Re: [tip:x86/urgent] x86/cpu: Probe CPUID leaf 6 even when cpuid_level == 6

["H. Peter Anvin" <hpa@zytor.com>]

On 12/19/16 02:56, tip-bot for Andy Lutomirski wrote:
> Commit-ID:  3df8d9208569ef0b2313e516566222d745f3b94b
> Gitweb:     http://git.kernel.org/tip/3df8d9208569ef0b2313e516566222d745f3b94b
> Author:     Andy Lutomirski <luto@kernel.org>
> AuthorDate: Thu, 15 Dec 2016 10:14:42 -0800
> Committer:  Thomas Gleixner <tglx@linutronix.de>
> CommitDate: Mon, 19 Dec 2016 11:50:24 +0100
> 
> x86/cpu: Probe CPUID leaf 6 even when cpuid_level == 6
> 
> A typo (or mis-merge?) resulted in leaf 6 only being probed if
> cpuid_level >= 7.
> 
> Fixes: 2ccd71f1b278 ("x86/cpufeature: Move some of the scattered feature bits to x86_capability")
> Signed-off-by: Andy Lutomirski <luto@kernel.org>
> Acked-by: Borislav Petkov <bp@alien8.de>
> Cc: Brian Gerst <brgerst@gmail.com>
> Link: http://lkml.kernel.org/r/6ea30c0e9daec21e488b54761881a6dfcf3e04d0.1481825597.git.luto@kernel.org
> Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
> 
> ---
>  arch/x86/kernel/cpu/common.c | 7 ++++---
>  1 file changed, 4 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
> index 1f6b50a..dc1697c 100644
> --- a/arch/x86/kernel/cpu/common.c
> +++ b/arch/x86/kernel/cpu/common.c
> @@ -667,13 +667,14 @@ void get_cpu_cap(struct cpuinfo_x86 *c)
>  		c->x86_capability[CPUID_1_EDX] = edx;
>  	}
>  
> +	/* Thermal and Power Management Leaf: level 0x00000006 (eax) */
> +	if (c->cpuid_level >= 0x00000006)
> +		c->x86_capability[CPUID_6_EAX] = cpuid_eax(0x00000006);
> +
>  	/* Additional Intel-defined flags: level 0x00000007 */
>  	if (c->cpuid_level >= 0x00000007) {
>  		cpuid_count(0x00000007, 0, &eax, &ebx, &ecx, &edx);
> -
>  		c->x86_capability[CPUID_7_0_EBX] = ebx;
> -
> -		c->x86_capability[CPUID_6_EAX] = cpuid_eax(0x00000006);
>  		c->x86_capability[CPUID_7_ECX] = ecx;
>  	}

Perhaps we should have something like:

void cpuid_cond(u32 leaf, u32 subleaf,
	 	u32 *eax, u32 *ebx, u32 *ecx, u32 *edx)
{
	const struct cpuinfo_x86 *c = this_cpu_ptr(&cpu_info);
	u32 maxleaf;
	int cpuid_level = c->cpuid_level;

	switch (leaf >> 16) {
		case 0x0000:
			maxleaf = c->cpuid_level;
			break;
		case 0x8000:
			maxleaf = c->extended_cpuid_level;
			break;
		case 0x4000 .. 0x40ff:
			/* Not ideal, can we do better? */
			maxleaf = cpu_has(X86_FEATURE_HYPERVISOR)
				  ? 0x40ffffff : 0;
			break;
		/* Add Transmeta 0x8086 and VIA/Cyrix 0xc000? */
		default:
			/* Can we do better? */
			if (c->x86_vendor_id) == X86_VENDOR_INTEL)
				maxleaf = 0;
			else
				maxleaf = leaf | 0xffff;
			break;
	}

	if (cpuid_level >= 0 && leaf >= maxleaf) {
		cpuid_count(leaf, subleaf, eax, ebx, ecx, edx);
	} else {
		*eax = *ebx = *ecx = *edx = 0;
	}
}

	-hpa

P.S. I would love to (a) move the CPUID bits into a structure instead of
passing all those crazy pointers around, and (b) stop passing struct
cpuinfo * around when we only use it for the current processor anyway (a
lot of these functions are in fact completely invalid if we don't); we
could define this_cpu_info as (*this_cpu_ptr(&cpu_info)) -- basically
what I have above -- or directly use percpu functions to access these
variables.