[PATCH] x86/uaccess: Avoid barrier_nospec() in copy_from_user()

[Linus Torvalds]

On Mon, 21 Oct 2024 at 03:48, Kirill A. Shutemov <kirill at shutemov.name> wrote:
>
> LAM brings own speculation issues[1] that is going to be addressed by
> LASS[2]. There was a patch[3] to disable LAM until LASS is landed, but it
> never got applied for some reason.

Bah., I think the reason was that nobody knew what - if any - any hardware.

I did find an Intel "Product Brief" for "Intel® Xeon® 6 processors
with E-cores" that claim both LASS and LAM, but that one puts LAM as a
"security feature", so that seems a bit confused. That's Redwood Cove
cores, I think.

I guess that should mean that Meteor Lake ("Core Ultra") should also
have it? Or is it then turned off because the E cores don't do it?

I do find LASS - but not LAM - mentioned in the "Intel Core Ultra (PS
series)" datasheet.

And there's a Intel Core Ultra 200V datasheet (1 of 2) that does
mention both LAM and LASS. So I guess now it *really* exists.

Can somebody confirm?

But yeah, I guess we should just do this:

> [3] https://lore.kernel.org/all/5373262886f2783f054256babdf5a98545dc986b.1706068222.git.pawan.kumar.gupta@linux.intel.com

for now.

                  Linus