All of lore.kernel.org
 help / color / mirror / Atom feed
From: "Matias Bjørling" <m@bjorling.me>
To: hch@infradead.org, axboe@fb.com, linux-fsdevel@vger.kernel.org,
	linux-kernel@vger.kernel.org, linux-nvme@lists.infradead.org
Cc: javier@paletta.io, keith.busch@intel.com,
	"Matias Bjørling" <m@bjorling.me>
Subject: [PATCH 0/5 v2] Support for Open-Channel SSDs
Date: Wed, 15 Apr 2015 14:34:39 +0200	[thread overview]
Message-ID: <1429101284-19490-1-git-send-email-m@bjorling.me> (raw)

A problem with SSDs is that they expose a narrow read/write interface,
where the host and device must communicate their intent. The narrow
interface leaves little information to be carried down from file-systems
and applications, and therefore performance guarantees in these devices
are best-efforts.

In the case of SSDs, there are various approaches to mitigate it.
Examples include trim and multi-streams. However, these approaches are
specialized by each vendor, each having their own behavior. More
importantly, they do not allow the host to completely control data
placement, parallelism and garbage collection.

By exposing the physical characteristics of an SSD to the host,
file-systems and applications can directly place data and control when
and where garbage collection should be applied. We call the class of
SSDs that expose these physical characteristics Open-Channel SSDs.

For this class of SSDs, LightNVM is a specification that defines a
common interface. It allows the host to manage data placement, garbage
collection, and parallelism. With it, the kernel can expose a building
block for further integration into file-systems and applications.
Immediate benefits include strict control of access latency and IO
response variance.

This patchset wires up support in the block layer, introduces a simple
block device target called rrpc, and at last adds support in the
null_blk and NVMe drivers.

Patches are against v4.0.

Development and further information on LightNVM can be found at:

  https://github.com/OpenChannelSSD/linux

Changes since v1:

 - Splitted LightNVM into two parts. A get/put interface for flash
   blocks and the respective targets that implement flash translation
   layer logic.
 - Updated the patches accordring to the LightNVM specification changes.
 - Added interface to add/remove targets for a block device.

Matias Bjørling (5):
  blk-mq: Add prep/unprep support
  blk-mq: Support for Open-Channel SSDs
  lightnvm: RRPC target
  null_blk: LightNVM support
  nvme: LightNVM support

 Documentation/block/null_blk.txt |    8 +
 block/Kconfig                    |   12 +
 block/Makefile                   |    2 +-
 block/blk-mq.c                   |   40 +-
 block/blk-nvm.c                  |  722 ++++++++++++++++++++++
 block/blk-sysfs.c                |   11 +
 block/blk.h                      |   18 +
 drivers/Kconfig                  |    2 +
 drivers/Makefile                 |    2 +
 drivers/block/null_blk.c         |   89 ++-
 drivers/block/nvme-core.c        |  380 +++++++++++-
 drivers/lightnvm/Kconfig         |   29 +
 drivers/lightnvm/Makefile        |    5 +
 drivers/lightnvm/rrpc.c          | 1222 ++++++++++++++++++++++++++++++++++++++
 drivers/lightnvm/rrpc.h          |  203 +++++++
 include/linux/bio.h              |    9 +
 include/linux/blk-mq.h           |    3 +
 include/linux/blk_types.h        |   12 +-
 include/linux/blkdev.h           |  218 +++++++
 include/linux/lightnvm.h         |   55 ++
 include/linux/nvme.h             |    2 +
 include/uapi/linux/nvm.h         |   70 +++
 include/uapi/linux/nvme.h        |  116 ++++
 23 files changed, 3217 insertions(+), 13 deletions(-)
 create mode 100644 block/blk-nvm.c
 create mode 100644 drivers/lightnvm/Kconfig
 create mode 100644 drivers/lightnvm/Makefile
 create mode 100644 drivers/lightnvm/rrpc.c
 create mode 100644 drivers/lightnvm/rrpc.h
 create mode 100644 include/linux/lightnvm.h
 create mode 100644 include/uapi/linux/nvm.h

-- 
1.9.1


WARNING: multiple messages have this Message-ID (diff)
From: m@bjorling.me (Matias Bjørling)
Subject: [PATCH 0/5 v2] Support for Open-Channel SSDs
Date: Wed, 15 Apr 2015 14:34:39 +0200	[thread overview]
Message-ID: <1429101284-19490-1-git-send-email-m@bjorling.me> (raw)

A problem with SSDs is that they expose a narrow read/write interface,
where the host and device must communicate their intent. The narrow
interface leaves little information to be carried down from file-systems
and applications, and therefore performance guarantees in these devices
are best-efforts.

In the case of SSDs, there are various approaches to mitigate it.
Examples include trim and multi-streams. However, these approaches are
specialized by each vendor, each having their own behavior. More
importantly, they do not allow the host to completely control data
placement, parallelism and garbage collection.

By exposing the physical characteristics of an SSD to the host,
file-systems and applications can directly place data and control when
and where garbage collection should be applied. We call the class of
SSDs that expose these physical characteristics Open-Channel SSDs.

For this class of SSDs, LightNVM is a specification that defines a
common interface. It allows the host to manage data placement, garbage
collection, and parallelism. With it, the kernel can expose a building
block for further integration into file-systems and applications.
Immediate benefits include strict control of access latency and IO
response variance.

This patchset wires up support in the block layer, introduces a simple
block device target called rrpc, and at last adds support in the
null_blk and NVMe drivers.

Patches are against v4.0.

Development and further information on LightNVM can be found at:

  https://github.com/OpenChannelSSD/linux

Changes since v1:

 - Splitted LightNVM into two parts. A get/put interface for flash
   blocks and the respective targets that implement flash translation
   layer logic.
 - Updated the patches accordring to the LightNVM specification changes.
 - Added interface to add/remove targets for a block device.

