On Tue, Jun 25, 2024 at 12:09 PM Oleg Nesterov [off-list ref] wrote:

On 06/25, Andrii Nakryiko wrote:

quoted

On Tue, Jun 25, 2024 at 7:51 AM Oleg Nesterov [off-list ref] wrote:

quoted

Why?

So far I don't understand this change. Quite possibly I missed something,
but in this case the changelog should explain the problem more clearly.

I just went off of "Called with mm->mmap_lock held for write." comment
in uprobe_write_opcode(), tbh.

Ah, indeed... and git blame makes me sad ;)

I _think_ that 29dedee0e693a updated this comment without any thinking,
but today I can't recall. In any case, today this nothing to do with
mem_cgroup_charge(). Not sure __replace_page() is correct (in this respect)
when it returns -EAGAIN but this is another story.

quoted

If we don't actually need writer
mmap_lock, we should probably update at least that comment.

Agreed.

Ok, I'll drop this change and will just update the comment.

quoted

There is a
lot going on in uprobe_write_opcode(), and I don't understand all the
requirements there.

Heh. Neither me today ;)

Oleg.