2018-10-09 11:12:06 +00:00
|
|
|
/* SPDX-License-Identifier: GPL-2.0 */
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
/*
|
|
|
|
* Copyright (C) 2015 IT University of Copenhagen (rrpc.h)
|
|
|
|
* Copyright (C) 2016 CNEX Labs
|
|
|
|
* Initial release: Matias Bjorling <matias@cnexlabs.com>
|
|
|
|
* Write buffering: Javier Gonzalez <javier@cnexlabs.com>
|
|
|
|
*
|
|
|
|
* This program is free software; you can redistribute it and/or
|
|
|
|
* modify it under the terms of the GNU General Public License version
|
|
|
|
* 2 as published by the Free Software Foundation.
|
|
|
|
*
|
|
|
|
* This program is distributed in the hope that it will be useful, but
|
|
|
|
* WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
|
|
* General Public License for more details.
|
|
|
|
*
|
|
|
|
* Implementation of a Physical Block-device target for Open-channel SSDs.
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
|
|
|
#ifndef PBLK_H_
|
|
|
|
#define PBLK_H_
|
|
|
|
|
|
|
|
#include <linux/blkdev.h>
|
|
|
|
#include <linux/blk-mq.h>
|
|
|
|
#include <linux/bio.h>
|
|
|
|
#include <linux/module.h>
|
|
|
|
#include <linux/kthread.h>
|
|
|
|
#include <linux/vmalloc.h>
|
|
|
|
#include <linux/crc32.h>
|
|
|
|
#include <linux/uuid.h>
|
|
|
|
|
|
|
|
#include <linux/lightnvm.h>
|
|
|
|
|
|
|
|
/* Run only GC if less than 1/X blocks are free */
|
|
|
|
#define GC_LIMIT_INVERSE 5
|
|
|
|
#define GC_TIME_MSECS 1000
|
|
|
|
|
|
|
|
#define PBLK_SECTOR (512)
|
|
|
|
#define PBLK_EXPOSED_PAGE_SIZE (4096)
|
|
|
|
|
2017-06-26 09:57:28 +00:00
|
|
|
#define PBLK_NR_CLOSE_JOBS (4)
|
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
#define PBLK_CACHE_NAME_LEN (DISK_NAME_LEN + 16)
|
|
|
|
|
|
|
|
#define PBLK_COMMAND_TIMEOUT_MS 30000
|
|
|
|
|
|
|
|
/* Max 512 LUNs per device */
|
|
|
|
#define PBLK_MAX_LUNS_BITMAP (4)
|
|
|
|
|
|
|
|
#define NR_PHY_IN_LOG (PBLK_EXPOSED_PAGE_SIZE / PBLK_SECTOR)
|
|
|
|
|
2017-10-13 12:46:08 +00:00
|
|
|
/* Static pool sizes */
|
2017-10-13 12:46:07 +00:00
|
|
|
#define PBLK_GEN_WS_POOL_SIZE (2)
|
|
|
|
|
2018-01-05 13:16:14 +00:00
|
|
|
#define PBLK_DEFAULT_OP (11)
|
|
|
|
|
2017-10-13 12:46:19 +00:00
|
|
|
enum {
|
|
|
|
PBLK_READ = READ,
|
|
|
|
PBLK_WRITE = WRITE,/* Write from write buffer */
|
|
|
|
PBLK_WRITE_INT, /* Internal write - no write buffer */
|
2018-01-05 13:16:16 +00:00
|
|
|
PBLK_READ_RECOV, /* Recovery read - errors allowed */
|
2017-10-13 12:46:19 +00:00
|
|
|
PBLK_ERASE,
|
|
|
|
};
|
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
enum {
|
|
|
|
/* IO Types */
|
|
|
|
PBLK_IOTYPE_USER = 1 << 0,
|
|
|
|
PBLK_IOTYPE_GC = 1 << 1,
|
|
|
|
|
|
|
|
/* Write buffer flags */
|
|
|
|
PBLK_FLUSH_ENTRY = 1 << 2,
|
|
|
|
PBLK_WRITTEN_DATA = 1 << 3,
|
|
|
|
PBLK_SUBMITTED_ENTRY = 1 << 4,
|
|
|
|
PBLK_WRITABLE_ENTRY = 1 << 5,
|
|
|
|
};
|
|
|
|
|
|
|
|
enum {
|
|
|
|
PBLK_BLK_ST_OPEN = 0x1,
|
|
|
|
PBLK_BLK_ST_CLOSED = 0x2,
|
|
|
|
};
|
|
|
|
|
2018-10-09 11:11:55 +00:00
|
|
|
enum {
|
|
|
|
PBLK_CHUNK_RESET_START,
|
|
|
|
PBLK_CHUNK_RESET_DONE,
|
|
|
|
PBLK_CHUNK_RESET_FAILED,
|
|
|
|
};
|
|
|
|
|
lightnvm: pblk: redesign GC algorithm
At the moment, in order to get enough read parallelism, we have recycled
several lines at the same time. This approach has proven not to work
well when reaching capacity, since we end up mixing valid data from all
lines, thus not maintaining a sustainable free/recycled line ratio.
The new design, relies on a two level workqueue mechanism. In the first
level, we read the metadata for a number of lines based on the GC list
they reside on (this is governed by the number of valid sectors in each
line). In the second level, we recycle a single line at a time. Here, we
issue reads in parallel, while a single GC write thread places data in
the write buffer. This design allows to (i) only move data from one line
at a time, thus maintaining a sane free/recycled ration and (ii)
maintain the GC writer busy with recycled data.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2017-06-26 09:57:27 +00:00
|
|
|
struct pblk_sec_meta {
|
|
|
|
u64 reserved;
|
|
|
|
__le64 lba;
|
|
|
|
};
|
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
/* The number of GC lists and the rate-limiter states go together. This way the
|
|
|
|
* rate-limiter can dictate how much GC is needed based on resource utilization.
