Re: [PATCH v3] EDAC/ghes: Setup DIMM label from DMI and use it in error reports
From: Robert Richter
Date: Wed May 20 2020 - 10:12:13 EST
Thanks for testing.
On 19.05.20 22:25:35, Borislav Petkov wrote:
> -/sys/devices/system/edac/mc/mc0/csrow15/ch0_dimm_label:1:mc#0memory#15
> +/sys/devices/system/edac/mc/mc0/csrow15/ch0_dimm_label:1:unknown memory (handle: 0x0030)
Looks like on that system device locator or bank locator (or both) are
empty. What shows this (esp. the Locator fields):?
dmidecode -t 17
So maybe the check is too strict and we should allow one of both being
empty?
If the strings are missing, shouldn't we still provide the handle in
the label information?
The string 'mc#0memory#15' is also of no use as it just takes the
mc_num and dimm_num as a reference, which can be determined from sysfs
without that label.
Your add on patch to just ignore the write does not revert to we old
behavior as you will see now 'mc#0memory#15' in the error reports
where you have seen 'unknown label' before.
So the question is what to do if that information is missing?
Note: it should better show "unknown label ..." instead of "unknown
memory ...".
Thanks,
-Robert