From: Hugh Dickins <hugh@veritas.com>
Date: Thu, 7 Feb 2008 08:13:48 +0000 (-0800)
Subject: memcgroup: temporarily revert swapoff mod
X-Git-Tag: v2.6.25-rc1~479
X-Git-Url: https://err.no/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=59bd26582de660d4c9c26125747f1b4a5eb40d1e;p=linux-2.6

memcgroup: temporarily revert swapoff mod

This patch precisely reverts the "swapoff: scan ptes preemptibly" patch
just presented.  It's a temporary measure to allow existing memory
controller patches to apply without rejects: in due course they should be
rendered down into one sensible patch, and this reversion disappear.

Signed-off-by: Hugh Dickins <hugh@veritas.com>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
---

diff --git a/mm/swapfile.c b/mm/swapfile.c
index eade24da93..afae7b1f68 100644
--- a/mm/swapfile.c
+++ b/mm/swapfile.c
@@ -506,19 +506,9 @@ unsigned int count_swap_pages(int type, int free)
  * just let do_wp_page work it out if a write is requested later - to
  * force COW, vm_page_prot omits write permission from any private vma.
  */
-static int unuse_pte(struct vm_area_struct *vma, pmd_t *pmd,
+static void unuse_pte(struct vm_area_struct *vma, pte_t *pte,
 		unsigned long addr, swp_entry_t entry, struct page *page)
 {
-	spinlock_t *ptl;
-	pte_t *pte;
-	int found = 1;
-
-	pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
-	if (unlikely(!pte_same(*pte, swp_entry_to_pte(entry)))) {
-		found = 0;
-		goto out;
-	}
-
 	inc_mm_counter(vma->vm_mm, anon_rss);
 	get_page(page);
 	set_pte_at(vma->vm_mm, addr, pte,
@@ -530,9 +520,6 @@ static int unuse_pte(struct vm_area_struct *vma, pmd_t *pmd,
 	 * immediately swapped out again after swapon.
 	 */
 	activate_page(page);
-out:
-	pte_unmap_unlock(pte, ptl);
-	return found;
 }
 
 static int unuse_pte_range(struct vm_area_struct *vma, pmd_t *pmd,
@@ -541,33 +528,22 @@ static int unuse_pte_range(struct vm_area_struct *vma, pmd_t *pmd,
 {
 	pte_t swp_pte = swp_entry_to_pte(entry);
 	pte_t *pte;
+	spinlock_t *ptl;
 	int found = 0;
 
-	/*
-	 * We don't actually need pte lock while scanning for swp_pte: since
-	 * we hold page lock and mmap_sem, swp_pte cannot be inserted into the
-	 * page table while we're scanning; though it could get zapped, and on
-	 * some architectures (e.g. x86_32 with PAE) we might catch a glimpse
-	 * of unmatched parts which look like swp_pte, so unuse_pte must
-	 * recheck under pte lock.  Scanning without pte lock lets it be
-	 * preemptible whenever CONFIG_PREEMPT but not CONFIG_HIGHPTE.
-	 */
-	pte = pte_offset_map(pmd, addr);
+	pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
 	do {
 		/*
 		 * swapoff spends a _lot_ of time in this loop!
 		 * Test inline before going to call unuse_pte.
 		 */
 		if (unlikely(pte_same(*pte, swp_pte))) {
-			pte_unmap(pte);
-			found = unuse_pte(vma, pmd, addr, entry, page);
-			if (found)
-				goto out;
-			pte = pte_offset_map(pmd, addr);
+			unuse_pte(vma, pte++, addr, entry, page);
+			found = 1;
+			break;
 		}
 	} while (pte++, addr += PAGE_SIZE, addr != end);
-	pte_unmap(pte - 1);
-out:
+	pte_unmap_unlock(pte - 1, ptl);
 	return found;
 }