[PATCH] dm9601: warn on invalid mac address

[PATCH] dm9601: warn on invalid mac address

[Wu Fengguang]

Add warnings on invalid mac address to help disclose/debug problems.

Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
---
 drivers/net/usb/dm9601.c |   12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

--- linux-2.6.orig/drivers/net/usb/dm9601.c
+++ linux-2.6/drivers/net/usb/dm9601.c
@@ -401,8 +401,12 @@ static int dm9601_set_mac_address(struct
 	struct sockaddr *addr = p;
 	struct usbnet *dev = netdev_priv(net);
 
-	if (!is_valid_ether_addr(addr->sa_data))
+	if (!is_valid_ether_addr(addr->sa_data)) {
+		DECLARE_MAC_BUF(mac_buf);
+		print_mac(mac_buf, addr->sa_data);
+		dev_warn(&net->dev, "not setting invalid mac address %s\n", mac_buf);
 		return -EINVAL;
+	}
 
 	memcpy(net->dev_addr, addr->sa_data, net->addr_len);
 	dm_write_async(dev, DM_PHY_ADDR, net->addr_len, net->dev_addr);
@@ -449,6 +453,12 @@ static int dm9601_bind(struct usbnet *de
 	 */
 	if (is_valid_ether_addr(mac))
 		memcpy(dev->net->dev_addr, mac, ETH_ALEN);
+	else {
+		DECLARE_MAC_BUF(mac_buf);
+		print_mac(mac_buf, mac);
+		devdbg(dev, "EEPROM reported mac address %s is invalid,"
+				" use the randomly generated one.", mac_buf);
+	}
 
 	/* power up phy */
 	dm_write_reg(dev, DM_GPR_CTRL, 1);

Re: [PATCH] dm9601: warn on invalid mac address

[Peter Korsgaard]

>>>>> "Wu" == Wu Fengguang <wfg@linux.intel.com> writes:

Hi,

 Wu> Add warnings on invalid mac address to help disclose/debug problems.
 Wu> Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
 Wu> ---
 Wu>  drivers/net/usb/dm9601.c |   12 +++++++++++-
 Wu>  1 file changed, 11 insertions(+), 1 deletion(-)

 Wu> --- linux-2.6.orig/drivers/net/usb/dm9601.c
 Wu> +++ linux-2.6/drivers/net/usb/dm9601.c
 Wu> @@ -401,8 +401,12 @@ static int dm9601_set_mac_address(struct
 Wu>  	struct sockaddr *addr = p;
 Wu>  	struct usbnet *dev = netdev_priv(net);
 
 Wu> -	if (!is_valid_ether_addr(addr->sa_data))
 Wu> +	if (!is_valid_ether_addr(addr->sa_data)) {
 Wu> +		DECLARE_MAC_BUF(mac_buf);
 Wu> +		print_mac(mac_buf, addr->sa_data);
 Wu> +		dev_warn(&net->dev, "not setting invalid mac address %s\n", mac_buf);

This should be an error and not a warning.
Notice that print_mac returns the string, so you can do:

dev_err(&net->dev, "... %s", print_mac(mac_buf, addr->sa_data));
 
 Wu>  	memcpy(net->dev_addr, addr->sa_data, net->addr_len);
 Wu>  	dm_write_async(dev, DM_PHY_ADDR, net->addr_len, net->dev_addr);
 Wu> @@ -449,6 +453,12 @@ static int dm9601_bind(struct usbnet *de
 Wu>  	 */
 Wu>  	if (is_valid_ether_addr(mac))
 Wu>  		memcpy(dev->net->dev_addr, mac, ETH_ALEN);
 Wu> +	else {
 Wu> +		DECLARE_MAC_BUF(mac_buf);
 Wu> +		print_mac(mac_buf, mac);
 Wu> +		devdbg(dev, "EEPROM reported mac address %s is invalid,"
 Wu> +				" use the randomly generated one.", mac_buf);

And this should be a warning.

