Re: ie31200_edac missing PCI ID for i3-4370

[Rick Moritz <rahvin@gmail.com>]

Hi List, Hi Jason, Hi Paul,

Sorry for reviving this relatively old thread, but it brought back
memories from 2016, when I tried to add the Skylake i3's (PCI-ID
190f).

I've been running a modified kernel since then, with no issues (but no
proof that any error reporting is being done).

Output from ecc-utils is merely:

edac-util -v
mc0: 0 Uncorrected Errors with no DIMM info
mc0: 0 Corrected Errors with no DIMM info
edac-util: No errors to report.

without any of the DIMM-information I would expect..
The /sys tree for edac also contains Unknown for e.g. dimm_edac_node,
which isn't quite promising either.
The doc should be here:
https://www.intel.com/content/dam/www/public/us/en/documents/datasheets/desktop-6th-gen-core-family-datasheet-vol-2.pdf
Comparing the contents of the DID register does show a different
layout, something to keep in mind for the Haswell i3's as well.

Also, I'm not sure how this change will behave on same-generation
desktop i5s/i7s - they may just report nothing. Even the i3s may not
report any EDAC, depending how deep the ECC implementation really
goes.

Finally: did the original thread ever end in a kernel patch? I
couldn't find anything that looks related.

As I am back in kernel compiling, I will try and get my 6th gen i3 to
work - but I guess first I will need to dive into the code and data
sheet some more. I'd be glad for any pointers.

I would really like to hear back from Paul, how far he managed to get
with Haswell.

Cheers,


Rick


Full quote of last post in thread for reference:

On 2/9/21 10:27 PM, Jason Baron wrote:
>
>
> On 2/9/21 6:58 PM, Paul Marks wrote:
>> On Tue, Feb 9, 2021 at 2:25 PM Jason Baron <jbaron@akamai.com> wrote:
>>> On 2/4/21 6:22 PM, Paul Marks wrote:
>>>> On Thu, Feb 4, 2021 at 2:59 PM Jason Baron <jbaron@akamai.com> wrote:
>>>>> On 1/31/21 7:07 PM, Paul Marks wrote:
>>>>>> I have an ASRock C226M WS with an i3-4370 CPU.
>>>>>>
>>>>>> # lspci -vnn
>>>>>> 00:00.0 Host bridge [0600]: Intel Corporation 4th Gen Core Processor
>>>>>> Â Â Â Â Â Â Â Â Â Â Â Â  DRAM Controller [8086:0c00] (rev 06)
>>>>>> Â Â Â Â Â Â Â Â  Subsystem: ASRock Incorporation 4th Gen Core Processor
>>>>>> Â Â Â Â Â Â Â Â Â Â Â Â  DRAM Controller [1849:0c00]
>>>>>> Â Â Â Â Â Â Â Â  Flags: bus master, fast devsel, latency 0
>>>>>> Â Â Â Â Â Â Â Â  Capabilities: [e0] Vendor Specific Information: Len=0c <?>
>>>>>> Â Â Â Â Â Â Â Â  Kernel driver in use: hsw_uncore
>>>>>>
>>>>>> But edac-util doesn't work:
>>>>>>
>>>>>> # edac-util -v
>>>>>> edac-util: Fatal: Unable to get EDAC data: Unable to find EDAC data in sysfs
>>>>>>
>>>>>> I tried this ham-fisted patch:
>>>>>>
>>>>>> # diff -u ./drivers/edac/ie31200_edac.c{.old,}
>>>>>> --- ./drivers/edac/ie31200_edac.c.old
>>>>>> +++ ./drivers/edac/ie31200_edac.c
>>>>>> @@ -58,7 +58,7 @@
>>>>>> Â  #define PCI_DEVICE_ID_INTEL_IE31200_HB_3 0x0150
>>>>>> Â  #define PCI_DEVICE_ID_INTEL_IE31200_HB_4 0x0158
>>>>>> Â  #define PCI_DEVICE_ID_INTEL_IE31200_HB_5 0x015c
>>>>>> -#define PCI_DEVICE_ID_INTEL_IE31200_HB_6 0x0c04
>>>>>> +#define PCI_DEVICE_ID_INTEL_IE31200_HB_6 0x0c00
>>>>>> Â  #define PCI_DEVICE_ID_INTEL_IE31200_HB_7 0x0c08
>>>>>> Â  #define PCI_DEVICE_ID_INTEL_IE31200_HB_8 0x1918
>>>>>> Â  #define PCI_DEVICE_ID_INTEL_IE31200_HB_9 0x5918
>>>>> just curious why you removed here and didn't just add?
>>>> This is not a serious patch, just a one-liner to demonstrate the problem.
>>> Ok. Any chance you can find the datasheet that shows that this
>>> driver is using the appropriate registers for this hw? I didn't
>>> find it quickly looking...
>>>
>> I wouldn't know where to begin.  Do you have an example of a similar
>> datasheet from one of the known-good devices?
>>
>> I left "memtester" running on this machine, because it might increase
>> the odds of generating an ECC error someday.
> Hi Paul,
>
> I have a list of them at the top of:
> drivers/edac/ie31200_edac.c
>
> According to the following intel link it looks
> like '0xc[0-f]' is valid (page 52):

Sorry meant to write that as: '0x0c0[0-f]'.


> https://www.intel.com/content/dam/www/public/us/en/documents/datasheets/xeon-e3-1200v3-vol-2-datasheet.pdf
>
> So I'm fine with this patch (assuming it just
> becomes an addition).
>
> Thanks,
>
> -Jason
>