|
|
|
|
*/
|
2018-06-01 14:41:06 +00:00
|
|
|
#define PBLK_GC_NR_LISTS 4
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
|
|
|
enum {
|
2018-06-01 14:41:06 +00:00
|
|
|
PBLK_RL_OFF = 0,
|
|
|
|
PBLK_RL_WERR = 1,
|
|
|
|
PBLK_RL_HIGH = 2,
|
|
|
|
PBLK_RL_MID = 3,
|
|
|
|
PBLK_RL_LOW = 4
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
};
|
|
|
|
|
2018-10-09 11:11:37 +00:00
|
|
|
#define pblk_dma_meta_size (sizeof(struct pblk_sec_meta) * NVM_MAX_VLBA)
|
|
|
|
#define pblk_dma_ppa_size (sizeof(u64) * NVM_MAX_VLBA)
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
2017-06-26 22:27:13 +00:00
|
|
|
/* write buffer completion context */
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
struct pblk_c_ctx {
|
|
|
|
struct list_head list; /* Head for out-of-order completion */
|
|
|
|
|
|
|
|
unsigned long *lun_bitmap; /* Luns used on current request */
|
|
|
|
unsigned int sentry;
|
|
|
|
unsigned int nr_valid;
|
|
|
|
unsigned int nr_padded;
|
|
|
|
};
|
|
|
|
|
2017-10-13 12:46:22 +00:00
|
|
|
/* read context */
|
2017-06-26 22:27:13 +00:00
|
|
|
struct pblk_g_ctx {
|
|
|
|
void *private;
|
2018-01-05 13:16:20 +00:00
|
|
|
unsigned long start_time;
|
2017-10-13 12:46:22 +00:00
|
|
|
u64 lba;
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
};
|
|
|
|
|
2018-07-13 08:48:44 +00:00
|
|
|
/* partial read context */
|
|
|
|
struct pblk_pr_ctx {
|
|
|
|
struct bio *orig_bio;
|
|
|
|
DECLARE_BITMAP(bitmap, NVM_MAX_VLBA);
|
|
|
|
unsigned int orig_nr_secs;
|
|
|
|
unsigned int bio_init_idx;
|
|
|
|
void *ppa_ptr;
|
|
|
|
dma_addr_t dma_ppa_list;
|
|
|
|
};
|
|
|
|
|
2017-06-30 15:56:40 +00:00
|
|
|
/* Pad context */
|
|
|
|
struct pblk_pad_rq {
|
|
|
|
struct pblk *pblk;
|
|
|
|
struct completion wait;
|
|
|
|
struct kref ref;
|
|
|
|
};
|
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
/* Recovery context */
|
|
|
|
struct pblk_rec_ctx {
|
|
|
|
struct pblk *pblk;
|
|
|
|
struct nvm_rq *rqd;
|
|
|
|
struct work_struct ws_rec;
|
|
|
|
};
|
|
|
|
|
|
|
|
/* Write context */
|
|
|
|
struct pblk_w_ctx {
|
|
|
|
struct bio_list bios; /* Original bios - used for completion
|
|
|
|
* in REQ_FUA, REQ_FLUSH case
|
|
|
|
*/
|
2017-04-19 17:39:13 +00:00
|
|
|
u64 lba; /* Logic addr. associated with entry */
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
struct ppa_addr ppa; /* Physic addr. associated with entry */
|
|
|
|
int flags; /* Write context flags */
|
|
|
|
};
|
|
|
|
|
|
|
|
struct pblk_rb_entry {
|
|
|
|
struct ppa_addr cacheline; /* Cacheline for this entry */
|
|
|
|
void *data; /* Pointer to data on this entry */
|
|
|
|
struct pblk_w_ctx w_ctx; /* Context for this entry */
|
|
|
|
struct list_head index; /* List head to enable indexes */
|
|
|
|
};
|
|
|
|
|
|
|
|
#define EMPTY_ENTRY (~0U)
|
|
|
|
|
|
|
|
struct pblk_rb_pages {
|
|
|
|
struct page *pages;
|
|
|
|
int order;
|
|
|
|
struct list_head list;
|
|
|
|
};
|
|
|
|
|
|
|
|
struct pblk_rb {
|
|
|
|
struct pblk_rb_entry *entries; /* Ring buffer entries */
|
|
|
|
unsigned int mem; /* Write offset - points to next
|
|
|
|
* writable entry in memory
|
|
|
|
*/
|
|
|
|
unsigned int subm; /* Read offset - points to last entry
|
|
|
|
* that has been submitted to the media
|
|
|
|
* to be persisted
|
|
|
|
*/
|
|
|
|
unsigned int sync; /* Synced - backpointer that signals
|
|
|
|
* the last submitted entry that has
|
|
|
|
* been successfully persisted to media
|
|
|
|
*/
|
2018-01-05 13:16:09 +00:00
|
|
|
unsigned int flush_point; /* Sync point - last entry that must be
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
* flushed to the media. Used with
|
|
|
|
* REQ_FLUSH and REQ_FUA
|
|
|
|
*/
|
|
|
|
unsigned int l2p_update; /* l2p update point - next entry for
|
|
|
|
* which l2p mapping will be updated to
|
|
|
|
* contain a device ppa address (instead
|
|
|
|
* of a cacheline
|
|
|
|
*/
|
|
|
|
unsigned int nr_entries; /* Number of entries in write buffer -
|
|
|
|
* must be a power of two
|
|
|
|
*/
|
|
|
|
unsigned int seg_size; /* Size of the data segments being
|
|
|
|
* stored on each entry. Typically this
|
|
|
|
* will be 4KB
|
|
|
|
*/
|
|
|
|
|
2018-10-09 11:12:15 +00:00
|
|
|
unsigned int back_thres; /* Threshold that shall be maintained by
|
|
|
|
* the backpointer in order to respect
|
|
|
|
* geo->mw_cunits on a per chunk basis
|
|
|
|
*/
|
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
struct list_head pages; /* List of data pages */
|
|
|
|
|
|
|
|
spinlock_t w_lock; /* Write lock */
|
|
|
|
spinlock_t s_lock; /* Sync lock */
|
|
|
|
|
2018-07-13 08:48:37 +00:00
|
|
|
#ifdef CONFIG_NVM_PBLK_DEBUG
|
2018-01-05 13:16:09 +00:00
|
|
|
atomic_t inflight_flush_point; /* Not served REQ_FLUSH | REQ_FUA */
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
#endif
|
|
|
|
};
|
|
|
|
|
|
|
|
#define PBLK_RECOVERY_SECTORS 16
|
|
|
|
|
|
|
|
struct pblk_lun {
|
|
|
|
struct ppa_addr bppa;
|
|
|
|
struct semaphore wr_sem;
|
|
|
|
};
|
|
|
|
|
|
|
|
struct pblk_gc_rq {
|
|
|
|
struct pblk_line *line;
|
|
|
|
void *data;
|
2018-10-09 11:11:37 +00:00
|
|
|
u64 paddr_list[NVM_MAX_VLBA];
|
|
|
|
u64 lba_list[NVM_MAX_VLBA];
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
int nr_secs;
|
|
|
|
int secs_to_gc;
|
|
|
|
struct list_head list;
|
|
|
|
};
|
|
|
|
|
|
|
|
struct pblk_gc {
|
lightnvm: pblk: redesign GC algorithm
At the moment, in order to get enough read parallelism, we have recycled
several lines at the same time. This approach has proven not to work
well when reaching capacity, since we end up mixing valid data from all
lines, thus not maintaining a sustainable free/recycled line ratio.
The new design, relies on a two level workqueue mechanism. In the first
level, we read the metadata for a number of lines based on the GC list
they reside on (this is governed by the number of valid sectors in each
line). In the second level, we recycle a single line at a time. Here, we
issue reads in parallel, while a single GC write thread places data in
the write buffer. This design allows to (i) only move data from one line
at a time, thus maintaining a sane free/recycled ration and (ii)
maintain the GC writer busy with recycled data.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2017-06-26 09:57:27 +00:00
|
|
|
/* These states are not protected by a lock since (i) they are in the
|
|
|
|
* fast path, and (ii) they are not critical.
|
|
|
|
*/
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
int gc_active;
|
|
|
|
int gc_enabled;
|
|
|
|
int gc_forced;
|
|
|
|
|
|
|
|
struct task_struct *gc_ts;
|
|
|
|
struct task_struct *gc_writer_ts;
|
lightnvm: pblk: redesign GC algorithm
At the moment, in order to get enough read parallelism, we have recycled
several lines at the same time. This approach has proven not to work
well when reaching capacity, since we end up mixing valid data from all
lines, thus not maintaining a sustainable free/recycled line ratio.
The new design, relies on a two level workqueue mechanism. In the first
level, we read the metadata for a number of lines based on the GC list
they reside on (this is governed by the number of valid sectors in each
line). In the second level, we recycle a single line at a time. Here, we
issue reads in parallel, while a single GC write thread places data in
the write buffer. This design allows to (i) only move data from one line
at a time, thus maintaining a sane free/recycled ration and (ii)
maintain the GC writer busy with recycled data.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2017-06-26 09:57:27 +00:00
|
|
|
struct task_struct *gc_reader_ts;
|
|
|
|
|
|
|
|
struct workqueue_struct *gc_line_reader_wq;
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
struct workqueue_struct *gc_reader_wq;
|
lightnvm: pblk: redesign GC algorithm
At the moment, in order to get enough read parallelism, we have recycled
several lines at the same time. This approach has proven not to work
well when reaching capacity, since we end up mixing valid data from all
lines, thus not maintaining a sustainable free/recycled line ratio.
