[PATCH v4 1/1] IPoIB: Avoid multicast join attempts when having invalid p_key

[PATCH v4 1/1] IPoIB: Avoid multicast join attempts when having invalid p_key

[Alex Estrin]

Currently, the parent interface keeps sending broadcast group join
requests even if p_key index 0 is invalid, which for itself is
possible/common in virtualized environment where a VF has been probed to
VM but the actual p_key configuration has not yet been assigned by the
management software. This creates unnecessary noise on the fabric and in
the kernel logs:

ib0: multicast join failed for ff12:401b:8000:0000:0000:0000:ffff:ffff,
status -22

The original code run the multicast task regardless of the actual
p_key value, which can be avoided. The fix is to re-init resources  and
bring interface up only if p_key index 0 is valid either when starting
up or on PKEY_CHANGE event.

Fixes: c290414169 ('IPoIB: Fix pkey change flow for virtualization environments')

Reviewed-by: Ira Weiny <ira.weiny-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Signed-off-by: Alex Estrin <alex.estrin-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 drivers/infiniband/ulp/ipoib/ipoib_ib.c |   35 +++++++++++++++++++------------
 1 files changed, 21 insertions(+), 14 deletions(-)

diff --git a/drivers/infiniband/ulp/ipoib/ipoib_ib.c b/drivers/infiniband/ulp/ipoib/ipoib_ib.c
index 6a7003d..952db0b 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib_ib.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_ib.c
@@ -52,6 +52,7 @@ MODULE_PARM_DESC(data_debug_level,
 #endif
 
 static DEFINE_MUTEX(pkey_mutex);
+static void ipoib_pkey_dev_check_presence(struct net_device *dev);
 
 struct ipoib_ah *ipoib_create_ah(struct net_device *dev,
 				 struct ib_pd *pd, struct ib_ah_attr *attr)
@@ -669,12 +670,13 @@ int ipoib_ib_dev_open(struct net_device *dev)
 	struct ipoib_dev_priv *priv = netdev_priv(dev);
 	int ret;
 
-	if (ib_find_pkey(priv->ca, priv->port, priv->pkey, &priv->pkey_index)) {
-		ipoib_warn(priv, "P_Key 0x%04x not found\n", priv->pkey);
-		clear_bit(IPOIB_PKEY_ASSIGNED, &priv->flags);
+	ipoib_pkey_dev_check_presence(dev);
+
+	if (!test_bit(IPOIB_PKEY_ASSIGNED, &priv->flags)) {
+		ipoib_warn(priv, "P_Key 0x%04x is %s\n", priv->pkey,
+			   !(priv->pkey & 0x7fff) ? "Invalid" : "not found");
 		return -1;
 	}
-	set_bit(IPOIB_PKEY_ASSIGNED, &priv->flags);
 
 	ret = ipoib_init_qp(dev);
 	if (ret) {
@@ -712,9 +714,10 @@ dev_stop:
 static void ipoib_pkey_dev_check_presence(struct net_device *dev)
 {
 	struct ipoib_dev_priv *priv = netdev_priv(dev);
-	u16 pkey_index = 0;
 
-	if (ib_find_pkey(priv->ca, priv->port, priv->pkey, &pkey_index))
+	if (!(priv->pkey & 0x7fff) ||
+	    ib_find_pkey(priv->ca, priv->port, priv->pkey,
+			 &priv->pkey_index))
 		clear_bit(IPOIB_PKEY_ASSIGNED, &priv->flags);
 	else
 		set_bit(IPOIB_PKEY_ASSIGNED, &priv->flags);
@@ -987,15 +990,17 @@ static void __ipoib_ib_dev_flush(struct ipoib_dev_priv *priv,
 	up_read(&priv->vlan_rwsem);
 
