On Mon, Oct 22, 2018 at 08:59:46PM +0300, snikolov wrote: > On Mon, 2018-10-22 at 10:27 -0700, Mike Larkin wrote: > > On Mon, Oct 22, 2018 at 10:13:14AM -0700, Mike Larkin wrote: > > > On Mon, Oct 22, 2018 at 08:01:13PM +0300, snikolov wrote: > > > > > This appears to be related to the LFENCE serializing MSR change > > > > > that > > > > > went in > > > > > during the last round of side channel analysis fixes: > > > > > > > > > > ffffffff811c3037: b9 29 10 01 > > > > > c0 mov $0xc0011029,%ecx > > > > > ffffffff811c303c: 0f 32 rdmsr > > > > > > > > > > According to the commit, "This MSR is available on all AMD > > > > > families > > > > > > = 10h...", > > > > > > > > > > and since yours is family 15h, it should work. Maybe that > > > > > assumption > > > > > was wrong? > > > > > > > > > > -ml > > > > > > > > > > > > > The Host's CPU is FX-8350 ,so you assumed right. Yet, I am new to > > > > openBSD so I have no clue what approach to be taken. > > > > > > > > Strahil > > > > > > Maybe a BIOS update is available? The date on yours is 2014. Maybe > > > they > > > added that MSR after? > > > > > > > As brynet@ pointed out in a later reply, BIOS update probably won't > > help. > > > > -ml > > > > BIOS update is unavailable, yet the microcode patch_level is > '0x06000852'. The kernel is: 3.10.0-862.3.3.el7.x86_64 > > Any kernel parameters to disable the mitigations and still leave the > CPU at Opteron_G5 ? > > Strahil >
Not from the OpenBSD side. -ml