The new design, relies on a two level workqueue mechanism. In the first
level, we read the metadata for a number of lines based on the GC list
they reside on (this is governed by the number of valid sectors in each
line). In the second level, we recycle a single line at a time. Here, we
issue reads in parallel, while a single GC write thread places data in
the write buffer. This design allows to (i) only move data from one line
at a time, thus maintaining a sane free/recycled ration and (ii)
maintain the GC writer busy with recycled data.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2017-06-26 09:57:27 +00:00
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
struct timer_list gc_timer;
|
|
|
|
|
lightnvm: pblk: redesign GC algorithm
At the moment, in order to get enough read parallelism, we have recycled
several lines at the same time. This approach has proven not to work
well when reaching capacity, since we end up mixing valid data from all
lines, thus not maintaining a sustainable free/recycled line ratio.
The new design, relies on a two level workqueue mechanism. In the first
level, we read the metadata for a number of lines based on the GC list
they reside on (this is governed by the number of valid sectors in each
line). In the second level, we recycle a single line at a time. Here, we
issue reads in parallel, while a single GC write thread places data in
the write buffer. This design allows to (i) only move data from one line
at a time, thus maintaining a sane free/recycled ration and (ii)
maintain the GC writer busy with recycled data.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2017-06-26 09:57:27 +00:00
|
|
|
struct semaphore gc_sem;
|
2017-10-13 12:46:41 +00:00
|
|
|
atomic_t read_inflight_gc; /* Number of lines with inflight GC reads */
|
|
|
|
atomic_t pipeline_gc; /* Number of lines in the GC pipeline -
|
|
|
|
* started reads to finished writes
|
|
|
|
*/
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
int w_entries;
|
lightnvm: pblk: redesign GC algorithm
At the moment, in order to get enough read parallelism, we have recycled
several lines at the same time. This approach has proven not to work
well when reaching capacity, since we end up mixing valid data from all
lines, thus not maintaining a sustainable free/recycled line ratio.
The new design, relies on a two level workqueue mechanism. In the first
level, we read the metadata for a number of lines based on the GC list
they reside on (this is governed by the number of valid sectors in each
line). In the second level, we recycle a single line at a time. Here, we
issue reads in parallel, while a single GC write thread places data in
the write buffer. This design allows to (i) only move data from one line
at a time, thus maintaining a sane free/recycled ration and (ii)
maintain the GC writer busy with recycled data.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2017-06-26 09:57:27 +00:00
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
struct list_head w_list;
|
lightnvm: pblk: redesign GC algorithm
At the moment, in order to get enough read parallelism, we have recycled
several lines at the same time. This approach has proven not to work
well when reaching capacity, since we end up mixing valid data from all
lines, thus not maintaining a sustainable free/recycled line ratio.
The new design, relies on a two level workqueue mechanism. In the first
level, we read the metadata for a number of lines based on the GC list
they reside on (this is governed by the number of valid sectors in each
line). In the second level, we recycle a single line at a time. Here, we
issue reads in parallel, while a single GC write thread places data in
the write buffer. This design allows to (i) only move data from one line
at a time, thus maintaining a sane free/recycled ration and (ii)
maintain the GC writer busy with recycled data.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2017-06-26 09:57:27 +00:00
|
|
|
struct list_head r_list;
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
|
|
|
spinlock_t lock;
|
|
|
|
spinlock_t w_lock;
|
lightnvm: pblk: redesign GC algorithm
At the moment, in order to get enough read parallelism, we have recycled
several lines at the same time. This approach has proven not to work
well when reaching capacity, since we end up mixing valid data from all
lines, thus not maintaining a sustainable free/recycled line ratio.
The new design, relies on a two level workqueue mechanism. In the first
level, we read the metadata for a number of lines based on the GC list
they reside on (this is governed by the number of valid sectors in each
line). In the second level, we recycle a single line at a time. Here, we
issue reads in parallel, while a single GC write thread places data in
the write buffer. This design allows to (i) only move data from one line
at a time, thus maintaining a sane free/recycled ration and (ii)
maintain the GC writer busy with recycled data.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2017-06-26 09:57:27 +00:00
|
|
|
spinlock_t r_lock;
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
};
|
|
|
|
|
|
|
|
struct pblk_rl {
|
|
|
|
unsigned int high; /* Upper threshold for rate limiter (free run -
|
|
|
|
* user I/O rate limiter
|
|
|
|
*/
|
|
|
|
unsigned int high_pw; /* High rounded up as a power of 2 */
|
|
|
|
|
lightnvm: pblk: redesign GC algorithm
At the moment, in order to get enough read parallelism, we have recycled
several lines at the same time. This approach has proven not to work
well when reaching capacity, since we end up mixing valid data from all
lines, thus not maintaining a sustainable free/recycled line ratio.
The new design, relies on a two level workqueue mechanism. In the first
level, we read the metadata for a number of lines based on the GC list
they reside on (this is governed by the number of valid sectors in each
line). In the second level, we recycle a single line at a time. Here, we
issue reads in parallel, while a single GC write thread places data in
the write buffer. This design allows to (i) only move data from one line
at a time, thus maintaining a sane free/recycled ration and (ii)
maintain the GC writer busy with recycled data.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2017-06-26 09:57:27 +00:00
|
|
|
#define PBLK_USER_HIGH_THRS 8 /* Begin write limit at 12% available blks */
|
|
|
|
#define PBLK_USER_LOW_THRS 10 /* Aggressive GC at 10% available blocks */
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
|
|
|
int rb_windows_pw; /* Number of rate windows in the write buffer
|
|
|
|
* given as a power-of-2. This guarantees that
|
|
|
|
* when user I/O is being rate limited, there
|
|
|
|
* will be reserved enough space for the GC to
|
|
|
|
* place its payload. A window is of
|
|
|
|
* pblk->max_write_pgs size, which in NVMe is
|
|
|
|
* 64, i.e., 256kb.
|
|
|
|
*/
|
|
|
|
int rb_budget; /* Total number of entries available for I/O */
|
|
|
|
int rb_user_max; /* Max buffer entries available for user I/O */
|
|
|
|
int rb_gc_max; /* Max buffer entries available for GC I/O */
|
|
|
|
int rb_gc_rsv; /* Reserved buffer entries for GC I/O */
|
|
|
|
int rb_state; /* Rate-limiter current state */
|
2017-10-13 12:46:05 +00:00
|
|
|
int rb_max_io; /* Maximum size for an I/O giving the config */
|
2017-06-26 09:57:29 +00:00
|
|
|
|
|
|
|
atomic_t rb_user_cnt; /* User I/O buffer counter */
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
atomic_t rb_gc_cnt; /* GC I/O buffer counter */
|
2017-06-26 09:57:29 +00:00
|
|
|
atomic_t rb_space; /* Space limit in case of reaching capacity */
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
lightnvm: pblk: redesign GC algorithm
At the moment, in order to get enough read parallelism, we have recycled
several lines at the same time. This approach has proven not to work
well when reaching capacity, since we end up mixing valid data from all
lines, thus not maintaining a sustainable free/recycled line ratio.
The new design, relies on a two level workqueue mechanism. In the first
level, we read the metadata for a number of lines based on the GC list
they reside on (this is governed by the number of valid sectors in each
line). In the second level, we recycle a single line at a time. Here, we
issue reads in parallel, while a single GC write thread places data in
the write buffer. This design allows to (i) only move data from one line
at a time, thus maintaining a sane free/recycled ration and (ii)
maintain the GC writer busy with recycled data.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2017-06-26 09:57:27 +00:00
|
|
|
int rsv_blocks; /* Reserved blocks for GC */
|
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
int rb_user_active;
|
lightnvm: pblk: redesign GC algorithm
At the moment, in order to get enough read parallelism, we have recycled
several lines at the same time. This approach has proven not to work
well when reaching capacity, since we end up mixing valid data from all
lines, thus not maintaining a sustainable free/recycled line ratio.
The new design, relies on a two level workqueue mechanism. In the first
level, we read the metadata for a number of lines based on the GC list
they reside on (this is governed by the number of valid sectors in each
line). In the second level, we recycle a single line at a time. Here, we
issue reads in parallel, while a single GC write thread places data in
the write buffer. This design allows to (i) only move data from one line
at a time, thus maintaining a sane free/recycled ration and (ii)
maintain the GC writer busy with recycled data.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2017-06-26 09:57:27 +00:00
|
|
|
int rb_gc_active;
|
|
|
|
|
2018-06-01 14:41:06 +00:00
|
|
|
atomic_t werr_lines; /* Number of write error lines that needs gc */
|
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
struct timer_list u_timer;
|
|
|
|
|
|
|
|
unsigned long long nr_secs;
|
|
|
|
unsigned long total_blocks;
|
2018-01-05 13:16:13 +00:00
|
|
|
|
|
|
|
atomic_t free_blocks; /* Total number of free blocks (+ OP) */
|
|
|
|
atomic_t free_user_blocks; /* Number of user free blocks (no OP) */
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
};
|
|
|
|
|
|
|
|
#define PBLK_LINE_EMPTY (~0U)
|
|
|
|
|
|
|
|
enum {
|
|
|
|
/* Line Types */
|
|
|
|
PBLK_LINETYPE_FREE = 0,
|
|
|
|
PBLK_LINETYPE_LOG = 1,
|
|
|
|
PBLK_LINETYPE_DATA = 2,
|
|
|
|
|
|
|
|
/* Line state */
|
2018-03-29 22:05:20 +00:00
|
|
|
PBLK_LINESTATE_NEW = 9,
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
PBLK_LINESTATE_FREE = 10,
|
|
|
|
PBLK_LINESTATE_OPEN = 11,
|
|
|
|
PBLK_LINESTATE_CLOSED = 12,
|
|
|
|
PBLK_LINESTATE_GC = 13,
|
|
|
|
PBLK_LINESTATE_BAD = 14,
|
|
|
|
PBLK_LINESTATE_CORRUPT = 15,
|
|
|
|
|
|
|
|
/* GC group */
|
|
|
|
PBLK_LINEGC_NONE = 20,
|
|
|
|
PBLK_LINEGC_EMPTY = 21,
|
|
|
|
PBLK_LINEGC_LOW = 22,
|
|
|
|
PBLK_LINEGC_MID = 23,
|
|
|
|
PBLK_LINEGC_HIGH = 24,
|
|
|
|
PBLK_LINEGC_FULL = 25,
|
2018-06-01 14:41:06 +00:00
|
|
|
PBLK_LINEGC_WERR = 26
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
};
|
|
|
|
|
|
|
|
#define PBLK_MAGIC 0x70626c6b /*pblk*/
|
2018-03-29 22:04:51 +00:00
|
|
|
|
|
|
|
/* emeta/smeta persistent storage format versions:
|
|
|
|
* Changes in major version requires offline migration.
|
|
|
|
* Changes in minor version are handled automatically during
|
|
|
|
* recovery.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#define SMETA_VERSION_MAJOR (0)
|
|
|
|
#define SMETA_VERSION_MINOR (1)
|
|
|
|
|
|
|
|
#define EMETA_VERSION_MAJOR (0)
|
2018-03-29 22:04:52 +00:00
|
|
|
#define EMETA_VERSION_MINOR (2)
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
|
|
|
struct line_header {
|
|
|
|
__le32 crc;
|
|
|
|
__le32 identifier; /* pblk identifier */
|
|
|
|
__u8 uuid[16]; /* instance uuid */
|
|
|
|
__le16 type; /* line type */
|
2018-03-29 22:04:51 +00:00
|
|
|
__u8 version_major; /* version major */
|
|
|
|
__u8 version_minor; /* version minor */
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
__le32 id; /* line id for current line */
|
|
|
|
};
|
|
|
|
|
|
|
|
struct line_smeta {
|
|
|
|
struct line_header header;
|
|
|
|
|
|
|
|
__le32 crc; /* Full structure including struct crc */
|
|
|
|
/* Previous line metadata */
|
|
|
|
__le32 prev_id; /* Line id for previous line */
|
|
|
|
|
|
|
|
/* Current line metadata */
|
|
|
|
__le64 seq_nr; /* Sequence number for current line */
|
|
|
|
|
|
|
|
/* Active writers */
|
|
|
|
__le32 window_wr_lun; /* Number of parallel LUNs to write */
|
|
|
|
|
|
|
|
__le32 rsvd[2];
|
2017-06-26 09:57:17 +00:00
|
|
|
|
|
|
|
__le64 lun_bitmap[];
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
};
|
|
|
|
|
2018-03-29 22:04:52 +00:00
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
/*
|
2017-06-26 09:57:17 +00:00
|
|
|
* Metadata layout in media:
|
|
|
|
* First sector:
|
|
|
|
* 1. struct line_emeta
|
|
|
|
* 2. bad block bitmap (u64 * window_wr_lun)
|
2018-03-29 22:04:52 +00:00
|
|
|
* 3. write amplification counters
|
2017-06-26 09:57:17 +00:00
|
|
|
* Mid sectors (start at lbas_sector):
|
|
|
|
* 3. nr_lbas (u64) forming lba list
|
|
|
|
* Last sectors (start at vsc_sector):
|
|
|
|
* 4. u32 valid sector count (vsc) for all lines (~0U: free line)
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
*/
|
|
|
|
struct line_emeta {
|
|
|
|
struct line_header header;
|
|
|
|
|
|
|
|
__le32 crc; /* Full structure including struct crc */
|
|
|
|
|
|
|
|
/* Previous line metadata */
