Re: [Qemu-devel] [RFC untested PATCH] i386/cpu: make -cpu host support monitor/mwait

[Igor Mammedov <imammedo@redhat.com>]

On Fri, 8 Jun 2018 23:59:19 +0300
"Michael S. Tsirkin" <mst@redhat.com> wrote:

> When guest CPU PM is enabled, and with -cpu host, expose the host CPU
> MWAIT leaf to guest so guest can make good PM decisions.
> 
> Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
> ---
> 
> This builds but is untested. Is this a reasonable way to go about it?
> 
>  target/i386/cpu.h |  9 +++++++++
>  target/i386/cpu.c | 18 +++++++++++++-----
>  2 files changed, 22 insertions(+), 5 deletions(-)
> 
> diff --git a/target/i386/cpu.h b/target/i386/cpu.h
> index 664504610e..309f804573 100644
> --- a/target/i386/cpu.h
> +++ b/target/i386/cpu.h
> @@ -1378,6 +1378,15 @@ struct X86CPU {
>      /* if true the CPUID code directly forward host cache leaves to the guest */
>      bool cache_info_passthrough;
>  
> +    /* if true the CPUID code directly forwards
> +     * host monitor/mwait leaves to the guest */
> +    struct {
> +        uint32_t eax;
> +        uint32_t ebx;
> +        uint32_t ecx;
> +        uint32_t edx;
> +    } mwait;
> +
>      /* Features that were filtered out because of missing host capabilities */
>      uint32_t filtered_features[FEATURE_WORDS];
>  
> diff --git a/target/i386/cpu.c b/target/i386/cpu.c
> index 94260412e2..a49443de56 100644
> --- a/target/i386/cpu.c
> +++ b/target/i386/cpu.c
> @@ -3760,11 +3760,11 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,
>          }
>          break;
>      case 5:
> -        /* mwait info: needed for Core compatibility */
> -        *eax = 0; /* Smallest monitor-line size in bytes */
> -        *ebx = 0; /* Largest monitor-line size in bytes */
> -        *ecx = CPUID_MWAIT_EMX | CPUID_MWAIT_IBE;
> -        *edx = 0;
> +        /* MONITOR/MWAIT Leaf */
> +        *eax = cpu->mwait.eax; /* Smallest monitor-line size in bytes */
> +        *ebx = cpu->mwait.ebx; /* Largest monitor-line size in bytes */
> +        *ecx = cpu->mwait.ecx; /* flags */
> +        *edx = cpu->mwait.edx; /* mwait substates */
>          break;
>      case 6:
>          /* Thermal and Power Leaf */
> @@ -4595,6 +4595,14 @@ static void x86_cpu_realizefn(DeviceState *dev, Error **errp)
>          goto out;
>      }
>  
> +    if (xcc->host_cpuid_required && enable_cpu_pm) {
> +        host_cpuid(5, 0, &cpu->mwait.eax, &cpu->mwait.ebx,
> +                   &cpu->mwait.ecx, &cpu->mwait.edx);
> +    }
could this state be migrated? or 'host' is still unmigratable?

also max_x86_cpu_initfn() might be better place for filling it up.

> +    /* We always wake on interrupt even if host does not have the capability */
> +    /* mwait extended info: needed for Core compatibility */
> +    cpu->mwait.ecx |= CPUID_MWAIT_EMX | CPUID_MWAIT_IBE;
> +
>      if (cpu->apic_id == UNASSIGNED_APIC_ID) {
>          error_setg(errp, "apic-id property was not initialized properly");
>          return;