Setting alternate MAC addresses on e1000

Setting alternate MAC addresses on e1000

[Carsten Aulbert]

Hi,

I know that quite a few e1000 experts are reading here, so I hope to 
have some luck here:

I need to reprogram the MAC addresses of several 82573E/L NICs. I tried 
the obvious approach first using ethtool -E to write the numbers into 
the first six bytes and this works quite ok [1], except that I need to 
specifically change the second MAC not necessarily the first one. This 
approach only changed the first NIC and left the second unchanged.

I've browsed http://download.intel.com/design/network/manuals/31608004.pdf
quite a bit, but except finding that I seem to need to set word 0x37 to 
point to 12 free bytes I have not come up with much. For example, I 
don't know where I would find those 12 free bytes.

Anyone with a different approach or a hint how to succeed?

TIA

Carsten

[1] Along the lines of
ethtool -E eth0 magic $MAGIC offset 0x0 value 0x${numbers[0]}
[...]
where magic is MAGIC=0x108c8086

However this does not work on eth1 :(

RE: Setting alternate MAC addresses on e1000

[Brandeburg, Jesse]

Carsten Aulbert wrote:
> where magic is MAGIC=0x108c8086
> 
> However this does not work on eth1 :(

I think you need a different MAGIC for eth1 since they are separate
parts (different device ID) with separate eeproms, then the first 6 byte
change should work for eth1 too.

in the future e1000-devel@lists.sf.net might be more appropriate to
Intel specific hardware questions.  but copying netdev is still okay.

Thanks,
  Jesse

Re: Setting alternate MAC addresses on e1000

[Carsten Aulbert]

Hi Jesse,

thanks for the hint, I was stupid to think these two NIC are just using 
the very same EEPROM.

Brandeburg, Jesse wrote:
> Carsten Aulbert wrote:
>> where magic is MAGIC=0x108c8086
>>
>> However this does not work on eth1 :(
> 
> I think you need a different MAGIC for eth1 since they are separate
> parts (different device ID) with separate eeproms, then the first 6 byte
> change should work for eth1 too.
> 

Yes, with MAGIC=0x109a8086 I can set the MAC address of eth1.

> in the future e1000-devel@lists.sf.net might be more appropriate to
> Intel specific hardware questions.  but copying netdev is still okay.

Will do, starting with a closely related question:

The Intel specs state that word 37H is a pointer to an "alternate MAC 
address". My boss things it would be wise to let the original MAC 
address in place and just add another MAC address, so that we can easily 
revert to the old "official" one by writing 0xffff into that word.

Has anyone experience with that approach? In principle that should be 
easy, however, I don't know where/how to find 6 free bytes from the 
table given in Intel's docs.

Thanks again and hopefully someone has some light for my brain ;)

Cheers

Carsten

Re: Setting alternate MAC addresses on e1000

[Brandeburg, Jesse]

Carsten Aulbert wrote:
> Yes, with MAGIC=0x109a8086 I can set the MAC address of eth1.

good, I think you need to just use that way.
 
> The Intel specs state that word 37H is a pointer to an "alternate MAC
> address". My boss things it would be wise to let the original MAC
> address in place and just add another MAC address, so that we can
> easily revert to the old "official" one by writing 0xffff into that
> word. 

The alternate MAC address typically refers to a MAC used only for
management/ASF/IPMI traffic.  I don't think that is what you want.

According to our internal documents that I have access to, word 37h is
not what you think it is on 82573V/L.

PXE Words (words 30h thru 3Eh)
Words 30h through 3Eh have been reserved for configuration and version
values to be used by PXE code.
 
> Has anyone experience with that approach? In principle that should be
> easy, however, I don't know where/how to find 6 free bytes from the
> table given in Intel's docs.
> 
> Thanks again and hopefully someone has some light for my brain ;)

I think that you should stick with programming the original MAC address.
Start messing around with other words and you're bound to get yourself
into trouble.

Jesse

-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace

Re: Setting alternate MAC addresses on e1000

[Andy Gospodarek]

On Tue, Apr 01, 2008 at 03:15:54PM +0200, Carsten Aulbert wrote:
> Hi,
> 
> I know that quite a few e1000 experts are reading here, so I hope to 
> have some luck here:
> 
> I need to reprogram the MAC addresses of several 82573E/L NICs. I tried 
> the obvious approach first using ethtool -E to write the numbers into 
> the first six bytes and this works quite ok [1], except that I need to 
> specifically change the second MAC not necessarily the first one. This 
> approach only changed the first NIC and left the second unchanged.
> 
> I've browsed http://download.intel.com/design/network/manuals/31608004.pdf
> quite a bit, but except finding that I seem to need to set word 0x37 to 
> point to 12 free bytes I have not come up with much. For example, I 
> don't know where I would find those 12 free bytes.
> 
> Anyone with a different approach or a hint how to succeed?
> 
> TIA
> 
> Carsten
> 

Bill posted a patch for this a while back (but it was not accepted since
support for that hardware is offically moving to e1000e).  This may be
what you want....




