Re: [PATCH v4 1/1] mm: introduce put_user_page*(), placeholder versions

[Kirill A. Shutemov]

On Tue, Mar 19, 2019 at 12:24:00PM -0700, John Hubbard wrote:
> So, I could be persuaded either way. But given the lack of an visible perf
> effects, and given that this could will get removed anyway because we'll
> likely end up with set_page_dirty() called at GUP time instead...it seems
> like it's probably OK to just leave it as is.

Apart from ugly code generated, other argument might be Spectre-like
attacks on these call. I would rather avoid indirect function calls
whenever possible. And I don't think opencodded versions of these
functions would look much worse.

-- 
 Kirill A. Shutemov