[PATCH] rmaplock: swapoff use anon_vma

Swapoff can make good use of a page's anon_vma and index, while it's still
left in swapcache, or once it's brought back in and the first pte mapped back:
unuse_vma go directly to just one page of only those vmas with the same
anon_vma.  And unuse_process can skip any vmas without an anon_vma (extending
the hugetlb check: hugetlb vmas have no anon_vma).

This just hacks in on top of the existing procedure, still going through all
the vmas of all the mms in mmlist.  A more elegant procedure might replace
mmlist by a list of anon_vmas: but that would be more work to implement, with
apparently more overhead in the common paths.
Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

include/linux/rmap.h

@@ -91,6 +91,11 @@ static inline void page_dup_rmap(struct page *page)
 int page_referenced(struct page *, int is_locked);
 int try_to_unmap(struct page *);
 
+/*
+ * Used by swapoff to help locate where page is expected in vma.
+ */
+unsigned long page_address_in_vma(struct page *, struct vm_area_struct *);
+
 #else	/* !CONFIG_MMU */
 
 #define anon_vma_init()		do {} while (0)

mm/rmap.c

@@ -230,6 +230,24 @@ vma_address(struct page *page, struct vm_area_struct *vma)
 	return address;
 }
 
+/*
+ * At what user virtual address is page expected in vma? checking that the
+ * page matches the vma: currently only used by unuse_process, on anon pages.
+ */
+unsigned long page_address_in_vma(struct page *page, struct vm_area_struct *vma)
+{
+	if (PageAnon(page)) {
+		if ((void *)vma->anon_vma !=
+		    (void *)page->mapping - PAGE_MAPPING_ANON)
+			return -EFAULT;
+	} else if (page->mapping && !(vma->vm_flags & VM_NONLINEAR)) {
+		if (vma->vm_file->f_mapping != page->mapping)
+			return -EFAULT;
+	} else
+		return -EFAULT;
+	return vma_address(page, vma);
+}
+
 /*
  * Subfunctions of page_referenced: page_referenced_one called
  * repeatedly from either page_referenced_anon or page_referenced_file.
@@ -459,6 +477,8 @@ void page_remove_rmap(struct page *page)
 		 * which increments mapcount after us but sets mapping
 		 * before us: so leave the reset to free_hot_cold_page,
 		 * and remember that it's only reliable while mapped.
+		 * Leaving it set also helps swapoff to reinstate ptes
+		 * faster for those pages still in swapcache.
 		 */
 		if (page_test_and_clear_dirty(page))
 			set_page_dirty(page);

mm/swapfile.c

@@ -520,14 +520,24 @@ static unsigned long unuse_pgd(struct vm_area_struct * vma, pgd_t *dir,
 }
 
 /* vma->vm_mm->page_table_lock is held */
-static unsigned long unuse_vma(struct vm_area_struct * vma, pgd_t *pgdir,
+static unsigned long unuse_vma(struct vm_area_struct * vma,
 	swp_entry_t entry, struct page *page)
 {
-	unsigned long start = vma->vm_start, end = vma->vm_end;
+	pgd_t *pgdir;
+	unsigned long start, end;
 	unsigned long foundaddr;
 
-	if (start >= end)
-		BUG();
+	if (page->mapping) {
+		start = page_address_in_vma(page, vma);
+		if (start == -EFAULT)
+			return 0;
+		else
+			end = start + PAGE_SIZE;
+	} else {
+		start = vma->vm_start;
+		end = vma->vm_end;
+	}
+	pgdir = pgd_offset(vma->vm_mm, start);
 	do {
 		foundaddr = unuse_pgd(vma, pgdir, start, end - start,
 						entry, page);
@@ -559,9 +569,8 @@ static int unuse_process(struct mm_struct * mm,
 	}
 	spin_lock(&mm->page_table_lock);
 	for (vma = mm->mmap; vma; vma = vma->vm_next) {
-		if (!is_vm_hugetlb_page(vma)) {
-			pgd_t * pgd = pgd_offset(mm, vma->vm_start);
-			foundaddr = unuse_vma(vma, pgd, entry, page);
+		if (vma->anon_vma) {
+			foundaddr = unuse_vma(vma, entry, page);
 			if (foundaddr)
 				break;
 		}