[solved] Random "-- Reboot --" => Freezes with Ryzen

whoops · 2017-05-28 14:31:02

Hi!

I've had random "-- Reboot --" for a few days now (2-3 times that I'm sure of so far, approx. 30 hours apart), always while the PC was (supposed to be) idle so far. The one time I actually saw it, it happens very quickly - probably faster than rebooting in a conventional way (one moment there's my Desktop, 2 seconds later there's GRUB).

The "-- Reboot --" line is the only thing in the journal - no date, no time, no system unit, no "Journal stopped", no "SIGTERM", nothing.

- I upgraded to a Ryzen CPU about a month ago and to Linux 4.11 shortly before I noticed the problem for the first time. The problem might have started to occur any time in between though.
- My sensors still aren't working in Linux, but after the reboot, the CPU temp was usually around 70°C (read somewhere that that up to 90°C should be fine for Ryzen CPUs)

Haven't found a way to reproduce the problem, so I guess I have to leave the PC running for a week every time I try to change something? For now, I have started to pipe dmesg to a file... now I'm waiting for another random reboot.

I have no idea where to start. Should the "-- Reboot --" message (or the absence of more details prior to / after the message) tell me anything that could narrow down the problem at least a tiny bit?
I assume that - since the message is present - it's probably not the PSU? And I also probably would not get a message at all, if the reset button pins were somehow triggered / short circuited (I don't have a reset button, but the contacts on the mainboard were a bit loose fiddly and hard to reach...) .

And since there aren't any additional messages, it's probably nothing in user-space (and if it is an evil spirit, it has root access at the very least or it's in the kernel)? Broken RAM also probably would result in either no message at all or in something more sinister looking than a sole "-- Reboot --"?

Any hints?

Last edited by whoops (2019-06-10 10:15:38)

ephemeralCuriosities · 2017-06-01 00:13:24

Just posting to say I have the exact same issue (been having it since I set up this rig, which was a couple weeks ago while the kernel was still 4.10). Might be related to Ryzen, but I don't know what the cause is, exactly.

This is the only weird thing I found on my journalctl output, in case it's related:

May 31 19:55:37 AycayiArch kernel: mce: [Hardware Error]: Machine check events logged
May 31 19:55:37 AycayiArch kernel: mce: [Hardware Error]: CPU 10: Machine Check: 0 Bank 0: baa0200000070135
May 31 19:55:37 AycayiArch kernel: mce: [Hardware Error]: TSC 0 MISC d012000101000000 SYND 144c180167 IPID b000000000 
May 31 19:55:37 AycayiArch kernel: mce: [Hardware Error]: PROCESSOR 2:800f11 TIME 1496274934 SOCKET 0 APIC a microcode 800111c

Ropid · 2017-06-01 00:40:19

Here's a thread about AMD Ryzen with people discussing instability:

https://bbs.archlinux.org/viewtopic.php?id=224907

It seemed to change with UEFI/BIOS updates and different memory kits or different memory settings.

ephemeralCuriosities · 2017-06-01 01:09:57

Based on that I looked around a bit, and according to one of my motherboard's reviews it probably is due to memory / BIOS issues. Relevant quote:

Cons: -Old BIOS, and that's a problem because...
-...Memory 2666Mhz or superior only recognized at 2133Mhz but it's unstable and can't update the BIOS because of random frozen/restart.
Other Thoughts: Notes:
-The memory i buy was selected according the compatibility list from MSI site. They say it's supported, but no with which BIOS version.
-Im stuck with this unstable motherboard with memory running at 2133Mhz, windows run and I can work doing basic stuff, but I can't enjoy my 64MB of RAM in programs for rendering, 3D modeling, etc. because of random restarts.

According to one of the users in the above thread there'll be a BIOS update for this mobo in a couple weeks, so hopefully that'll fix it. Don't know if that'll help with OP's issues but in case you're reading this I'd check to see if yours has a similar issue (we definitely have the same symptoms, at least).

whoops · 2017-06-01 08:39:42

I thought the other thread was about freezes and crashes? Didn't see anyone mention the semi-orderly "-- Reboot --", but it's a long thread, so maybe I overlooked it?

The timing of the mce error messages (I found 2 in my logs, too) is weird - might be unrelated. The first occurrence of the mce error in my logs was a few days ago. But the first occurrence of a reboot without prior "journal stopped" dates back further (was still using kernel 4.10 back then). I installed the mcelog service anyway, maybe it'll tell me something next time that happens.

