On 04/09/2014 05:33 PM, Luck, Tony wrote: >> Unfortunately, the box reporting the ue errors just went into transit (so >> that I can better examine this issue), so I will probably not be able to >> run this experiment on that specific box until next week. > > Do you have any other logs from this machine. Is there something > logged in one (or more) of the machine check banks when your EDAC > driver says that there are uncorrected errors? > > When the box is back online again - I'd be interested to know if mcelog(8) > daemon reports any errors. Grab the latest from mcelog.org, compile > and run as "mcelog --daemon". Logs show up in /var/log/mcelog >
I finally got the bad dimm in hand - running the latest version of mcelog, I don't see anything reported in the log. (I verified that mcelog was reporting events via mce-inject.) >> # ./rdmsr 0x179 >> c09 > > So this processor does support CMCI - next question is whether each > bank support it (and got enabled by Linux) [can run on any system ... don't > need to wait for the one to finish transit)] > > # for I in `seq 0 8` > do > ./rdmsr 0x28$i > done > > will print the MCi_CTL2 registers from each bank. Bit 30 (0x40000000) > shows CMCI enabled. > > On the name of the driver - can you throw in an underscore: ie3_12xx.c ? > > Do you have systems from Sandy Bridge, Ivy Bridge and Haswell generations > (no suffix for Sandy Bridge, then v2 and v3) ... and does this driver work > across all of > them? If it is just for Haswell ... then "ie3_12xx_v3.c" might be a better > name. > So I've moved the dimm around to all three systems Sandy, Ivy Bridge and Haswell. And this driver is populating '/sys/devices/system/edac/mc/mc0/ue_count' correctly when a userspace memory scanning program also report errors. So the driver appears to be doing the right thing. I've tested: CPU E3-1270 v3 @ 3.50GHz : 8086:0c08 (haswell) CPU E3-1270 V2 @ 3.50GHz : 8086:0158 (ivy bridge) CPU E31270 @ 3.40GHz : 8086:0108 (sandy bridge) I still don't see any evidence of CMCI doing anything as the "THR" entry in /proc/interrupts doesn't budge off of 0. Let me know if you'd like me to look at anything else... Thanks, -Jason -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