-- 
Bye, Peter Korsgaard

Re: [PATCH] dm9601: warn on invalid mac address

[Wu Fengguang]

On Tue, Jan 06, 2009 at 10:18:17AM +0100, Peter Korsgaard wrote:
> >>>>> "Wu" == Wu Fengguang <wfg@linux.intel.com> writes:
> 
> Hi,
> 
>  Wu> Add warnings on invalid mac address to help disclose/debug problems.
>  Wu> Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
>  Wu> ---
>  Wu>  drivers/net/usb/dm9601.c |   12 +++++++++++-
>  Wu>  1 file changed, 11 insertions(+), 1 deletion(-)
> 
>  Wu> --- linux-2.6.orig/drivers/net/usb/dm9601.c
>  Wu> +++ linux-2.6/drivers/net/usb/dm9601.c
>  Wu> @@ -401,8 +401,12 @@ static int dm9601_set_mac_address(struct
>  Wu>  	struct sockaddr *addr = p;
>  Wu>  	struct usbnet *dev = netdev_priv(net);
>  
>  Wu> -	if (!is_valid_ether_addr(addr->sa_data))
>  Wu> +	if (!is_valid_ether_addr(addr->sa_data)) {
>  Wu> +		DECLARE_MAC_BUF(mac_buf);
>  Wu> +		print_mac(mac_buf, addr->sa_data);
>  Wu> +		dev_warn(&net->dev, "not setting invalid mac address %s\n", mac_buf);
> 
> This should be an error and not a warning.
> Notice that print_mac returns the string, so you can do:
> 
> dev_err(&net->dev, "... %s", print_mac(mac_buf, addr->sa_data));

OK.

>  Wu>  	memcpy(net->dev_addr, addr->sa_data, net->addr_len);
>  Wu>  	dm_write_async(dev, DM_PHY_ADDR, net->addr_len, net->dev_addr);
>  Wu> @@ -449,6 +453,12 @@ static int dm9601_bind(struct usbnet *de
>  Wu>  	 */
>  Wu>  	if (is_valid_ether_addr(mac))
>  Wu>  		memcpy(dev->net->dev_addr, mac, ETH_ALEN);
>  Wu> +	else {
>  Wu> +		DECLARE_MAC_BUF(mac_buf);
>  Wu> +		print_mac(mac_buf, mac);
>  Wu> +		devdbg(dev, "EEPROM reported mac address %s is invalid,"
>  Wu> +				" use the randomly generated one.", mac_buf);
> 
> And this should be a warning.

Then let the warning message appear repeatedly for some devices?

Also dev_warn() won't be able to show the device name at that time,
like this:

[28489.062180]  : EEPROM reported mac address ff:ff:ff:ff:ff:ff is invalid, use the randomly generated one.

Thanks,
Fengguang

Re: [PATCH] dm9601: warn on invalid mac address

[Peter Korsgaard]

>>>>> "Wu" == Wu Fengguang <wfg@linux.intel.com> writes:

Hi,

 Wu> if (is_valid_ether_addr(mac))
 Wu> memcpy(dev->net->dev_addr, mac, ETH_ALEN);
 Wu> +	else {
 Wu> +		DECLARE_MAC_BUF(mac_buf);
 Wu> +		print_mac(mac_buf, mac);
 Wu> +		devdbg(dev, "EEPROM reported mac address %s is invalid,"
 Wu> +				" use the randomly generated one.", mac_buf);
 >> 
 >> And this should be a warning.

 Wu> Then let the warning message appear repeatedly for some devices?

This is called at probe time - But yes, I think it makes sense to print
it.

We should print the random address instead of the ff's though.

 Wu> Also dev_warn() won't be able to show the device name at that time,
 Wu> like this:

Ah yes, that's presumably why I used a raw printk just above.

 Wu> [28489.062180] : EEPROM reported mac address ff:ff:ff:ff:ff:ff is
 Wu> invalid, use the randomly generated one.

I would prefer something like:

printk(KERN_WARNING "dm9601: No valid MAC address in EEPROM, using %s\n",
       print_mac(..));

Also, it seems like you're not writing the random address to the
hardware registers, so you won't be able to receive any unicast -
You'll need to add a call to dm9601_set_mac_address() or similar.

