Re: [Xen-devel] [PATCH] VMX: wbinvd when vmentry under UC

[Andrew Cooper <andrew.cooper3@xxxxxxxxxx>]

On 25/11/13 16:14, Liu, Jinsong wrote:
> From e2d47e2f75bac6876b7c2eaecfe946966bf27516 Mon Sep 17 00:00:00 2001
> From: Liu Jinsong <jinsong.liu@xxxxxxxxx>
> Date: Tue, 26 Nov 2013 04:53:17 +0800
> Subject: [PATCH] VMX: wbinvd when vmentry under UC
>
> This patch flush cache when vmentry back to UC guest, to prevent
> cache polluted by hypervisor access guest memory during UC mode.
>
> However, wbinvd is a _very_ time consuming operation, so
> 1. wbinvd ... timer has a good possibility to expire while
>    irq disabled, it then would be delayed until
> 2. ... vmentry back to guest (and irq enalbed), timer interrupt
>    then occurs and drops guest at once;
> 3. drop to hypervisor ... then vmentry and wbinvd again;
>
> This loop will run again and again, until lucky enough wbinvd
> happens not to expire timer and then loop break, usually it would
> occur 10K~60K times, blocking guest 10s~60s.
>
> reprogram timer to avoid dead_like_loop.
>
> Signed-off-by: Liu Jinsong <jinsong.liu@xxxxxxxxx>
> ---
>  xen/arch/x86/hvm/vmx/vmx.c |   32 ++++++++++++++++++++++++++++----
>  1 files changed, 28 insertions(+), 4 deletions(-)
>
> diff --git a/xen/arch/x86/hvm/vmx/vmx.c b/xen/arch/x86/hvm/vmx/vmx.c
> index 75be62e..4768c9b 100644
> --- a/xen/arch/x86/hvm/vmx/vmx.c
> +++ b/xen/arch/x86/hvm/vmx/vmx.c
> @@ -642,10 +642,6 @@ static void vmx_ctxt_switch_to(struct vcpu *v)
>              __invept(INVEPT_SINGLE_CONTEXT, ept_get_eptp(ept_data), 0);
>      }
>  
> -    /* For guest cr0.cd setting, do not use potentially polluted cache */
> -    if ( unlikely(v->arch.hvm_vcpu.cache_mode == NO_FILL_CACHE_MODE) )
> -        wbinvd();
> -
>      vmx_restore_guest_msrs(v);
>      vmx_restore_dr(v);
>  }
> @@ -2967,6 +2963,27 @@ out:
>          nvmx_idtv_handling();
>  }
>  
> +/*
> + * wbinvd is a _very_ time consuming operation, so
> + * 1. wbinvd ... timer has a good possibility to expire while
> + *    irq disabled, it then would be delayed until
> + * 2. ... vmentry back to guest (and irq enalbed), timer interrupt
> + *    then occurs and drops guest at once;
> + * 3. drop to hypervisor ... then vmentry and wbinvd again;
> + *
> + * This loop will run again and again, until lucky enough wbinvd
> + * happens not to expire timer and then loop break, usually it would
> + * occur 10K~60K times, blocking guest 10s~60s.
> + *
> + * reprogram timer to avoid dead_like_loop.
> + */
> +static inline void uc_wbinvd_and_timer_adjust(void)
> +{
> +    reprogram_timer(0);
> +    wbinvd();
> +    reprogram_timer(NOW() + MILLISECS(1));

Do you have any number of the time delta across the wbinvd() ?

As it currently stands, I do not think it is reasonable to reprogram the
timer like this.

~Andrew

> +}
> +
>  void vmx_vmenter_helper(const struct cpu_user_regs *regs)
>  {
>      struct vcpu *curr = current;
> @@ -2974,6 +2991,13 @@ void vmx_vmenter_helper(const struct cpu_user_regs 
> *regs)
>      struct hvm_vcpu_asid *p_asid;
>      bool_t need_flush;
>  
> +    /*
> +     * In case hypervisor may access hvm guest memory, and then
> +     * cache line polluted under UC mode.
> +     */
> +    if ( unlikely(curr->arch.hvm_vcpu.cache_mode == NO_FILL_CACHE_MODE) )
> +        uc_wbinvd_and_timer_adjust();
> +
>      if ( !cpu_has_vmx_vpid )
>          goto out;
>      if ( nestedhvm_vcpu_in_guestmode(curr) )


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