Comment 659 for bug 1690085

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

(In reply to Borislav Petkov from comment #571)
> (In reply to Lars Viklund from comment #570)
> > rdmsr yields 6800000000010, which has bit 4 set.
>
> Looks like your BIOS applies the fix. Now, does the patch in comment #526
> fix your freezes?

I'm sorry, I don't currently suffer from any significant hangs on my machines when running stock Ubuntu 18.04.02 LTS (4.15.0-43-generic).

I mostly meant #570 to demonstrate that some firmware does indeed implement the errata.

While I've had core hangs and hard freezes in the past, the machine has enough workarounds applied to make it sufficiently stable. Current uptime is something like 81 days since the last incident and I can honestly not remember what it did then, but it's been used heavily as a build machine.

From my memory: sufficiently RMA:d CPUs, slight overclocking, change the current idle setting in firmware, boot idle=nomwait.

Back when I had significant problems, the best way to mitigate them was to disable SMT.

It's possible that my increased stability has come from firmware implementing this errata, or it might be coincidental, I don't have enough data to tell.

(You people should be happy you're not running FreeBSD, there I can reasonably reliably hang Ryzens within hours by sending ZFS snapshots :D )