Comment 669 for bug 1690085

I've spent the last few days reading through this thread.

I have Fedora 28, ryzen 1700x, and ASRock Steel Legend B450M. It's a server mostly used for files at the moment. Bone stock Fedora 28, samba, ZFS.

With GPU plugged in, hangs within 6-10 hours. Replicating this too often since March 27th.

With GPU unplugged, hangs happen within 10-20 minutes. Replicated 3 times.

Does not hang while SSH'd in.

Ran Memtest and tested for the segfault issue. Hardware is fine. Bought 'open box' on Amazon; this was in case they were re-selling an affected CPU without knowing.

Have not tried alfie's patch. Do have kernel 4.20 compiled with alfie's newest rifw patch as per Comment 580.

With Borislav's patch:
- Modifications to the OS are Samba, ZFS, Borislav's patch (comment #526 on kernel 5.0.1)
- booted March 30th at ~22:00
- Hung March 31st at ~09:40
- This is slightly longer than other tests
- I used the instructions he provided in Comment 511 without checking out 4.20. It compiles, but doesn't fix the issue, at least with linux 5.0.1-rc2 (apologies for not following steps and using 4.20)

'Typical Current Idle' set:
- Modifications to the OS are only Samba and ZFS
- Booted at 20:45 EST
- Still on now at time of this post 11:40 EST(I'll check before I hit send)
- This is the record up-time for this server
- Haven't tested anything else because I needed to have reliable access to my shares for my days off. This was 1 seemingly sure-fire way of making that happen.

Will update with more results once linux 4.20 with borislav and Alfie's patches tested. Could be a couple days. Less if 'Typical Current Idle' doesn't work, as I'll likely be RMAing my mobo and CPU.