Virtex-4FX12MM: Any hardware MAC address accessable?

Hi all,

my hardware guys are on vacation but I need to clarify this:

The EDK soft cores like opb_ethernetlite need the MAC address as
parameter. They do not provide a unique MAC address as default. You can
imagine, what management effort would be necessary to maintain a unique
MAC address for every device we deliver.

Is there any MAC address available at the Virtex-4FX12MM that
- is unique for every board
- is accessable (read-only) by the PowerPC where our code runs

Maybe the hard_temac core?

Thank you for help on this!

No, Xilinx don't provide mac address. It's your job to either :
 - Choose some random ones but then it's your responsability
 - Request a range for your company.

And the hard_temac core has no MAC provided either.

    Sylvain

henryk.mueller@gmx.de wrote:
> Hi all,
>
> my hardware guys are on vacation but I need to clarify this:
>
> The EDK soft cores like opb_ethernetlite need the MAC address as
> parameter. They do not provide a unique MAC address as default. You can
> imagine, what management effort would be necessary to maintain a unique
> MAC address for every device we deliver.
>
> Is there any MAC address available at the Virtex-4FX12MM that
> - is unique for every board
> - is accessable (read-only) by the PowerPC where our code runs
> 
> Maybe the hard_temac core?
> 
> Thank you for help on this!

Hi Sylvian,

that is bad news.

Even if we request for an OUI, it means a lot of additional management
effort. We can not just copy the same code to every device, but need to
patch every single one with it's unique MAC address. That applies to
the first copy and to every software update that some service guy has
to do later.

Basically, every customer who uses the Virtex-4FX12MM and the ethernet
interface faces the same problems as we do now. Is there no elegant
solution out there? I hoped so much...

Another approach would be to use some unique serial number from the
board that we could combine with our OUI to form a unique address. Does
the PowerPC or FPGA or any other part have a software-readable ID?

Henryk, who's desperately trying to avoid managing its own MAC
addresses


Sylvain Munaut <SomeOne@SomeDomain.com> schrieb:

> No, Xilinx don't provide mac address. It's your job to either :
>  - Choose some random ones but then it's your responsability
>  - Request a range for your company.
>
> And the hard_temac core has no MAC provided either.
>
>     Sylvain
>
> henryk.mueller@gmx.de wrote:
> > Hi all,
> >
> > my hardware guys are on vacation but I need to clarify this:
> >
> > The EDK soft cores like opb_ethernetlite need the MAC address as
> > parameter. They do not provide a unique MAC address as default. You can
> > imagine, what management effort would be necessary to maintain a unique
> > MAC address for every device we deliver.
> >
> > Is there any MAC address available at the Virtex-4FX12MM that
> > - is unique for every board
> > - is accessable (read-only) by the PowerPC where our code runs
> > 
> > Maybe the hard_temac core?
> > 
> > Thank you for help on this!

henryk.mueller@gmx.de schrieb:

> Hi Sylvian,
>
> that is bad news.
>
> Even if we request for an OUI, it means a lot of additional management
> effort. We can not just copy the same code to every device, but need to
> patch every single one with it's unique MAC address. That applies to
> the first copy and to every software update that some service guy has
> to do later.
>
> Basically, every customer who uses the Virtex-4FX12MM and the ethernet
> interface faces the same problems as we do now. Is there no elegant
> solution out there? I hoped so much...

are you talking about the memec mini module?
if so then the only place to store the MAC would be onboard flash

I would have assumed Memec gives a MAC with the module !
bad luck for you I guess

http://www.hydraxc.com

modules as example are all coming with unique MAC the OUI range
was specially purchased for this purpose. The MAC there is stored
in proteced area of the boot processor and can not be changed -
but the user can use its own MAC # if he wants.

possible memec doesnt care.
some Xilinx boards did have MAC assigne (ML300 at least had)

so if you did not get the MAC with the memec MM when purchased
then you dont have a valid MAC, meaning to be safe you need to
purchase the OUIs and store them in on module flash

Antti

henryk  wrote

> Even if we request for an OUI, it means a lot of additional management
> effort. We can not just copy the same code to every device, but need to
> patch every single one with it's unique MAC address. That applies to
> the first copy and to every software update that some service guy has
> to do later.

AFAIR one of the Dallas 1-wire chips supplies a valid MAC address if you 
have a spare pin and some glue to add one to your board.

Hi Antti,

correct, I'm talking about the Memec Mini-Module. Somehow I find no
information at the new avnet memec website.

