Re: [PATCH v2 33/39] x86/cpufeatures: Limit shadow stack to Intel CPUs

[Nathan Chancellor]

Hi Kees,

On Mon, Oct 03, 2022 at 09:54:26PM -0700, Kees Cook wrote:
> On Mon, Oct 03, 2022 at 05:09:04PM -0700, Dave Hansen wrote:
> > On 10/3/22 16:57, Kees Cook wrote:
> > > On Thu, Sep 29, 2022 at 03:29:30PM -0700, Rick Edgecombe wrote:
> > >> Shadow stack is supported on newer AMD processors, but the kernel
> > >> implementation has not been tested on them. Prevent basic issues from
> > >> showing up for normal users by disabling shadow stack on all CPUs except
> > >> Intel until it has been tested. At which point the limitation should be
> > >> removed.
> > >>
> > >> Signed-off-by: Rick Edgecombe <rick.p.edgecombe@xxxxxxxxx>
> > > So running the selftests on an AMD system is sufficient to drop this
> > > patch?
> > 
> > Yes, that's enough.
> > 
> > I _thought_ the AMD folks provided some tested-by's at some point in the
> > past.  But, maybe I'm confusing this for one of the other shared
> > features.  Either way, I'm sure no tested-by's were dropped on purpose.
> > 
> > I'm sure Rick is eager to trim down his series and this would be a great
> > patch to drop.  Does anyone want to make that easy for Rick?
> > 
> > <hint> <hint>
> 
> Hey Gustavo, Nathan, or Nick! I know y'all have some fancy AMD testing
> rigs. Got a moment to spin up this series and run the selftests? :)

I do have access to a system with an EPYC 7513, which does have Shadow
Stack support (I can see 'shstk' in the "Flags" section of lscpu with
this series). As far as I understand it, AMD only added Shadow Stack
with Zen 3; my regular AMD test system is Zen 2 (probably should look at
procurring a Zen 3 or Zen 4 one at some point).

I applied this series on top of 6.0 and reverted this change then booted
it on that system. After building the selftest (which did require
'make headers_install' and a small addition to make it build beyond
that, see below), I ran it and this was the result. I am not sure if
that is expected or not but the other results seem promising for
dropping this patch.

  $ ./test_shadow_stack_64
  [INFO]  new_ssp = 7f8a36c9fff8, *new_ssp = 7f8a36ca0001
  [INFO]  changing ssp from 7f8a374a0ff0 to 7f8a36c9fff8
  [INFO]  ssp is now 7f8a36ca0000
  [OK]    Shadow stack pivot
  [OK]    Shadow stack faults
  [INFO]  Corrupting shadow stack
  [INFO]  Generated shadow stack violation successfully
  [OK]    Shadow stack violation test
  [INFO]  Gup read -> shstk access success
  [INFO]  Gup write -> shstk access success
  [INFO]  Violation from normal write
  [INFO]  Gup read -> write access success
  [INFO]  Violation from normal write
  [INFO]  Gup write -> write access success
  [INFO]  Cow gup write -> write access success
  [OK]    Shadow gup test
  [INFO]  Violation from shstk access
  [OK]    mprotect() test
  [OK]    Userfaultfd test
  [FAIL]  Alt shadow stack test

  $ echo $?
  1

I am happy to provide any information that would be useful for exploring
this failure and test further iterations of this series as necessary.

Cheers,
Nathan

test_shadow_stack.c: In function ‘create_shstk’:
test_shadow_stack.c:86:70: error: ‘SHADOW_STACK_SET_TOKEN’ undeclared (first use in this function)
   86 |         return (void *)syscall(__NR_map_shadow_stack, addr, SS_SIZE, SHADOW_STACK_SET_TOKEN);
      |                                                                      ^~~~~~~~~~~~~~~~~~~~~~
test_shadow_stack.c:86:70: note: each undeclared identifier is reported only once for each function it appears in
test_shadow_stack.c:87:1: warning: control reaches end of non-void function [-Wreturn-type]
   87 | }
      | ^

diff --git a/tools/testing/selftests/x86/test_shadow_stack.c b/tools/testing/selftests/x86/test_shadow_stack.c
index 22b856de5cdd..958dbb248518 100644
--- a/tools/testing/selftests/x86/test_shadow_stack.c
+++ b/tools/testing/selftests/x86/test_shadow_stack.c
@@ -11,6 +11,7 @@
 #define _GNU_SOURCE
 
 #include <sys/syscall.h>
+#include <asm/mman.h>
 #include <sys/mman.h>
 #include <sys/stat.h>
 #include <sys/wait.h>