[fm-discuss] AMD Chipkill Syndrome related query

[Kenny Gibbons]

Srihari.Venkatesan at Sun.COM wrote:
> Thanks Gavin,
>
>   another basic question please..
>
>   ECC bits are used to correct/detect data bits. How does the ECC 
> mechanism work
>   for the ECC bits themselves, when they are affected?
>
>   i.e., is there a second level protection for the meta-data(ECC bits) 
> themselves..? is there a
>   short table/mapping maintained for an Address-Data and its recent ECC 
> code and then compared
>   on a read, or do the ECC bits themselves get stored in DRAMs and are 
> treated as
>   regular data, during ECC execution?
>   
the latter
>
> Thank you
> Srihari
>
>
> Gavin Maltby wrote:
>
>   
>> On 01/23/08 04:47, Srihari.Venkatesan at Sun.COM wrote:
>>
>>     
>>> Hi,
>>>
>>>   On AMD (with Chipkill ECC support), if there is a syndrome(CE) that 
>>> maps to
>>>   a symbol 22h - which according to the doc - " the symbols 22h-23h 
>>> maps to
>>>   ECC check bits for data bits from 64-127 " -  does this mean that 
>>> the ECC
>>>   check bits themselves are in trouble/affected  ?
>>>
>>>   please advise..
>>>       
>> Yes, that is correct.
>>
>> Gavin
>>
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> fm-discuss mailing list
>> fm-discuss at opensolaris.org
>>  
>>
>>     
> _______________________________________________
> fm-discuss mailing list
> fm-discuss at opensolaris.org
>