How to check that ECC is working in Linux?

Ive got a Phenom X4 with ECC ram on a Gigabyte motherboard. What can I do to verify ECC is working in Linux?

Ive done this to hopefully enable ECC:
http://www.reddit.com/r/linux/related/19gcx8/protip_linux_can_forceenable_ecc_on_amd/

I need to test it somehow.

Is it un-buffered, unregistered ecc ram?
EDIT: And is it a Phenom II cpu that you have?
EDIT 2: from the link you provided:
To actually use the ECC stuff, install edac-utils and check out the manpages for edac-util and edac-ctl.
 
Yes, unbuffered unregistered on a Phenom X4 9550.

Ive tried to use edac-utils after installing it but it says command not found so Im not really sure what I should do. Been Googling for an hour.

Edac-ctl does work.
 
I got edac-util to work after uninstalling/reinstalling it. I just verified the kernel has the module loaded as well.

I am pretty certain AMD has supported ECC in all of their CPUs with built in memory controller for a long time... only excluding some of their new stuff - I think the A series.
 
I got edac-util to work after uninstalling/reinstalling it. I just verified the kernel has the module loaded as well.

I am pretty certain AMD has supported ECC in all of their CPUs with built in memory controller for a long time... only excluding some of their new stuff - I think the A series.

Does your motherboard support ecc?
 
In Linux, you will get dmesg output similar to this after the amd64_edac_mod was loaded:
Code:
EDAC amd64: DRAM ECC enabled.

If you don't trust this message and your kernel has CONFIG_EDAC_AMD64_ERROR_INJECTION enabled, you can use the controls in /sys/devices/system/edac/mc/ to inject errors into your memory. Then you can verify that a Machine Check Error occurs on correctable errors and a Machine Check Exception (careful, will cause kernel panic by default) on uncorrectable errors.

For more details, read edac.txt from the kernel documentation directory.
 
Back
Top