Re: [PATCH RFC v1 02/11] x86/fpu/xstate: Introduce xstate order table and accessor macro

[Dave Hansen]

On 2/27/25 12:02, Ingo Molnar wrote:
>>> There's a 4th option:
>>>
>>>    4. Reuse XFEATURES 3/4 (MPX) only on APX-aware kernels, keep it 
>>>       disabled for old kernels.
>>>
>>> Problem solved.
>> The worry boils down to code in the kernel or userspace doing this:
>>
>> 	if (XGETBV(0) & 0x11)
>> 		do_mpx_gunk();
> New, APX-aware kernels wouldn't be doing this, why would they?

New, updated software is rarely the problem. It would not be a problem
in this case either.

The problem is with old OSes/VMMs and even old userspace on new kernels.
XGETBV() is unprivileged, so the concern extends to userspace and kernel.

>> So, sure, we could try to make sure that new kernels don't have any 
>> do_mpx_gunk() in them, but that doesn't help old kernels or other 
>> OSes/VMMs.
> Old kernels would *never* see this bit enabled if it's disabled by 
> default on bootup ...
> 
> VMMs would boot with it default-disabled as well, they can enable it 
> themselves.

The problem is with code which is logically like this:

        cpuid_count(CPUID_LEAF_XSTATE, 0, &eax, &ebx, &ecx, &edx);
        fpu_kernel_cfg.max_features = eax + ((u64)edx << 32);
	fpu_kernel_cfg.max_features &= SUPPORTED_FEATURE_MASK;
	xsetbv(fpu_kernel_cfg.max_features);

In that code, the kernel asks the CPU which features it supports and
then it enables all those features. In an old kernel,
SUPPORTED_FEATURE_MASK would contain the two MPX XFEATURES.

So I'm not _quite_ sure what you meant when you said:

> Old kernels would *never* see this bit enabled if it's disabled by
> default on bootup ...

Because old kernel very much do see the MPX bits in CPUID and very much
do enable them by default via XSETBV.

But I suspect I'm misunderstanding what you mean by "this bit" in first
place. Could you explain in some more detail, please?