From: "Rudoff, Andy" <andy.rudoff@intel.com>
To: Dave Chinner <david@fromorbit.com>
Cc: "Williams, Dan J" <dan.j.williams@intel.com>,
"hch@infradead.org" <hch@infradead.org>,
"jack@suse.cz" <jack@suse.cz>, "axboe@fb.com" <axboe@fb.com>,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
"xfs@oss.sgi.com" <xfs@oss.sgi.com>,
"linux-block@vger.kernel.org" <linux-block@vger.kernel.org>,
"linux-mm@kvack.org" <linux-mm@kvack.org>,
"viro@zeniv.linux.org.uk" <viro@zeniv.linux.org.uk>,
"linux-nvdimm@ml01.01.org" <linux-nvdimm@ml01.01.org>,
"linux-fsdevel@vger.kernel.org" <linux-fsdevel@vger.kernel.org>,
"akpm@linux-foundation.org" <akpm@linux-foundation.org>,
"linux-ext4@vger.kernel.org" <linux-ext4@vger.kernel.org>,
"Wilcox, Matthew R" <matthew.r.wilcox@intel.com>
Subject: Re: [PATCH v2 5/5] dax: handle media errors in dax_do_io
Date: Tue, 3 May 2016 18:30:04 +0000 [thread overview]
Message-ID: <FBB11841-7DFE-4223-9973-3457034260C2@intel.com> (raw)
In-Reply-To: <20160503024948.GT26977@dastard>
>
>And when the filesystem says no because the fs devs don't want to
>have to deal with broken apps because app devs learn that "this is a
>go fast knob" and data integrity be damned? It's "fsync is slow so I
>won't use it" all over again...
...
>
>And, please keep in mind: many application developers will not
>design for pmem because they also have to support traditional
>storage backed by page cache. If they use msync(), the app will work
>on any storage stack, but just be much, much faster on pmem+DAX. So,
>really, we have to make the msync()-only model work efficiently, so
>we may as well design for that in the first place....
Both of these snippets seem to be arguing that we should make msync/fsync
more efficient. But I don't think anyone is arguing the opposite. Is
someone saying we shouldn't make the msync()-only model work efficiently?
Said another way: the common case for DAX will be applications simply
following the POSIX model. open, mmap, msync... That will work fine
and of course we should optimize that path as much as possible. Less
common are latency-sensitive applications built to leverage to byte-
addressable nature of pmem. File systems supporting this model will
indicate it using a new ioctl that says doing CPU cache flushes is
sufficient to flush stores to persistence. But I don't see how that
direction is getting turned into an argument against msync() efficiency.
>Which brings up another point: advanced new functionality
>is going to require native pmem filesystems.
I agree there's opportunity for new filesystems (and old) to leverage
what pmem provides. But the word "require" implies that's the only
way to go and we know that's not the case. Using ext4+dax to map
pmem into an application allows that application to use the pmem
directly and a good number of software projects are doing exactly that.
-andy
WARNING: multiple messages have this Message-ID (diff)
From: "Rudoff, Andy" <andy.rudoff@intel.com>
To: Dave Chinner <david@fromorbit.com>
Cc: "Williams, Dan J" <dan.j.williams@intel.com>,
"hch@infradead.org" <hch@infradead.org>,
"jack@suse.cz" <jack@suse.cz>, "axboe@fb.com" <axboe@fb.com>,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
"xfs@oss.sgi.com" <xfs@oss.sgi.com>,
"linux-block@vger.kernel.org" <linux-block@vger.kernel.org>,
"linux-mm@kvack.org" <linux-mm@kvack.org>,
"viro@zeniv.linux.org.uk" <viro@zeniv.linux.org.uk>,
"linux-nvdimm@ml01.01.org" <linux-nvdimm@ml01.01.org>,
"linux-fsdevel@vger.kernel.org" <linux-fsdevel@vger.kernel.org>,
"akpm@linux-foundation.org" <akpm@linux-foundation.org>,
"linux-ext4@vger.kernel.org" <linux-ext4@vger.kernel.org>,
"Wilcox, Matthew R" <matthew.r.wilcox@intel.com>
Subject: Re: [PATCH v2 5/5] dax: handle media errors in dax_do_io
Date: Tue, 3 May 2016 18:30:04 +0000 [thread overview]
Message-ID: <FBB11841-7DFE-4223-9973-3457034260C2@intel.com> (raw)
In-Reply-To: <20160503024948.GT26977@dastard>
[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset="utf-8", Size: 2010 bytes --]
>
>And when the filesystem says no because the fs devs don't want to
>have to deal with broken apps because app devs learn that "this is a
>go fast knob" and data integrity be damned? It's "fsync is slow so I
>won't use it" all over again...
