Hopefully this isn't a dumb question ...but I can't find anything that talks about the format of the frame buffers that are pointed to by the Buffer Descriptors. I'm guessing that the Frame buffers hold the following: Dest MAC Address Source MAC address Length Data Field ..but then wouldn't the source address for transmit buffers be filled in by the controller? Hopefully, another simple question about transmit buffers: Is their a one to one associating between buffer desciptors and packets. Can a packet span multiple buffer descriptors. If so, how does the controller know when the packet is complete and to start adding the CRC. (...from the length field maybe in the Frame buffer??)

--- bsibilsky@comcast.net wrote:
Your guess is right. I had also the similar doubt
earlier. I have verified it through simulation.

Yes, we have to fill the SourceHA field in the buffer.
Also we have to write the Host MAC address in the MAC1
and MAC0 registers of EthMAC. (I hope I am right!)

There is a one to one association between a BD and a
Frame. A Frame never spans multiple BDs. There is no
Fragmentation Concept (like IP)at the Ethernet layer.
The Ethernet MAC knows about the length in bytes from
the LENGTH field of the TxBD. (bits 31..16). And
EthMAC adds the CRC accordingly.

Mayur __________________________________ Do you Yahoo!? Friends. Fun. Try the all-new Yahoo! Messenger. http://messenger.yahoo.com/

Thanks for the help! ----- Original Message ----- From: Mayur M Vegadmayurmvegad@y...> To: Date: Thu May 27 07:48:07 CEST 2004 Subject: [ethmac] Frame buffer format

Just a general comment, and not "target" to your own solution, but when you say that in Eth there is no fragmentation like in IP you probably refer to Layer 2 as compare to Layer 3 or above but notice even in Layer 2 while fragmentation don;t have the same implentation as in IP there is such concept as well as the size field in many protocol are not size at all but rather Etype (Size > 0x600 like for example MPLS 0x8847/8 AH 0x8809 and so on). Also notice that due to increase demand for jambo packet some sort of support in fragmentation is needed as not all switchs can handle such packet, so you might want to consider adding such support tho' keep in mind that while segmentation will not "cost" you too much reasmable will, and will most likely require external memorey. have a nice day Illan -----Original Message----- From: Mayur M Vegad [mailto:mayurmvegad@yahoo.com] Sent: Wednesday, May 26, 2004 10:48 PM To: List about open source Ethernet MAC core Subject: Re: [ethmac] Frame buffer format --- bsibilsky@comcast.net wrote:
Your guess is right. I had also the similar doubt
earlier. I have verified it through simulation.

Yes, we have to fill the SourceHA field in the buffer.
Also we have to write the Host MAC address in the MAC1
and MAC0 registers of EthMAC. (I hope I am right!)

There is a one to one association between a BD and a
Frame. A Frame never spans multiple BDs. There is no
Fragmentation Concept (like IP)at the Ethernet layer.
The Ethernet MAC knows about the length in bytes from
the LENGTH field of the TxBD. (bits 31..16). And
EthMAC adds the CRC accordingly.

Mayur __________________________________ Do you Yahoo!? Friends. Fun. Try the all-new Yahoo! Messenger. http://messenger.yahoo.com/ _______________________________________________ http://www.opencores.org/mailman/listinfo/ethmac