Re: [PATCH v6 05/10] x86/tdx: Handle port I/O

From: Kuppuswamy, Sathyanarayanan
Date: Thu Sep 23 2021 - 13:24:30 EST




On 9/23/21 9:32 AM, Tom Lendacky wrote:
On 9/22/21 5:52 PM, Kuppuswamy Sathyanarayanan wrote:
From: "Kirill A. Shutemov" <kirill.shutemov@xxxxxxxxxxxxxxx>

TDX hypervisors cannot emulate instructions directly. This includes
port IO which is normally emulated in the hypervisor. All port IO
instructions inside TDX trigger the #VE exception in the guest and
would be normally emulated there.

Also string I/O is not supported in TDX guest. So, unroll the string
I/O operation into a loop operating on one element at a time. This
method is similar to AMD SEV, so just extend the support for TDX guest
platform.

Add a new confidential guest flag CC_ATTR_GUEST_UNROLL_STRING_IO to
add string unroll support in asm/io.h

Co-developed-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@xxxxxxxxxxxxxxx>
Signed-off-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@xxxxxxxxxxxxxxx>
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@xxxxxxxxxxxxxxx>
Reviewed-by: Andi Kleen <ak@xxxxxxxxxxxxxxx>
Reviewed-by: Dan Williams <dan.j.williams@xxxxxxxxx>
---

Changes since v5:
  * Changed prot_guest_has() to cc_platform_has().

Changes since v4:
  * Changed order of variable declaration in tdx_handle_io().
  * Changed tdg_* prefix with tdx_*.

Changes since v3:
  * Included PATTR_GUEST_UNROLL_STRING_IO protected guest flag
    addition change in this patch.
  * Rebased on top of Tom Lendacks protected guest change.

Changes since v2:
  * None

Changes since v1:
  * Fixed comments for tdg_handle_io().
  * Used _tdx_hypercall() instead of __tdx_hypercall() in tdg_handle_io().

  arch/x86/include/asm/io.h   |  7 +++++--
  arch/x86/kernel/cpu/intel.c |  1 +
  arch/x86/kernel/tdx.c       | 35 +++++++++++++++++++++++++++++++++++
  include/linux/cc_platform.h | 11 +++++++++++
  4 files changed, 52 insertions(+), 2 deletions(-)

diff --git a/arch/x86/include/asm/io.h b/arch/x86/include/asm/io.h
index fa6aa43e5dc3..67e0c4a0a0f4 100644
--- a/arch/x86/include/asm/io.h
+++ b/arch/x86/include/asm/io.h
@@ -40,6 +40,7 @@
  #include <linux/string.h>
  #include <linux/compiler.h>
+#include <linux/cc_platform.h>
  #include <asm/page.h>
  #include <asm/tdx.h>
  #include <asm/early_ioremap.h>
@@ -310,7 +311,8 @@ static inline unsigned type in##bwl##_p(int port)            \
                                      \
  static inline void outs##bwl(int port, const void *addr, unsigned long count) \
  {                                    \
-    if (sev_key_active()) {                        \ > +    if (sev_key_active() ||                        \
+        cc_platform_has(CC_ATTR_GUEST_UNROLL_STRING_IO)) {        \

Would it make sense to make sev_key_active() and sev_enable_key generic and just re-use those instead of adding CC_ATTR_GUEST_UNROLL_STRING_IO and having multiple conditions here?

You can set the key in the TDX init routine just like SEV does.

Any reason for using sev_enable_key over CC attribute? IMO, CC attribute exist
to generalize the common feature code. My impression is SEV is specific to AMD
code.


Thanks,
Tom


--
Sathyanarayanan Kuppuswamy
Linux Kernel Developer