...
>
>And, please keep in mind: many application developers will not
>design for pmem because they also have to support traditional
>storage backed by page cache. If they use msync(), the app will work
>on any storage stack, but just be much, much faster on pmem+DAX. So,
>really, we have to make the msync()-only model work efficiently, so
>we may as well design for that in the first place....
Both of these snippets seem to be arguing that we should make msync/fsync
more efficient. But I don't think anyone is arguing the opposite. Is
someone saying we shouldn't make the msync()-only model work efficiently?
Said another way: the common case for DAX will be applications simply
following the POSIX model. open, mmap, msync... That will work fine
and of course we should optimize that path as much as possible. Less
common are latency-sensitive applications built to leverage to byte-
addressable nature of pmem. File systems supporting this model will
indicate it using a new ioctl that says doing CPU cache flushes is
sufficient to flush stores to persistence. But I don't see how that
direction is getting turned into an argument against msync() efficiency.
>Which brings up another point: advanced new functionality
>is going to require native pmem filesystems.
I agree there's opportunity for new filesystems (and old) to leverage
what pmem provides. But the word "require" implies that's the only
way to go and we know that's not the case. Using ext4+dax to map
pmem into an application allows that application to use the pmem
directly and a good number of software projects are doing exactly that.
-andy
N‹§²æìr¸›zǧu©ž²Æ {\b†éì¹»\x1c®&Þ–)îÆi¢žØ^n‡r¶‰šŽŠÝ¢j$½§$¢¸\x05¢¹¨è§~Š'.)îÄÃ,yèm¶ŸÿÃ\f%Š{±šj+ƒðèž×¦j)Z†·Ÿ
WARNING: multiple messages have this Message-ID (diff)
From: "Rudoff, Andy" <andy.rudoff@intel.com>
To: Dave Chinner <david@fromorbit.com>
Cc: "axboe@fb.com" <axboe@fb.com>, "jack@suse.cz" <jack@suse.cz>,
"linux-nvdimm@ml01.01.org" <linux-nvdimm@ml01.01.org>,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
"xfs@oss.sgi.com" <xfs@oss.sgi.com>,
"hch@infradead.org" <hch@infradead.org>,
"linux-mm@kvack.org" <linux-mm@kvack.org>,
"linux-block@vger.kernel.org" <linux-block@vger.kernel.org>,
"viro@zeniv.linux.org.uk" <viro@zeniv.linux.org.uk>,
"linux-fsdevel@vger.kernel.org" <linux-fsdevel@vger.kernel.org>,
"Williams, Dan J" <dan.j.williams@intel.com>,
"linux-ext4@vger.kernel.org" <linux-ext4@vger.kernel.org>,
"akpm@linux-foundation.org" <akpm@linux-foundation.org>,
"Wilcox, Matthew R" <matthew.r.wilcox@intel.com>
Subject: Re: [PATCH v2 5/5] dax: handle media errors in dax_do_io
Date: Tue, 3 May 2016 18:30:04 +0000 [thread overview]
Message-ID: <FBB11841-7DFE-4223-9973-3457034260C2@intel.com> (raw)
In-Reply-To: <20160503024948.GT26977@dastard>
>
>And when the filesystem says no because the fs devs don't want to
>have to deal with broken apps because app devs learn that "this is a
>go fast knob" and data integrity be damned? It's "fsync is slow so I
>won't use it" all over again...
...
>
>And, please keep in mind: many application developers will not
>design for pmem because they also have to support traditional
>storage backed by page cache. If they use msync(), the app will work
>on any storage stack, but just be much, much faster on pmem+DAX. So,
>really, we have to make the msync()-only model work efficiently, so
>we may as well design for that in the first place....
Both of these snippets seem to be arguing that we should make msync/fsync
more efficient. But I don't think anyone is arguing the opposite. Is
someone saying we shouldn't make the msync()-only model work efficiently?
Said another way: the common case for DAX will be applications simply
following the POSIX model. open, mmap, msync... That will work fine
and of course we should optimize that path as much as possible. Less
common are latency-sensitive applications built to leverage to byte-
addressable nature of pmem. File systems supporting this model will
indicate it using a new ioctl that says doing CPU cache flushes is
sufficient to flush stores to persistence. But I don't see how that
direction is getting turned into an argument against msync() efficiency.
