Comment 204 for bug 1690085

Revision history for this message
In , oyvinds (oyvinds-linux-kernel-bugs) wrote :

>> This issue is related to the power supply.
> You guys buy this?

No. This is hogwash.

EVGA SuperNOVA 750 G3 is a rather new PSU. Corsair RMi Series RM650i is also fairly new. Both of these were released after Haswell was in 2013. These are the ones I (ab)use for my Ryzen 1600X's. AMD's "engineer" can forget about selling me some story about how both of these PSUs are somehow flawed in a way that makes these CPUs hang shortly after boot if I leave the system idle (when not using the right kernel options/parameter).

If everyone with this problem was re-using a ancient PSU from an earlier build then there would have been a chance that this would hold water. That's just not the case and that makes it obvious that AMD story is hogwash.

It is still the case that kernel configuration CONFIG_RCU_NOCB_CPU=y and the kernel option rcu_nocbs=0-11 fixes this issue - completely on both my Ryzen systems. This tells me that a kernel patch can fix this (or permanently mask it) because it's plainly clear that software does fix it.