[PATCH 4/9] tmpfs: allow filepage alongside swappage

[Hugh Dickins <hugh@veritas.com>]

tmpfs has long allowed for a fresh filepage to be created in pagecache,
just before shmem_getpage gets the chance to match it up with the swappage
which already belongs to that offset.  But unionfs_writepage now does a
find_or_create_page, divorced from shmem_getpage, which leaves conflicting
filepage and swappage outstanding indefinitely, when unionfs is over tmpfs.

Therefore shmem_writepage (where a page is swizzled from file to swap) must
now be on the lookout for existing swap, ready to free it in favour of the
more uptodate filepage, instead of BUGging on that clash.  And when the
add_to_page_cache fails in shmem_unuse_inode, it must defer to an uptodate
filepage, otherwise swapoff would hang.  Whereas when add_to_page_cache
fails in shmem_getpage, it should retry in the same way it already does.

Signed-off-by: Hugh Dickins <hugh@veritas.com>
---

 mm/shmem.c |   69 ++++++++++++++++++++++++++++++++-------------------
 1 file changed, 44 insertions(+), 25 deletions(-)

--- tmpfs3/mm/shmem.c	2007-12-05 16:42:18.000000000 +0000
+++ tmpfs4/mm/shmem.c	2007-12-05 16:42:19.000000000 +0000
@@ -827,6 +827,7 @@ static int shmem_unuse_inode(struct shme
 	struct page *subdir;
 	swp_entry_t *ptr;
 	int offset;
+	int error;
 
 	idx = 0;
 	ptr = info->i_direct;
@@ -884,7 +885,20 @@ lost2:
 found:
 	idx += offset;
 	inode = &info->vfs_inode;
-	if (add_to_page_cache(page, inode->i_mapping, idx, GFP_ATOMIC) == 0) {
+	error = add_to_page_cache(page, inode->i_mapping, idx, GFP_ATOMIC);
+	if (error == -EEXIST) {
+		struct page *filepage = find_get_page(inode->i_mapping, idx);
+		if (filepage) {
+			/*
+			 * There might be a more uptodate page coming down
+			 * from a stacked writepage: forget our swappage if so.
+			 */
+			if (PageUptodate(filepage))
+				error = 0;
+			page_cache_release(filepage);
+		}
+	}
+	if (!error) {
 		delete_from_swap_cache(page);
 		set_page_dirty(page);
 		info->flags |= SHMEM_PAGEIN;
@@ -937,44 +951,45 @@ static int shmem_writepage(struct page *
 	struct inode *inode;
 
 	BUG_ON(!PageLocked(page));
-	/*
-	 * shmem_backing_dev_info's capabilities prevent regular writeback or
-	 * sync from ever calling shmem_writepage; but a stacking filesystem
-	 * may use the ->writepage of its underlying filesystem, in which case
-	 * we want to do nothing when that underlying filesystem is tmpfs
-	 * (writing out to swap is useful as a response to memory pressure, but
-	 * of no use to stabilize the data) - just redirty the page, unlock it
-	 * and claim success in this case.  AOP_WRITEPAGE_ACTIVATE, and the
-	 * page_mapped check below, must be avoided unless we're in reclaim.
-	 */
-	if (!wbc->for_reclaim) {
-		set_page_dirty(page);
-		unlock_page(page);
-		return 0;
-	}
-	BUG_ON(page_mapped(page));
-
 	mapping = page->mapping;
 	index = page->index;
 	inode = mapping->host;
 	info = SHMEM_I(inode);
 	if (info->flags & VM_LOCKED)
 		goto redirty;
-	swap = get_swap_page();
-	if (!swap.val)
+	if (!total_swap_pages)
 		goto redirty;
 
+	/*
+	 * shmem_backing_dev_info's capabilities prevent regular writeback or
+	 * sync from ever calling shmem_writepage; but a stacking filesystem
+	 * may use the ->writepage of its underlying filesystem, in which case
+	 * tmpfs should write out to swap only in response to memory pressure,
+	 * and not for pdflush or sync.  However, in those cases, we do still
+	 * want to check if there's a redundant swappage to be discarded.
+	 */
+	if (wbc->for_reclaim)
+		swap = get_swap_page();
+	else
+		swap.val = 0;
+
 	spin_lock(&info->lock);
-	shmem_recalc_inode(inode);
 	if (index >= info->next_index) {
 		BUG_ON(!(info->flags & SHMEM_TRUNCATE));
 		goto unlock;
 	}
 	entry = shmem_swp_entry(info, index, NULL);
-	BUG_ON(!entry);
-	BUG_ON(entry->val);
+	if (entry->val) {
+		/*
+		 * The more uptodate page coming down from a stacked
+		 * writepage should replace our old swappage.
+		 */
+		free_swap_and_cache(*entry);
+		shmem_swp_set(info, entry, 0);
+	}
+	shmem_recalc_inode(inode);
 
-	if (add_to_swap_cache(page, swap, GFP_ATOMIC) == 0) {
+	if (swap.val && add_to_swap_cache(page, swap, GFP_ATOMIC) == 0) {
 		remove_from_page_cache(page);
 		shmem_swp_set(info, entry, swap.val);
 		shmem_swp_unmap(entry);
@@ -986,6 +1001,7 @@ static int shmem_writepage(struct page *
 			spin_unlock(&shmem_swaplist_lock);
 		}
 		swap_duplicate(swap);
+		BUG_ON(page_mapped(page));
 		page_cache_release(page);	/* pagecache ref */
 		set_page_dirty(page);
 		unlock_page(page);
@@ -998,7 +1014,10 @@ unlock:
 	swap_free(swap);
 redirty:
 	set_page_dirty(page);
-	return AOP_WRITEPAGE_ACTIVATE;	/* Return with the page locked */
+	if (wbc->for_reclaim)
+		return AOP_WRITEPAGE_ACTIVATE;	/* Return with page locked */
+	unlock_page(page);
+	return 0;
 }
 
 #ifdef CONFIG_NUMA
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/