 	if (!test_bit(IPOIB_FLAG_INITIALIZED, &priv->flags)) {
-		/* for non-child devices must check/update the pkey value here */
-		if (level == IPOIB_FLUSH_HEAVY &&
-		    !test_bit(IPOIB_FLAG_SUBINTERFACE, &priv->flags))
-			update_parent_pkey(priv);
-		ipoib_dbg(priv, "Not flushing - IPOIB_FLAG_INITIALIZED not set.\n");
-		return;
+		if (level != IPOIB_FLUSH_HEAVY) {
+			ipoib_dbg(priv, "Not flushing - IPOIB_FLAG_INITIALIZED not set.\n");
+			return;
+		}
 	}
 
 	if (!test_bit(IPOIB_FLAG_ADMIN_UP, &priv->flags)) {
+		/* interface is down. update pkey and leave. */
+		if (level == IPOIB_FLUSH_HEAVY &&
+		    !test_bit(IPOIB_FLAG_SUBINTERFACE, &priv->flags))
+			update_parent_pkey(priv);
 		ipoib_dbg(priv, "Not flushing - IPOIB_FLAG_ADMIN_UP not set.\n");
 		return;
 	}
@@ -1038,8 +1043,10 @@ static void __ipoib_ib_dev_flush(struct ipoib_dev_priv *priv,
 		ipoib_ib_dev_down(dev, 0);
 
 	if (level == IPOIB_FLUSH_HEAVY) {
-		ipoib_ib_dev_stop(dev, 0);
-		ipoib_ib_dev_open(dev);
+		if (test_bit(IPOIB_FLAG_INITIALIZED, &priv->flags))
+			ipoib_ib_dev_stop(dev, 0);
+		if (ipoib_ib_dev_open(dev) != 0)
+			return;
 	}
 
 	/*

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH v4 1/1] IPoIB: Avoid multicast join attempts when having invalid p_key

[Erez Shitrit]

Hi Alex,
i played with your patch over 3.16-rc1 and got a strange behaviour (i 
think there is a problem with the patch, it doesn't happened without the 
patch).

please try the next scenario:
1. pkey table contains only the default pkey
2. echo 0x8001 > /sys/class/net/ib0/create_child ; ifconfig ib0.8001 
1.1.1.1 up
3. change the sm partiotion file to include the new pkey value 8001
4. force the sm to send the new event (PKEY_CHANGE_EVENT) via pkill -HUP 
opensm

The new interface ib0.8001 remains "down" and its carrier is 0.
without the patch the carrier is "1"

