linux-ext4 - Re: Discussion: is it time to remove dioread

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20200108174259.GD263696@mit.edu>
Date:   Wed, 8 Jan 2020 12:42:59 -0500
From:   "Theodore Y. Ts'o" <tytso@....edu>
To:     Ritesh Harjani <riteshh@...ux.ibm.com>
Cc:     Jan Kara <jack@...e.cz>,
        Xiaoguang Wang <xiaoguang.wang@...ux.alibaba.com>,
        Ext4 Developers List <linux-ext4@...r.kernel.org>,
        joseph.qi@...ux.alibaba.com, Liu Bo <bo.liu@...ux.alibaba.com>
Subject: Re: Discussion: is it time to remove dioread_nolock?

On Wed, Jan 08, 2020 at 04:15:13PM +0530, Ritesh Harjani wrote:
> Hello Ted/Jan,
> 
> On 1/7/20 10:52 PM, Jan Kara wrote:
> > On Tue 07-01-20 12:11:09, Theodore Y. Ts'o wrote:
> > > Hmm..... There's actually an even more radical option we could use,
> > > given that Ritesh has made dioread_nolock work on block sizes < page
> > > size.  We could make dioread_nolock the default, until we can revamp
> > > ext4_writepages() to write the data blocks first....
> 
> Agreed. I guess it should be a straight forward change to make.
> It should be just removing test_opt(inode->i_sb, DIOREAD_NOLOCK) condition
> from ext4_should_dioread_nolock().

Actually, it's simpler than that.  In fs/ext4/super.c, around line
3730, after the comment:

	/* Set defaults before we parse the mount options */

Just add:

     set_opt(sb, DIOREAD_NOLOCK);

This will allow system administrators to revert back to the original
method using the someone confusingly named mount option,
"dioread_lock".  (Maybe we can add a alias for that mount option so
it's less confusing).

> > Yes, that's a good point. And I'm not opposed to that if it makes the life
> > simpler. But I'd like to see some performance numbers showing how much is
> > writeback using unwritten extents slower so that we don't introduce too big
> > regression with this...
> > 
> 
> Yes, let me try to get some performance numbers with dioread_nolock as
> the default option for buffered write on my setup.

I started running some performance runs last night, and the
interesting thing that I found was that fs_mark actually *improved*
with dioread_nolock (with fsync enabled).  That may be an example of
where fixing the commit latency caused by writeback can actually show
up in a measurable way with benchmarks.

Dbench was slightly impacted; I didn't see any real differences with
compilebench or postmark.  dioread_nolock did improve fio with
sequential reads; which is interesting, since I would have expected
with the inode_lock improvements, there shouldn't have been any
difference.  So that may be a bit of wierdness that we should try to
understand.

See the attached tar file; open ext4-modes/index.html in a browser to
see the pretty graphs.  The raw numbers are in ext4/composite.xml.

Cheers,

						- Ted

Download attachment "pts.tar.xz" of type "application/x-xz" (27892 bytes)