From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jeff Moyer Subject: [patch] jbd: tag journal writes as metadata I/O Date: Tue, 06 Apr 2010 12:52:37 -0400 Message-ID: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-kernel@vger.kernel.org, jens.axboe@oracle.com, Jan Kara To: linux-ext4@vger.kernel.org, "Theodore Ts'o" Return-path: Received: from mx1.redhat.com ([209.132.183.28]:4826 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1757244Ab0DFQwt (ORCPT ); Tue, 6 Apr 2010 12:52:49 -0400 Sender: linux-ext4-owner@vger.kernel.org List-ID: Hi, In running iozone for writes to small files, we noticed a pretty big discrepency between the performance of the deadline and cfq I/O schedulers. Investigation showed that I/O was being issued from 2 different contexts: the iozone process itself, and the jbd/sdh-8 thread (as expected). Because of the way cfq performs slice idling, the delays introduced between the metadata and data I/Os were significant. For example, cfq would see about 7MB/s versus deadline's 35 for the same workload. I also tested fs_mark with writing and fsyncing 1000 64k files, and a similar 5x performance difference was observed. Eric Sandeen suggested that I flag the journal writes as metadata, and once I did that, the performance difference went away completely (cfq has special logic to prioritize metadata I/O). This is similar to the jbd2 patch I posted earlier that Ted has accepted into his patch queue. Comments, as always, are appreciated. Signed-off-by: Jeff Moyer diff --git a/fs/jbd/commit.c b/fs/jbd/commit.c index 4bd8825..e9f1369 100644 --- a/fs/jbd/commit.c +++ b/fs/jbd/commit.c @@ -318,7 +318,7 @@ void journal_commit_transaction(journal_t *journal) int first_tag = 0; int tag_flag; int i; - int write_op = WRITE; + int write_op = WRITE_META; /* * First job: lock down the current transaction and wait for @@ -357,7 +357,7 @@ void journal_commit_transaction(journal_t *journal) * instead we rely on sync_buffer() doing the unplug for us. */ if (commit_transaction->t_synchronous_commit) - write_op = WRITE_SYNC_PLUG; + write_op = WRITE_SYNC_PLUG | (1<t_handle_lock); while (commit_transaction->t_updates) { DEFINE_WAIT(wait);