On 02/08/2013 04:26 AM, Vladimir Mosgalin wrote:
> Hi Yasha Karant!
>
>   On 2013.02.07 at 22:16:53 -0800, Yasha Karant wrote next:
>
>> For an answer to the question as to which CPU is present on this
>> particular machine:
>>
>> Processor name string: AMD Phenom(tm) II X4 840 Processor
>
> [..skipped..]
>
> This is desktop CPU. Are you running it on desktop board? EDAC is kind
> of server technology and usually won't work on desktop board. If you are
> 100% sure that your motherboard declares ECC features (not just ability
> to install ECC ram - sometimes manufacturers don't trace extra lines
> required for ECC support or don't activate BIOS code for ECC support)
> and you are really using ECC ram, then you should investigate this
> matter further. If not, then your system simply doesn't support EDAC and
> it's not supposed to work.
>
> At very least, you should be able to see EDAC-related options in bios:
> ECC enable (for ram, not cache!) and ECC type and similar ones. They
> should be enabled for EDAC to work properly. But, like I said, many
> desktop boards won't support this..
>
> You can also check it quickly from linux with
> # dmidecode -t memory|grep 'Error Correction Type'
> command.
>
> On non-EDAC supported systems you'll see Error Correction Type: None
>
> On EDAC-supported you'll see Error Correction Type: Multi-bit ECC
> or
>      Error Correction Type: Single-bit ECC
>
>
> Seeing "None" sometimes means that it just wasn't enabled in BIOS, though, so
> check there first if you get that output.
>
>
> PS my previous similar reply to this list got filtered by
> "88.blocklist.zap" filter @messaging.microsoft.com; just how come
> that microsoft filter applies to SL mailing list??
>

Thank you for that clarification -- I was aware that this was primarily 
server technology.  However, the boot diagnostic concerning "unsupported 
CPU" does not appear on my desktop workstation (same base hardware 
platform) -- and thus I assumed that something was amiss.  Running the 
command you specified reveals

[root@ahprc2 ykarant]# dmidecode -t memory|grep 'Error Correction Type'
	Error Correction Type: None

May I thus assume that the module about which there is a diagnostic 
message cannot be loaded?  If so -- how does one communicate to the 
operating system implementers not to issue the diagnostic in this case?

Yasha Karant