Received-SPF: pass (google.com: best guess record for domain of linux-ext4-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Thu, 22 Aug 2019 15:40:01 +1000
From:   Dave Chinner <david@fromorbit.com>
To:     Joseph Qi <joseph.qi@linux.alibaba.com>
Cc:     "Theodore Y. Ts'o" <tytso@mit.edu>, Jan Kara <jack@suse.cz>,
        Joseph Qi <jiangqi903@gmail.com>,
        Andreas Dilger <adilger@dilger.ca>,
        Ext4 Developers List <linux-ext4@vger.kernel.org>,
        Xiaoguang Wang <xiaoguang.wang@linux.alibaba.com>,
        Liu Bo <bo.liu@linux.alibaba.com>
Subject: Re: [RFC] performance regression with "ext4: Allow parallel DIO
 reads"
Message-ID: <20190822054001.GT7777@dread.disaster.area>
References: <6DADA28C-542F-45F6-ADB0-870A81ABED23@dilger.ca>
 <15112e38-94fe-39d6-a8e2-064ff47187d5@linux.alibaba.com>
 <20190728225122.GG7777@dread.disaster.area>
 <960bb915-20cc-26a0-7abc-bfca01aa39c0@gmail.com>
 <20190815151336.GO14313@quack2.suse.cz>
 <075fd06f-b0b4-4122-81c6-e49200d5bd17@linux.alibaba.com>
 <20190816145719.GA3041@quack2.suse.cz>
 <a8ddec64-d87c-ae7a-9b02-2f24da2396e9@linux.alibaba.com>
 <20190820160805.GB10232@mit.edu>
 <f89131c9-6f84-ac3c-b53c-d3d55887ea89@linux.alibaba.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <f89131c9-6f84-ac3c-b53c-d3d55887ea89@linux.alibaba.com>
User-Agent: Mutt/1.10.1 (2018-07-13)
Sender: linux-ext4-owner@vger.kernel.org
Precedence: bulk

On Wed, Aug 21, 2019 at 09:04:57AM +0800, Joseph Qi wrote:
> Hi Ted，
> 
> On 19/8/21 00:08, Theodore Y. Ts'o wrote:
> > On Tue, Aug 20, 2019 at 11:00:39AM +0800, Joseph Qi wrote:
> >>
> >> I've tested parallel dio reads with dioread_nolock, it doesn't have
> >> significant performance improvement and still poor compared with reverting
> >> parallel dio reads. IMO, this is because with parallel dio reads, it take
> >> inode shared lock at the very beginning in ext4_direct_IO_read().
> > 
> > Why is that a problem?  It's a shared lock, so parallel threads should
> > be able to issue reads without getting serialized?
> > 
> The above just tells the result that even mounting with dioread_nolock,
> parallel dio reads still has poor performance than before (w/o parallel
> dio reads).
> 
> > Are you using sufficiently fast storage devices that you're worried
> > about cache line bouncing of the shared lock?  Or do you have some
> > other concern, such as some other thread taking an exclusive lock?
> > 
> The test case is random read/write described in my first mail. And

Regardless of dioread_nolock, ext4_direct_IO_read() is taking
inode_lock_shared() across the direct IO call.  And writes in ext4
_always_ take the inode_lock() in ext4_file_write_iter(), even
though it gets dropped quite early when overwrite && dioread_nolock
is set.  But just taking the lock exclusively in write fro a short
while is enough to kill all shared locking concurrency...

> from my preliminary investigation, shared lock consumes more in such
> scenario.

If the write lock is also shared, then there should not be a
scalability issue. The shared dio locking is only half-done in ext4,
so perhaps comparing your workload against XFS would be an
informative exercise...

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com