[PATCH] Fix a bad bug in read_cache_page_async() [try #2]

[David Howells <dhowells@redhat.com>]

Fix a bad bug in read_cache_page_async() introduced in commit:

	6fe6900e1e5b6fa9e5c59aa5061f244fe3f467e2

This adds:

	mark_page_accessed(page)

into the error handling path in read_cache_page_async().  In such a case,
'page' holds the error code.

Signed-off-by: David Howells <dhowells@redhat.com>
---

 mm/filemap.c |    6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/mm/filemap.c b/mm/filemap.c
index 9e56fd1..7b48b2a 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -1784,7 +1784,7 @@ struct page *read_cache_page_async(struct address_space *mapping,
 retry:
 	page = __read_cache_page(mapping, index, filler, data);
 	if (IS_ERR(page))
-		goto out;
+		return page;
 	mark_page_accessed(page);
 	if (PageUptodate(page))
 		goto out;
@@ -1802,9 +1802,9 @@ retry:
 	err = filler(data, page);
 	if (err < 0) {
 		page_cache_release(page);
-		page = ERR_PTR(err);
+		return ERR_PTR(err);
 	}
- out:
+out:
 	mark_page_accessed(page);
 	return page;
 }