mm: add_to_swap_cache() does not return -EEXIST
authorDaisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Tue, 22 Sep 2009 00:02:52 +0000 (17:02 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Tue, 22 Sep 2009 14:17:35 +0000 (07:17 -0700)
After commit 355cfa73 ("mm: modify swap_map and add SWAP_HAS_CACHE flag"),
only the context which have set SWAP_HAS_CACHE flag by swapcache_prepare()
or get_swap_page() would call add_to_swap_cache().  So add_to_swap_cache()
doesn't return -EEXIST any more.

Even though it doesn't return -EEXIST, it's not good behavior conceptually
to call swapcache_prepare() in the -EEXIST case, because it means clearing
SWAP_HAS_CACHE flag while the entry is on swap cache.

This patch removes redundant codes and comments from callers of it, and
adds VM_BUG_ON() in error path of add_to_swap_cache() and some comments.

Signed-off-by: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Balbir Singh <balbir@linux.vnet.ibm.com>
Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/shmem.c
mm/swap_state.c

index bd20f8b..376d8f0 100644 (file)
@@ -1097,6 +1097,10 @@ static int shmem_writepage(struct page *page, struct writeback_control *wbc)
        shmem_swp_unmap(entry);
 unlock:
        spin_unlock(&info->lock);
+       /*
+        * add_to_swap_cache() doesn't return -EEXIST, so we can safely
+        * clear SWAP_HAS_CACHE flag.
+        */
        swapcache_free(swap, NULL);
 redirty:
        set_page_dirty(page);
index b076a1a..6d1daeb 100644 (file)
@@ -92,6 +92,12 @@ static int __add_to_swap_cache(struct page *page, swp_entry_t entry)
        spin_unlock_irq(&swapper_space.tree_lock);
 
        if (unlikely(error)) {
+               /*
+                * Only the context which have set SWAP_HAS_CACHE flag
+                * would call add_to_swap_cache().
+                * So add_to_swap_cache() doesn't returns -EEXIST.
+                */
+               VM_BUG_ON(error == -EEXIST);
                set_page_private(page, 0UL);
                ClearPageSwapCache(page);
                page_cache_release(page);
@@ -146,38 +152,34 @@ int add_to_swap(struct page *page)
        VM_BUG_ON(!PageLocked(page));
        VM_BUG_ON(!PageUptodate(page));
 
-       for (;;) {
-               entry = get_swap_page();
-               if (!entry.val)
-                       return 0;
+       entry = get_swap_page();
+       if (!entry.val)
+               return 0;
 
+       /*
+        * Radix-tree node allocations from PF_MEMALLOC contexts could
+        * completely exhaust the page allocator. __GFP_NOMEMALLOC
+        * stops emergency reserves from being allocated.
+        *
+        * TODO: this could cause a theoretical memory reclaim
+        * deadlock in the swap out path.
+        */
+       /*
+        * Add it to the swap cache and mark it dirty
+        */
+       err = add_to_swap_cache(page, entry,
+                       __GFP_HIGH|__GFP_NOMEMALLOC|__GFP_NOWARN);
+
+       if (!err) {     /* Success */
+               SetPageDirty(page);
+               return 1;
+       } else {        /* -ENOMEM radix-tree allocation failure */
                /*
-                * Radix-tree node allocations from PF_MEMALLOC contexts could
-                * completely exhaust the page allocator. __GFP_NOMEMALLOC
-                * stops emergency reserves from being allocated.
-                *
-                * TODO: this could cause a theoretical memory reclaim
-                * deadlock in the swap out path.
-                */
-               /*
-                * Add it to the swap cache and mark it dirty
+                * add_to_swap_cache() doesn't return -EEXIST, so we can safely
+                * clear SWAP_HAS_CACHE flag.
                 */
-               err = add_to_swap_cache(page, entry,
-                               __GFP_HIGH|__GFP_NOMEMALLOC|__GFP_NOWARN);
-
-               switch (err) {
-               case 0:                         /* Success */
-                       SetPageDirty(page);
-                       return 1;
-               case -EEXIST:
-                       /* Raced with "speculative" read_swap_cache_async */
-                       swapcache_free(entry, NULL);
-                       continue;
-               default:
-                       /* -ENOMEM radix-tree allocation failure */
-                       swapcache_free(entry, NULL);
-                       return 0;
-               }
+               swapcache_free(entry, NULL);
+               return 0;
        }
 }
 
@@ -318,14 +320,7 @@ struct page *read_swap_cache_async(swp_entry_t entry, gfp_t gfp_mask,
                        break;
                }
 
-               /*
-                * Associate the page with swap entry in the swap cache.
-                * May fail (-EEXIST) if there is already a page associated
-                * with this entry in the swap cache: added by a racing
-                * read_swap_cache_async, or add_to_swap or shmem_writepage
-                * re-using the just freed swap entry for an existing page.
-                * May fail (-ENOMEM) if radix-tree node allocation failed.
-                */
+               /* May fail (-ENOMEM) if radix-tree node allocation failed. */
                __set_page_locked(new_page);
                SetPageSwapBacked(new_page);
                err = __add_to_swap_cache(new_page, entry);
@@ -341,6 +336,10 @@ struct page *read_swap_cache_async(swp_entry_t entry, gfp_t gfp_mask,
                radix_tree_preload_end();
                ClearPageSwapBacked(new_page);
                __clear_page_locked(new_page);
+               /*
+                * add_to_swap_cache() doesn't return -EEXIST, so we can safely
+                * clear SWAP_HAS_CACHE flag.
+                */
                swapcache_free(entry, NULL);
        } while (err != -ENOMEM);