NFS: writepage of a single page should not be synchronous
authorTrond Myklebust <trond.myklebust@primarydata.com>
Wed, 1 Jun 2016 22:25:56 +0000 (18:25 -0400)
committerTrond Myklebust <trond.myklebust@primarydata.com>
Wed, 22 Jun 2016 13:59:43 +0000 (09:59 -0400)
It is almost always better to wait for more so that we can issue a
bulk commit.

Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
fs/nfs/write.c

index 980d44f3a84cdd50dd5c50ba93ad60ded7879999..b13d48881d3ae3c5d09b4fc88bda3b49740bb3ab 100644 (file)
@@ -625,7 +625,7 @@ static int nfs_writepage_locked(struct page *page,
        int err;
 
        nfs_inc_stats(inode, NFSIOS_VFSWRITEPAGE);
-       nfs_pageio_init_write(&pgio, inode, wb_priority(wbc),
+       nfs_pageio_init_write(&pgio, inode, 0,
                                false, &nfs_async_write_completion_ops);
        err = nfs_do_writepage(page, wbc, &pgio, launder);
        nfs_pageio_complete(&pgio);