Re: [Xen-devel] [PATCH 4/8] x86: dynamically attach/detach CQM service for a guest

[Andrew Cooper <andrew.cooper3@xxxxxxxxxx>]

On 20/11/13 03:27, dongxiao.xu@xxxxxxxxx wrote:
> From: Dongxiao Xu <dongxiao.xu@xxxxxxxxx>
>
> Add hypervisor side support for dynamically attach and detach CQM
> services for a certain guest.
>
> Signed-off-by: Jiongxi Li <jiongxi.li@xxxxxxxxx>
> Signed-off-by: Dongxiao Xu <dongxiao.xu@xxxxxxxxx>
> ---
>  xen/arch/x86/domctl.c       |   66 
> +++++++++++++++++++++++++++++++++++++++++++
>  xen/include/public/domctl.h |   15 ++++++++++
>  2 files changed, 81 insertions(+)
>
> diff --git a/xen/arch/x86/domctl.c b/xen/arch/x86/domctl.c
> index f7e4586..df6a373 100644
> --- a/xen/arch/x86/domctl.c
> +++ b/xen/arch/x86/domctl.c
> @@ -35,6 +35,7 @@
>  #include <asm/mem_sharing.h>
>  #include <asm/xstate.h>
>  #include <asm/debugger.h>
> +#include <asm/pqos.h>
>  
>  static int gdbsx_guest_mem_io(
>      domid_t domid, struct xen_domctl_gdbsx_memio *iop)
> @@ -1223,6 +1224,71 @@ long arch_do_domctl(
>      }
>      break;
>  
> +    case XEN_DOMCTL_attach_pqos:
> +    {
> +        struct domain *d;
> +
> +        if ( (d = rcu_lock_domain_by_id(domctl->domain)) == NULL )
> +        {
> +            ret = -EBUSY;
> +            break;
> +        }

do_domctl has done this for us.  There is also no need to shadow 'd'.

> +
> +        if ( domctl->u.qos_res.flags & XEN_DOMCTL_ADF_pqos_cqm )
> +        {
> +            if ( !system_support_cqm() )
> +                ret = -ENODEV;

Surely this is generic, rather than condition on the caller specifying
XEN_DOMCTL_ADF_pqos_cqm

> +            else if ( d->arch.pqos_cqm_rmid > 0 )
> +                ret = -EINVAL;
> +            else
> +            {
> +                ret = 0;
> +                d->arch.pqos_cqm_rmid = alloc_cqm_resource(d->domain_id);
> +                if ( d->arch.pqos_cqm_rmid <= 0 )
> +                {
> +                    /* set to default value */
> +                    d->arch.pqos_cqm_rmid = 0;
> +                    ret = -ENODEV;

Perhaps E2BIG or EBUSY ?  This error can be fixed by detaching cqm from
other domains.

> +                }
> +            }
> +        }
> +        else
> +            ret = -EINVAL;
> +
> +        rcu_unlock_domain(d);
> +    }
> +    break;
> +
> +    case XEN_DOMCTL_detach_pqos:
> +    {
> +        struct domain *d;
> +
> +        if ( (d = rcu_lock_domain_by_id(domctl->domain)) == NULL )
> +        {
> +            ret = -EBUSY;
> +            break;
> +        }
> +
> +        if ( domctl->u.qos_res.flags & XEN_DOMCTL_ADF_pqos_cqm )
> +        {
> +            if ( !system_support_cqm() )
> +                ret = -ENODEV;
> +            else if ( d->arch.pqos_cqm_rmid > 0 )
> +            {
> +                free_cqm_resource(d->domain_id);
> +                d->arch.pqos_cqm_rmid = 0;
> +                ret = 0;
> +            }
> +            else
> +                ret = -EINVAL;
> +        }
> +        else
> +            ret = -EINVAL;
> +
> +        rcu_unlock_domain(d);
> +    }
> +    break;
> +
>      default:
>          ret = iommu_do_domctl(domctl, d, u_domctl);
>          break;
> diff --git a/xen/include/public/domctl.h b/xen/include/public/domctl.h
> index 47a850a..4fe21db 100644
> --- a/xen/include/public/domctl.h
> +++ b/xen/include/public/domctl.h
> @@ -872,6 +872,17 @@ struct xen_domctl_set_max_evtchn {
>  typedef struct xen_domctl_set_max_evtchn xen_domctl_set_max_evtchn_t;
>  DEFINE_XEN_GUEST_HANDLE(xen_domctl_set_max_evtchn_t);
>  
> +/* XEN_DOMCTL_attach_pqos */
> +/* XEN_DOMCTL_detach_pqos */
> +struct xen_domctl_qos_resource {
> + /* Attach or detach flag for cqm */
> +#define _XEN_DOMCTL_ADF_pqos_cqm      0
> +#define XEN_DOMCTL_ADF_pqos_cqm       (1U<<_XEN_DOMCTL_ADF_pqos_cqm)
> +    uint32_t flags;

The implementation semantics of this flag is redundant.  You
differentiate attach/detach both on the domctl subop, and the flag in
this structure, leaving two invalid states.

It might be better to have a XEN_DOMCTL_qpos_op and a sub command, if
more of these hypercalls are on their way.

> +};
> +typedef struct xen_domctl_qos_resource xen_domctl_qos_resource_t;
> +DEFINE_XEN_GUEST_HANDLE(xen_domctl_qos_resource_t);
> +
>  struct xen_domctl {
>      uint32_t cmd;
>  #define XEN_DOMCTL_createdomain                   1
> @@ -941,6 +952,9 @@ struct xen_domctl {
>  #define XEN_DOMCTL_setnodeaffinity               68
>  #define XEN_DOMCTL_getnodeaffinity               69
>  #define XEN_DOMCTL_set_max_evtchn                70
> +#define XEN_DOMCTL_attach_pqos                   71
> +#define XEN_DOMCTL_detach_pqos                   72
> +#define XEN_DOMCTL_list_pqos                     73

You introduce this list_pqos domctl with no implementation.

>  #define XEN_DOMCTL_gdbsx_guestmemio            1000
>  #define XEN_DOMCTL_gdbsx_pausevcpu             1001
>  #define XEN_DOMCTL_gdbsx_unpausevcpu           1002
> @@ -1001,6 +1015,7 @@ struct xen_domctl {
>          struct xen_domctl_set_broken_page_p2m set_broken_page_p2m;
>          struct xen_domctl_gdbsx_pauseunp_vcpu gdbsx_pauseunp_vcpu;
>          struct xen_domctl_gdbsx_domstatus   gdbsx_domstatus;
> +        struct xen_domctl_qos_resource      qos_res;
>          uint8_t                             pad[128];
>      } u;
>  };


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel