[PATCH] drop_monitor: make last_rx timestamp private

[PATCH] drop_monitor: make last_rx timestamp private

[Neil Horman]

Hey all-


It was recently pointed out to me that the last_rx field of the net_device
structure wasn't updated regularly.  In fact only the bonding driver really uses
it currently.  Since the drop_monitor code relies on the last_rx field to detect
drops on recevie in hardware, We need to find a more reliable way to rate limit
our drop checks (so that we don't check for drops on every frame recevied, which
would be inefficient.  This patch makes a last_rx timestamp that is private to
the drop monitor code and is updated for every device that we track.

Neil

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


 drop_monitor.c |   15 ++++++++++-----
 1 file changed, 10 insertions(+), 5 deletions(-)

diff --git a/net/core/drop_monitor.c b/net/core/drop_monitor.c
index 9d66fa9..34a05ce 100644
--- a/net/core/drop_monitor.c
+++ b/net/core/drop_monitor.c
@@ -52,6 +52,7 @@ struct per_cpu_dm_data {
 
 struct dm_hw_stat_delta {
 	struct net_device *dev;
+	unsigned long last_rx;
 	struct list_head list;
 	struct rcu_head rcu;
 	unsigned long last_drop_val;
@@ -179,18 +180,21 @@ static void trace_napi_poll_hit(struct napi_struct *napi)
 {
 	struct dm_hw_stat_delta *new_stat;
 
-	/*
-	 * Ratelimit our check time to dm_hw_check_delta jiffies
-	 */
-	if (!time_after(jiffies, napi->dev->last_rx + dm_hw_check_delta))
-		return;
 
 	rcu_read_lock();
 	list_for_each_entry_rcu(new_stat, &hw_stats_list, list) {
+		/*
+		 * only add a note to our monitor buffer if:
+		 * 1) this is the dev we received on
+		 * 2) its after the last_rx delta
+		 * 3) our rx_dropped count has gone up
+		 */
 		if ((new_stat->dev == napi->dev)  &&
+		    (time_after(jiffies, new_stat->last_rx + dm_hw_check_delta)) &&
 		    (napi->dev->stats.rx_dropped != new_stat->last_drop_val)) {
 			trace_drop_common(NULL, NULL);
 			new_stat->last_drop_val = napi->dev->stats.rx_dropped;
+			new_stat->last_rx = jiffies;
 			break;
 		}
 	}
@@ -286,6 +290,7 @@ static int dropmon_net_event(struct notifier_block *ev_block,
 			goto out;
 
 		new_stat->dev = dev;
+		new_stat->last_rx = jiffies;
 		INIT_RCU_HEAD(&new_stat->rcu);
 		spin_lock(&trace_state_lock);
 		list_add_rcu(&new_stat->list, &hw_stats_list);

Re: [PATCH] drop_monitor: make last_rx timestamp private

[David Miller]

From: Neil Horman <nhorman@tuxdriver.com>
Date: Mon, 31 Aug 2009 15:58:47 -0400

> It was recently pointed out to me that the last_rx field of the net_device
> structure wasn't updated regularly.  In fact only the bonding driver really uses
> it currently.  Since the drop_monitor code relies on the last_rx field to detect
> drops on recevie in hardware, We need to find a more reliable way to rate limit
> our drop checks (so that we don't check for drops on every frame recevied, which
> would be inefficient.  This patch makes a last_rx timestamp that is private to
> the drop monitor code and is updated for every device that we track.
> 
> Signed-off-by: Neil Horman <nhorman@tuxdriver.com>

Neil, this doesn't apply to net-next-2.6:

> diff --git a/net/core/drop_monitor.c b/net/core/drop_monitor.c
> index 9d66fa9..34a05ce 100644
> --- a/net/core/drop_monitor.c
> +++ b/net/core/drop_monitor.c
 ...
> @@ -179,18 +180,21 @@ static void trace_napi_poll_hit(struct napi_struct *napi)
>  {
>  	struct dm_hw_stat_delta *new_stat;
>  
> -	/*
> -	 * Ratelimit our check time to dm_hw_check_delta jiffies
> -	 */
> -	if (!time_after(jiffies, napi->dev->last_rx + dm_hw_check_delta))
> -		return;
>  
>  	rcu_read_lock();
>  	list_for_each_entry_rcu(new_stat, &hw_stats_list, list) {

In net-next-2.6 this test reads:

	/*
	 * Ratelimit our check time to dm_hw_check_delta jiffies
	 */
	if (!napi->dev ||
	    !time_after(jiffies, napi->dev->last_rx + dm_hw_check_delta))
		return;

and you must retain the napi->dev NULL check there as otherwise
the list traversal tests will blindly dereference that pointer.