|
|
|
|
__le32 prev_id; /* Line id for prev line */
|
|
|
|
|
|
|
|
/* Current line metadata */
|
|
|
|
__le64 seq_nr; /* Sequence number for current line */
|
|
|
|
|
|
|
|
/* Active writers */
|
|
|
|
__le32 window_wr_lun; /* Number of parallel LUNs to write */
|
|
|
|
|
|
|
|
/* Bookkeeping for recovery */
|
|
|
|
__le32 next_id; /* Line id for next line */
|
|
|
|
__le64 nr_lbas; /* Number of lbas mapped in line */
|
|
|
|
__le64 nr_valid_lbas; /* Number of valid lbas mapped in line */
|
2018-03-29 22:04:52 +00:00
|
|
|
__le64 bb_bitmap[]; /* Updated bad block bitmap for line */
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
|
|
/* Write amplification counters stored on media */
|
|
|
|
struct wa_counters {
|
|
|
|
__le64 user; /* Number of user written sectors */
|
|
|
|
__le64 gc; /* Number of sectors written by GC*/
|
|
|
|
__le64 pad; /* Number of padded sectors */
|
2017-06-26 09:57:17 +00:00
|
|
|
};
|
|
|
|
|
|
|
|
struct pblk_emeta {
|
|
|
|
struct line_emeta *buf; /* emeta buffer in media format */
|
|
|
|
int mem; /* Write offset - points to next
|
|
|
|
* writable entry in memory
|
|
|
|
*/
|
|
|
|
atomic_t sync; /* Synced - backpointer that signals the
|
|
|
|
* last entry that has been successfully
|
|
|
|
* persisted to media
|
|
|
|
*/
|
|
|
|
unsigned int nr_entries; /* Number of emeta entries */
|
|
|
|
};
|
|
|
|
|
|
|
|
struct pblk_smeta {
|
|
|
|
struct line_smeta *buf; /* smeta buffer in persistent format */
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
};
|
|
|
|
|
2018-06-01 14:41:06 +00:00
|
|
|
struct pblk_w_err_gc {
|
|
|
|
int has_write_err;
|
|
|
|
__le64 *lba_list;
|
|
|
|
};
|
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
struct pblk_line {
|
|
|
|
struct pblk *pblk;
|
|
|
|
unsigned int id; /* Line number corresponds to the
|
|
|
|
* block line
|
|
|
|
*/
|
|
|
|
unsigned int seq_nr; /* Unique line sequence number */
|
|
|
|
|
|
|
|
int state; /* PBLK_LINESTATE_X */
|
|
|
|
int type; /* PBLK_LINETYPE_X */
|
|
|
|
int gc_group; /* PBLK_LINEGC_X */
|
|
|
|
struct list_head list; /* Free, GC lists */
|
|
|
|
|
|
|
|
unsigned long *lun_bitmap; /* Bitmap for LUNs mapped in line */
|
|
|
|
|
2018-03-29 22:05:20 +00:00
|
|
|
struct nvm_chk_meta *chks; /* Chunks forming line */
|
|
|
|
|
2017-06-26 09:57:17 +00:00
|
|
|
struct pblk_smeta *smeta; /* Start metadata */
|
|
|
|
struct pblk_emeta *emeta; /* End medatada */
|
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
int meta_line; /* Metadata line id */
|
2017-06-26 09:57:17 +00:00
|
|
|
int meta_distance; /* Distance between data and metadata */
|
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
u64 smeta_ssec; /* Sector where smeta starts */
|
|
|
|
u64 emeta_ssec; /* Sector where emeta starts */
|
|
|
|
|
|
|
|
unsigned int sec_in_line; /* Number of usable secs in line */
|
|
|
|
|
2017-04-21 23:32:49 +00:00
|
|
|
atomic_t blk_in_line; /* Number of good blocks in line */
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
unsigned long *blk_bitmap; /* Bitmap for valid/invalid blocks */
|
|
|
|
unsigned long *erase_bitmap; /* Bitmap for erased blocks */
|
|
|
|
|
|
|
|
unsigned long *map_bitmap; /* Bitmap for mapped sectors in line */
|
|
|
|
unsigned long *invalid_bitmap; /* Bitmap for invalid sectors in line */
|
|
|
|
|
2017-04-21 23:32:49 +00:00
|
|
|
atomic_t left_eblks; /* Blocks left for erasing */
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
atomic_t left_seblks; /* Blocks left for sync erasing */
|
|
|
|
|
|
|
|
int left_msecs; /* Sectors left for mapping */
|
|
|
|
unsigned int cur_sec; /* Sector map pointer */
|
2017-06-26 09:57:17 +00:00
|
|
|
unsigned int nr_valid_lbas; /* Number of valid lbas in line */
|
|
|
|
|
|
|
|
__le32 *vsc; /* Valid sector count in line */
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
|
|
|
struct kref ref; /* Write buffer L2P references */
|
|
|
|
|
2018-06-01 14:41:06 +00:00
|
|
|
struct pblk_w_err_gc *w_err_gc; /* Write error gc recovery metadata */
|
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
spinlock_t lock; /* Necessary for invalid_bitmap only */
|
|
|
|
};
|
|
|
|
|
2017-04-21 23:32:49 +00:00
|
|
|
#define PBLK_DATA_LINES 4
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
2017-06-26 09:57:17 +00:00
|
|
|
enum {
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
PBLK_KMALLOC_META = 1,
|
|
|
|
PBLK_VMALLOC_META = 2,
|
|
|
|
};
|
|
|
|
|
2017-06-26 09:57:17 +00:00
|
|
|
enum {
|
|
|
|
PBLK_EMETA_TYPE_HEADER = 1, /* struct line_emeta first sector */
|
|
|
|
PBLK_EMETA_TYPE_LLBA = 2, /* lba list - type: __le64 */
|
|
|
|
PBLK_EMETA_TYPE_VSC = 3, /* vsc list - type: __le32 */
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
};
|
|
|
|
|
|
|
|
struct pblk_line_mgmt {
|
|
|
|
int nr_lines; /* Total number of full lines */
|
|
|
|
int nr_free_lines; /* Number of full lines in free list */
|
|
|
|
|
|
|
|
/* Free lists - use free_lock */
|
|
|
|
struct list_head free_list; /* Full lines ready to use */
|
|
|
|
struct list_head corrupt_list; /* Full lines corrupted */
|
|
|
|
struct list_head bad_list; /* Full lines bad */
|
|
|
|
|
|
|
|
/* GC lists - use gc_lock */
|
lightnvm: pblk: redesign GC algorithm
At the moment, in order to get enough read parallelism, we have recycled
several lines at the same time. This approach has proven not to work
well when reaching capacity, since we end up mixing valid data from all
lines, thus not maintaining a sustainable free/recycled line ratio.
