On 6/20/2025 5:27 PM, Zhao Liu wrote:
> As preparation for merging cache_info_cpuid4 and cache_info_amd in
> X86CPUState, set legacy cache model based on vendor in the CPUID
> 0x80000005 leaf. For AMD CPU, select legacy AMD cache model (in
> cache_info_amd) as the default cache model like before, otherwise,
> select legacy Intel cache model (in cache_info_cpuid4).
>
> To ensure compatibility is not broken, add an enable_legacy_vendor_cache
> flag based on x-vendor-only-v2 to indicate cases where the legacy cache
> model should be used regardless of the vendor. For CPUID 0x80000005
> leaf, enable_legacy_vendor_cache flag indicates to pick legacy AMD cache
> model, which is for compatibility with the behavior of PC machine v10.0
> and older.
>
> The following explains how current vendor-based default legacy cache
> model ensures correctness without breaking compatibility.
>
> * For the PC machine v6.0 and older, vendor_cpuid_only=false, and
> vendor_cpuid_only_v2=false.
>
> - If the named CPU model has its own cache model, and doesn't use
> legacy cache model (legacy_cache=false), then cache_info_cpuid4 and
> cache_info_amd are same, so 0x80000005 leaf uses its own cache model
> regardless of the vendor.
>
> - For max/host/named CPU (without its own cache model), then the flag
> enable_legacy_vendor_cache is true, they will use legacy AMD cache
> model just like their previous behavior.
>
> * For the PC machine v10.0 and older (to v6.1), vendor_cpuid_only=true,
> and vendor_cpuid_only_v2=false.
>
> - No change, since this leaf doesn't aware vendor_cpuid_only.
>
> * For the PC machine v10.1 and newer, vendor_cpuid_only=true, and
> vendor_cpuid_only_v2=true.
>
> - If the named CPU model has its own cache model (legacy_cache=false),
> then cache_info_cpuid4 & cache_info_amd both equal to its own cache
> model, so it uses its own cache model in 0x80000005 leaf regardless
> of the vendor. Only Intel CPUs have all-0 leaf due to
> vendor_cpuid_only_2=true, and this is exactly the expected behavior.
>
> - For max/host/named CPU (without its own cache model), then the flag
> enable_legacy_vendor_cache is false, the legacy cache model is
> selected based on vendor.
>
> For AMD CPU, it will use legacy AMD cache as expected.
>
> For Intel CPU, it will use legacy Intel cache but still get all-0
> leaf due to vendor_cpuid_only_2=true as expected.
>
> (Note) And for Zhaoxin CPU, it will use legacy Intel cache model
> instead of AMD's. This is the difference brought by this change! But
> it's correct since then Zhaoxin could have the consistent cache info
> in CPUID 0x2, 0x4 and 0x80000005 leaves.
>
> Here, except Zhaoxin, selecting the legacy cache model based on the
> vendor does not change the previous (before the change) behavior.
> And the change for Zhaoxin is also a good improvement.
>
> Therefore, the above analysis proves that, with the help of the flag
> enable_legacy_vendor_cache, it is acceptable to select the default
> legacy cache model based on the vendor.
>
> For the CPUID 0x80000005 leaf, in X86CPUState, a unified cache_info is
> enough. It only needs to be initialized and configured with the
> corresponding legacy cache model based on the vendor.
>
> Cc: EwanHai <ewanhai-oc@zhaoxin.com>
> Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
> ---
> Note, side effect of this patch: fix the inconsistency cache info for
> Zhaoxin. For more details, see the commit message above.
> ---
> target/i386/cpu.c | 36 ++++++++++++++++++++++++++++++++----
> 1 file changed, 32 insertions(+), 4 deletions(-)
>
> diff --git a/target/i386/cpu.c b/target/i386/cpu.c
> index afbf11569ab4..16b4ecb76113 100644
> --- a/target/i386/cpu.c
> +++ b/target/i386/cpu.c
> @@ -7945,8 +7945,35 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,
> *ecx = env->cpuid_model[(index - 0x80000002) * 4 + 2];
> *edx = env->cpuid_model[(index - 0x80000002) * 4 + 3];
> break;
> - case 0x80000005:
> - /* cache info (L1 cache/TLB Associativity Field) */
> + case 0x80000005: { /* cache info (L1 cache/TLB Associativity Field) */
Better put the comment into an independent line.
> + const CPUCaches *caches;
> +
> + if (env->enable_legacy_vendor_cache) {
> + caches = &legacy_amd_cache_info;
> + } else {
> + /*
> + * FIXME: Temporarily select cache info model here based on
> + * vendor, and merge these 2 cache info models later.
> + *
> + * This condition covers the following cases (with
> + * enable_legacy_vendor_cache=false):
> + * - When CPU model has its own cache model and doesn't uses legacy
> + * cache model (legacy_model=off). Then cache_info_amd and
> + * cache_info_cpuid4 are the same.
> + *
> + * - For v10.1 and newer machines, when CPU model uses legacy cache
> + * model. AMD CPUs use cache_info_amd like before and non-AMD
> + * CPU will use cache_info_cpuid4. But this doesn't matter,
> + * because for Intel CPU, it will get all-0 leaf, and Zhaoxin CPU
> + * will get correct cache info. Both are expected.
> + */
> + if (IS_AMD_CPU(env)) {
> + caches = &env->cache_info_amd;
> + } else {
> + caches = &env->cache_info_cpuid4;
> + }
> + }
> +
> if (cpu->cache_info_passthrough) {
> x86_cpu_get_cache_cpuid(index, 0, eax, ebx, ecx, edx);
> break;
> @@ -7961,9 +7988,10 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,
> (L1_ITLB_2M_ASSOC << 8) | (L1_ITLB_2M_ENTRIES);
> *ebx = (L1_DTLB_4K_ASSOC << 24) | (L1_DTLB_4K_ENTRIES << 16) |
> (L1_ITLB_4K_ASSOC << 8) | (L1_ITLB_4K_ENTRIES);
> - *ecx = encode_cache_cpuid80000005(env->cache_info_amd.l1d_cache);
> - *edx = encode_cache_cpuid80000005(env->cache_info_amd.l1i_cache);
> + *ecx = encode_cache_cpuid80000005(caches->l1d_cache);
> + *edx = encode_cache_cpuid80000005(caches->l1i_cache);
> break;
> + }
> case 0x80000006:
> /* cache info (L2 cache/TLB/L3 cache) */
> if (cpu->cache_info_passthrough) {