mm: hugetlb: fix pgoff computation when unmapping page from vma

The computation for pgoff is incorrect, at least with

(vma->vm_pgoff >> PAGE_SHIFT)

involved. It is fixed with the available method if HPAGE_SIZE is
concerned in page cache lookup.

[[e-mail]: use vma_hugecache_offset() directly, per Michal]
Signed-off-by: Hillf Danton <[e-mail]>
Cc: Mel Gorman <[e-mail]>
Cc: Michal Hocko <[e-mail]>
Reviewed-by: KAMEZAWA Hiroyuki <[e-mail]>
Cc: Andrea Arcangeli <[e-mail]>
Cc: David Rientjes <[e-mail]>
Reviewed-by: Michal Hocko <[e-mail]>
Signed-off-by: Andrew Morton <[e-mail]>
Signed-off-by: Linus Torvalds <[e-mail]>

mm/hugetlb.c

···
 	 * from page cache lookup which is in HPAGE_SIZE units.
 	 */
 	address = address & huge_page_mask(h);
-	pgoff = ((address - vma->vm_start) >> PAGE_SHIFT)
-		+ (vma->vm_pgoff >> PAGE_SHIFT);
+	pgoff = vma_hugecache_offset(h, vma, address);
 	mapping = (struct address_space *)page_private(page);
 
 	/*