Comment 147 for bug 1690085

Another affected user here. Got my RMA week 33 segfault-free Ryzen 7 1700 a few weeks ago, running Gentoo on a custom vanilla kernel 4.13.10 atm. After 10 days of uptime, I now consider it stable. I did have freezes before, also with my pre-RMA CPU, but rcu_nocbs boot parameter seems to have fixed it, together with CONFIG_RCU_NOCB_CPU of course.

What I don't understand is why this should be another hardware bug? Ok, we don't know, maybe it is, but even then, this time there should be some software workaround. And I don't necessarily mean this RCU offloading-thing, which probably just masks the issue, but more in the terms of some underlying RCU (?) kernel issue triggering on this platform...?

Also consider that there are no freeze-issues on Windows as far as I know. Ok, AMD quickly put out their AMD Ryzen Balanced power plan, but I think it's mostly about SMT-scheduling, although it does disable C6 (I think, not sure). But anyhow, even on default Windows power plans, there are no freezes. So... either Microsoft quickly and silently incorporated some patch fixing this issue a long time ago, or it never was an issue for Windows in the first place. Doesn't matter however, the point is that something can be done in software in order to make this "bug" not triggering.

Forgive my ignorance, but why is it so difficult to assign this bug to the correct kernel dev? I'm just a user really and wondering... thanks.