Re: [PATCH v4 3/4] target/i386: KVM: allow fast string operations if host supports them

[Xiaoyao Li <xiaoyao.li@intel.com>]

On 2/27/2023 6:13 PM, Paolo Bonzini wrote:
> These are just a flag that documents the performance characteristic of
> an instruction; it needs no hypervisor support.  So include them even
> if KVM does not show them.  In particular, FZRM/FSRS/FSRC have only
> been added very recently, but they are available on Sapphire Rapids
> processors.
> 
> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
> ---
>   target/i386/kvm/kvm.c | 17 ++++++++++++++++-
>   1 file changed, 16 insertions(+), 1 deletion(-)
> 
> diff --git a/target/i386/kvm/kvm.c b/target/i386/kvm/kvm.c
> index 587030199192..fe66a4953d41 100644
> --- a/target/i386/kvm/kvm.c
> +++ b/target/i386/kvm/kvm.c
> @@ -352,7 +352,7 @@ uint32_t kvm_arch_get_supported_cpuid(KVMState *s, uint32_t function,
>   {
>       struct kvm_cpuid2 *cpuid;
>       uint32_t ret = 0;
> -    uint32_t cpuid_1_edx;
> +    uint32_t cpuid_1_edx, unused;
>       uint64_t bitmask;
>   
>       cpuid = get_supported_cpuid(s);
> @@ -399,10 +399,20 @@ uint32_t kvm_arch_get_supported_cpuid(KVMState *s, uint32_t function,
>       } else if (function == 6 && reg == R_EAX) {
>           ret |= CPUID_6_EAX_ARAT; /* safe to allow because of emulated APIC */
>       } else if (function == 7 && index == 0 && reg == R_EBX) {
> +        /* Not new instructions, just an optimization.  */
> +        uint32_t ebx;
> +        host_cpuid(1, 0, &unused, &ebx, &unused, &unused);
                       ^

It should be leaf 7, not 1.

> +        ret |= ebx & CPUID_7_0_EBX_ERMS;
> +
>           if (host_tsx_broken()) {
>               ret &= ~(CPUID_7_0_EBX_RTM | CPUID_7_0_EBX_HLE);
>           }
>       } else if (function == 7 && index == 0 && reg == R_EDX) {
> +        /* Not new instructions, just an optimization.  */
> +        uint32_t edx;
> +        host_cpuid(1, 0, &unused, &unused, &unused, &edx);

Ditto.

> +        ret |= edx & CPUID_7_0_EDX_FSRM;
> +
>           /*
>            * Linux v4.17-v4.20 incorrectly return ARCH_CAPABILITIES on SVM hosts.
>            * We can detect the bug by checking if MSR_IA32_ARCH_CAPABILITIES is
> @@ -411,6 +421,11 @@ uint32_t kvm_arch_get_supported_cpuid(KVMState *s, uint32_t function,
>           if (!has_msr_arch_capabs) {
>               ret &= ~CPUID_7_0_EDX_ARCH_CAPABILITIES;
>           }
> +    } else if (function == 7 && index == 1 && reg == R_EAX) {
> +        /* Not new instructions, just an optimization.  */
> +        uint32_t eax;
> +        host_cpuid(1, 0, &eax, &unused, &unused, &unused);

Ditto.

After them fixed,

Reviewed-by: Xiaoyao Li <xiaoyao.li@intel.com>

> +        ret |= eax & (CPUID_7_1_EAX_FZRM | CPUID_7_1_EAX_FSRS | CPUID_7_1_EAX_FSRC);
>       } else if (function == 0xd && index == 0 &&
>                  (reg == R_EAX || reg == R_EDX)) {
>           /*