The new design, relies on a two level workqueue mechanism. In the first
level, we read the metadata for a number of lines based on the GC list
they reside on (this is governed by the number of valid sectors in each
line). In the second level, we recycle a single line at a time. Here, we
issue reads in parallel, while a single GC write thread places data in
the write buffer. This design allows to (i) only move data from one line
at a time, thus maintaining a sane free/recycled ration and (ii)
maintain the GC writer busy with recycled data.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2017-06-26 09:57:27 +00:00
|
|
|
struct list_head *gc_lists[PBLK_GC_NR_LISTS];
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
struct list_head gc_high_list; /* Full lines ready to GC, high isc */
|
|
|
|
struct list_head gc_mid_list; /* Full lines ready to GC, mid isc */
|
|
|
|
struct list_head gc_low_list; /* Full lines ready to GC, low isc */
|
|
|
|
|
2018-06-01 14:41:06 +00:00
|
|
|
struct list_head gc_werr_list; /* Write err recovery list */
|
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
struct list_head gc_full_list; /* Full lines ready to GC, no valid */
|
|
|
|
struct list_head gc_empty_list; /* Full lines close, all valid */
|
|
|
|
|
|
|
|
struct pblk_line *log_line; /* Current FTL log line */
|
|
|
|
struct pblk_line *data_line; /* Current data line */
|
|
|
|
struct pblk_line *log_next; /* Next FTL log line */
|
|
|
|
struct pblk_line *data_next; /* Next data line */
|
|
|
|
|
2017-06-26 09:57:17 +00:00
|
|
|
struct list_head emeta_list; /* Lines queued to schedule emeta */
|
|
|
|
|
|
|
|
__le32 *vsc_list; /* Valid sector counts for all lines */
|
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
/* Metadata allocation type: VMALLOC | KMALLOC */
|
|
|
|
int emeta_alloc_type;
|
|
|
|
|
|
|
|
/* Pre-allocated metadata for data lines */
|
2017-06-26 09:57:17 +00:00
|
|
|
struct pblk_smeta *sline_meta[PBLK_DATA_LINES];
|
|
|
|
struct pblk_emeta *eline_meta[PBLK_DATA_LINES];
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
unsigned long meta_bitmap;
|
|
|
|
|
2018-10-09 11:11:47 +00:00
|
|
|
/* Cache and mempool for map/invalid bitmaps */
|
|
|
|
struct kmem_cache *bitmap_cache;
|
|
|
|
mempool_t *bitmap_pool;
|
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
/* Helpers for fast bitmap calculations */
|
|
|
|
unsigned long *bb_template;
|
|
|
|
unsigned long *bb_aux;
|
|
|
|
|
|
|
|
unsigned long d_seq_nr; /* Data line unique sequence number */
|
|
|
|
unsigned long l_seq_nr; /* Log line unique sequence number */
|
|
|
|
|
|
|
|
spinlock_t free_lock;
|
2017-06-26 09:57:17 +00:00
|
|
|
spinlock_t close_lock;
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
spinlock_t gc_lock;
|
|
|
|
};
|
|
|
|
|
|
|
|
struct pblk_line_meta {
|
|
|
|
unsigned int smeta_len; /* Total length for smeta */
|
2017-06-26 09:57:17 +00:00
|
|
|
unsigned int smeta_sec; /* Sectors needed for smeta */
|
|
|
|
|
|
|
|
unsigned int emeta_len[4]; /* Lengths for emeta:
|
2018-03-29 22:04:52 +00:00
|
|
|
* [0]: Total
|
|
|
|
* [1]: struct line_emeta +
|
|
|
|
* bb_bitmap + struct wa_counters
|
|
|
|
* [2]: L2P portion
|
|
|
|
* [3]: vsc
|
2017-06-26 09:57:17 +00:00
|
|
|
*/
|
|
|
|
unsigned int emeta_sec[4]; /* Sectors needed for emeta. Same layout
|
|
|
|
* as emeta_len
|
|
|
|
*/
|
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
unsigned int emeta_bb; /* Boundary for bb that affects emeta */
|
2017-06-26 09:57:17 +00:00
|
|
|
|
|
|
|
unsigned int vsc_list_len; /* Length for vsc list */
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
unsigned int sec_bitmap_len; /* Length for sector bitmap in line */
|
|
|
|
unsigned int blk_bitmap_len; /* Length for block bitmap in line */
|
|
|
|
unsigned int lun_bitmap_len; /* Length for lun bitmap in line */
|
|
|
|
|
|
|
|
unsigned int blk_per_line; /* Number of blocks in a full line */
|
|
|
|
unsigned int sec_per_line; /* Number of sectors in a line */
|
2017-06-26 09:57:17 +00:00
|
|
|
unsigned int dsec_per_line; /* Number of data sectors in a line */
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
unsigned int min_blk_line; /* Min. number of good blocks in line */
|
|
|
|
|
|
|
|
unsigned int mid_thrs; /* Threshold for GC mid list */
|
|
|
|
unsigned int high_thrs; /* Threshold for GC high list */
|
2017-06-26 09:57:17 +00:00
|
|
|
|
|
|
|
unsigned int meta_distance; /* Distance between data and metadata */
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
};
|
|
|
|
|
2017-06-26 09:57:29 +00:00
|
|
|
enum {
|
|
|
|
PBLK_STATE_RUNNING = 0,
|
|
|
|
PBLK_STATE_STOPPING = 1,
|
|
|
|
PBLK_STATE_RECOVERING = 2,
|
|
|
|
PBLK_STATE_STOPPED = 3,
|
|
|
|
};
|
|
|
|
|
2018-03-29 22:05:21 +00:00
|
|
|
/* Internal format to support not power-of-2 device formats */
|
|
|
|
struct pblk_addrf {
|
|
|
|
/* gen to dev */
|
|
|
|
int sec_stripe;
|
|
|
|
int ch_stripe;
|
|
|
|
int lun_stripe;
|
|
|
|
|
|
|
|
/* dev to gen */
|
|
|
|
int sec_lun_stripe;
|
|
|
|
int sec_ws_stripe;
|
|
|
|
};
|
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
struct pblk {
|
|
|
|
struct nvm_tgt_dev *dev;
|
|
|
|
struct gendisk *disk;
|
|
|
|
|
|
|
|
struct kobject kobj;
|
|
|
|
|
|
|
|
struct pblk_lun *luns;
|
|
|
|
|
|
|
|
struct pblk_line *lines; /* Line array */
|
|
|
|
struct pblk_line_mgmt l_mg; /* Line management */
|
|
|
|
struct pblk_line_meta lm; /* Line metadata */
|
|
|
|
|
2018-03-29 22:05:21 +00:00
|
|
|
struct nvm_addrf addrf; /* Aligned address format */
|
|
|
|
struct pblk_addrf uaddrf; /* Unaligned address format */
|
2018-03-29 22:05:19 +00:00
|
|
|
int addrf_len;
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
|
|
|
struct pblk_rb rwb;
|
|
|
|
|
2017-06-26 09:57:29 +00:00
|
|
|
int state; /* pblk line state */
|
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
int min_write_pgs; /* Minimum amount of pages required by controller */
|
|
|
|
int max_write_pgs; /* Maximum amount of pages supported by controller */
|
|
|
|
|
|
|
|
sector_t capacity; /* Device capacity when bad blocks are subtracted */
|
2018-01-05 13:16:13 +00:00
|
|
|
|
|
|
|
int op; /* Percentage of device used for over-provisioning */
|
|
|
|
int op_blks; /* Number of blocks used for over-provisioning */
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
|
|
|
/* pblk provisioning values. Used by rate limiter */
|
|
|
|
struct pblk_rl rl;
|
|
|
|
|
2017-06-26 09:57:14 +00:00
|
|
|
int sec_per_write;
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
|
|
|
unsigned char instance_uuid[16];
|
2018-03-29 22:04:52 +00:00
|
|
|
|
|
|
|
/* Persistent write amplification counters, 4kb sector I/Os */
|
|
|
|
atomic64_t user_wa; /* Sectors written by user */
|
|
|
|
atomic64_t gc_wa; /* Sectors written by GC */
|
|
|
|
atomic64_t pad_wa; /* Padded sectors written */
|
|
|
|
|
|
|
|
/* Reset values for delta write amplification measurements */
|
|
|
|
u64 user_rst_wa;
|
|
|
|
u64 gc_rst_wa;
|
|
|
|
u64 pad_rst_wa;
|
|
|
|
|
2018-03-29 22:04:55 +00:00
|
|
|
/* Counters used for calculating padding distribution */
|
|
|
|
atomic64_t *pad_dist; /* Padding distribution buckets */
|
|
|
|
u64 nr_flush_rst; /* Flushes reset value for pad dist.*/
|
|
|
|
atomic64_t nr_flush; /* Number of flush/fua I/O */
|
|
|
|
|
2018-07-13 08:48:37 +00:00
|
|
|
#ifdef CONFIG_NVM_PBLK_DEBUG
|
2018-03-29 22:04:52 +00:00
|
|
|
/* Non-persistent debug counters, 4kb sector I/Os */
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
atomic_long_t inflight_writes; /* Inflight writes (user and gc) */
|
|
|
|
atomic_long_t padded_writes; /* Sectors padded due to flush/fua */
|
|
|
|
atomic_long_t padded_wb; /* Sectors padded in write buffer */
|
|
|
|
atomic_long_t req_writes; /* Sectors stored on write buffer */
|
|
|
|
atomic_long_t sub_writes; /* Sectors submitted from buffer */
|
|
|
|
atomic_long_t sync_writes; /* Sectors synced to media */
|
|
|
|
atomic_long_t inflight_reads; /* Inflight sector read requests */
|
2017-06-26 09:57:13 +00:00
|
|
|
atomic_long_t cache_reads; /* Read requests that hit the cache */
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
atomic_long_t sync_reads; /* Completed sector read requests */
|
|
|
|
atomic_long_t recov_writes; /* Sectors submitted from recovery */
|
|
|
|
atomic_long_t recov_gc_writes; /* Sectors submitted from write GC */
|
|
|
|
atomic_long_t recov_gc_reads; /* Sectors submitted from read GC */
|
|
|
|
#endif
|
|
|
|
|
|
|
|
spinlock_t lock;
|
|
|
|
|
|
|
|
atomic_long_t read_failed;
|
|
|
|
atomic_long_t read_empty;
|
|
|
|
atomic_long_t read_high_ecc;
|
|
|
|
atomic_long_t read_failed_gc;
|
|
|
|
atomic_long_t write_failed;
|
|
|
|
atomic_long_t erase_failed;
|
|
|
|
|
2017-06-26 09:57:29 +00:00
|
|
|
atomic_t inflight_io; /* General inflight I/O counter */
|
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
struct task_struct *writer_ts;
|
|
|
|
|
|
|
|
/* Simple translation map of logical addresses to physical addresses.
