On 11/13/19 11:09 AM, Ira Weiny wrote: ...
diff --git a/mm/gup.c b/mm/gup.c index 82e7e4ce5027..90f5f95ee7ac 100644 --- a/mm/gup.c +++ b/mm/gup.c @@ -1756,11 +1756,11 @@ long get_user_pages(unsigned long start, unsigned long nr_pages, struct vm_area_struct **vmas) { /* - * FOLL_PIN must only be set internally by the pin_user_page*() and - * pin_longterm_*() APIs, never directly by the caller, so enforce that - * with an assertion: + * FOLL_PIN and FOLL_LONGTERM must only be set internally by the + * pin_user_page*() and pin_longterm_*() APIs, never directly by the + * caller, so enforce that with an assertion: */ - if (WARN_ON_ONCE(gup_flags & FOLL_PIN)) + if (WARN_ON_ONCE(gup_flags & (FOLL_PIN | FOLL_LONGTERM)))Don't we want to block FOLL_LONGTERM in get_user_pages_fast() as well after all this?
Yes. But with the latest idea to restore FOLL_LONGTERM to its original glory, that won't be an issue in the next version. heh. thanks, -- John Hubbard NVIDIA
