Re: [PATCH 1/2] selftests: harness: remove unneeded __constructor_order_last()

[Kees Cook]

On Sat, May 18, 2024 at 12:29:00PM +0900, Masahiro Yamada wrote:
> It will be set to "true" eventually,
> but __LIST_APPEND() still sees "false"
> on backward-order systems.

Ah, yes -- you are right. I looked through the commit history (I had
to go back to when the seccomp test, and the harness, was out of tree).
There was a time when the logic happened during the list walking, rather
than during list _creation_. I was remembering the former.

So, yes, let's make this change. As you say, it also solves for defining
TEST_HARNESS_MAIN before the tests. Thank you! I'd still like to replace
all the open-coded TEST_HARNESS_MAIN calls, though.

-- 
Kees Cook