-- 
Bye, Peter Korsgaard

Re: [PATCH] dm9601: warn on invalid mac address

[Ben Hutchings]

On Tue, 2009-01-06 at 10:18 +0100, Peter Korsgaard wrote:
> >>>>> "Wu" == Wu Fengguang <wfg@linux.intel.com> writes:
> 
> Hi,
> 
>  Wu> Add warnings on invalid mac address to help disclose/debug problems.
>  Wu> Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
>  Wu> ---
>  Wu>  drivers/net/usb/dm9601.c |   12 +++++++++++-
>  Wu>  1 file changed, 11 insertions(+), 1 deletion(-)
> 
>  Wu> --- linux-2.6.orig/drivers/net/usb/dm9601.c
>  Wu> +++ linux-2.6/drivers/net/usb/dm9601.c
>  Wu> @@ -401,8 +401,12 @@ static int dm9601_set_mac_address(struct
>  Wu>  	struct sockaddr *addr = p;
>  Wu>  	struct usbnet *dev = netdev_priv(net);
>  
>  Wu> -	if (!is_valid_ether_addr(addr->sa_data))
>  Wu> +	if (!is_valid_ether_addr(addr->sa_data)) {
>  Wu> +		DECLARE_MAC_BUF(mac_buf);
>  Wu> +		print_mac(mac_buf, addr->sa_data);
>  Wu> +		dev_warn(&net->dev, "not setting invalid mac address %s\n", mac_buf);
> 
> This should be an error and not a warning.
> Notice that print_mac returns the string, so you can do:
> 
> dev_err(&net->dev, "... %s", print_mac(mac_buf, addr->sa_data));

print_mac() is already obsolete; use %pM in the format string instead.

Ben.

-- 
Ben Hutchings, Senior Software Engineer, Solarflare Communications
Not speaking for my employer; that's the marketing department's job.
They asked us to note that Solarflare product names are trademarked.

Re: [PATCH] dm9601: warn on invalid mac address

[David Miller]

From: Ben Hutchings <bhutchings@solarflare.com>
Date: Tue, 06 Jan 2009 11:52:35 +0000

> On Tue, 2009-01-06 at 10:18 +0100, Peter Korsgaard wrote:
> > >>>>> "Wu" == Wu Fengguang <wfg@linux.intel.com> writes:
> > 
> > Hi,
> > 
> >  Wu> Add warnings on invalid mac address to help disclose/debug problems.
> >  Wu> Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
> >  Wu> ---
> >  Wu>  drivers/net/usb/dm9601.c |   12 +++++++++++-
> >  Wu>  1 file changed, 11 insertions(+), 1 deletion(-)
> > 
> >  Wu> --- linux-2.6.orig/drivers/net/usb/dm9601.c
> >  Wu> +++ linux-2.6/drivers/net/usb/dm9601.c
> >  Wu> @@ -401,8 +401,12 @@ static int dm9601_set_mac_address(struct
> >  Wu>  	struct sockaddr *addr = p;
> >  Wu>  	struct usbnet *dev = netdev_priv(net);
> >  
> >  Wu> -	if (!is_valid_ether_addr(addr->sa_data))
> >  Wu> +	if (!is_valid_ether_addr(addr->sa_data)) {
> >  Wu> +		DECLARE_MAC_BUF(mac_buf);
> >  Wu> +		print_mac(mac_buf, addr->sa_data);
> >  Wu> +		dev_warn(&net->dev, "not setting invalid mac address %s\n", mac_buf);
> > 
> > This should be an error and not a warning.
> > Notice that print_mac returns the string, so you can do:
> > 
> > dev_err(&net->dev, "... %s", print_mac(mac_buf, addr->sa_data));
> 
> print_mac() is already obsolete; use %pM in the format string instead.

Yep, please fix this.

Re: [PATCH] dm9601: warn on invalid mac address

[Wu Fengguang]

On Tue, Jan 06, 2009 at 12:17:56PM +0100, Peter Korsgaard wrote:
> >>>>> "Wu" == Wu Fengguang <wfg@linux.intel.com> writes:
> 
> Hi,
> 
>  Wu> if (is_valid_ether_addr(mac))
>  Wu> memcpy(dev->net->dev_addr, mac, ETH_ALEN);
>  Wu> +	else {
>  Wu> +		DECLARE_MAC_BUF(mac_buf);
>  Wu> +		print_mac(mac_buf, mac);
>  Wu> +		devdbg(dev, "EEPROM reported mac address %s is invalid,"
>  Wu> +				" use the randomly generated one.", mac_buf);
>  >> 
>  >> And this should be a warning.
> 
>  Wu> Then let the warning message appear repeatedly for some devices?
> 
> This is called at probe time - But yes, I think it makes sense to print
> it.
> 
> We should print the random address instead of the ff's though.
> 
>  Wu> Also dev_warn() won't be able to show the device name at that time,
>  Wu> like this:
> 
> Ah yes, that's presumably why I used a raw printk just above.
> 
>  Wu> [28489.062180] : EEPROM reported mac address ff:ff:ff:ff:ff:ff is
>  Wu> invalid, use the randomly generated one.
> 
> I would prefer something like:
> 
> printk(KERN_WARNING "dm9601: No valid MAC address in EEPROM, using %s\n",
>        print_mac(..));

This looks better :-)

> Also, it seems like you're not writing the random address to the
> hardware registers, so you won't be able to receive any unicast -
> You'll need to add a call to dm9601_set_mac_address() or similar.

