Re: [PATCHv2] mm: Fix warning in move_normal_pmd()

[Linus Torvalds]

On Wed, Jul 15, 2020 at 1:55 PM Kirill A. Shutemov <kirill@xxxxxxxxxxxxx> wrote:
>
> I don't understand 'len' calculation in try_to_align_end().

Yeah, Joel found the same thing. You don't understand it, because it's garbage.

> BUT
>
> I *think* there's a bigger problem with the patch:
>
> For stack relocation case both VMAs are the same and always(?) the only
> VMA around at the time. It means none of ADDR_BEFORE_PREV and
> ADDR_AFTER_NEXT are going to stop us.

But for the stack relocation case, that should actually be fine. We
are moving the whole thing.

Or maybe I missed something.

> Consider the following case, before and after try_to_align_start():
>
>                 before          after
> old_addr:       0x0123000       0x0000000
> new_addr:       0x1123000       0x1000000
> len:            0x1000000       0x1123000

That's the "move up" case that fundamentally doesn't work anyway
because it will corrupt the data as it moves it.

The stack relocation only moves down.

               Linus