On Sat, Oct 18, 2025 at 4:41 AM Dave Hansen <[email protected]> wrote:
> Yeah, that 9=>8 change is exactly what I suspected. The low bit got > cleared. The first CPU has X86_FEATURE_REP_GOOD and the second one > doesn't. This is completely nonsensical on a HT pair. > > So, it's probably a BIOS bug. The BIOS forgot to set up the second > thread's MSR correctly. > > I'm not quite sure why newer kernels are complaining. The > "alternatives_patched" check has been around since ~6.10. If someone has > some spare time on their hands, they could bisect it. But I'm not sure > it will change anything. > > The warning might be annoying, but I do think it's probably harmless in > your case. X86_FEATURE_REP_GOOD is really just an optimization rather > than something functional. So even if it's wrong, I don't think it will > do any harm. > > Basically, the warning is working as intended. It caught a bug in > another piece of software that is putting absolutely incoherent data > into MSRs. If the warning really bothers you, I think setcpuid=0x70 on > the kernel command-line will work around it. > Many thanks! The warning still appears with setcpuid=0x70, but if it's harmless, it's ok.
