Re: [PATCH] x86/mm: Unbreak modules that rely on external PAGE_KERNEL availability

[Linus Torvalds]

On Wed, Nov 8, 2017 at 1:23 PM, Linus Torvalds
<torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:
>
> So _this_ is the underlying problem:
>
>   #define _PAGE_ENC  (_AT(pteval_t, sme_me_mask))
>
> because that is simply not how the _PAGE_xyz macros should work!
>
> So it should have been a fixed bit to begin with, and the dynamic part
> should have been elsewhere.

Hmm. It's not an entirely new problem. We have that
"cachemode2protval()" thing, which causes he exact same thing, except
it accesses the __cachemode2pte_tbl[] array instead.

Which we also EXPORT_SYMBOL().

So I guess _PAGE_ENC isn't any worse than what we already had.

Of course, that at least doesn't trigger for the simple cases - only
_PAGE_CACHE_WP and _PAGE_NOCACHE end up triggering that
"cachemode2protval()" case.

I do wonder if we could perhaps at least try to unify these things a
bit, and export just one thing.

And maybe avoid accessing two completely different memory locasions
every time we use _PAGE_KERNEL or whatever.

But it all looks rather nasty, so for 4.14 clearly I should just apply
that trivial one-liner patch for now.

           Linus