Matias Bj?rling (5):
  blk-mq: Add prep/unprep support
  blk-mq: Support for Open-Channel SSDs
  lightnvm: RRPC target
  null_blk: LightNVM support
  nvme: LightNVM support

 Documentation/block/null_blk.txt |    8 +
 block/Kconfig                    |   12 +
 block/Makefile                   |    2 +-
 block/blk-mq.c                   |   40 +-
 block/blk-nvm.c                  |  722 ++++++++++++++++++++++
 block/blk-sysfs.c                |   11 +
 block/blk.h                      |   18 +
 drivers/Kconfig                  |    2 +
 drivers/Makefile                 |    2 +
 drivers/block/null_blk.c         |   89 ++-
 drivers/block/nvme-core.c        |  380 +++++++++++-
 drivers/lightnvm/Kconfig         |   29 +
 drivers/lightnvm/Makefile        |    5 +
 drivers/lightnvm/rrpc.c          | 1222 ++++++++++++++++++++++++++++++++++++++
 drivers/lightnvm/rrpc.h          |  203 +++++++
 include/linux/bio.h              |    9 +
 include/linux/blk-mq.h           |    3 +
 include/linux/blk_types.h        |   12 +-
 include/linux/blkdev.h           |  218 +++++++
 include/linux/lightnvm.h         |   55 ++
 include/linux/nvme.h             |    2 +
 include/uapi/linux/nvm.h         |   70 +++
 include/uapi/linux/nvme.h        |  116 ++++
 23 files changed, 3217 insertions(+), 13 deletions(-)
 create mode 100644 block/blk-nvm.c
 create mode 100644 drivers/lightnvm/Kconfig
 create mode 100644 drivers/lightnvm/Makefile
 create mode 100644 drivers/lightnvm/rrpc.c
 create mode 100644 drivers/lightnvm/rrpc.h
 create mode 100644 include/linux/lightnvm.h
 create mode 100644 include/uapi/linux/nvm.h

-- 
1.9.1

             reply	other threads:[~2015-04-15 12:36 UTC|newest]

Thread overview: 53+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-04-15 12:34 Matias Bjørling [this message]
2015-04-15 12:34 ` [PATCH 0/5 v2] Support for Open-Channel SSDs Matias Bjørling
2015-04-15 12:34 ` [PATCH 1/5 v2] blk-mq: Add prep/unprep support Matias Bjørling
2015-04-15 12:34   ` Matias Bjørling
2015-04-17  6:34   ` Christoph Hellwig
2015-04-17  6:34     ` Christoph Hellwig
2015-04-17  8:15     ` Matias Bjørling
2015-04-17  8:15       ` Matias Bjørling
2015-04-17 17:46       ` Christoph Hellwig
2015-04-17 17:46         ` Christoph Hellwig
2015-04-18  6:45         ` Matias Bjorling
2015-04-18  6:45           ` Matias Bjorling
2015-04-18 20:16           ` Christoph Hellwig
2015-04-18 20:16             ` Christoph Hellwig
2015-04-19 18:12             ` Matias Bjorling
2015-04-19 18:12               ` Matias Bjorling
2015-04-15 12:34 ` [PATCH 2/5 v2] blk-mq: Support for Open-Channel SSDs Matias Bjørling
2015-04-15 12:34   ` Matias Bjørling
2015-04-15 12:34   ` Matias Bjørling
2015-04-16  9:10   ` Paul Bolle
2015-04-16  9:10     ` Paul Bolle
2015-04-16 10:23     ` Matias Bjørling
2015-04-16 10:23       ` Matias Bjørling
2015-04-16 10:23       ` Matias Bjørling
2015-04-16 11:34       ` Paul Bolle
2015-04-16 11:34         ` Paul Bolle
2015-04-16 11:34         ` Paul Bolle
2015-04-16 13:29         ` Matias Bjørling
2015-04-16 13:29           ` Matias Bjørling
2015-04-16 13:29           ` Matias Bjørling
2015-04-15 12:34 ` [PATCH 3/5 v2] lightnvm: RRPC target Matias Bjørling
2015-04-15 12:34   ` Matias Bjørling
2015-04-15 12:34   ` Matias Bjørling
2015-04-16  9:12   ` Paul Bolle
2015-04-16  9:12     ` Paul Bolle
2015-04-15 12:34 ` [PATCH 4/5 v2] null_blk: LightNVM support Matias Bjørling
2015-04-15 12:34   ` Matias Bjørling
2015-04-15 12:34 ` [PATCH 5/5 v2] nvme: " Matias Bjørling
2015-04-15 12:34   ` Matias Bjørling
2015-04-16 14:55   ` Keith Busch
2015-04-16 14:55     ` Keith Busch
2015-04-16 15:14     ` Javier González
2015-04-16 15:14       ` Javier González
2015-04-16 15:52       ` Keith Busch
2015-04-16 15:52         ` Keith Busch
2015-04-16 16:01         ` James R. Bergsten
2015-04-16 16:01           ` James R. Bergsten
2015-04-16 16:01           ` James R. Bergsten
2015-04-16 16:12           ` Keith Busch
2015-04-16 16:12             ` Keith Busch
2015-04-16 17:17     ` Matias Bjorling
2015-04-16 17:17       ` Matias Bjorling
2015-04-16 17:17       ` Matias Bjorling

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=1429101284-19490-1-git-send-email-m@bjorling.me \
    --to=m@bjorling.me \
    --cc=axboe@fb.com \
    --cc=hch@infradead.org \
    --cc=javier@paletta.io \
    --cc=keith.busch@intel.com \
    --cc=linux-fsdevel@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-nvme@lists.infradead.org \
    /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.