[PATCH net-next] net: dsa: set random switch address

[PATCH net-next] net: dsa: set random switch address

[Vivien Didelot]

An Ethernet switch may support having a MAC address, which can be used
as the switch's source address in transmitted full-duplex Pause frames.

If a DSA switch supports the related .set_addr operation, the DSA core
sets the master's MAC address on the switch.

This won't make sense anymore in a multi-CPU ports system, because there
won't be a unique master device assigned to a switch tree.

To fix this, assign a random MAC address to the switch chip instead.

Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
---
 net/dsa/dsa2.c     |  8 +++-----
 net/dsa/dsa_priv.h |  1 +
 net/dsa/legacy.c   |  8 +++-----
 net/dsa/switch.c   | 17 +++++++++++++++++
 4 files changed, 24 insertions(+), 10 deletions(-)

diff --git a/net/dsa/dsa2.c b/net/dsa/dsa2.c
index 54ed054777bd..8e5780ddd7f9 100644
--- a/net/dsa/dsa2.c
+++ b/net/dsa/dsa2.c
@@ -336,11 +336,9 @@ static int dsa_ds_apply(struct dsa_switch_tree *dst, struct dsa_switch *ds)
 	if (err)
 		return err;
 
-	if (ds->ops->set_addr) {
-		err = ds->ops->set_addr(ds, dst->cpu_dp->netdev->dev_addr);
-		if (err < 0)
-			return err;
-	}
+	err = dsa_switch_set_addr(ds);
+	if (err)
+		return err;
 
 	if (!ds->slave_mii_bus && ds->ops->phy_read) {
 		ds->slave_mii_bus = devm_mdiobus_alloc(ds->dev);
diff --git a/net/dsa/dsa_priv.h b/net/dsa/dsa_priv.h
index 2850077cc9cc..9c4c17a4bd6b 100644
--- a/net/dsa/dsa_priv.h
+++ b/net/dsa/dsa_priv.h
@@ -172,6 +172,7 @@ void dsa_slave_unregister_notifier(void);
 /* switch.c */
 int dsa_switch_register_notifier(struct dsa_switch *ds);
 void dsa_switch_unregister_notifier(struct dsa_switch *ds);
+int dsa_switch_set_addr(struct dsa_switch *ds);
 
 /* tag_brcm.c */
 extern const struct dsa_device_ops brcm_netdev_ops;
diff --git a/net/dsa/legacy.c b/net/dsa/legacy.c
index 19ff6e0a21dc..340ca7997271 100644
--- a/net/dsa/legacy.c
+++ b/net/dsa/legacy.c
@@ -172,11 +172,9 @@ static int dsa_switch_setup_one(struct dsa_switch *ds,
 	if (ret)
 		return ret;
 
-	if (ops->set_addr) {
-		ret = ops->set_addr(ds, master->dev_addr);
-		if (ret < 0)
-			return ret;
-	}
+	ret = dsa_switch_set_addr(ds);
+	if (ret)
+		return ret;
 
 	if (!ds->slave_mii_bus && ops->phy_read) {
 		ds->slave_mii_bus = devm_mdiobus_alloc(ds->dev);
diff --git a/net/dsa/switch.c b/net/dsa/switch.c
index e6c06aa349a6..b45a26b006af 100644
--- a/net/dsa/switch.c
+++ b/net/dsa/switch.c
@@ -10,6 +10,7 @@
  * (at your option) any later version.
  */
 
+#include <linux/etherdevice.h>
 #include <linux/netdevice.h>
 #include <linux/notifier.h>
 #include <net/switchdev.h>
@@ -267,3 +268,19 @@ void dsa_switch_unregister_notifier(struct dsa_switch *ds)
 	if (err)
 		dev_err(ds->dev, "failed to unregister notifier (%d)\n", err);
 }
+
+int dsa_switch_set_addr(struct dsa_switch *ds)
+{
+	u8 addr[6];
+	int err;
+
+	if (ds->ops->set_addr) {
+		eth_random_addr(addr);
+
+		err = ds->ops->set_addr(ds, addr);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
-- 
2.14.2

Re: [PATCH net-next] net: dsa: set random switch address

[Florian Fainelli]

On 10/12/2017 03:10 PM, Vivien Didelot wrote:
> An Ethernet switch may support having a MAC address, which can be used
> as the switch's source address in transmitted full-duplex Pause frames.
> 
> If a DSA switch supports the related .set_addr operation, the DSA core
> sets the master's MAC address on the switch.
> 
> This won't make sense anymore in a multi-CPU ports system, because there
> won't be a unique master device assigned to a switch tree.