Port alternate MAC address support from the sourceforge 
e1000 driver to the upstream e1000 driver.

Signed-off-by: Bill Hayes <bill.hayes@hp.com>
---

 drivers/net/e1000/e1000_hw.c |   42 +++++++++++++++++++++++++++++++++++++++---
 drivers/net/e1000/e1000_hw.h |    2 ++
 2 files changed, 41 insertions(+), 3 deletions(-)

diff --git a/drivers/net/e1000/e1000_hw.c b/drivers/net/e1000/e1000_hw.c
index 8fa0fe4..07e3178 100644
--- a/drivers/net/e1000/e1000_hw.c
+++ b/drivers/net/e1000/e1000_hw.c
@@ -719,6 +719,11 @@ e1000_reset_hw(struct e1000_hw *hw)
     DEBUGOUT("Masking off all interrupts\n");
     E1000_WRITE_REG(hw, IMC, 0xffffffff);
 
+    if (hw->mac_type == e1000_82571 && hw->alt_mac_addr_is_present) {
+    	hw->laa_is_present = 1;
+    	e1000_rar_set(hw, hw->mac_addr, E1000_RAR_ENTRIES - 1);
+    }
+
     /* Clear any pending interrupt events. */
     icr = E1000_READ_REG(hw, ICR);
 
@@ -5693,11 +5698,41 @@ e1000_read_mac_addr(struct e1000_hw * hw)
 {
     uint16_t offset;
     uint16_t eeprom_data, i;
+    u16 mac_addr_offset = 0;
 
     DEBUGFUNC("e1000_read_mac_addr");
 
+    if (hw->mac_type == e1000_82571) {
+        /* Check for an alternate MAC address.  An alternate MAC address can
+         * be setup by pre-boot software and must be treated like a permanent
+         * address and must override the actual permanent MAC address. */
+        if (e1000_read_eeprom(hw, EEPROM_ALT_MAC_ADDR_PTR, 1,
+                              &mac_addr_offset) < 0) {
+            DEBUGOUT("EEPROM Read Error\n");
+            return -E1000_ERR_EEPROM;
+        }
+        if (mac_addr_offset == 0xFFFF)
+            mac_addr_offset = 0;
+
+        if (mac_addr_offset) {
+            if (E1000_READ_REG(hw, STATUS) & E1000_STATUS_FUNC_1)
+                mac_addr_offset += NODE_ADDRESS_SIZE/sizeof(u16);
+
+            /* make sure we have a valid mac address here before using it */
+            if (e1000_read_eeprom(hw, mac_addr_offset, 1, &eeprom_data) < 0) {
+                DEBUGOUT("EEPROM Read Error\n");
+                return -E1000_ERR_EEPROM;
+            }
+            if (eeprom_data & 0x0001)
+                mac_addr_offset = 0;
+	}
+
+        if (mac_addr_offset) 
+                hw->alt_mac_addr_is_present = 1;
+    }
+
     for (i = 0; i < NODE_ADDRESS_SIZE; i += 2) {
-        offset = i >> 1;
+        offset = mac_addr_offset + (i >> 1);
         if (e1000_read_eeprom(hw, offset, 1, &eeprom_data) < 0) {
             DEBUGOUT("EEPROM Read Error\n");
             return -E1000_ERR_EEPROM;
@@ -5713,8 +5748,9 @@ e1000_read_mac_addr(struct e1000_hw * hw)
     case e1000_82546_rev_3:
     case e1000_82571:
     case e1000_80003es2lan:
-        if (E1000_READ_REG(hw, STATUS) & E1000_STATUS_FUNC_1)
-            hw->perm_mac_addr[5] ^= 0x01;
+        if (!mac_addr_offset &&
+            (E1000_READ_REG(hw, STATUS) & E1000_STATUS_FUNC_1))
+                hw->perm_mac_addr[5] ^= 0x01;
         break;
     }
 
