Re: [Xen-devel] [PATCH v1 07/14] x86/vvmx: restart nested vmentry in case of stale_np2m

[George Dunlap <george.dunlap@xxxxxxxxxx>]

On 09/04/2017 09:14 AM, Sergey Dyasli wrote:
> If an IPI flushes vCPU's np2m object just before nested vmentry, there
> will be a stale shadow EPTP value in VMCS02. Allow vmentry to be
> restarted in such cases and add nvmx_eptp_update() to perform an update.
> 
> Signed-off-by: Sergey Dyasli <sergey.dyasli@xxxxxxxxxx>
> ---
>  xen/arch/x86/hvm/vmx/entry.S |  6 ++++++
>  xen/arch/x86/hvm/vmx/vmx.c   |  8 +++++++-
>  xen/arch/x86/hvm/vmx/vvmx.c  | 14 ++++++++++++++
>  3 files changed, 27 insertions(+), 1 deletion(-)
> 
> diff --git a/xen/arch/x86/hvm/vmx/entry.S b/xen/arch/x86/hvm/vmx/entry.S
> index 53eedc6363..9fb8f89220 100644
> --- a/xen/arch/x86/hvm/vmx/entry.S
> +++ b/xen/arch/x86/hvm/vmx/entry.S
> @@ -79,6 +79,8 @@ UNLIKELY_END(realmode)
>  
>          mov  %rsp,%rdi
>          call vmx_vmenter_helper
> +        cmp  $0,%eax
> +        jne .Lvmx_vmentry_restart
>          mov  VCPU_hvm_guest_cr2(%rbx),%rax
>  
>          pop  %r15
> @@ -117,6 +119,10 @@ ENTRY(vmx_asm_do_vmentry)
>          GET_CURRENT(bx)
>          jmp  .Lvmx_do_vmentry
>  
> +.Lvmx_vmentry_restart:
> +        sti
> +        jmp  .Lvmx_do_vmentry
> +
>  .Lvmx_goto_emulator:
>          sti
>          mov  %rsp,%rdi
> diff --git a/xen/arch/x86/hvm/vmx/vmx.c b/xen/arch/x86/hvm/vmx/vmx.c
> index f6da119c9f..06509590b7 100644
> --- a/xen/arch/x86/hvm/vmx/vmx.c
> +++ b/xen/arch/x86/hvm/vmx/vmx.c
> @@ -4223,13 +4223,17 @@ static void lbr_fixup(void)
>          bdw_erratum_bdf14_fixup();
>  }
>  
> -void vmx_vmenter_helper(const struct cpu_user_regs *regs)
> +int vmx_vmenter_helper(const struct cpu_user_regs *regs)
>  {
>      struct vcpu *curr = current;
>      u32 new_asid, old_asid;
>      struct hvm_vcpu_asid *p_asid;
>      bool_t need_flush;
>  
> +    /* Shadow EPTP can't be updated here because irqs are disabled */
> +     if ( nestedhvm_vcpu_in_guestmode(curr) && 
> vcpu_nestedhvm(curr).stale_np2m )
> +         return 1;
> +
>      if ( curr->domain->arch.hvm_domain.pi_ops.do_resume )
>          curr->domain->arch.hvm_domain.pi_ops.do_resume(curr);
>  
> @@ -4290,6 +4294,8 @@ void vmx_vmenter_helper(const struct cpu_user_regs 
> *regs)
>      __vmwrite(GUEST_RIP,    regs->rip);
>      __vmwrite(GUEST_RSP,    regs->rsp);
>      __vmwrite(GUEST_RFLAGS, regs->rflags | X86_EFLAGS_MBS);
> +
> +    return 0;
>  }
>  
>  /*
> diff --git a/xen/arch/x86/hvm/vmx/vvmx.c b/xen/arch/x86/hvm/vmx/vvmx.c
> index ea2da14489..26ce349c76 100644
> --- a/xen/arch/x86/hvm/vmx/vvmx.c
> +++ b/xen/arch/x86/hvm/vmx/vvmx.c
> @@ -1405,12 +1405,26 @@ static void virtual_vmexit(struct cpu_user_regs *regs)
>      vmsucceed(regs);
>  }
>  
> +static void nvmx_eptp_update(void)
> +{
> +    if ( !nestedhvm_vcpu_in_guestmode(current) ||
> +          vcpu_nestedhvm(current).nv_vmexit_pending ||
> +         !vcpu_nestedhvm(current).stale_np2m ||
> +         !nestedhvm_paging_mode_hap(current) )
> +        return;
> +
> +    __vmwrite(EPT_POINTER, get_shadow_eptp(current));
> +    vcpu_nestedhvm(current).stale_np2m = false;

Hmm, so interrupts are enabled here.  What happens if a flush IPI occurs
between these two lines of code?  Won't we do the vmenter with a stale np2m?

It seems like we should clear stale_np2m first.  If an IPI occurs then,
we'll end up re-executing the vmenter unnecessarily, but it's better to
do that than to not re-execute it when we need to.

 -George

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