Thus far, everything you have said is true, but why we should do it,
that is: what if we don't, needs to be explained. Does that create a
problem with the generation of pause frames throughout the switch fabric?

> 
> To fix this, assign a random MAC address to the switch chip instead.

Maybe this is something that should be removed entirely from the DSA
core and pushed into the individual switch drivers instead. dsa_loop
implements it for code coverage, but that does not do anything.

set_addr is confusing in that you may think it could be used to program
the switch with the MAC address of the CPU/management port such that you
can disable MAC address learning on said port, but in fact, that's not
how it is used.

> 
> Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
> ---
>  net/dsa/dsa2.c     |  8 +++-----
>  net/dsa/dsa_priv.h |  1 +
>  net/dsa/legacy.c   |  8 +++-----
>  net/dsa/switch.c   | 17 +++++++++++++++++
>  4 files changed, 24 insertions(+), 10 deletions(-)
> 
> diff --git a/net/dsa/dsa2.c b/net/dsa/dsa2.c
> index 54ed054777bd..8e5780ddd7f9 100644
> --- a/net/dsa/dsa2.c
> +++ b/net/dsa/dsa2.c
> @@ -336,11 +336,9 @@ static int dsa_ds_apply(struct dsa_switch_tree *dst, struct dsa_switch *ds)
>  	if (err)
>  		return err;
>  
> -	if (ds->ops->set_addr) {
> -		err = ds->ops->set_addr(ds, dst->cpu_dp->netdev->dev_addr);
> -		if (err < 0)
> -			return err;
> -	}
> +	err = dsa_switch_set_addr(ds);
> +	if (err)
> +		return err;
>  
>  	if (!ds->slave_mii_bus && ds->ops->phy_read) {
>  		ds->slave_mii_bus = devm_mdiobus_alloc(ds->dev);
> diff --git a/net/dsa/dsa_priv.h b/net/dsa/dsa_priv.h
> index 2850077cc9cc..9c4c17a4bd6b 100644
> --- a/net/dsa/dsa_priv.h
> +++ b/net/dsa/dsa_priv.h
> @@ -172,6 +172,7 @@ void dsa_slave_unregister_notifier(void);
>  /* switch.c */
>  int dsa_switch_register_notifier(struct dsa_switch *ds);
>  void dsa_switch_unregister_notifier(struct dsa_switch *ds);
> +int dsa_switch_set_addr(struct dsa_switch *ds);
>  
>  /* tag_brcm.c */
>  extern const struct dsa_device_ops brcm_netdev_ops;
> diff --git a/net/dsa/legacy.c b/net/dsa/legacy.c
> index 19ff6e0a21dc..340ca7997271 100644
> --- a/net/dsa/legacy.c
> +++ b/net/dsa/legacy.c
> @@ -172,11 +172,9 @@ static int dsa_switch_setup_one(struct dsa_switch *ds,
>  	if (ret)
>  		return ret;
>  
> -	if (ops->set_addr) {
> -		ret = ops->set_addr(ds, master->dev_addr);
> -		if (ret < 0)
> -			return ret;
> -	}
> +	ret = dsa_switch_set_addr(ds);
> +	if (ret)
> +		return ret;
>  
>  	if (!ds->slave_mii_bus && ops->phy_read) {
>  		ds->slave_mii_bus = devm_mdiobus_alloc(ds->dev);
> diff --git a/net/dsa/switch.c b/net/dsa/switch.c
> index e6c06aa349a6..b45a26b006af 100644
> --- a/net/dsa/switch.c
> +++ b/net/dsa/switch.c
> @@ -10,6 +10,7 @@
>   * (at your option) any later version.
>   */
>  
> +#include <linux/etherdevice.h>
>  #include <linux/netdevice.h>
>  #include <linux/notifier.h>
>  #include <net/switchdev.h>
> @@ -267,3 +268,19 @@ void dsa_switch_unregister_notifier(struct dsa_switch *ds)
>  	if (err)
>  		dev_err(ds->dev, "failed to unregister notifier (%d)\n", err);
>  }
> +
> +int dsa_switch_set_addr(struct dsa_switch *ds)
> +{
> +	u8 addr[6];
> +	int err;
> +
> +	if (ds->ops->set_addr) {
> +		eth_random_addr(addr);
> +
> +		err = ds->ops->set_addr(ds, addr);
> +		if (err)
> +			return err;
> +	}
> +
> +	return 0;
> +}
> 