>Which brings up another point: advanced new functionality
>is going to require native pmem filesystems.
I agree there's opportunity for new filesystems (and old) to leverage
what pmem provides. But the word "require" implies that's the only
way to go and we know that's not the case. Using ext4+dax to map
pmem into an application allows that application to use the pmem
directly and a good number of software projects are doing exactly that.
-andy
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
next prev parent reply other threads:[~2016-05-03 18:30 UTC|newest]
Thread overview: 216+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-03-30 1:59 [PATCH v2 0/5] dax: handling of media errors Vishal Verma
2016-03-30 1:59 ` Vishal Verma
2016-03-30 1:59 ` Vishal Verma
2016-03-30 1:59 ` Vishal Verma
2016-03-30 1:59 ` [PATCH v2 1/5] block, dax: pass blk_dax_ctl through to drivers Vishal Verma
2016-03-30 1:59 ` Vishal Verma
2016-03-30 1:59 ` Vishal Verma
2016-03-30 1:59 ` Vishal Verma
2016-03-30 4:19 ` kbuild test robot
2016-03-30 4:19 ` kbuild test robot
2016-03-30 4:19 ` kbuild test robot
2016-03-30 4:19 ` kbuild test robot
2016-04-15 14:55 ` Jeff Moyer
2016-04-15 14:55 ` Jeff Moyer
2016-04-15 14:55 ` Jeff Moyer
2016-03-30 1:59 ` [PATCH v2 2/5] dax: fallback from pmd to pte on error Vishal Verma
2016-03-30 1:59 ` Vishal Verma
2016-03-30 1:59 ` Vishal Verma
2016-03-30 1:59 ` Vishal Verma
2016-03-30 1:59 ` Vishal Verma
2016-04-15 14:55 ` Jeff Moyer
2016-04-15 14:55 ` Jeff Moyer
2016-04-15 14:55 ` Jeff Moyer
2016-03-30 1:59 ` [PATCH v2 3/5] dax: enable dax in the presence of known media errors (badblocks) Vishal Verma
2016-03-30 1:59 ` Vishal Verma
2016-03-30 1:59 ` Vishal Verma
2016-03-30 1:59 ` Vishal Verma
2016-04-15 14:56 ` Jeff Moyer
2016-04-15 14:56 ` Jeff Moyer
2016-04-15 14:56 ` Jeff Moyer
2016-03-30 1:59 ` [PATCH v2 4/5] dax: use sb_issue_zerout instead of calling dax_clear_sectors Vishal Verma
2016-03-30 1:59 ` Vishal Verma
2016-03-30 1:59 ` Vishal Verma
2016-03-30 1:59 ` Vishal Verma
2016-04-15 15:18 ` Jeff Moyer
2016-04-15 15:18 ` Jeff Moyer
2016-04-15 15:18 ` Jeff Moyer
2016-03-30 1:59 ` [PATCH v2 5/5] dax: handle media errors in dax_do_io Vishal Verma
2016-03-30 1:59 ` Vishal Verma
2016-03-30 1:59 ` Vishal Verma
2016-03-30 1:59 ` Vishal Verma
2016-03-30 3:00 ` kbuild test robot
2016-03-30 3:00 ` kbuild test robot
2016-03-30 3:00 ` kbuild test robot
2016-03-30 3:00 ` kbuild test robot
2016-03-30 3:00 ` kbuild test robot
2016-03-30 3:00 ` kbuild test robot
2016-03-30 6:34 ` Christoph Hellwig
2016-03-30 6:34 ` Christoph Hellwig
2016-03-30 6:34 ` Christoph Hellwig
2016-03-30 6:54 ` Vishal Verma
2016-03-30 6:54 ` Vishal Verma
2016-03-30 6:54 ` Vishal Verma
2016-03-30 6:56 ` Christoph Hellwig
2016-03-30 6:56 ` Christoph Hellwig
2016-03-30 6:56 ` Christoph Hellwig
