Re: [PATCH v4 24/71] perf/x86/msr: Switch to new Intel CPU model defines

[Dave Hansen]

On 4/25/24 09:43, Luck, Tony wrote:
>>> @@ -43,75 +43,75 @@ static bool test_intel(int idx, void *data)
>>>          boot_cpu_data.x86 != 6)
>>>              return false;
>> It arguably makes these easier to review when you _aren't_ removing the
>> explicit family 6 checks, but what's the plan for these?  They can go
>> away now, right?
> Yes. I expect that one will have to go if some non-family-6 CPUs are added
> to the switch. I didn't dig into what that function is testing for. But very recent
> CPUs have been added, so it seems likely that future ones will be added too.

Well, my point is that a .x86_vfm switch:

> +	switch (boot_cpu_data.x86_vfm) {
> +	case INTEL_NEHALEM:
> +	case INTEL_NEHALEM_G:

implicitly checks boot_cpu_data.x86==6.  So the explicit check can now
go away, no matter what CPUs are being checked.