I haven't encountered one of the reboots since I updated the BIOS 3 days ago - but according to my journal, that doesn't mean anything (...yet. if I trust my good old rusty "statistics sense", I need ~12 days without the problem occurring before it would be reasonable to draw any conclusions?):
- Upgraded Mainboard 1 month ago
- First reboot without prior "journal stopped" logged 8 days later, still kernel 4.10
- Over the last 30 days, it happened on average every 5-6 days (/ PC running ~17 hours per day and hibernating rest of the time)

Or are there any other circumstances under which I would get a "-- Reboot --" without a "journal stopped" in my logs?

whoops · 2017-07-24 16:40:54

For a while now it stopped rebooting and is freezing instead. This also happens every few days and I haven't had reboots since, so I guess it's the same problem in another guise (possibly thanks to BIOS or kernel updates).

"Should" it still do that (as in BIOS / Kernel bug) or am I alone with that problem (as in: probably broken hardware or something)?

drcouzelis · 2017-07-24 17:29:41

Haha, maybe it's not just my PSU after all!

I just wish it was reproducible, or at least crashed on a regular schedule!

I'm guessing, but I think "-- Reboot -- " gets inserted into journalctl when the computer is turned out without ever having been turned on (I make this guess based on the known behavior of the "last" command).

Last edited by drcouzelis (2017-07-24 17:31:32)

whoops · 2017-08-14 22:22:43

Now it's rebooting again, not freezing. Not sure why (maybe kernel or bios update?).

For a while I was able to trigger the reboots with a bash script (every time after only a few minutes, script never finished). Tried it 4 or 5 times. Then I disabled opcache and the script ran flawlessly. 5 minutes later though: "random" reboot. Enabled opcache again. I have no idea what's happening anymore -.-

Someone, somewhere, suggested setting RAM to 2400mhz might help - tried but can't even boot with that. RAM runs only @2133Mhz (autodetect) even though it's supposed to be 3200.

drcouzelis · 2017-08-15 12:36:53

Please stop driving yourself crazy trying to figure out what's wrong with your computer. AMD confirmed that there are problems with Ryzen processors that are seen when using Linux.

It sounds like you have a particularly bad situation. I suggest you go to the AMD website and open a service ticket. You will probably be asked to mail in your processor and they will mail you back a new one that (hopefully) will work better.

I will wait a little bit longer before I look into returning my processor, to see if AMD says anything else or releases a software update that makes things better. But my computer only reboots every 2 to 6 days... Yours sounds completely unusable!

Good luck!

Last edited by drcouzelis (2017-08-15 12:37:12)

smnpl · 2017-09-11 16:40:26

Hi, I'm actually experiencing the same on my Asus 370 Prime with Ryzen 1700x. Same MCE error sometimes in the journal and sudden "-- Reboot --"s. As well there is this thread: https://community.amd.com/thread/216084 … 5&tstart=0
Did you get around it somehow? I'm thinking about opening a service ticket, but would like to avoid the hassle if possible

drcouzelis · 2017-09-11 16:47:38

smnpl wrote:

Hi, I'm actually experiencing the same on my Asus 370 Prime with Ryzen 1700x. Same MCE error sometimes in the journal and sudden "-- Reboot --"s. As well there is this thread: https://community.amd.com/thread/216084 … 5&tstart=0
Did you get around it somehow?

Yep! Just go into BIOS, find the setting for "C6 states" (a CPU advanced setting), and set it to "Disabled".

...I'm actually halfway through the RMA process with AMD. The good news is, people who have done an RMA have reported more stable and cooler CPUs. The bad news is, some people are still experiencing the random reboots. So I still might be doing the "C6" workaround, described above. I don't think the idea of a kernel / BIOS patch has been ruled out for that issue yet, so there's still hope...

whoops · 2017-09-11 17:19:37

I think I already tried turning off C6 + voltage tweaks in one go... Didn't seem to help, but maybe I should try it again with default settings for the rest, just to make sure.

Did you leave it running like that for some weeks already? Mine made it up to 10 days without reboot so far just due to luck alone - although usually it's less...

( I'm currently on "day 3" of BIOS 0812 with default settings... so it could take a while before I try C6 again )

drcouzelis · 2017-09-11 17:51:44

I made it to 10 days with C6 states disabled. Which is hard to do, due to the nature of Arch Linux and constant updates...