|
|
|
|
* The logical addresses is known by the host system, while the physical
|
|
|
|
* addresses are used when writing to the disk block device.
|
|
|
|
*/
|
|
|
|
unsigned char *trans_map;
|
|
|
|
spinlock_t trans_lock;
|
|
|
|
|
|
|
|
struct list_head compl_list;
|
|
|
|
|
2018-06-01 14:41:05 +00:00
|
|
|
spinlock_t resubmit_lock; /* Resubmit list lock */
|
|
|
|
struct list_head resubmit_list; /* Resubmit list for failed writes*/
|
|
|
|
|
2018-05-20 22:25:50 +00:00
|
|
|
mempool_t page_bio_pool;
|
|
|
|
mempool_t gen_ws_pool;
|
|
|
|
mempool_t rec_pool;
|
|
|
|
mempool_t r_rq_pool;
|
|
|
|
mempool_t w_rq_pool;
|
|
|
|
mempool_t e_rq_pool;
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
2017-06-26 09:57:28 +00:00
|
|
|
struct workqueue_struct *close_wq;
|
|
|
|
struct workqueue_struct *bb_wq;
|
2017-10-13 12:46:23 +00:00
|
|
|
struct workqueue_struct *r_end_wq;
|
2017-06-26 09:57:28 +00:00
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
struct timer_list wtimer;
|
|
|
|
|
|
|
|
struct pblk_gc gc;
|
|
|
|
};
|
|
|
|
|
|
|
|
struct pblk_line_ws {
|
|
|
|
struct pblk *pblk;
|
|
|
|
struct pblk_line *line;
|
|
|
|
void *priv;
|
|
|
|
struct work_struct ws;
|
|
|
|
};
|
|
|
|
|
2017-06-26 22:27:13 +00:00
|
|
|
#define pblk_g_rq_size (sizeof(struct nvm_rq) + sizeof(struct pblk_g_ctx))
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
#define pblk_w_rq_size (sizeof(struct nvm_rq) + sizeof(struct pblk_c_ctx))
|
|
|
|
|
2018-07-13 08:48:42 +00:00
|
|
|
#define pblk_err(pblk, fmt, ...) \
|
|
|
|
pr_err("pblk %s: " fmt, pblk->disk->disk_name, ##__VA_ARGS__)
|
|
|
|
#define pblk_info(pblk, fmt, ...) \
|
|
|
|
pr_info("pblk %s: " fmt, pblk->disk->disk_name, ##__VA_ARGS__)
|
|
|
|
#define pblk_warn(pblk, fmt, ...) \
|
|
|
|
pr_warn("pblk %s: " fmt, pblk->disk->disk_name, ##__VA_ARGS__)
|
|
|
|
#define pblk_debug(pblk, fmt, ...) \
|
|
|
|
pr_debug("pblk %s: " fmt, pblk->disk->disk_name, ##__VA_ARGS__)
|
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
/*
|
|
|
|
* pblk ring buffer operations
|
|
|
|
*/
|
2018-10-09 11:12:15 +00:00
|
|
|
int pblk_rb_init(struct pblk_rb *rb, unsigned int size, unsigned int threshold,
|
|
|
|
unsigned int seg_sz);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
int pblk_rb_may_write_user(struct pblk_rb *rb, struct bio *bio,
|
|
|
|
unsigned int nr_entries, unsigned int *pos);
|
|
|
|
int pblk_rb_may_write_gc(struct pblk_rb *rb, unsigned int nr_entries,
|
|
|
|
unsigned int *pos);
|
|
|
|
void pblk_rb_write_entry_user(struct pblk_rb *rb, void *data,
|
|
|
|
struct pblk_w_ctx w_ctx, unsigned int pos);
|
|
|
|
void pblk_rb_write_entry_gc(struct pblk_rb *rb, void *data,
|
2017-10-13 12:46:14 +00:00
|
|
|
struct pblk_w_ctx w_ctx, struct pblk_line *line,
|
|
|
|
u64 paddr, unsigned int pos);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
struct pblk_w_ctx *pblk_rb_w_ctx(struct pblk_rb *rb, unsigned int pos);
|
2017-06-26 09:57:29 +00:00
|
|
|
void pblk_rb_flush(struct pblk_rb *rb);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
|
|
|
void pblk_rb_sync_l2p(struct pblk_rb *rb);
|
2017-06-26 09:57:15 +00:00
|
|
|
unsigned int pblk_rb_read_to_bio(struct pblk_rb *rb, struct nvm_rq *rqd,
|
2017-10-13 12:46:18 +00:00
|
|
|
unsigned int pos, unsigned int nr_entries,
|
|
|
|
unsigned int count);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
int pblk_rb_copy_to_bio(struct pblk_rb *rb, struct bio *bio, sector_t lba,
|
2017-07-28 13:13:16 +00:00
|
|
|
struct ppa_addr ppa, int bio_iter, bool advanced_bio);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
unsigned int pblk_rb_read_commit(struct pblk_rb *rb, unsigned int entries);
|
|
|
|
|
|
|
|
unsigned int pblk_rb_sync_init(struct pblk_rb *rb, unsigned long *flags);
|
|
|
|
unsigned int pblk_rb_sync_advance(struct pblk_rb *rb, unsigned int nr_entries);
|
2018-10-09 11:12:09 +00:00
|
|
|
unsigned int pblk_rb_ptr_wrap(struct pblk_rb *rb, unsigned int p,
|
|
|
|
unsigned int nr_entries);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
void pblk_rb_sync_end(struct pblk_rb *rb, unsigned long *flags);
|
2018-01-05 13:16:09 +00:00
|
|
|
unsigned int pblk_rb_flush_point_count(struct pblk_rb *rb);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
|
|
|
unsigned int pblk_rb_read_count(struct pblk_rb *rb);
|
2017-06-30 15:56:40 +00:00
|
|
|
unsigned int pblk_rb_sync_count(struct pblk_rb *rb);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
unsigned int pblk_rb_wrap_pos(struct pblk_rb *rb, unsigned int pos);
|
|
|
|
|
|
|
|
int pblk_rb_tear_down_check(struct pblk_rb *rb);
|
|
|
|
int pblk_rb_pos_oob(struct pblk_rb *rb, u64 pos);
|
2018-10-09 11:12:10 +00:00
|
|
|
void pblk_rb_free(struct pblk_rb *rb);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
ssize_t pblk_rb_sysfs(struct pblk_rb *rb, char *buf);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* pblk core
|
|
|
|
*/
|
2017-10-13 12:46:20 +00:00
|
|
|
struct nvm_rq *pblk_alloc_rqd(struct pblk *pblk, int type);
|
|
|
|
void pblk_free_rqd(struct pblk *pblk, struct nvm_rq *rqd, int type);
|
2018-10-09 11:12:02 +00:00
|
|
|
int pblk_alloc_rqd_meta(struct pblk *pblk, struct nvm_rq *rqd);
|
|
|
|
void pblk_free_rqd_meta(struct pblk *pblk, struct nvm_rq *rqd);
|
2017-06-26 09:57:14 +00:00
|
|
|
void pblk_set_sec_per_write(struct pblk *pblk, int sec_per_write);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
int pblk_setup_w_rec_rq(struct pblk *pblk, struct nvm_rq *rqd,
|
|
|
|
struct pblk_c_ctx *c_ctx);
|
|
|
|
void pblk_discard(struct pblk *pblk, struct bio *bio);