please check that,

Thanks, Erez

6/17/2014 6:06 PM, Alex Estrin:
> Currently, the parent interface keeps sending broadcast group join
> requests even if p_key index 0 is invalid, which for itself is
> possible/common in virtualized environment where a VF has been probed to
> VM but the actual p_key configuration has not yet been assigned by the
> management software. This creates unnecessary noise on the fabric and in
> the kernel logs:
>
> ib0: multicast join failed for ff12:401b:8000:0000:0000:0000:ffff:ffff,
> status -22
>
> The original code run the multicast task regardless of the actual
> p_key value, which can be avoided. The fix is to re-init resources  and
> bring interface up only if p_key index 0 is valid either when starting
> up or on PKEY_CHANGE event.
>
> Fixes: c290414169 ('IPoIB: Fix pkey change flow for virtualization environments')
>
> Reviewed-by: Ira Weiny <ira.weiny-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
> Signed-off-by: Alex Estrin <alex.estrin-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
> ---
>   drivers/infiniband/ulp/ipoib/ipoib_ib.c |   35 +++++++++++++++++++------------
>   1 files changed, 21 insertions(+), 14 deletions(-)
>
> diff --git a/drivers/infiniband/ulp/ipoib/ipoib_ib.c b/drivers/infiniband/ulp/ipoib/ipoib_ib.c
> index 6a7003d..952db0b 100644
> --- a/drivers/infiniband/ulp/ipoib/ipoib_ib.c
> +++ b/drivers/infiniband/ulp/ipoib/ipoib_ib.c
> @@ -52,6 +52,7 @@ MODULE_PARM_DESC(data_debug_level,
>   #endif
>   
>   static DEFINE_MUTEX(pkey_mutex);
> +static void ipoib_pkey_dev_check_presence(struct net_device *dev);
>   
>   struct ipoib_ah *ipoib_create_ah(struct net_device *dev,
>   				 struct ib_pd *pd, struct ib_ah_attr *attr)
> @@ -669,12 +670,13 @@ int ipoib_ib_dev_open(struct net_device *dev)
>   	struct ipoib_dev_priv *priv = netdev_priv(dev);
>   	int ret;
>   
> -	if (ib_find_pkey(priv->ca, priv->port, priv->pkey, &priv->pkey_index)) {
> -		ipoib_warn(priv, "P_Key 0x%04x not found\n", priv->pkey);
> -		clear_bit(IPOIB_PKEY_ASSIGNED, &priv->flags);
> +	ipoib_pkey_dev_check_presence(dev);
> +
> +	if (!test_bit(IPOIB_PKEY_ASSIGNED, &priv->flags)) {
> +		ipoib_warn(priv, "P_Key 0x%04x is %s\n", priv->pkey,
> +			   !(priv->pkey & 0x7fff) ? "Invalid" : "not found");
>   		return -1;
>   	}
> -	set_bit(IPOIB_PKEY_ASSIGNED, &priv->flags);
>   
>   	ret = ipoib_init_qp(dev);
>   	if (ret) {
> @@ -712,9 +714,10 @@ dev_stop:
>   static void ipoib_pkey_dev_check_presence(struct net_device *dev)
>   {
>   	struct ipoib_dev_priv *priv = netdev_priv(dev);
> -	u16 pkey_index = 0;
>   
> -	if (ib_find_pkey(priv->ca, priv->port, priv->pkey, &pkey_index))
> +	if (!(priv->pkey & 0x7fff) ||
> +	    ib_find_pkey(priv->ca, priv->port, priv->pkey,
> +			 &priv->pkey_index))
>   		clear_bit(IPOIB_PKEY_ASSIGNED, &priv->flags);
>   	else
>   		set_bit(IPOIB_PKEY_ASSIGNED, &priv->flags);
> @@ -987,15 +990,17 @@ static void __ipoib_ib_dev_flush(struct ipoib_dev_priv *priv,
>   	up_read(&priv->vlan_rwsem);
>   
>   	if (!test_bit(IPOIB_FLAG_INITIALIZED, &priv->flags)) {
> -		/* for non-child devices must check/update the pkey value here */
> -		if (level == IPOIB_FLUSH_HEAVY &&
> -		    !test_bit(IPOIB_FLAG_SUBINTERFACE, &priv->flags))
> -			update_parent_pkey(priv);
> -		ipoib_dbg(priv, "Not flushing - IPOIB_FLAG_INITIALIZED not set.\n");
> -		return;
> +		if (level != IPOIB_FLUSH_HEAVY) {
> +			ipoib_dbg(priv, "Not flushing - IPOIB_FLAG_INITIALIZED not set.\n");
> +			return;
> +		}
>   	}
>   
>   	if (!test_bit(IPOIB_FLAG_ADMIN_UP, &priv->flags)) {
> +		/* interface is down. update pkey and leave. */
> +		if (level == IPOIB_FLUSH_HEAVY &&
> +		    !test_bit(IPOIB_FLAG_SUBINTERFACE, &priv->flags))
> +			update_parent_pkey(priv);
>   		ipoib_dbg(priv, "Not flushing - IPOIB_FLAG_ADMIN_UP not set.\n");
>   		return;
>   	}
> @@ -1038,8 +1043,10 @@ static void __ipoib_ib_dev_flush(struct ipoib_dev_priv *priv,
>   		ipoib_ib_dev_down(dev, 0);
>   
>   	if (level == IPOIB_FLUSH_HEAVY) {
> -		ipoib_ib_dev_stop(dev, 0);
> -		ipoib_ib_dev_open(dev);
> +		if (test_bit(IPOIB_FLAG_INITIALIZED, &priv->flags))
> +			ipoib_ib_dev_stop(dev, 0);
> +		if (ipoib_ib_dev_open(dev) != 0)
> +			return;
>   	}
>   
>   	/*
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html