On Thu, Nov 19, 2020 at 03:15:12PM +0100, Borislav Petkov wrote: > On Thu, Nov 19, 2020 at 09:50:10PM +0800, Feng Tang wrote: > > That's really odd. I tried on 3 baremetal machines: one Skylake NUC device, > > one Xeon E5-2699 and one Xeon E5-2680. > > Ah, sorry, not virt, virt is 0x4000_0000. Yeah, I remember now. It is > function 4 which AMD doesn't implement and I'm running this on AMD:
Great! That's the trick :) I will be more careful with these special ranges from different vendors. Thanks, Feng > $ cpuid -1r > CPU: > 0x00000000 0x00: eax=0x0000000d ebx=0x68747541 ecx=0x444d4163 > edx=0x69746e65 > 0x00000001 0x00: eax=0x00800f82 ebx=0x0c100800 ecx=0x7ed8320b > edx=0x178bfbff > 0x00000002 0x00: eax=0x00000000 ebx=0x00000000 ecx=0x00000000 > edx=0x00000000 > 0x00000003 0x00: eax=0x00000000 ebx=0x00000000 ecx=0x00000000 > edx=0x00000000 > > <-- no function 4. > > 0x00000005 0x00: eax=0x00000040 ebx=0x00000040 ecx=0x00000003 > edx=0x00000011 > ... > > That's why. :-) > > Btw, there are other funny ranges on Intel: > > ./cpuid -1r > CPU: > 0x20000000 0x00: eax=0x00000000 ebx=0x00000001 ecx=0x00000001 > edx=0x00000000 > > That one has 2 bits set. > > 0x80860000 0x00: eax=0x00000000 ebx=0x00000001 ecx=0x00000001 > edx=0x00000000 > 0xc0000000 0x00: eax=0x00000000 ebx=0x00000001 ecx=0x00000001 > edx=0x00000000 > > And those too. > > Fun. > > -- > Regards/Gruss, > Boris. > > https://people.kernel.org/tglx/notes-about-netiquette