2016-04-15 16:11 ` Jeff Moyer
2016-04-15 16:11 ` Jeff Moyer
2016-04-15 16:11 ` Jeff Moyer
2016-04-15 16:54 ` Verma, Vishal L
2016-04-15 16:54 ` Verma, Vishal L
2016-04-15 16:54 ` Verma, Vishal L
2016-04-15 17:11 ` Jeff Moyer
2016-04-15 17:11 ` Jeff Moyer
2016-04-15 17:11 ` Jeff Moyer
2016-04-15 17:11 ` Jeff Moyer
2016-04-15 17:37 ` Verma, Vishal L
2016-04-15 17:37 ` Verma, Vishal L
2016-04-15 17:57 ` Dan Williams
2016-04-15 17:57 ` Dan Williams
2016-04-15 17:57 ` Dan Williams
2016-04-15 18:06 ` Jeff Moyer
2016-04-15 18:06 ` Jeff Moyer
2016-04-15 18:06 ` Jeff Moyer
2016-04-15 18:06 ` Jeff Moyer
2016-04-15 18:17 ` Dan Williams
2016-04-15 18:17 ` Dan Williams
2016-04-15 18:17 ` Dan Williams
2016-04-15 18:24 ` Jeff Moyer
2016-04-15 18:24 ` Jeff Moyer
2016-04-15 18:24 ` Jeff Moyer
2016-04-15 18:24 ` Jeff Moyer
2016-04-15 18:56 ` Dan Williams
2016-04-15 18:56 ` Dan Williams
2016-04-15 18:56 ` Dan Williams
2016-04-15 19:13 ` Jeff Moyer
2016-04-15 19:13 ` Jeff Moyer
2016-04-15 19:13 ` Jeff Moyer
2016-04-15 19:13 ` Jeff Moyer
2016-04-15 19:01 ` Toshi Kani
2016-04-15 19:01 ` Toshi Kani
2016-04-15 19:01 ` Toshi Kani
2016-04-15 19:01 ` Toshi Kani
2016-04-15 19:08 ` Toshi Kani
2016-04-15 19:08 ` Toshi Kani
2016-04-15 19:08 ` Toshi Kani
2016-04-15 19:08 ` Toshi Kani
2016-04-20 20:59 ` Christoph Hellwig
2016-04-20 20:59 ` Christoph Hellwig
2016-04-20 20:59 ` Christoph Hellwig
2016-04-23 18:08 ` Verma, Vishal L
2016-04-23 18:08 ` Verma, Vishal L
2016-04-25 8:31 ` hch
2016-04-25 8:31 ` hch
2016-04-25 8:31 ` hch
2016-04-25 15:32 ` Jeff Moyer
2016-04-25 15:32 ` Jeff Moyer
2016-04-25 15:32 ` Jeff Moyer
2016-04-25 15:32 ` Jeff Moyer
2016-04-26 8:32 ` hch
2016-04-26 8:32 ` hch
2016-04-26 8:32 ` hch
2016-04-25 17:14 ` Verma, Vishal L
2016-04-25 17:14 ` Verma, Vishal L
2016-04-25 17:21 ` Dan Williams
2016-04-25 17:21 ` Dan Williams
2016-04-25 17:21 ` Dan Williams
2016-04-25 23:25 ` Dave Chinner
2016-04-25 23:25 ` Dave Chinner
2016-04-25 23:25 ` Dave Chinner
2016-04-25 23:25 ` Dave Chinner
2016-04-25 23:25 ` Dave Chinner
2016-04-25 23:34 ` Darrick J. Wong
2016-04-25 23:34 ` Darrick J. Wong
2016-04-25 23:34 ` Darrick J. Wong
2016-04-25 23:34 ` Darrick J. Wong
2016-04-25 23:34 ` Darrick J. Wong
2016-04-25 23:43 ` Dan Williams
2016-04-25 23:43 ` Dan Williams
2016-04-25 23:43 ` Dan Williams
2016-04-26 0:11 ` Dave Chinner
2016-04-26 0:11 ` Dave Chinner
2016-04-26 0:11 ` Dave Chinner
2016-04-26 1:45 ` Dan Williams
2016-04-26 1:45 ` Dan Williams
2016-04-26 1:45 ` Dan Williams
2016-04-26 2:56 ` Dave Chinner
2016-04-26 2:56 ` Dave Chinner
2016-04-26 2:56 ` Dave Chinner
2016-04-26 4:18 ` Dan Williams
2016-04-26 4:18 ` Dan Williams
2016-04-26 4:18 ` Dan Williams
2016-04-26 8:27 ` Dave Chinner
2016-04-26 8:27 ` Dave Chinner
2016-04-26 8:27 ` Dave Chinner
2016-04-26 14:59 ` Dan Williams
