Ryzen 7 Lockups

[miscous]

Yesterday was day 4 without a lockup .... for four hours or so.... until the cursor disappeared at a window boundary ... no keyboard response, only reset!

Today, an interesting anomaly "Driver Manager" indicates I am using the proprietary Nvidia 340 driver, the command "inix -Fxz" indicates "Display Server: X.Org 1.18.4 drivers: nouveau (unloaded: fbdev,vesa)" ; "lshw -c video" confirms "configuration: driver=nouveau latency=0". Had not spotted that Many Thanks to ClixTrix and mr_raider. ...... after that I will go after the kernel installation errors ..... I used Ugrade Manager/View.

Looks like actually using the Nvidia driver could change a great deal.

I really appreciate the the help and suggestions.

[mr_raider]

The driver manager is not reliable. Run the following command:


sudo apt-get reinstall nvidia-3xx


replace 3xx with the appropriate driver for your card, eithr 340 or 375.

[miscous]

As suggested, I tried "sudo apt-get reinstall nvidia340" apparently the reinstall option is not longer supported. Then I tried "purge/install" .... worked fine but the nouveau still loads even after blacklisting in grub2. Driver Manager show the nvidia driver and various command line programs show nouveau is loaded.

A few days ago I disconnected the ssd and loaded Mint 18.2 on a 320Gb HD I pulled from a USB case. Everything went fine .... I got a lockup quickly and went back to the ssd. Tried that disk and I could load the Nvidia 340 driver with driver manager. The Nvidia software was active the the command line programs showed the nvidia driver loaded.

All my observations are now with little meaning since I don't know what video driver was operative.... though I suspect it was nouveau .... and what I experienced is not a surprise in that context. I am probably heading for a reload .......

[miscous]

This is a few lines from dmesg:

6.021863] nvidia: module verification failed: signature and/or required key missing - tainting kernel
[ 6.026549] NVRM: The NVIDIA probe routine was not called for 1 device(s).
[ 6.026550] NVRM: This can occur when a driver such as:
NVRM: nouveau, rivafb, nvidiafb or rivatv
NVRM: was loaded and obtained ownership of the NVIDIA device(s).
[ 6.026551] NVRM: Try unloading the conflicting kernel module (and/or
NVRM: reconfigure your kernel without the conflicting
NVRM: driver(s)), then try loading the NVIDIA kernel module
NVRM: again.
[ 6.026552] NVRM: No NVIDIA graphics adapter probed!
[ 6.026553] [drm] Module unloaded

I am actually relieved that this seems to have resolved into a video driver issue. Knowing that a different loading of Mint 18.2 allows the Nvidia driver to load is also a plus.

The Ryzen Realm seems to included frequent bios, kernel, and driver updates. My choice is to have as clean an install as possible for those items.

Next step will be to collect the various Tips and Tricks for setting up Mint 18.2 on my laptop with Mint 17.3 and reload.

[mr_raider]

Make sure nouveau is blacklisted:

what is the content of
/etc/modprobe.d/blacklist.conf

Also did you check if your GPU requires nvidia 340 (legacy) or nvidia 375 (current)?

[miscous]

Ah ... mr_raider ..THANK YOU!

I will add "nouveau" to the modprobe blacklist...

Nvidia 340 was the recommended driver in update manager .... I'll check this out if I can get nouveau out

[miscous]

My Ryzen 1700 is now running the Nvidia driver. Apparently the nouveau driver was being loaded due to a problem with the "key" for the Nvidia driver which would "taint" the kernel(4.10). I purged nvidea with apt-get and attempted to install the latest Nvidia driver downloaded directly from Nvidia. This is a binary (run) file. After making it executable and switching to a terminal, and stopping the xserver, I ran the installation which failed due to an improper key. After restarting in Low-Res mode, I was able to select and download the Nvidia driver from the repositories. A number on command utilities show the Nvidia driver loaded which agrees with Driver Manager. Dmesg indicates a "tainted" kernel. There was a lockup after about 4 hours!

[ClixTrix]

No lockups due to idle for past 5 days on my system, and that's with 24 hour up-time.

The report at kernel.org has a recent update with excellent explanation of the problem.

https://bugzilla.kernel.org/show_bug.cgi?id=196683

The problem appears to have started with this kernel change last June.

https://git.kernel.org/pub/scm/linux/ke ... 0602c25764

[miscous]

I began this thread on October 30. Thanks to all for the suggestions. One week ago after 4 days lockup free... a lockup! I was ready to reload, but I took one more pass through all the posts, and happened on a BIOS setting. In looking at C-states, i followed a hint on a gaming website to this:

"In my ASUS PRIME B350-PLUS AM4 BIOS, I went to "Advanced" -> "AMD CBS" (near the bottom) -> "Global C-state Control" -> and set it to "Disabled"." Since I had an Asus PRIMEX370a, I decided to try this change. My Ryzen 7 has not had a lockup for a week: 12 hours or more per day, under load, idle, and me doing surfing too! I have used my original load, much tinkered with, a separate mint 18.2 load, Ubuntu 17.10(Mate and Gnome) .....NO Lockups!!!!

This is the link to Hardforum: https://hardforum.com/threads/how-to-st ... s.1931143/


At this point, I consider this solved. Thanks to all on Mint Forum. I'm looking forward to 18.3 ....