diff --git a/drivers/net/e1000/e1000_hw.h b/drivers/net/e1000/e1000_hw.h
index a2a86c5..e18760d 100644
--- a/drivers/net/e1000/e1000_hw.h
+++ b/drivers/net/e1000/e1000_hw.h
@@ -1456,6 +1456,7 @@ struct e1000_hw {
 	boolean_t		tbi_compatibility_en;
 	boolean_t		tbi_compatibility_on;
 	boolean_t		laa_is_present;
+	boolean_t		alt_mac_addr_is_present;
 	boolean_t		phy_reset_disable;
 	boolean_t		initialize_hw_bits_disable;
 	boolean_t		fc_send_xon;
@@ -2286,6 +2287,7 @@ struct e1000_host_command_info {
 #define EEPROM_INIT_CONTROL3_PORT_A   0x0024
 #define EEPROM_CFG                    0x0012
 #define EEPROM_FLASH_VERSION          0x0032
+#define EEPROM_ALT_MAC_ADDR_PTR       0x0037
 #define EEPROM_CHECKSUM_REG           0x003F
 
 #define E1000_EEPROM_CFG_DONE         0x00040000   /* MNG config cycle done */

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

Re: Setting alternate MAC addresses on e1000

[Kok, Auke]

Andy Gospodarek wrote:
> On Tue, Apr 01, 2008 at 03:15:54PM +0200, Carsten Aulbert wrote:
>> Hi,
>>
>> I know that quite a few e1000 experts are reading here, so I hope to 
>> have some luck here:
>>
>> I need to reprogram the MAC addresses of several 82573E/L NICs. I tried 
>> the obvious approach first using ethtool -E to write the numbers into 
>> the first six bytes and this works quite ok [1], except that I need to 
>> specifically change the second MAC not necessarily the first one. This 
>> approach only changed the first NIC and left the second unchanged.
>>
>> I've browsed http://download.intel.com/design/network/manuals/31608004.pdf
>> quite a bit, but except finding that I seem to need to set word 0x37 to 
>> point to 12 free bytes I have not come up with much. For example, I 
>> don't know where I would find those 12 free bytes.
>>
>> Anyone with a different approach or a hint how to succeed?
>>
>> TIA
>>
>> Carsten
>>
> 
> Bill posted a patch for this a while back (but it was not accepted since
> support for that hardware is offically moving to e1000e).  This may be
> what you want....

yeah I thought of that as well and after digging through the documentation I don't
think this will work for 82573's - it seems that only 82571 hardware supports this
feature, which is not what Carsten needs.

as stated above, the alternate eeprom layout of the 82573 seems to also indicate that.

Auke



> Port alternate MAC address support from the sourceforge 
> e1000 driver to the upstream e1000 driver.
> 
> Signed-off-by: Bill Hayes <bill.hayes@hp.com>
> ---
> 
>  drivers/net/e1000/e1000_hw.c |   42 +++++++++++++++++++++++++++++++++++++++---
>  drivers/net/e1000/e1000_hw.h |    2 ++
>  2 files changed, 41 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/net/e1000/e1000_hw.c b/drivers/net/e1000/e1000_hw.c
> index 8fa0fe4..07e3178 100644
> --- a/drivers/net/e1000/e1000_hw.c
> +++ b/drivers/net/e1000/e1000_hw.c
> @@ -719,6 +719,11 @@ e1000_reset_hw(struct e1000_hw *hw)
>      DEBUGOUT("Masking off all interrupts\n");
>      E1000_WRITE_REG(hw, IMC, 0xffffffff);
>  
> +    if (hw->mac_type == e1000_82571 && hw->alt_mac_addr_is_present) {
> +    	hw->laa_is_present = 1;
> +    	e1000_rar_set(hw, hw->mac_addr, E1000_RAR_ENTRIES - 1);
> +    }
> +
>      /* Clear any pending interrupt events. */
>      icr = E1000_READ_REG(hw, ICR);
>  
> @@ -5693,11 +5698,41 @@ e1000_read_mac_addr(struct e1000_hw * hw)
>  {
>      uint16_t offset;
>      uint16_t eeprom_data, i;
> +    u16 mac_addr_offset = 0;
>  
>      DEBUGFUNC("e1000_read_mac_addr");
>  
> +    if (hw->mac_type == e1000_82571) {
> +        /* Check for an alternate MAC address.  An alternate MAC address can
> +         * be setup by pre-boot software and must be treated like a permanent
> +         * address and must override the actual permanent MAC address. */
> +        if (e1000_read_eeprom(hw, EEPROM_ALT_MAC_ADDR_PTR, 1,
> +                              &mac_addr_offset) < 0) {
> +            DEBUGOUT("EEPROM Read Error\n");
> +            return -E1000_ERR_EEPROM;
> +        }
> +        if (mac_addr_offset == 0xFFFF)
> +            mac_addr_offset = 0;
> +
> +        if (mac_addr_offset) {
> +            if (E1000_READ_REG(hw, STATUS) & E1000_STATUS_FUNC_1)
> +                mac_addr_offset += NODE_ADDRESS_SIZE/sizeof(u16);
> +
> +            /* make sure we have a valid mac address here before using it */
> +            if (e1000_read_eeprom(hw, mac_addr_offset, 1, &eeprom_data) < 0) {
> +                DEBUGOUT("EEPROM Read Error\n");
> +                return -E1000_ERR_EEPROM;
> +            }
> +            if (eeprom_data & 0x0001)
> +                mac_addr_offset = 0;
> +	}
> +
> +        if (mac_addr_offset) 
> +                hw->alt_mac_addr_is_present = 1;
> +    }
> +
>      for (i = 0; i < NODE_ADDRESS_SIZE; i += 2) {
> -        offset = i >> 1;
> +        offset = mac_addr_offset + (i >> 1);
>          if (e1000_read_eeprom(hw, offset, 1, &eeprom_data) < 0) {
>              DEBUGOUT("EEPROM Read Error\n");
>              return -E1000_ERR_EEPROM;
> @@ -5713,8 +5748,9 @@ e1000_read_mac_addr(struct e1000_hw * hw)
>      case e1000_82546_rev_3:
>      case e1000_82571:
>      case e1000_80003es2lan:
> -        if (E1000_READ_REG(hw, STATUS) & E1000_STATUS_FUNC_1)
> -            hw->perm_mac_addr[5] ^= 0x01;
> +        if (!mac_addr_offset &&
> +            (E1000_READ_REG(hw, STATUS) & E1000_STATUS_FUNC_1))
> +                hw->perm_mac_addr[5] ^= 0x01;
>          break;
>      }
>  
> diff --git a/drivers/net/e1000/e1000_hw.h b/drivers/net/e1000/e1000_hw.h
> index a2a86c5..e18760d 100644
> --- a/drivers/net/e1000/e1000_hw.h
> +++ b/drivers/net/e1000/e1000_hw.h
> @@ -1456,6 +1456,7 @@ struct e1000_hw {
>  	boolean_t		tbi_compatibility_en;
>  	boolean_t		tbi_compatibility_on;
>  	boolean_t		laa_is_present;
> +	boolean_t		alt_mac_addr_is_present;
>  	boolean_t		phy_reset_disable;
>  	boolean_t		initialize_hw_bits_disable;
>  	boolean_t		fc_send_xon;
> @@ -2286,6 +2287,7 @@ struct e1000_host_command_info {
>  #define EEPROM_INIT_CONTROL3_PORT_A   0x0024
>  #define EEPROM_CFG                    0x0012
>  #define EEPROM_FLASH_VERSION          0x0032
> +#define EEPROM_ALT_MAC_ADDR_PTR       0x0037
>  #define EEPROM_CHECKSUM_REG           0x003F
>  
>  #define E1000_EEPROM_CFG_DONE         0x00040000   /* MNG config cycle done */
> 
> -
> To unsubscribe from this list: send the line "unsubscribe netdev" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> --
> To unsubscribe from this list: send the line "unsubscribe netdev" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: Setting alternate MAC addresses on e1000

[Carsten Aulbert]

Hi all,

Kok, Auke schrieb:
>> Bill posted a patch for this a while back (but it was not accepted since
>> support for that hardware is offically moving to e1000e).  This may be
>> what you want....
> 
> yeah I thought of that as well and after digging through the documentation I don't
> think this will work for 82573's - it seems that only 82571 hardware supports this
> feature, which is not what Carsten needs.
> 
> as stated above, the alternate eeprom layout of the 82573 seems to also indicate that.

Thanks for not forgetting me :)

I wasn't able to reply yet because I'm buried in work, but the initial 
tip, to just use a different ID for the second device worked already. 
Though I don't use the alternate MAC possibility, I can change the MAC 
address easily - and already did that on ~280 nodes on Sunday.

So, I would say, please consider my problem solved.

Thanks to all of you who replied and the rest for the thoughts and the 
software!

Cheers

Carsten