2016-04-26 14:59 ` Dan Williams
2016-04-26 14:59 ` Dan Williams
2016-04-26 15:31 ` Jan Kara
2016-04-26 15:31 ` Jan Kara
2016-04-26 15:31 ` Jan Kara
2016-04-26 17:16 ` Dan Williams
2016-04-26 17:16 ` Dan Williams
2016-04-26 17:16 ` Dan Williams
2016-04-25 23:53 ` Verma, Vishal L
2016-04-25 23:53 ` Verma, Vishal L
2016-04-25 23:53 ` Verma, Vishal L
2016-04-26 0:41 ` Dave Chinner
2016-04-26 0:41 ` Dave Chinner
2016-04-26 0:41 ` Dave Chinner
2016-04-26 0:41 ` Dave Chinner
2016-04-26 0:41 ` Dave Chinner
2016-04-26 14:58 ` Vishal Verma
2016-04-26 14:58 ` Vishal Verma
2016-04-26 14:58 ` Vishal Verma
2016-04-26 14:58 ` Vishal Verma
2016-05-02 15:18 ` Jeff Moyer
2016-05-02 15:18 ` Jeff Moyer
2016-05-02 15:18 ` Jeff Moyer
2016-05-02 15:18 ` Jeff Moyer
2016-05-02 17:53 ` Dan Williams
2016-05-02 17:53 ` Dan Williams
2016-05-02 17:53 ` Dan Williams
2016-05-03 0:42 ` Dave Chinner
2016-05-03 0:42 ` Dave Chinner
2016-05-03 0:42 ` Dave Chinner
2016-05-03 1:26 ` Rudoff, Andy
2016-05-03 1:26 ` Rudoff, Andy
2016-05-03 2:49 ` Dave Chinner
2016-05-03 2:49 ` Dave Chinner
2016-05-03 2:49 ` Dave Chinner
2016-05-03 18:30 ` Rudoff, Andy [this message]
2016-05-03 18:30 ` Rudoff, Andy
2016-05-03 18:30 ` Rudoff, Andy
2016-05-04 1:36 ` Dave Chinner
2016-05-04 1:36 ` Dave Chinner
2016-05-04 1:36 ` Dave Chinner
2016-05-02 23:04 ` Dave Chinner
2016-05-02 23:04 ` Dave Chinner
2016-05-02 23:04 ` Dave Chinner
2016-05-02 23:04 ` Dave Chinner
2016-05-02 23:04 ` Dave Chinner
2016-05-02 23:17 ` Verma, Vishal L
2016-05-02 23:17 ` Verma, Vishal L
2016-05-02 23:25 ` Dan Williams
2016-05-02 23:25 ` Dan Williams
2016-05-02 23:25 ` Dan Williams
2016-05-03 1:51 ` Dave Chinner
2016-05-03 1:51 ` Dave Chinner
2016-05-03 1:51 ` Dave Chinner
2016-05-03 17:28 ` Dan Williams
2016-05-03 17:28 ` Dan Williams
2016-05-03 17:28 ` Dan Williams
2016-05-04 3:18 ` Dave Chinner
2016-05-04 3:18 ` Dave Chinner
2016-05-04 3:18 ` Dave Chinner
2016-05-04 5:05 ` Dan Williams
2016-05-04 5:05 ` Dan Williams
2016-05-04 5:05 ` Dan Williams
2016-04-26 8:33 ` hch
2016-04-26 8:33 ` hch
2016-04-26 8:33 ` hch
2016-04-26 15:01 ` Vishal Verma
2016-04-26 15:01 ` Vishal Verma
2016-04-26 15:01 ` Vishal Verma
2016-04-26 15:01 ` Vishal Verma
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=FBB11841-7DFE-4223-9973-3457034260C2@intel.com \
--to=andy.rudoff@intel.com \
--cc=akpm@linux-foundation.org \
--cc=axboe@fb.com \
--cc=dan.j.williams@intel.com \
--cc=david@fromorbit.com \
--cc=hch@infradead.org \
--cc=jack@suse.cz \
--cc=linux-block@vger.kernel.org \
--cc=linux-ext4@vger.kernel.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=linux-nvdimm@ml01.01.org \
--cc=matthew.r.wilcox@intel.com \
--cc=viro@zeniv.linux.org.uk \
--cc=xfs@oss.sgi.com \
/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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.