-- 
Florian

Re: [PATCH net-next] net: dsa: set random switch address

[Vivien Didelot]

Hi Florian,

Florian Fainelli <f.fainelli@gmail.com> writes:

> On 10/12/2017 03:10 PM, Vivien Didelot wrote:
>> An Ethernet switch may support having a MAC address, which can be used
>> as the switch's source address in transmitted full-duplex Pause frames.
>> 
>> If a DSA switch supports the related .set_addr operation, the DSA core
>> sets the master's MAC address on the switch.
>> 
>> This won't make sense anymore in a multi-CPU ports system, because there
>> won't be a unique master device assigned to a switch tree.
>
> Thus far, everything you have said is true, but why we should do it,
> that is: what if we don't, needs to be explained. Does that create a
> problem with the generation of pause frames throughout the switch fabric?
>
>> 
>> To fix this, assign a random MAC address to the switch chip instead.
>
> Maybe this is something that should be removed entirely from the DSA
> core and pushed into the individual switch drivers instead. dsa_loop
> implements it for code coverage, but that does not do anything.
>
> set_addr is confusing in that you may think it could be used to program
> the switch with the MAC address of the CPU/management port such that you
> can disable MAC address learning on said port, but in fact, that's not
> how it is used.

You are correct. So what I can do is assign a random MAC address in the
Marvell driver, remove the .set_addr implementation of mv88e6xxx and
dsa_loop, and finally remove this code from DSA core completely.


Thanks,

        Vivien

Re: [PATCH net-next] net: dsa: set random switch address

[Florian Fainelli]

On 10/12/2017 03:35 PM, Vivien Didelot wrote:
> Hi Florian,
> 
> Florian Fainelli <f.fainelli@gmail.com> writes:
> 
>> On 10/12/2017 03:10 PM, Vivien Didelot wrote:
>>> An Ethernet switch may support having a MAC address, which can be used
>>> as the switch's source address in transmitted full-duplex Pause frames.
>>>
>>> If a DSA switch supports the related .set_addr operation, the DSA core
>>> sets the master's MAC address on the switch.
>>>
>>> This won't make sense anymore in a multi-CPU ports system, because there
>>> won't be a unique master device assigned to a switch tree.
>>
>> Thus far, everything you have said is true, but why we should do it,
>> that is: what if we don't, needs to be explained. Does that create a
>> problem with the generation of pause frames throughout the switch fabric?
>>
>>>
>>> To fix this, assign a random MAC address to the switch chip instead.
>>
>> Maybe this is something that should be removed entirely from the DSA
>> core and pushed into the individual switch drivers instead. dsa_loop
>> implements it for code coverage, but that does not do anything.
>>
>> set_addr is confusing in that you may think it could be used to program
>> the switch with the MAC address of the CPU/management port such that you
>> can disable MAC address learning on said port, but in fact, that's not
>> how it is used.
> 
> You are correct. So what I can do is assign a random MAC address in the
> Marvell driver, remove the .set_addr implementation of mv88e6xxx and
> dsa_loop, and finally remove this code from DSA core completely.

Works for me, thanks!
-- 
Florian