Re: [PATCH 0/1] uprobes: Kill __replace_page(), change uprobe_write_opcode() to rely on gup(WRITE)

[Oleg Nesterov <oleg@redhat.com>]

On 12/10, Linus Torvalds wrote:
>
> On Tue, Dec 10, 2013 at 12:04 PM, Oleg Nesterov <oleg@redhat.com> wrote:
> >> We'd be *much* better off using get_user_pages_fast() if possible -
> >> and I bet _is_ possible in 99% of all cases.
> >
> > We can't. get_user_pages_fast() takes mmap_sem.
>
> Yeah, and we need to look up the page table entry anyway, so what we
> actually want here is just the page table walker,

Yes, this is clear.

And damn, somehow I forgot that _fast() can (obviously) only use
current->mm.

> none of the "get
> page" crap at all.
>
> So the core function should (I think) just do something like:

I'll try to think, but this is what I actually tried to avoid.

I mean,

> Put another way: I actually think the existing "__replace_page()" code
> is closer to being good than that disgusting uprobe_write_opcode()
> function. I think you may be getting rid of the wrong ugly function.

And perhaps you are right.

But my only motivation was: rely on gup() to simplify this code. Yes,
gup() is slow, but it works and we can avoid the "nontrivial" things
like page_remove_rmap/munlock_vma_page.

If we want to optimize this code, then this patch obviously goes to
the wrong direction, I agree.

Oleg.