Good catch. I had wanted to ask why it only works with promisc mode ;-)

Here is the tested patch.

Thanks,
Fengguang
---
dm9601: tell HW about random generated mac address

Otherwise unicast RX will only work in promisc mode.

Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
---
 drivers/net/usb/dm9601.c |    9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

--- linux-2.6.orig/drivers/net/usb/dm9601.c
+++ linux-2.6/drivers/net/usb/dm9601.c
@@ -396,6 +396,11 @@ static void dm9601_set_multicast(struct 
 	dm_write_reg_async(dev, DM_RX_CTRL, rx_ctl);
 }
 
+static void __dm9601_set_mac_address(struct usbnet *dev)
+{
+	dm_write_async(dev, DM_PHY_ADDR, ETH_ALEN, dev->net->dev_addr);
+}
+
 static int dm9601_set_mac_address(struct net_device *net, void *p)
 {
 	struct sockaddr *addr = p;
@@ -405,7 +410,7 @@ static int dm9601_set_mac_address(struct
 		return -EINVAL;
 
 	memcpy(net->dev_addr, addr->sa_data, net->addr_len);
-	dm_write_async(dev, DM_PHY_ADDR, net->addr_len, net->dev_addr);
+	__dm9601_set_mac_address(dev);
 
 	return 0;
 }
@@ -449,6 +454,8 @@ static int dm9601_bind(struct usbnet *de
 	 */
 	if (is_valid_ether_addr(mac))
 		memcpy(dev->net->dev_addr, mac, ETH_ALEN);
+	else
+		__dm9601_set_mac_address(dev);
 
 	/* power up phy */
 	dm_write_reg(dev, DM_GPR_CTRL, 1);

Re: [PATCH] dm9601: warn on invalid mac address

[Wu Fengguang]

On Tue, Jan 06, 2009 at 10:52:26AM -0800, David Miller wrote:
> From: Ben Hutchings <bhutchings@solarflare.com>
> Date: Tue, 06 Jan 2009 11:52:35 +0000
> 
> > On Tue, 2009-01-06 at 10:18 +0100, Peter Korsgaard wrote:
> > > >>>>> "Wu" == Wu Fengguang <wfg@linux.intel.com> writes:
> > > 
> > > Hi,
> > > 
> > >  Wu> Add warnings on invalid mac address to help disclose/debug problems.
> > >  Wu> Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
> > >  Wu> ---
> > >  Wu>  drivers/net/usb/dm9601.c |   12 +++++++++++-
> > >  Wu>  1 file changed, 11 insertions(+), 1 deletion(-)
> > > 
> > >  Wu> --- linux-2.6.orig/drivers/net/usb/dm9601.c
> > >  Wu> +++ linux-2.6/drivers/net/usb/dm9601.c
> > >  Wu> @@ -401,8 +401,12 @@ static int dm9601_set_mac_address(struct
> > >  Wu>  	struct sockaddr *addr = p;
> > >  Wu>  	struct usbnet *dev = netdev_priv(net);
> > >  
> > >  Wu> -	if (!is_valid_ether_addr(addr->sa_data))
> > >  Wu> +	if (!is_valid_ether_addr(addr->sa_data)) {
> > >  Wu> +		DECLARE_MAC_BUF(mac_buf);
> > >  Wu> +		print_mac(mac_buf, addr->sa_data);
> > >  Wu> +		dev_warn(&net->dev, "not setting invalid mac address %s\n", mac_buf);
> > > 
> > > This should be an error and not a warning.
> > > Notice that print_mac returns the string, so you can do:
> > > 
> > > dev_err(&net->dev, "... %s", print_mac(mac_buf, addr->sa_data));
> > 
> > print_mac() is already obsolete; use %pM in the format string instead.
> 
> Yep, please fix this.

