On Wed, May 20, 2020 at 12:32 AM John Hubbard <jhubbard@xxxxxxxxxx> wrote: > On 2020-05-19 19:39, Michel Lespinasse wrote: > >> That gives you additional options inside internal_get_user_pages_fast(), such > >> as, approximately: > >> > >> if (!(gup_flags & FOLL_FAST_ONLY)) > >> might_lock_read(¤t->mm->mmap_lock); > >> > >> ...not that that is necessarily a great idea, seeing as how it merely changes > >> "might lock" into "maybe might lock". :) > > > > I think that is completely fine, makes sure everyone not using > > FOLL_FAST_ONLY realizes that the call could block. > > > > Can I ask you to add that assertion in your patchset ? Based on > > Matthew's feedback, I would do it in my patchset, but it doesn't seem > > worth doing if we know this will conflict with your changes. > > Sure, that's no problem. Although it looks like my changes may land > in mmotm first, and then your patchset, so maybe the right move is to > make this change *after* both of those things happen, yes? I don't have a strong opinion on this. I would suggest you add the might_lock_read() assertion and whoever comes in second will deal with the conflict by changing mmap_sem to mmap_lock in that assertion. -- Michel "Walken" Lespinasse A program is never fully debugged until the last user dies.
