto out_unreg;
 		}
 	}
 
+	rc = register_netdevice_notifier(&dropmon_net_notifier);
+	if (rc < 0) {
+		printk(KERN_CRIT "Failed to register netdevice notifier\n");
+		goto out_unreg;
+	}
+
 	rc = 0;
 
 	for_each_present_cpu(cpu) {
@@ -250,6 +370,7 @@ static int __init init_net_drop_monitor(void)
 		data->send_timer.data = cpu;
 		data->send_timer.function = sched_send_work;
 	}
+
 	goto out;
 
 out_unreg:
diff --git a/net/core/net-traces.c b/net/core/net-traces.c
index c8fb456..b07b25b 100644
--- a/net/core/net-traces.c
+++ b/net/core/net-traces.c
@@ -20,6 +20,7 @@
 #include <linux/netlink.h>
 #include <linux/net_dropmon.h>
 #include <trace/skb.h>
+#include <trace/napi.h>
 
 #include <asm/unaligned.h>
 #include <asm/bitops.h>
@@ -27,3 +28,6 @@
 
 DEFINE_TRACE(kfree_skb);
 EXPORT_TRACEPOINT_SYMBOL_GPL(kfree_skb);
+
+DEFINE_TRACE(napi_poll);
+EXPORT_TRACEPOINT_SYMBOL_GPL(napi_poll);
diff --git a/net/core/netpoll.c b/net/core/netpoll.c
index 5049354..b2d41e9 100644
--- a/net/core/netpoll.c
+++ b/net/core/netpoll.c
@@ -25,6 +25,7 @@
 #include <net/tcp.h>
 #include <net/udp.h>
 #include <asm/unaligned.h>
+#include <trace/napi.h>
 
 /*
  * We maintain a small pool of fully-sized skbs, to make sure the
@@ -147,6 +148,7 @@ static void poll_napi(struct netpoll *np)
 		atomic_inc(&trapped);
 
 		np->dev->poll(np->dev, &budget);
+		trace_napi_poll(np->dev);
 
 		atomic_dec(&trapped);
 		npinfo->rx_flags &= ~NETPOLL_RX_DROP;