Re: [PATCH 2/2] selftests/x86: add generic prctl shadow stack test

[Bill Roberts]

On 7/29/26 8:31 AM, Edgecombe, Rick P wrote:
> On Tue, 2026-07-14 at 10:47 -0500, Bill Roberts wrote:
> > diff --git a/tools/testing/selftests/x86/test_shadow_stack_prctl.c
> > b/tools/testing/selftests/x86/test_shadow_stack_prctl.c
> > new file mode 100644
> > index 000000000000..9c9e2728a9f9
> > --- /dev/null
> > +++ b/tools/testing/selftests/x86/test_shadow_stack_prctl.c
> > @@ -0,0 +1,3 @@
> > +// SPDX-License-Identifier: GPL-2.0
> > +#define BUILD_PRCTL 1
> > +#include "test_shadow_stack.c"
> > --
> Hmm, do we really need a full re-test for the different prctl routing?

Bear with me, I am quite perplexed, usually tests are meant with fanfare.
Yes we want this! It ensures all regressions are caught and features are 
working
as intended through both interfaces. This means, nothing broke for x86 
in this change.

Tests are cheap, I have no idea why we wouldn't want this.

Now, what I did want to propose in the future, is moving a generic test 
up out of
arch specific tests that would run on all systems. This will require me 
to look at all
3 arches and build a test suite. However, I wanted to ensure that 
nothing broke for x86,
this ensures that. I was going to roll those patches after the LSM 
changes go up.

As an aside and hearkening back to your comment on UNLOCK in patch 1, 
the test never tests UNLOCK.