public inbox for devel@edk2.groups.io
 help / color / mirror / Atom feed
From: Laszlo Ersek <lersek@redhat.com>
To: Brijesh Singh <brijesh.singh@amd.com>, edk2-devel@lists.01.org
Cc: Tom Lendacky <thomas.lendacky@amd.com>,
	Jordan Justen <jordan.l.justen@intel.com>,
	Ard Biesheuvel <ard.biesheuvel@linaro.org>
Subject: Re: [PATCH v1 3/4] OvmfPkg: IommuDxe: Zero the shared page(s) on Unmap()
Date: Wed, 2 Aug 2017 09:37:07 +0200	[thread overview]
Message-ID: <58dc46c4-59ae-4743-214a-df25b138303b@redhat.com> (raw)
In-Reply-To: <1501529474-20550-4-git-send-email-brijesh.singh@amd.com>

On 07/31/17 21:31, Brijesh Singh wrote:
> To support the Map(), we allocate bounce buffer with C-bit cleared,
> the buffer is referred as a DeviceAddress. Typically, DeviceAddress
> is used as communication block between guest and hypervisor. When
> guest is done with communication block, it calls Unmap().The Unmap()
> free's the DeviceAddress, if we do not clear the content of shared
> communication block during Unmap() then data remains readble to the
> hypervisor for an unpredicatable time. Let's zero the bounce buffer
> after we are done using it.
> 
> I did some benchmark and did not see any measure perform impact of
> zeroing the page(s).
> 
> Suggested-by: Laszlo Ersek <lersek@redhat.com>
> Contributed-under: TianoCore Contribution Agreement 1.0
> Cc: Laszlo Ersek <lersek@redhat.com>
> Cc: Jordan Justen <jordan.l.justen@intel.com>
> Signed-off-by: Brijesh Singh <brijesh.singh@amd.com>
> ---
>  OvmfPkg/IoMmuDxe/AmdSevIoMmu.c | 18 ++++++++++--------
>  1 file changed, 10 insertions(+), 8 deletions(-)
> 
> diff --git a/OvmfPkg/IoMmuDxe/AmdSevIoMmu.c b/OvmfPkg/IoMmuDxe/AmdSevIoMmu.c
> index 5ae54482fffe..04e3725ff7e6 100644
> --- a/OvmfPkg/IoMmuDxe/AmdSevIoMmu.c
> +++ b/OvmfPkg/IoMmuDxe/AmdSevIoMmu.c
> @@ -67,8 +67,7 @@ SetBufferAsEncDec (
>    // buffer matches with same encryption mask.
>    //
>    if (!Enc) {
> -    Status = MemEncryptSevClearPageEncMask (0, MapInfo->DeviceAddress,
> -               MapInfo->NumberOfPages, TRUE);
> +    Status = MemEncryptSevClearPageEncMask (0, TempBuffer, MapInfo->NumberOfPages, TRUE);
>      ASSERT_EFI_ERROR (Status);
>    }
>  
> @@ -79,7 +78,7 @@ SetBufferAsEncDec (
>    //
>    CopyMem (
>      (VOID *) (UINTN) TempBuffer,
> -    (VOID *) (UINTN)MapInfo->HostAddress,
> +    (VOID *) (UINTN) MapInfo->HostAddress,
>      MapInfo->NumberOfBytes);
>  
>    //
> @@ -109,11 +108,8 @@ SetBufferAsEncDec (
>    //
>    // Restore the encryption mask of the intermediate buffer
>    //
> -  if (!Enc) {
> -    Status = MemEncryptSevSetPageEncMask (0, MapInfo->DeviceAddress,
> -               MapInfo->NumberOfPages, TRUE);
> -    ASSERT_EFI_ERROR (Status);
> -  }
> +  Status = MemEncryptSevSetPageEncMask (0, TempBuffer, MapInfo->NumberOfPages, TRUE);
> +  ASSERT_EFI_ERROR (Status);
>  
>    //
>    // Free the intermediate buffer
> @@ -386,6 +382,12 @@ IoMmuUnmap (
>    ASSERT_EFI_ERROR(Status);
>  
>    //
> +  // Zero the shared memory so that hypervisor no longer able to get intelligentable
> +  // data.
> +  //
> +  SetMem ((VOID *) (UINTN)MapInfo->DeviceAddress, MapInfo->NumberOfBytes, 0);

Please use ZeroMem().

Furthermore, ZeroMem() should occur just before every FreePages() call:
- when Unmap() releases the implicitly allocated bounce buffer
- when FreeBuffer() releases the explicitly allocated common buffer
  (I thought I spelled this out in my previous email(s), but in
  retrospect it seems I only intended to :/ )
- in the virtio drivers' exit-boot-services callbacks, FreeBuffer()
  can't be called (only Unmap(), after the virtio reset), so the
  ZeroMem() should be done manually there.

Thanks
Laszlo

> +
> +  //
>    // Free the bounce buffer
>    //
>    gBS->FreePages (MapInfo->DeviceAddress, MapInfo->NumberOfPages);
> 



  parent reply	other threads:[~2017-08-02  7:35 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-07-31 19:31 [PATCH v1 0/4] OvmfPkg : IoMmuDxe: BusMasterCommonBuffer support when SEV is active Brijesh Singh
2017-07-31 19:31 ` [PATCH v1 1/4] OvmfPkg: IommuDxe: Do not clear C-bit in Allocate() and Free() Brijesh Singh
2017-08-01 20:42   ` Laszlo Ersek
2017-07-31 19:31 ` [PATCH v1 2/4] OvmfPkg: IommuDxe: Provide support for mapping BusMasterCommonBuffer operation Brijesh Singh
2017-07-31 19:49   ` Ard Biesheuvel
2017-07-31 20:27     ` Brijesh Singh
2017-08-01 20:52       ` Laszlo Ersek
2017-08-01 21:59   ` Laszlo Ersek
2017-08-01 23:51     ` Brijesh Singh
2017-08-02  8:41       ` Laszlo Ersek
2017-07-31 19:31 ` [PATCH v1 3/4] OvmfPkg: IommuDxe: Zero the shared page(s) on Unmap() Brijesh Singh
2017-07-31 19:38   ` Brijesh Singh
2017-08-02  7:37   ` Laszlo Ersek [this message]
2017-08-02 11:22     ` Brijesh Singh
2017-08-02 12:52       ` Laszlo Ersek
2017-07-31 19:31 ` [PATCH v1 4/4] OvmfPkg : QemuFwCfgLib: Map DMA buffer with CommonBuffer when SEV is enable Brijesh Singh
2017-08-02  8:09   ` Laszlo Ersek

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-list from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=58dc46c4-59ae-4743-214a-df25b138303b@redhat.com \
    --to=devel@edk2.groups.io \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox