userfaultfd: don't BUG_ON() if khugepaged yanks our page table

Since khugepaged was changed to allow retracting page tables in file
mappings without holding the mmap lock, these BUG_ON()s are wrong - get
rid of them.

We could also remove the preceding "if (unlikely(...))" block, but then we
could reach pte_offset_map_lock() with transhuge pages not just for file
mappings but also for anonymous mappings - which would probably be fine
but I think is not necessarily expected.

Link: https://lkml.kernel.org/r/20240813-uffd-thp-flip-fix-v2-2-5efa61078a41@google.com
Fixes: 1d65b77 ("mm/khugepaged: retract_page_tables() without mmap or vma lock")
Signed-off-by: Jann Horn <jannh@google.com>
Reviewed-by: Qi Zheng <zhengqi.arch@bytedance.com>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Pavel Emelyanov <xemul@virtuozzo.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm/userfaultfd.c

@@ -807,9 +807,10 @@ static __always_inline ssize_t mfill_atomic(struct userfaultfd_ctx *ctx,
 			err = -EFAULT;
 			break;
 		}
-
-		BUG_ON(pmd_none(*dst_pmd));
-		BUG_ON(pmd_trans_huge(*dst_pmd));
+		/*
+		 * For shmem mappings, khugepaged is allowed to remove page
+		 * tables under us; pte_offset_map_lock() will deal with that.
+		 */
 
 		err = mfill_atomic_pte(dst_pmd, dst_vma, dst_addr,
 				       src_addr, flags, &folio);