Re: [PATCH 01/31] mm, gup: introduce concept of "foreign" get_user_pages()

[Dave Hansen]

On 01/13/2016 11:00 AM, Vlastimil Babka wrote:
>> > We leave a stub get_user_pages() around with a __deprecated
>> > warning.
> Hm when replying to previous version I assumed this is because there are many
> get_user_pages() callers remaining. But now I see there are just 3 drivers not
> converted by this patch? In that case I would favor to convert get_user_pages()
> to become what is now get_current_user_pages(). This would be much more
> consistent IMHO. We don't need to cater to out-of-tree modules?
> 
> Sorry, I should have looked thoroughly on the previous reply, not just assume.

It's really hard to submit a set of patches that remove a well-known
API.  New (in-tree) callers are always popping up, and you can see that
a few have popped up since I updated this the last time.  Without
leaving the old stub around, it virtually guarantees that this patch
will cause breakage in -next for a release or two.

I'll fix up the other bits you commented on, btw!