Re: [PATCH] drop_monitor: make last_rx timestamp private

[Neil Horman]

On Tue, Sep 01, 2009 at 06:21:27PM -0700, David Miller wrote:
> From: Neil Horman <nhorman@tuxdriver.com>
> Date: Mon, 31 Aug 2009 15:58:47 -0400
> 
> > It was recently pointed out to me that the last_rx field of the net_device
> > structure wasn't updated regularly.  In fact only the bonding driver really uses
> > it currently.  Since the drop_monitor code relies on the last_rx field to detect
> > drops on recevie in hardware, We need to find a more reliable way to rate limit
> > our drop checks (so that we don't check for drops on every frame recevied, which
> > would be inefficient.  This patch makes a last_rx timestamp that is private to
> > the drop monitor code and is updated for every device that we track.
> > 
> > Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
> 
> Neil, this doesn't apply to net-next-2.6:
> 
> > diff --git a/net/core/drop_monitor.c b/net/core/drop_monitor.c
> > index 9d66fa9..34a05ce 100644
> > --- a/net/core/drop_monitor.c
> > +++ b/net/core/drop_monitor.c
>  ...
> > @@ -179,18 +180,21 @@ static void trace_napi_poll_hit(struct napi_struct *napi)
> >  {
> >  	struct dm_hw_stat_delta *new_stat;
> >  
> > -	/*
> > -	 * Ratelimit our check time to dm_hw_check_delta jiffies
> > -	 */
> > -	if (!time_after(jiffies, napi->dev->last_rx + dm_hw_check_delta))
> > -		return;
> >  
> >  	rcu_read_lock();
> >  	list_for_each_entry_rcu(new_stat, &hw_stats_list, list) {
> 
> In net-next-2.6 this test reads:
> 
> 	/*
> 	 * Ratelimit our check time to dm_hw_check_delta jiffies
> 	 */
> 	if (!napi->dev ||
> 	    !time_after(jiffies, napi->dev->last_rx + dm_hw_check_delta))
> 		return;
> 
> and you must retain the napi->dev NULL check there as otherwise
> the list traversal tests will blindly dereference that pointer.
> 

Apologies, this raced with Xiao fix to trace_napi_poll hit, which introduced
that null check.  I'll rediff/repost shortly.

Neil

Re: [PATCH] drop_monitor: make last_rx timestamp private

[Neil Horman]

Ok, heres a repost, unchanged saved for being massaged to fit in with the recent
changes that my origional patch raced against.  Thanks!
Neil


It was recently pointed out to me that the last_rx field of the net_device
structure wasn't updated regularly.  In fact only the bonding driver really uses
it currently.  Since the drop_monitor code relies on the last_rx field to detect
drops on recevie in hardware, We need to find a more reliable way to rate limit
our drop checks (so that we don't check for drops on every frame recevied, which
would be inefficient.  This patch makes a last_rx timestamp that is private to
the drop monitor code and is updated for every device that we track.

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


 drop_monitor.c |   15 ++++++++++++---
 1 file changed, 12 insertions(+), 3 deletions(-)

diff --git a/net/core/drop_monitor.c b/net/core/drop_monitor.c
index d311202..0a113f2 100644
--- a/net/core/drop_monitor.c
+++ b/net/core/drop_monitor.c
@@ -52,6 +52,7 @@ struct per_cpu_dm_data {
 
 struct dm_hw_stat_delta {
 	struct net_device *dev;
+	unsigned long last_rx;
 	struct list_head list;
 	struct rcu_head rcu;
 	unsigned long last_drop_val;
@@ -180,18 +181,25 @@ static void trace_napi_poll_hit(struct napi_struct *napi)
 	struct dm_hw_stat_delta *new_stat;
 
 	/*
-	 * Ratelimit our check time to dm_hw_check_delta jiffies
+	 * Don't check napi structures with no associated device
 	 */
-	if (!napi->dev ||
-	    !time_after(jiffies, napi->dev->last_rx + dm_hw_check_delta))
+	if (!napi->dev)
 		return;
 
 	rcu_read_lock();
 	list_for_each_entry_rcu(new_stat, &hw_stats_list, list) {
+		/*
+		 * only add a note to our monitor buffer if:
+		 * 1) this is the dev we received on
+		 * 2) its after the last_rx delta
+		 * 3) our rx_dropped count has gone up
+		 */
 		if ((new_stat->dev == napi->dev)  &&
+		    (time_after(jiffies, new_stat->last_rx + dm_hw_check_delta)) &&
 		    (napi->dev->stats.rx_dropped != new_stat->last_drop_val)) {
 			trace_drop_common(NULL, NULL);
 			new_stat->last_drop_val = napi->dev->stats.rx_dropped;
+			new_stat->last_rx = jiffies;
 			break;
 		}
 	}
@@ -287,6 +295,7 @@ static int dropmon_net_event(struct notifier_block *ev_block,
 			goto out;
 
 		new_stat->dev = dev;
+		new_stat->last_rx = jiffies;
 		INIT_RCU_HEAD(&new_stat->rcu);
 		spin_lock(&trace_state_lock);
 		list_add_rcu(&new_stat->list, &hw_stats_list);

Re: [PATCH] drop_monitor: make last_rx timestamp private

[David Miller]

From: Neil Horman <nhorman@tuxdriver.com>
Date: Wed, 2 Sep 2009 09:36:20 -0400

> Ok, heres a repost, unchanged saved for being massaged to fit in with the recent
> changes that my origional patch raced against.  Thanks!
> Neil
> 
> 
> It was recently pointed out to me that the last_rx field of the net_device
> structure wasn't updated regularly.  In fact only the bonding driver really uses
> it currently.  Since the drop_monitor code relies on the last_rx field to detect
> drops on recevie in hardware, We need to find a more reliable way to rate limit
> our drop checks (so that we don't check for drops on every frame recevied, which
> would be inefficient.  This patch makes a last_rx timestamp that is private to
> the drop monitor code and is updated for every device that we track.
> 
> Signed-off-by: Neil Horman <nhorman@tuxdriver.com>

Applied, thanks Neil.