teach /proc/$pid/numa_maps about transparent hugepages

This is modeled after the smaps code.

It detects transparent hugepages and then does a single gather_stats()
for the page as a whole.  This has two benifits:
 1. It is more efficient since it does many pages in a single shot.
 2. It does not have to break down the huge page.

Signed-off-by: Dave Hansen <dave@linux.vnet.ibm.com>
Acked-by: Hugh Dickins <hughd@google.com>
Acked-by: David Rientjes <rientjes@google.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

fs/proc/task_mmu.c

@@ -936,6 +936,26 @@ static int gather_pte_stats(pmd_t *pmd, unsigned long addr,
 	pte_t *pte;
 
 	md = walk->private;
+	spin_lock(&walk->mm->page_table_lock);
+	if (pmd_trans_huge(*pmd)) {
+		if (pmd_trans_splitting(*pmd)) {
+			spin_unlock(&walk->mm->page_table_lock);
+			wait_split_huge_page(md->vma->anon_vma, pmd);
+		} else {
+			pte_t huge_pte = *(pte_t *)pmd;
+			struct page *page;
+
+			page = can_gather_numa_stats(huge_pte, md->vma, addr);
+			if (page)
+				gather_stats(page, md, pte_dirty(huge_pte),
+						HPAGE_PMD_SIZE/PAGE_SIZE);
+			spin_unlock(&walk->mm->page_table_lock);
+			return 0;
+		}
+	} else {
+		spin_unlock(&walk->mm->page_table_lock);
+	}
+
 	orig_pte = pte = pte_offset_map_lock(walk->mm, pmd, addr, &ptl);
 	do {
 		struct page *page = can_gather_numa_stats(*pte, md->vma, addr);