Thanks for the tip. Here is the updated patch.

Thanks,
Fengguang
---
dm9601: warn on invalid mac address

Add warnings on invalid mac address to help disclose/debug problems.

Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
---
 drivers/net/usb/dm9601.c |   11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

--- linux-2.6.orig/drivers/net/usb/dm9601.c
+++ linux-2.6/drivers/net/usb/dm9601.c
@@ -406,8 +406,11 @@ static int dm9601_set_mac_address(struct
 	struct sockaddr *addr = p;
 	struct usbnet *dev = netdev_priv(net);
 
-	if (!is_valid_ether_addr(addr->sa_data))
+	if (!is_valid_ether_addr(addr->sa_data)) {
+		dev_err(&net->dev, "not setting invalid mac address %pM\n",
+								addr->sa_data);
 		return -EINVAL;
+	}
 
 	memcpy(net->dev_addr, addr->sa_data, net->addr_len);
 	__dm9601_set_mac_address(dev);
@@ -454,8 +457,12 @@ static int dm9601_bind(struct usbnet *de
 	 */
 	if (is_valid_ether_addr(mac))
 		memcpy(dev->net->dev_addr, mac, ETH_ALEN);
-	else
+	else {
+		printk(KERN_WARNING
+			"dm9601: No valid MAC address in EEPROM, using %pM\n",
+			dev->net->dev_addr);
 		__dm9601_set_mac_address(dev);
+	}
 
 	/* power up phy */
 	dm_write_reg(dev, DM_GPR_CTRL, 1);

Re: [PATCH] dm9601: warn on invalid mac address

[Peter Korsgaard]

>>>>> "Wu" == Wu Fengguang <wfg@linux.intel.com> writes:

Hi,

 >> Also, it seems like you're not writing the random address to the
 >> hardware registers, so you won't be able to receive any unicast -
 >> You'll need to add a call to dm9601_set_mac_address() or similar.

 Wu> Good catch. I had wanted to ask why it only works with promisc
 Wu> mode ;-)

;)

 Wu> Here is the tested patch.

 Wu> Thanks,
 Wu> Fengguang
 Wu> ---
 Wu> dm9601: tell HW about random generated mac address

 Wu> Otherwise unicast RX will only work in promisc mode.

 Wu> Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
 Wu> Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>

Acked-by: Peter Korsgaard <jacmet@sunsite.dk>

-- 
Bye, Peter Korsgaard

Re: [PATCH] dm9601: warn on invalid mac address

[Peter Korsgaard]

>>>>> "Wu" == Wu Fengguang <wfg@linux.intel.com> writes:

 Wu> Thanks for the tip. Here is the updated patch.

 Wu> Thanks,
 Wu> Fengguang
 Wu> ---
 Wu> dm9601: warn on invalid mac address

 Wu> Add warnings on invalid mac address to help disclose/debug problems.

 Wu> Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>

Acked-by: Peter Korsgaard <jacmet@sunsite.dk>

-- 
Bye, Peter Korsgaard

Re: [PATCH] dm9601: warn on invalid mac address

[David Miller]

From: Peter Korsgaard <jacmet@sunsite.dk>
Date: Wed, 07 Jan 2009 09:24:43 +0100

> >>>>> "Wu" == Wu Fengguang <wfg@linux.intel.com> writes:
> 
>  Wu> dm9601: tell HW about random generated mac address
> 
>  Wu> Otherwise unicast RX will only work in promisc mode.
> 
>  Wu> Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
>  Wu> Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
> 
> Acked-by: Peter Korsgaard <jacmet@sunsite.dk>

Applied, thanks everyone.

Re: [PATCH] dm9601: warn on invalid mac address

[David Miller]

From: Peter Korsgaard <jacmet@sunsite.dk>
Date: Wed, 07 Jan 2009 09:25:03 +0100

> >>>>> "Wu" == Wu Fengguang <wfg@linux.intel.com> writes:
> 
>  Wu> Thanks for the tip. Here is the updated patch.
> 
>  Wu> Thanks,
>  Wu> Fengguang
>  Wu> ---
>  Wu> dm9601: warn on invalid mac address
> 
>  Wu> Add warnings on invalid mac address to help disclose/debug problems.
> 
>  Wu> Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
> 
> Acked-by: Peter Korsgaard <jacmet@sunsite.dk>

Also applied, thanks.