NFS: Fix Oopsable condition in nfs_readpage_sync()

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

diff --git a/fs/nfs/read.c b/fs/nfs/read.c
index 7a9ee00..f0aff82 100644 (file)
--- a/fs/nfs/read.c
+++ b/fs/nfs/read.c
@@ -204,9 +204,11 @@ static int nfs_readpage_sync(struct nfs_open_context *ctx, struct inode *inode,
        NFS_I(inode)->cache_validity |= NFS_INO_INVALID_ATIME;
        spin_unlock(&inode->i_lock);
 
-       nfs_readpage_truncate_uninitialised_page(rdata);
-       if (rdata->res.eof || rdata->res.count == rdata->args.count)
+       if (rdata->res.eof || rdata->res.count == rdata->args.count) {
                SetPageUptodate(page);
+               if (rdata->res.eof && count != 0)
+                       memclear_highpage_flush(page, rdata->args.pgbase, count);
+       }
        result = 0;
 
 io_error: