1

May someone help me in decoding the exact significance of the following message I found in dmesg?

Jan 28 15:58:17 mint kernel: mce: [Hardware Error]: CPU 0: Machine Check: 0 Bank 7: 8c00004000010093
Jan 28 15:58:17 mint kernel: mce: [Hardware Error]: TSC 5159cf02b8 ADDR 529449f80 MISC 2040404086

There are many of them, but CPU 0: Machine Check: 0 Bank 7: 8c00004000010093 is constant among them.

I understand the CPU is detecting hardware problems, but I don't understand where. Maybe a failed memory stick on bank 7? (I got 8 banks fully populated).

I found another thread where it was advised to install mcelog, but apt finds no packages named mcelog.


EDIT:

The system completed one pass of memtest and was left running it for a whole night. It remained rock solid.

Here is what /var/log/syslog shows:

Sep 13 13:40:11 mint ntpd[1462]: kernel reports TIME_ERROR: 0x41: Clock Unsynchronized Sep 13 13:44:25 mint kernel: [ 648.820738] mce: [Hardware Error]: Machine check events logged Sep 13 13:44:25 mint kernel: [ 648.820770] EDAC sbridge MC0: HANDLING MCE MEMORY ERROR Sep 13 13:44:25 mint kernel: [ 648.820778] EDAC sbridge MC0: CPU 0: Machine Check Event: 0 Bank 7: c01fbb4000010093 Sep 13 13:44:25 mint kernel: [ 648.820778] EDAC sbridge MC0: TSC 0 Sep 13 13:44:25 mint kernel: [ 648.820779] EDAC sbridge MC0: ADDR 0 Sep 13 13:44:25 mint kernel: [ 648.820780] EDAC sbridge MC0: MISC 0 Sep 13 13:44:25 mint kernel: [ 648.820786] EDAC sbridge MC0: PROCESSOR 0:306e4 TIME 1536846265 SOCKET 0 APIC 0

It appears it's a memory related error, but it seems to me that the system reboots during I/O operations on disks.

The machine keeps rebooting randomly. Any help would be incredibly welcome.

closed as off-topic by guiverc, karel, Zanna, Eric Carvalho, N0rbert Sep 13 '18 at 21:01

This question appears to be off-topic. The users who voted to close gave this specific reason:

  • "This is not about Ubuntu. Questions about other Linux distributions can be asked on Unix & Linux, those about Windows on Super User, those about Apple products on Ask Different and generic programming questions on Stack Overflow." – guiverc, karel, Zanna, Eric Carvalho, N0rbert
If this question can be reworded to fit the rules in the help center, please edit the question.

  • What version of Ubuntu? mcelog was found in 'universe' packages.ubuntu.com/search?keywords=mcelog until it was removed upstream in debian. It was reported, confirmed (1752251) with three alternatives provided for bionic, but no work done; nor anything confirmed. – guiverc Sep 11 '18 at 12:04
  • 1
    You may have a bad RAM stick. Run memtest from the GRUB menu or after booting a Ubuntu Live DVD. Report back to @heynnema. – heynnema Sep 11 '18 at 21:25
  • 1
    1 or more complete passes is fine. You might also check that you have the latest BIOS installed, and the intel-microcode or amd64-microcode installed also. Also make sure that your {Bay Trail} processor chip isn't one of the ones that require a kernel boot flag modification to operate properly. – heynnema Sep 12 '18 at 15:19
  • 1
    Check /var/log/syslog for events around the time of the reboots. That should give us a clue. Make sure to place @heynnema into your comments, or I may miss them. – heynnema Sep 13 '18 at 2:21
  • 1
    I've added an answer for you. Please remember to accept it if it was helpful, and ultimately helped solve/identify your problem. Thanks! – heynnema Sep 13 '18 at 14:31
1

From the comments...

Although memtest seems to run OK, all indications are that you have an intermittent memory problem with Bank 7: c01fbb4000010093.

It may be a RAM memory module failure, or a RAM memory module seating failure, or a defective memory slot on your motherboard.

If your memory slots are identified with bank numbers on the logic board, mark the memory module in bank 7 with a black magic marker pen. Assuming that all of your memory modules are the same brand/make/model, rotate ALL of the memory modules ONLY ONE SLOT over, such that RAM memory module bank 7 ends up in bank 6. Make sure that you use proper ESD grounding techniques when handling RAM modules. Rerun memtest, noting any failures.

If memtest runs OK, then reboot the system, and if it runs fine, you've solved the problem. If it fails with a bank 6 error (or anything other than bank 7), you've identified the defective memory module.

Memory modules should be installed/removed in pairs. One module is normally on memory channel A, and the other on memory channel B. This is for memory interleaving purposes. If you wish to fully identify/eliminate a specific RAM module, remove it, and its interleaving mate, reboot the system, and retest.

  • Thanks, I'll do it as soon as I can do it calmly and without hurry. I know which bank is 7 by the mb manual. But Still, not managing to stay up for 10 minutes with IO operations while completing hours and hours of memtest makes me think I bent some IO pin on the cpu socket. Furthermore, note that my memory is buffered and ECC, so it's capable of correcting errors, up to some extent. It's worth trying, though. – MadHatter Sep 13 '18 at 16:46
  • 1
    Unused memory gets used as disk and I/O buffers, so failures during disk I/O may not be surprising... esp if you have a fast SSD. If you had a bent pin on the CPU, the CPU wouldn't fit nicely into its socket, and you'd have much bigger problems that what you have. – heynnema Sep 13 '18 at 17:16
  • Ok, I did as advised. Rotated memory modules. MCE error on bank 7 persists, as well as random reboots. Then I removed the memory modules on banks 7 and 8 (former 7). MCE errors on bank 7 persists. The memory modules seem to be blameless. Are we sure that that "bank 7" actually refers to memory bank 7? Anyway, note that the pins are on the mobo, it is the cpu that actually got "sockets" see i.ytimg.com/vi/2JAGNWve4X0/maxresdefault.jpg ; However yes, I got a fast ssd (nvme). I think I'll ditch the mainboard and buy another one. – MadHatter Sep 13 '18 at 23:03
  • 1
    Fair enough. Please keep me posted. – heynnema Sep 14 '18 at 17:01
  • 1
    @MadHatter good news! Thanks for the update! – heynnema Oct 10 '18 at 20:11

Not the answer you're looking for? Browse other questions tagged or ask your own question.