> I bet copy_hugetlb_page_range() is causing your complaints. It takes
Yes. Looking at the warning I'm not sure why lockdep doesn't filter
it out automatically. I cannot think of a legitimate case where
a "possible recursive lock" with different lock addresses would be
a genuine bug.
So instead of a false positive, it's more like a "always false" :)
I wasn't aware of that one.
-Andi
--