Re: [Qemu-devel] [PATCH] s390x: Enable KVM huge page backing support

[David Hildenbrand]

On 31.07.2018 10:42, Janosch Frank wrote:
> From: Janosch Frank <address@hidden>
> 
> QEMU has had huge page support for a longer time already, but KVM
> memory management under s390x needed some changes to work with huge
> backings.
> 
> Now that we have support, let's enable it if requested and
> available. Otherwise we now properly tell the user if there is no
> support and back out instead of failing to run the VM later on.
> 
> Signed-off-by: Janosch Frank <address@hidden>
> ---
> 
> Right now there's a conflict with arm for the capability, so I guess
> I'll have to touch the cap number soonish. :)
> 
> ---
>  linux-headers/linux/kvm.h | 1 +
>  target/s390x/kvm.c        | 6 ++++++
>  2 files changed, 7 insertions(+)
> 
> diff --git a/linux-headers/linux/kvm.h b/linux-headers/linux/kvm.h
> index 98f389a5a3..e3b1e3070c 100644
> --- a/linux-headers/linux/kvm.h
> +++ b/linux-headers/linux/kvm.h
> @@ -949,6 +949,7 @@ struct kvm_ppc_resize_hpt {
>  #define KVM_CAP_GET_MSR_FEATURES 153
>  #define KVM_CAP_HYPERV_EVENTFD 154
>  #define KVM_CAP_HYPERV_TLBFLUSH 155
> +#define KVM_CAP_S390_HPAGE 156
>  
>  #ifdef KVM_CAP_IRQ_ROUTING
>  
> diff --git a/target/s390x/kvm.c b/target/s390x/kvm.c
> index d923cf4240..1ce0b42039 100644
> --- a/target/s390x/kvm.c
> +++ b/target/s390x/kvm.c
> @@ -285,6 +285,12 @@ int kvm_arch_init(MachineState *ms, KVMState *s)
>  {
>      MachineClass *mc = MACHINE_GET_CLASS(ms);
>  
> +    if (mem_path && kvm_vm_enable_cap(s, KVM_CAP_S390_HPAGE, 0)) {
> +        error_report("Huge page backing was specified, "
> +                     "but this KVM does not support huge pages.");
> +        return -EINVAL;
> +    }
> +
>      mc->default_cpu_type = S390_CPU_TYPE_NAME("host");
>      cap_sync_regs = kvm_check_extension(s, KVM_CAP_SYNC_REGS);
>      cap_async_pf = kvm_check_extension(s, KVM_CAP_ASYNC_PF);
> 

Okay, I wonder if now is the right time to actually detect the pagesize
instead of relying on mempath pointing at hugetlbfs. I remember Power
already doing such a detection.

mempath is just an indicator.

-- 

Thanks,

David / dhildenb