RE: [ethmac] RX MAC

[Mahmud Galela <mgalela@vlsi.itb.ac.id>]

> 1.
> using a SDF detector that get a signal from the RX Machine which make it
> base on the rxdv is somewhat dangerus and wrong, you must take into
> considuration the case that you will get ONLT SFD without preamble and than
> the data.
> 
> now if you look on the rxdv in the state machine and clock after the raius
> of rxdv you generate a signla to the sdf detector to start looking you
> already MISSED this packet and lost it.
> 
> on the other hand if you make it combinatorial than why not siple pass the
> rxdv directly to the SDF detector ?

i got the point, thanks

> 
> 2.
> What is the purpose of the address match ?
> unless you are intent to work as a switch and if so you need more than
> single mac I see no reason for it as the switch that you are conneted to
> will already make the learning and filtering .


> 
> 3.
> in the crc machine , you wrote "Computes a CRC across the data and compares
> it with FCS field." I belive a easier way is simple to pass the crc field
> through the crc machine and simple see that you get the "magic number".

i don't understand, 
what do you mean about "magic number" ?
from some datasheets i've read, we computes the CRC accross the data and
compares it with FCS field (CRC received) to make sure that no error in
frame received. 

> 
> Also why do you send th 9 MSB bit of the crc ?
> 
to provide bits needed by host to make hash table for multicast
addressing.

> 4.
> a small typo in the front page in
> http://www.opencores.org/cores/ethmac/blocks_1/rxethmac/rxethmac.shtml where
> it say
> "Clock is provided by MII through rx_clk, which frequency is 25 MHz when
> operates at 10 Mbps and 2.5 MHz when operates at 100 Mbps."
> when of course the 25M is in 100Mb and 2.5 is in 10Mb.
i've corrected it, thanks

> have a nice day
> 
>    Illan
>