From: Tao Ma <tm@tao.ma>
To: Jiaying Zhang <jiayingz@google.com>
Cc: Michael Tokarev <mjt@tls.msk.ru>, Ted Ts'o <tytso@mit.edu>,
Jan Kara <jack@suse.cz>,
linux-ext4@vger.kernel.org, sandeen@redhat.com
Subject: Re: DIO process stuck apparently due to dioread_nolock (3.0)
Date: Fri, 19 Aug 2011 11:20:25 +0800 [thread overview]
Message-ID: <4E4DD679.8020603@tao.ma> (raw)
In-Reply-To: <CAFgt=MC0NJC9fgGaS7Oe7LTpjDgcN-53Wx7TV-1Q=X3zFv73Ew@mail.gmail.com>
Hi Ted and Jiaying,
On 08/19/2011 02:54 AM, Jiaying Zhang wrote:
> On Wed, Aug 17, 2011 at 11:49 PM, Michael Tokarev <mjt@tls.msk.ru> wrote:
>> 17.08.2011 21:02, Ted Ts'o wrote:
>> []
>>> What I'd like to do long-term here is to change things so that (a)
>>> instead of instantiating the extent as uninitialized, writing the
>>> data, and then doing the uninit->init conversion to writing the data
>>> and then instantiated the extent as initialzied. This would also
>>> allow us to get rid of data=ordered mode. And we should make it work
>>> for fs block size != page size.
>>>
>>> It means that we need a way of adding this sort of information into an
>>> in-memory extent cache but which isn't saved to disk until the data is
>>> written. We've also talked about adding the information about whether
>>> an extent is subject to delalloc as well, so we don't have to grovel
>>> through the page cache and look at individual buffers attached to the
>>> pages. And there are folks who have been experimenting with an
>>> in-memory extent tree cache to speed access to fast PCIe-attached
>>> flash.
>>>
>>> It seems to me that if we're careful a single solution should be able
>>> to solve all of these problems...
>>
>> What about current situation, how do you think - should it be ignored
>> for now, having in mind that dioread_nolock isn't used often (but it
>> gives _serious_ difference in read speed), or, short term, fix this
>> very case which have real-life impact already, while implementing a
>> long-term solution?
> I plan to send my patch as a bandaid fix. It doesn't solve the fundamental
> problem but I think it helps close the race you saw on your test. In the long
> term, I agree that we should think about implementing an extent tree cache
> and use it to hold pending uninitialized-to-initialized extent conversions.
Does Google has some plan of doing it recently? We used a large number
of direct read, and we can arrange some resources to try to work it out.
Thanks
Tao
next prev parent reply other threads:[~2011-08-19 3:20 UTC|newest]
Thread overview: 43+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-08-10 10:51 DIO process stuck apparently due to dioread_nolock (3.0) Michael Tokarev
2011-08-11 11:59 ` Jan Kara
2011-08-11 12:21 ` Michael Tokarev
2011-08-11 14:01 ` Jan Kara
2011-08-11 20:05 ` Michael Tokarev
2011-08-12 2:46 ` Jiaying Zhang
2011-08-12 6:23 ` Michael Tokarev
2011-08-12 7:07 ` Michael Tokarev
2011-08-12 13:07 ` Jan Kara
2011-08-12 15:55 ` Michael Tokarev
2011-08-12 17:01 ` Eric Sandeen
2011-08-12 17:34 ` Michael Tokarev
2011-08-13 16:02 ` Tao Ma
2011-08-14 20:57 ` Michael Tokarev
2011-08-14 21:07 ` Michael Tokarev
2011-08-15 2:36 ` Tao Ma
2011-08-15 8:00 ` Michael Tokarev
2011-08-15 8:56 ` Michael Tokarev
2011-08-15 9:03 ` Michael Tokarev
2011-08-15 10:28 ` Tao Ma
2011-08-15 23:53 ` Jiaying Zhang
2011-08-16 4:15 ` Tao Ma
2011-08-16 8:38 ` Michael Tokarev
2011-08-16 13:53 ` Jan Kara
2011-08-16 15:03 ` Tao Ma
2011-08-16 21:32 ` Jiaying Zhang
2011-08-16 22:28 ` Michael Tokarev
2011-08-16 23:07 ` Jiaying Zhang
2011-08-17 17:02 ` Ted Ts'o
2011-08-18 6:49 ` Michael Tokarev
2011-08-18 18:54 ` Jiaying Zhang
2011-08-19 3:20 ` Tao Ma [this message]
2011-08-19 3:18 ` Tao Ma
2011-08-19 7:05 ` Michael Tokarev
2011-08-19 17:55 ` Jiaying Zhang
2011-08-16 23:59 ` Dave Chinner
2011-08-17 0:08 ` Jiaying Zhang
2011-08-17 2:22 ` Tao Ma
2011-08-17 9:04 ` Jan Kara
2011-08-15 16:08 ` Eric Sandeen
2011-08-16 4:12 ` Tao Ma
2011-08-16 6:15 ` Tao Ma
2011-08-12 21:19 ` Jan Kara
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=4E4DD679.8020603@tao.ma \
--to=tm@tao.ma \
--cc=jack@suse.cz \
--cc=jiayingz@google.com \
--cc=linux-ext4@vger.kernel.org \
--cc=mjt@tls.msk.ru \
--cc=sandeen@redhat.com \
--cc=tytso@mit.edu \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).