Re: [PATCH v4 28/34] x86/cacheinfo: Use parsed CPUID(0x80000005) and CPUID(0x80000006)

[Dave Hansen <dave.hansen@intel.com>]

On 8/15/25 00:02, Ahmed S. Darwish wrote:
> +static void legacy_amd_cpuid4(struct cpuinfo_x86 *c, int index, struct leaf_0x4_0 *regs)
>  {
> -	unsigned int dummy, line_size, lines_per_tag, assoc, size_in_kb;
> -	union l1_cache l1i, l1d, *l1;
> -	union l2_cache l2;
> -	union l3_cache l3;
> +	const struct leaf_0x80000005_0 *el5 = cpuid_leaf(c, 0x80000005);
> +	const struct leaf_0x80000006_0 *el6 = cpuid_leaf(c, 0x80000006);
> +	const struct cpuid_regs *el5_raw = (const struct cpuid_regs *)el5;

Is there any way we could get rid of the casts? The lack of type safety
on those always worries me. Maybe a helper like this:

    const struct cpuid_regs *el5_raw = cpuid_leaf_raw(c, 0x80000006);

(although that would probably just do casts internally). The other way
would be to rig the macros up so that each 'struct leaf_$FOO' had a:

	union {
		struct leaf_0x80000005_0 l;
		struct cpuid_regs raw;
	};

and then the macros just selected one of the two.