Oh! That (s/__get_user_pages()/__get_user_pages_locked()/) seems to fix
it.

Attached a patch (gup-mlockall.txt, against v5.3.8) - although, I've
got no clue how patches work here.
gup-debug-mlockall.txt is the debug printk() I've added plus the code
change.
dmesg-out.txt.gz is the dmesg output of the latter for test.c


On Wed, 2019-11-06 at 12:26 +0100, Robert Stupp wrote:
> Maybe a native and wrong idea, but would it work to call
> __get_user_pages_locked() instead of __get_user_pages() in
> populate_vma_page_range() ?
>
> On Wed, 2019-11-06 at 11:25 +0100, Robert Stupp wrote:
> > Here's one more dmesg output with more information captured in
> > __get_user_pages() as well. It basically confirms that
> > handle_mm_fault() returns VM_FAULT_RETRY.
> >
> > I'm not sure where and what to change ("fix with a FOLL_TRIED
> > somewhere") to make it work. My (uneducated) impression is, that
> > only
> > __get_user_pages() needs to be changed - but I might be wrong.
> >