But reboots were pretty consistently every 2 to 6 days without that setting.

smnpl · 2017-09-11 18:16:05

Well I'll try that to see if it improves it. Though, I don't like to have features disabled to make it work ... I just contacted my vendor for a replacement. Eager to see if they will give a sh* on journalctl output xD
Thanks for the input!

whoops · 2017-10-09 08:46:31

I've had BIOS 0812 (and 0902) for a month now and no reboots happened. Still not totally sure it's fixed - mainly because the PC didn't spend much time totally idle during that month (which might be a prerequisite for the problem showing itself?), but it's really starting to look like it *might* be fixed.

drcouzelis · 2017-10-10 00:22:58

I finished my RMA process with AMD and am now using my new CPU. I'm on the default BIOS settings with the latest BIOS version and just got to the end of eight days of uptime without any random reboots or crashing, which, according to my memory, used to happen at least every six days. I'll see how long it keeps on kicking before Arch Linux yells at me too much for missed kernel updates.

whoops · 2017-10-10 08:55:33

Is there a reason why updating the kernel / rebooting manually would interfere with testing?

If it's just for logging purposes, you could look at p.E. the "journal stopped" message to tell the reboots apart:

journalctl -f --since "-3 months" | grep -i -- '-- reboot\|journal stopped\|linux version'

( I just write it down if there are any unclean reboots for other reasons... like power outages. )

drcouzelis · 2017-10-10 12:38:12

whoops wrote:

Is there a reason why updating the kernel / rebooting manually would interfere with testing?

Hmmm... No, I suppose it wouldn't interfere... That's a good point!

My desktop at home is also my HTTP and SSH server. I just want to be reasonably sure that if I'm ever away from my house for a week or more that my computer would still be up and running and stable.

EDIT:

FYI! https://bbs.archlinux.org/viewtopic.php … 0#p1742570

Last edited by drcouzelis (2017-10-13 18:04:48)

whoops · 2019-06-10 10:15:21

Just a quick summary / update because this took so long I forgot to update:
- First it kept rebooting every few days
- After the third BIOS update or so it kept freezing every day
- After the sixth BIOS update or so, the problem disappeared - as long as I turned off c-state switching in BIOS
- After a kernel update years later the freezes came back and I wasn't sure if it was Ryzen again
Rest kind of [solved] there: https://bbs.archlinux.org/viewtopic.php?id=246404 probably

Arch Linux

#1 2017-05-28 14:31:02

[solved] Random "-- Reboot --" => Freezes with Ryzen

#2 2017-06-01 00:13:24

Re: [solved] Random "-- Reboot --" => Freezes with Ryzen

#3 2017-06-01 00:40:19

Re: [solved] Random "-- Reboot --" => Freezes with Ryzen

#4 2017-06-01 01:09:57

Re: [solved] Random "-- Reboot --" => Freezes with Ryzen

#5 2017-06-01 08:39:42

Re: [solved] Random "-- Reboot --" => Freezes with Ryzen

#6 2017-07-24 16:40:54

Re: [solved] Random "-- Reboot --" => Freezes with Ryzen

#7 2017-07-24 17:29:41

Re: [solved] Random "-- Reboot --" => Freezes with Ryzen

#8 2017-08-14 22:22:43

Re: [solved] Random "-- Reboot --" => Freezes with Ryzen

#9 2017-08-15 12:36:53

Re: [solved] Random "-- Reboot --" => Freezes with Ryzen

#10 2017-09-11 16:40:26

Re: [solved] Random "-- Reboot --" => Freezes with Ryzen

#11 2017-09-11 16:47:38

Re: [solved] Random "-- Reboot --" => Freezes with Ryzen

#12 2017-09-11 17:19:37

Re: [solved] Random "-- Reboot --" => Freezes with Ryzen

#13 2017-09-11 17:51:44

Re: [solved] Random "-- Reboot --" => Freezes with Ryzen

#14 2017-09-11 18:16:05

Re: [solved] Random "-- Reboot --" => Freezes with Ryzen

#15 2017-10-09 08:46:31

Re: [solved] Random "-- Reboot --" => Freezes with Ryzen

#16 2017-10-10 00:22:58

Re: [solved] Random "-- Reboot --" => Freezes with Ryzen

#17 2017-10-10 08:55:33

Re: [solved] Random "-- Reboot --" => Freezes with Ryzen

#18 2017-10-10 12:38:12

Re: [solved] Random "-- Reboot --" => Freezes with Ryzen

#19 2019-06-10 10:15:21

Re: [solved] Random "-- Reboot --" => Freezes with Ryzen

Board footer