Return-Path: Received: from dkim1.fusionio.com ([66.114.96.53]:49770 "EHLO dkim1.fusionio.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1760177Ab3DDNMH (ORCPT ); Thu, 4 Apr 2013 09:12:07 -0400 Received: from mx2.fusionio.com (unknown [10.101.1.160]) by dkim1.fusionio.com (Postfix) with ESMTP id CBACF7C0427 for ; Thu, 4 Apr 2013 07:12:06 -0600 (MDT) Date: Thu, 4 Apr 2013 09:12:04 -0400 From: Josef Bacik To: Miao Xie CC: Linux Btrfs Subject: Re: [PATCH V2 1/2] Btrfs: improve the performance of the csums lookup Message-ID: <20130404131204.GQ1876@localhost.localdomain> References: <5153FB01.9030506@cn.fujitsu.com> <515BF138.90309@cn.fujitsu.com> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" In-Reply-To: <515BF138.90309@cn.fujitsu.com> Sender: linux-btrfs-owner@vger.kernel.org List-ID: On Wed, Apr 03, 2013 at 03:07:04AM -0600, Miao Xie wrote: > It is very likely that there are several blocks in bio, it is very > inefficient if we get their csums one by one. This patch improves > this problem by getting the csums in batch. > > According to the result of the following test, the execute time of > __btrfs_lookup_bio_sums() is down by ~28%(300us -> 217us). > > # dd if=/file of=/dev/null bs=1M count=1024 > > Signed-off-by: Miao Xie This is panicing when I run my fsync tests and when kdave runs xfstest 91. I don't have time to track it down now so I'm just dropping it from btrfs-next. I'll look at it after I figure out this fsync bug unless you beat me to it. Thanks, Josef