[edk2-devel] [PATCH 1/1] OvmfPkg/VmgExitLib: HALT on #VE when access to private memory

Yao, Jiewen jiewen.yao at intel.com
Fri Oct 28 07:14:40 UTC 2022


Thanks for the patch.

1) I suggest to say "EPT-violation #VE on private memory is not allowed!"

2) The VMM is not trusted. We need put an unconditional CpuDeadLoop() after TDVMCALL_HALT.

3) Please test both shared address and private address in EPT-violation case.

Thank you
Yao Jiewen


> +    DEBUG ((DEBUG_ERROR, "EPT-violation #VE on private memory is a bug
> or an attack."));
> +    ASSERT (FALSE);
> +    TdVmCall (TDVMCALL_HALT, 0, 0, 0, 0, 0);

> -----Original Message-----
> From: Xu, Min M <min.m.xu at intel.com>
> Sent: Friday, October 28, 2022 3:02 PM
> To: devel at edk2.groups.io
> Cc: Xu, Min M <min.m.xu at intel.com>; Aktas, Erdem
> <erdemaktas at google.com>; Gerd Hoffmann <kraxel at redhat.com>; James
> Bottomley <jejb at linux.ibm.com>; Yao, Jiewen <jiewen.yao at intel.com>;
> Tom Lendacky <thomas.lendacky at amd.com>
> Subject: [PATCH 1/1] OvmfPkg/VmgExitLib: HALT on #VE when access to
> private memory
> 
> From: Min M Xu <min.m.xu at intel.com>
> 
> BZ: https://bugzilla.tianocore.org/show_bug.cgi?id=4125
> 
> EPT-violation #VE should be always on shared memory, which means the
> shared bit of the GuestPA should be set. But in current #VE Handler
> it is not checked. When it occurs, stop TD immediately and log out
> the error.
> 
> Cc: Erdem Aktas <erdemaktas at google.com>
> Cc: Gerd Hoffmann <kraxel at redhat.com>
> Cc: James Bottomley <jejb at linux.ibm.com>
> Cc: Jiewen Yao <jiewen.yao at intel.com>
> Cc: Tom Lendacky <thomas.lendacky at amd.com>
> Signed-off-by: Min Xu <min.m.xu at intel.com>
> ---
>  .../Library/VmgExitLib/VmTdExitVeHandler.c    | 40 ++++++++++++++-----
>  1 file changed, 29 insertions(+), 11 deletions(-)
> 
> diff --git a/OvmfPkg/Library/VmgExitLib/VmTdExitVeHandler.c
> b/OvmfPkg/Library/VmgExitLib/VmTdExitVeHandler.c
> index b73e877c093b..5bc0e9b3aa74 100644
> --- a/OvmfPkg/Library/VmgExitLib/VmTdExitVeHandler.c
> +++ b/OvmfPkg/Library/VmgExitLib/VmTdExitVeHandler.c
> @@ -300,23 +300,41 @@ MmioExit (
>    IN TDCALL_VEINFO_RETURN_DATA   *Veinfo
>    )
>  {
> -  UINT64   Status;
> -  UINT32   MmioSize;
> -  UINT32   RegSize;
> -  UINT8    OpCode;
> -  BOOLEAN  SeenRex;
> -  UINT64   *Reg;
> -  UINT8    *Rip;
> -  UINT64   Val;
> -  UINT32   OpSize;
> -  MODRM    ModRm;
> -  REX      Rex;
> +  UINT64          Status;
> +  UINT32          MmioSize;
> +  UINT32          RegSize;
> +  UINT8           OpCode;
> +  BOOLEAN         SeenRex;
> +  UINT64          *Reg;
> +  UINT8           *Rip;
> +  UINT64          Val;
> +  UINT32          OpSize;
> +  MODRM           ModRm;
> +  REX             Rex;
> +  TD_RETURN_DATA  TdReturnData;
> +  UINT8           Gpaw;
> +  UINT64          TdSharedPageMask;
> 
>    Rip     = (UINT8 *)Regs->Rip;
>    Val     = 0;
>    Rex.Val = 0;
>    SeenRex = FALSE;
> 
> +  Status = TdCall (TDCALL_TDINFO, 0, 0, 0, &TdReturnData);
> +  if (Status == TDX_EXIT_REASON_SUCCESS) {
> +    Gpaw             = (UINT8)(TdReturnData.TdInfo.Gpaw & 0x3f);
> +    TdSharedPageMask = 1ULL << (Gpaw - 1);
> +  } else {
> +    DEBUG ((DEBUG_ERROR, "TDCALL failed with status=%llx\n", Status));
> +    return Status;
> +  }
> +
> +  if ((Veinfo->GuestPA & TdSharedPageMask) == 0) {
> +    DEBUG ((DEBUG_ERROR, "EPT-violation #VE on private memory is a bug
> or an attack."));
> +    ASSERT (FALSE);
> +    TdVmCall (TDVMCALL_HALT, 0, 0, 0, 0, 0);
> +  }
> +
>    //
>    // Default to 32bit transfer
>    //
> --
> 2.29.2.windows.2



-=-=-=-=-=-=-=-=-=-=-=-
Groups.io Links: You receive all messages sent to this group.
View/Reply Online (#95666): https://edk2.groups.io/g/devel/message/95666
Mute This Topic: https://groups.io/mt/94621644/1813853
Mute #ve:https://edk2.groups.io/g/devel/mutehashtag/ve
Group Owner: devel+owner at edk2.groups.io
Unsubscribe: https://edk2.groups.io/g/devel/unsub [edk2-devel-archive at redhat.com]
-=-=-=-=-=-=-=-=-=-=-=-




More information about the edk2-devel-archive mailing list