Comment # 15 on bug 1192288 from Borislav Petkov

(In reply to Olav Reinert from comment #13)
> Because of all this variety, any Leap user who invests in ECC memory really
> really wants to see the line in the log stating that it is in fact enabled.

And what happens if dmesg is overwritten because the ring buffer is too
small or some kernel printk is too chatty? That line is gone so Leap
user won't see a thing.

> Forcing Leap users to turn on kernel debug info to get it completely voids
> your intention of trimming down log noise (which is why you downgraded the
> messages, I assume).
> 
> Please undo that downgrade. I can (and will) file a separate bug report
> about it, if you prefer.

So all you want to know is whether your box has ECC enabled, correct?

Two ways:

1. amd64_edac loads *only* when it has detected ECC - not otherwise. So
the simple

$ lsmod | grep amd64_edac
amd64_edac             49152  0

already says that you have ECC.

2. Then, you can always do

$ grep -r . /sys/devices/system/edac/ 2>/dev/null | grep -i edac_mode
/sys/devices/system/edac/mc/mc0/rank7/dimm_edac_mode:SECDED
/sys/devices/system/edac/mc/mc0/rank5/dimm_edac_mode:SECDED
/sys/devices/system/edac/mc/mc0/csrow2/edac_mode:SECDED
/sys/devices/system/edac/mc/mc0/rank6/dimm_edac_mode:SECDED
/sys/devices/system/edac/mc/mc0/rank4/dimm_edac_mode:SECDED
/sys/devices/system/edac/mc/mc0/csrow3/edac_mode:SECDED

And that tells you, Single Error Correction, Double Error Detection

which is the type of ECC supported by those DIMMs.

And you can do that regardless of whether the dmesg circular buffer has
been overwritten or not.