Ryzen 1600 hang issue [solved] Disable all the P6!

Marlo · Veteran Joined: 26 Jul 2003 Posts: 1591

Your kernel is now running for 6 days. You originally wanted that.
Are you really sure that your motherbord has a watchdog SP5100?
The kernel says: "The Total Cost of Ownership is a watchdog timer that will reboot the machine after its expiration."
You dont receive any useful information or actions with this kerneloption.

Disable SP5100_TCO.
_________________
------------------------------------------------------------------
http://radio.garden/

RayDude · Posted: Mon Jul 23, 2018 2:00 pm Post subject:

Thanks Marlo. I really appreciate the help.

I have removed all the watchdog timer drivers and will see how it goes.
_________________
Some day there will only be free software.

RayDude · Posted: Sat Aug 04, 2018 4:44 pm Post subject:

It looks like that was indeed the problem.

I removed all watchdog modules and the system ran for 10 days without issue.

I have removed all the debugging I turned on and am running a cleaner kernel to see what happens. It's been up for a day so far with kernel gentoo-4.17.11.
_________________
Some day there will only be free software.

Marlo · Veteran Joined: 26 Jul 2003 Posts: 1591

In the meantime, I think that the bug at the beginning of this thread is not a CPU error.
It was definitely in the kernel. Maybe in Power Management and ACPI options.

RayDude, I can safely confirm what you are saying: 4.17.11 is stable for AMD Ryzen 5 1600.

I think you can close this thread with solved.

Thanks for this message.
Best regards
Ma
_________________
------------------------------------------------------------------
http://radio.garden/

RayDude · Posted: Sun Aug 05, 2018 3:46 pm Post subject:

Hi Marlo,

It hung overnight.

There was nothing in the remote logs to indicate why. In fact there was no data at all.

Can you or someone else please share your .config so I can compare it to mine?

I really want this to be stable.
_________________
Some day there will only be free software.

Marlo · Veteran Joined: 26 Jul 2003 Posts: 1591

Here is my .config.
But I have now developed this .config for qemu, kvm,xen,iptables and so on.
Remember. I have systemd and an AMD Polaris.

Good luck!
If you have any questions. please report here.
Ma
_________________
------------------------------------------------------------------
http://radio.garden/

RayDude · Posted: Sat Sep 22, 2018 11:01 pm Post subject:

It took me weeks to strip my kernel back but I did it. It's tiny and only has 50 megs of driver modules...

But it still crashes once in a while.

I think it's related to X or KDE...

It is a bit more stable though. Goes for around a week before a dead lock.
_________________
Some day there will only be free software.

Marlo · Veteran Joined: 26 Jul 2003 Posts: 1591

Which kernel?
_________________
------------------------------------------------------------------
http://radio.garden/

RayDude · Posted: Sun Sep 23, 2018 3:25 pm Post subject:

gentoo-sources-4.18.5 crashed twice in a couple of weeks.

I just updated to gentoo-sources-4.18.9
_________________
Some day there will only be free software.

Marlo · Veteran Joined: 26 Jul 2003 Posts: 1591

Only with the perspective of our topic I have tried these kernels. (Without bios settings and without CMD-Line parameters):

RayDude · Posted: Mon Sep 24, 2018 5:05 pm Post subject:

bammbamm808 · Guru Joined: 08 Dec 2002 Posts: 548 Location: Hawaii

My hardware is somewhat similar and I didn't see you reference one of the things that helped get my system rock solid. Not sure what it's called in your bios, or if it's even there, but under whatever passes for 'power supply idle' control, change it from 'low' to 'typical, 'common', 'normal' or whatever seems equivalent. I went through weeks of segfaults, reboots, hardlocks, both when loaded and idle before all my fiddling paid off.

On gentoo-sources-4.17.6 here, with the experimental use flag enabled. '-znver1' is my arch.
_________________
MSI MAG B550 Tomahawk
Ryzen 3900x
32Gb Samsung B-die (16GB dual rank x2) DDR4 @ 3200MHz, cl14
Geforce RTX 2070S 8GB
Samsung m.2 NVME pcie-3.0
Etc....

RayDude · Posted: Fri Oct 12, 2018 5:09 pm Post subject:

Tony0945 · Posted: Fri Oct 12, 2018 11:14 pm Post subject:

RayDude · Posted: Sat Oct 13, 2018 4:41 pm Post subject:

I have a Gigabyte AB350M-D3H, BIOS version F23d.

I looked for a setting like the one you described and didn't find anything.

If someone knows where it is, I'd appreciate the help.

Thanks.
_________________
Some day there will only be free software.

bammbamm808 · Guru Joined: 08 Dec 2002 Posts: 548 Location: Hawaii

Marlo · Veteran Joined: 26 Jul 2003 Posts: 1591

Maybe that's an alternative to the bios.
There is this little helper tool zenstates.py, to dynamically edit AMD Ryzen processor P-States.
_________________
------------------------------------------------------------------
http://radio.garden/

RayDude · Posted: Mon Oct 15, 2018 5:24 pm Post subject:

Thanks, I'll check it out tonight.
_________________
Some day there will only be free software.

krinn · Watchman Joined: 02 May 2003 Posts: 7470

Dunno if it has been ask before, but are you sure your cpu is not a faulty one?
You have a "huge" thread about ryzen, where users in it explains clearly how to ID your cpu and which cpu series are faulty, they even explains the RMA process to get a good cpu.

Because if your ryzen is a bad one, you could pass your life trying tricks in kernel without result

Look at that per example (rma because crach/lockup) https://forums.gentoo.org/viewtopic-p-8127190.html#8127190

RayDude · Posted: Tue Oct 16, 2018 5:18 pm Post subject:

RayDude · Posted: Sun Oct 21, 2018 5:33 pm Post subject:

Just to confirm, I'm disabling C6 with zenstates, is that the right thing to do?

I'm doing a two week test now, and if it makes it to two weeks it would be the first time in almost a year to do so.
_________________
Some day there will only be free software.

dewhite · Posted: Sat Jan 12, 2019 2:58 am Post subject:

RayDude · Posted: Sat May 25, 2019 3:30 pm Post subject:

Sorry I didn't reply.

This fell off my most recent posts screen.

The zenstates.py script fixed my problem when I disabled C6.

Now, I'm doing a fresh install and my kernel (using the same .config from the old system) hangs when called from grub. No output.

I'm in the process of adding nocbs to the kernel command line. I'm hoping that fixes this new install.
_________________
Some day there will only be free software.

dewhite · Posted: Thu May 30, 2019 3:23 pm Post subject: