CRC32 verification, was --> Re: [ethmac] what if DATA field is ofsize zero

[Mike Pratt <mpratt@anr.mcnc.org>]

Hi guys,

Thanks for your suggestions.  I thought I would post an update 
describing the solution I eventually ended up with.  

First I tried Kevin Kay's suggestion of using Ethereal.  Unfortunately, 
the packets I was able to analyze were missing the crc checksum.  The 
checksums must have been stripped off at some point and I couldn't 
figure out how to avoid this.

Then I found a java app which generates CRC32 from an input file, called 
jacksum:

http://www.jonelo.de/java/jacksum/

It can output the crc checksum as hex.  Then you can use xxd to view the 
original input file as hex, and now you have verification vectors 
suitable for testing any verilog crc32 implementation.

Also, FYI Xilinx has an appnote describing hardware crc implementation 
and a perl script which generates a verilog crc32 module for any given 
input data bus width.  http://www.xilinx.com/xapp/xapp209.pdf  

Thanks again for your help.  Regards,

Mike

Illan Glasner wrote:

>  Hi,
>
>        The simplest way to test for CRC is instantiate your MAC twice and
>send from one to the other as the first Mac will probably generate crc on
>the fly for the outgoing data and the second Mac will test it.
>
>I assume you use the same crc polynom which is standard and so if it pass
>you can be quite sure you are ok. also actualy the more "intresting" is the
>packet extending that is "unique" for the 1G.
>
>personaly I tested complient using if I recall right one of Intel MAC
>behavioral code. you might want to check for this and see if any of there
>vendor can get you there Behaviral code.
>
>than you can build a environemnt which include your code as well as the
>referance Mac and start sending data back and forth testing all the feature
>you are supporting.
>
>have a nice day
>
>   Illan
>
>
>
>
>
>-----Original Message-----
>From: Mike Pratt [mailto:mpratt@anr.mcnc.org]
>Sent: Friday, January 17, 2003 7:35 AM
>To: ethmac@opencores.org
>Subject: Re: [ethmac] what if DATA field is of size zero
>
>
>Hi all,
>
>I am working on a GigE MAC ... using applicable parts of the opencores 
>10/100 MAC as a guide.  Does anyone have some sample Ethernet packets 
>that would be suitable for testing this core?  I am particularly 
>interested in checking that I have implemented crc correctly ... so they 
>must have valid crc fields!  Thanks in advance for any assistance.  Regards,
>
>Mike
>
>Igor Mohor(opencores) wrote:
>
>  
>
>>Opencores Ethernet MAC needs a minimum size of 5 bytes for proper
>>functioning. By the standard it is between 64 and 1518.
>>
>>Regards,
>>	Igor
>>
>>
>>
>> 
>>
>>    
>>
>>>-----Original Message-----
>>>From: owner-ethmac@opencores.org [mailto:owner-ethmac@opencores.org]
>>>   
>>>
>>>      
>>>
>>On
>> 
>>
>>    
>>
>>>Behalf Of Kevin Kay
>>>Sent: Thursday, January 16, 2003 2:32 PM
>>>To: ethmac@opencores.org
>>>Subject: RE: [ethmac] what if DATA field is of size zero
>>>
>>>A minimum legal packet length is 64 bytes.  If you control your entire
>>>network, then I guess you can run just about any length packet you
>>>   
>>>
>>>      
>>>
>>wish (0
>> 
>>
>>    
>>
>>>length data field all the way to a jumbo packets of 64kbytes or so).
>>>However, if it's going to be on an open network, then the acceptable
>>>packet lengths are between 64 and 1518 bytes.
>>>
>>>Regards,
>>>Kevin
>>>
>>>-----Original Message-----
>>>From: Mohan Krishna [mailto:mkrishna@tataelxsi.co.in]
>>>Sent: Thursday, January 16, 2003 7:35 AM
>>>To: ethmac@opencores.org
>>>Subject: [ethmac] what if DATA field is of size zero
>>>
>>>
>>>Hi all,
>>>
>>>Can any one clarify me wether it is possible for the data field be of
>>>size zero.
>>>if so then if there is an option to disable padding in the transmitter
>>>then what is the expected behavior of the receiver.
>>>
>>>Thanks in advance
>>>
>>>--
>>>To unsubscribe from ethmac mailing list please visit
>>>http://www.opencores.org/mailinglists.shtml
>>>--
>>>To unsubscribe from ethmac mailing list please visit
>>>http://www.opencores.org/mailinglists.shtml
>>>   
>>>
>>>      
>>>
>>--
>>To unsubscribe from ethmac mailing list please visit
>>    
>>
>http://www.opencores.org/mailinglists.shtml
>  
>
>> 
>>
>>    
>>
>
>  
>


--
To unsubscribe from ethmac mailing list please visit http://www.opencores.org/mailinglists.shtml