I bet there is a MAC address on the board somewhere. I only need to
find it. :o(

Is there an Ethernet PHY chip that provides a MAC address? I'm not to
deep into such hardware issues. This information is usually provided to
me... ;o)

Storing our own unique MAC address into the onboard flash would be a
pain. We could not just write the same flash image to every device, but
need to patch every image before. And software updates could mess
everything up.

If I state now that there is no MAC address and we need to do all the
extra work, but later we find out that we just did not look carefully
enough. Then I get ... you know ...problems ;o)

Henryk

> henryk.mueller@gmx.de schrieb:
>
> > Hi Sylvian,
> >
> > that is bad news.
> >
> > Even if we request for an OUI, it means a lot of additional management
> > effort. We can not just copy the same code to every device, but need to
> > patch every single one with it's unique MAC address. That applies to
> > the first copy and to every software update that some service guy has
> > to do later.
> >
> > Basically, every customer who uses the Virtex-4FX12MM and the ethernet
> > interface faces the same problems as we do now. Is there no elegant
> > solution out there? I hoped so much...
>
> are you talking about the memec mini module?
> if so then the only place to store the MAC would be onboard flash
>
> I would have assumed Memec gives a MAC with the module !
> bad luck for you I guess
>
> http://www.hydraxc.com
>
> modules as example are all coming with unique MAC the OUI range
> was specially purchased for this purpose. The MAC there is stored
> in proteced area of the boot processor and can not be changed -
> but the user can use its own MAC # if he wants.
>
> possible memec doesnt care.
> some Xilinx boards did have MAC assigne (ML300 at least had)
>
> so if you did not get the MAC with the memec MM when purchased
> then you dont have a valid MAC, meaning to be safe you need to
> purchase the OUIs and store them in on module flash
> 
> Antti

"Tim" <tim@rockylogiccom.noooospam.com> schrieb im Newsbeitrag 
news:e5kfam$2hg$1$8300dec7@news.demon.co.uk...
>
> henryk  wrote
>
>> Even if we request for an OUI, it means a lot of additional management
>> effort. We can not just copy the same code to every device, but need to
>> patch every single one with it's unique MAC address. That applies to
>> the first copy and to every software update that some service guy has
>> to do later.
>
> AFAIR one of the Dallas 1-wire chips supplies a valid MAC address if you 
> have a spare pin and some glue to add one to your board.
>
if you refer to onewire unique ID then this is not valid MAC address

you can use onewire memory to store a valid MAC but you must
then yourself obtain the MAC and write it to memory location

Antti 

Antti  wrote
> "Tim" schrieb
>>
>> AFAIR one of the Dallas 1-wire chips supplies a valid MAC address if you 
>> have a spare pin and some glue to add one to your board.
>>
> if you refer to onewire unique ID then this is not valid MAC address
>
> you can use onewire memory to store a valid MAC but you must
> then yourself obtain the MAC and write it to memory location

I'm not an expert, but Maxim/Dallas advertise the DS2502-E48 as providing a 
unique and valid MAC address. I suppose they could be wrong. 

Tim schrieb:

> Antti  wrote
> > "Tim" schrieb
> >>
> >> AFAIR one of the Dallas 1-wire chips supplies a valid MAC address if you
> >> have a spare pin and some glue to add one to your board.
> >>
> > if you refer to onewire unique ID then this is not valid MAC address
> >
> > you can use onewire memory to store a valid MAC but you must
> > then yourself obtain the MAC and write it to memory location
>
> I'm not an expert, but Maxim/Dallas advertise the DS2502-E48 as providing a
> unique and valid MAC address. I suppose they could be wrong.

sorry, i did not pay attention to "one" dallas chip in your posting,
you are right the exactly part type you mentioned is that one chip that
has the MAC id in the user eeprom area (what is write protected)

Sorry, i should have read your posting more carefully
I assumed you referred to the 1-wire serial code 

Antti

Hey Tim,

thank you for the part number! A think that is the alternative we could
go for if we find no "built-in" MAC address. This question we have to
clarify with Memec directly first.

Greets

Henryk

Tim schrieb:
> Antti  wrote
> > "Tim" schrieb
> >>
> >> AFAIR one of the Dallas 1-wire chips supplies a valid MAC address if you
> >> have a spare pin and some glue to add one to your board.
> >>
> > if you refer to onewire unique ID then this is not valid MAC address
> >
> > you can use onewire memory to store a valid MAC but you must
> > then yourself obtain the MAC and write it to memory location
>
> I'm not an expert, but Maxim/Dallas advertise the DS2502-E48 as providing a
> unique and valid MAC address. I suppose they could be wrong.