|
2018-10-09 11:11:36 +00:00
|
|
|
struct nvm_chk_meta *pblk_get_chunk_meta(struct pblk *pblk);
|
2018-03-29 22:05:20 +00:00
|
|
|
struct nvm_chk_meta *pblk_chunk_get_off(struct pblk *pblk,
|
|
|
|
struct nvm_chk_meta *lp,
|
|
|
|
struct ppa_addr ppa);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
void pblk_log_write_err(struct pblk *pblk, struct nvm_rq *rqd);
|
|
|
|
void pblk_log_read_err(struct pblk *pblk, struct nvm_rq *rqd);
|
|
|
|
int pblk_submit_io(struct pblk *pblk, struct nvm_rq *rqd);
|
2017-10-13 12:46:47 +00:00
|
|
|
int pblk_submit_io_sync(struct pblk *pblk, struct nvm_rq *rqd);
|
2018-10-09 11:12:04 +00:00
|
|
|
int pblk_submit_io_sync_sem(struct pblk *pblk, struct nvm_rq *rqd);
|
2017-06-26 09:57:17 +00:00
|
|
|
int pblk_submit_meta_io(struct pblk *pblk, struct pblk_line *meta_line);
|
2018-10-09 11:11:52 +00:00
|
|
|
void pblk_check_chunk_state_update(struct pblk *pblk, struct nvm_rq *rqd);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
struct bio *pblk_bio_map_addr(struct pblk *pblk, void *data,
|
|
|
|
unsigned int nr_secs, unsigned int len,
|
2017-06-30 15:56:39 +00:00
|
|
|
int alloc_type, gfp_t gfp_mask);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
struct pblk_line *pblk_line_get(struct pblk *pblk);
|
|
|
|
struct pblk_line *pblk_line_get_first_data(struct pblk *pblk);
|
lightnvm: pblk: enable 1 LUN configuration
Metadata I/Os are scheduled to minimize their impact on user data I/Os.
When there are enough LUNs instantiated (i.e., enough bandwidth), it is
easy to interleave metadata and data one after the other so that
metadata I/Os are the ones being blocked and not vice-versa.
We do this by calculating the distance between the I/Os in terms of the
LUNs that are not in used, and selecting a free LUN that satisfies a
the simple heuristic that metadata is scheduled behind. The per-LUN
semaphores guarantee consistency. This works fine on >1 LUN
configuration. However, when a single LUN is instantiated, this design
leads to a deadlock, where metadata waits to be scheduled on a free LUN.
This patch implements the 1 LUN case by simply scheduling the metadada
I/O after the data I/O. In the process, we refactor the way a line is
replaced to ensure that metadata writes are submitted after data writes
in order to guarantee block sequentiality. Note that, since there is
only one LUN, both I/Os will block each other by design. However, such
configuration only pursues tight read latencies, not write bandwidth.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <m@bjorling.me>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2017-10-13 12:46:26 +00:00
|
|
|
struct pblk_line *pblk_line_replace_data(struct pblk *pblk);
|
2018-10-09 11:11:40 +00:00
|
|
|
void pblk_ppa_to_line_put(struct pblk *pblk, struct ppa_addr ppa);
|
|
|
|
void pblk_rq_to_line_put(struct pblk *pblk, struct nvm_rq *rqd);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
int pblk_line_recov_alloc(struct pblk *pblk, struct pblk_line *line);
|
|
|
|
void pblk_line_recov_close(struct pblk *pblk, struct pblk_line *line);
|
|
|
|
struct pblk_line *pblk_line_get_data(struct pblk *pblk);
|
2017-06-26 09:57:15 +00:00
|
|
|
struct pblk_line *pblk_line_get_erase(struct pblk *pblk);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
int pblk_line_erase(struct pblk *pblk, struct pblk_line *line);
|
|
|
|
int pblk_line_is_full(struct pblk_line *line);
|
2018-06-01 13:04:22 +00:00
|
|
|
void pblk_line_free(struct pblk_line *line);
|
2017-06-26 09:57:17 +00:00
|
|
|
void pblk_line_close_meta(struct pblk *pblk, struct pblk_line *line);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
void pblk_line_close(struct pblk *pblk, struct pblk_line *line);
|
2017-06-26 09:57:17 +00:00
|
|
|
void pblk_line_close_ws(struct work_struct *work);
|
2017-06-26 09:57:29 +00:00
|
|
|
void pblk_pipeline_stop(struct pblk *pblk);
|
2018-06-01 13:04:24 +00:00
|
|
|
void __pblk_pipeline_stop(struct pblk *pblk);
|
|
|
|
void __pblk_pipeline_flush(struct pblk *pblk);
|
2017-10-13 12:46:07 +00:00
|
|
|
void pblk_gen_run_ws(struct pblk *pblk, struct pblk_line *line, void *priv,
|
|
|
|
void (*work)(struct work_struct *), gfp_t gfp_mask,
|
|
|
|
struct workqueue_struct *wq);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
u64 pblk_line_smeta_start(struct pblk *pblk, struct pblk_line *line);
|
2018-10-09 11:12:03 +00:00
|
|
|
int pblk_line_smeta_read(struct pblk *pblk, struct pblk_line *line);
|
|
|
|
int pblk_line_emeta_read(struct pblk *pblk, struct pblk_line *line,
|
2017-06-26 09:57:17 +00:00
|
|
|
void *emeta_buf);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
int pblk_blk_erase_async(struct pblk *pblk, struct ppa_addr erase_ppa);
|
|
|
|
void pblk_line_put(struct kref *ref);
|
2017-10-13 12:46:23 +00:00
|
|
|
void pblk_line_put_wq(struct kref *ref);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
struct list_head *pblk_line_gc_list(struct pblk *pblk, struct pblk_line *line);
|
2017-06-26 09:57:17 +00:00
|
|
|
u64 pblk_lookup_page(struct pblk *pblk, struct pblk_line *line);
|
|
|
|
void pblk_dealloc_page(struct pblk *pblk, struct pblk_line *line, int nr_secs);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
u64 pblk_alloc_page(struct pblk *pblk, struct pblk_line *line, int nr_secs);
|
2017-06-26 09:57:17 +00:00
|
|
|
u64 __pblk_alloc_page(struct pblk *pblk, struct pblk_line *line, int nr_secs);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
int pblk_calc_secs(struct pblk *pblk, unsigned long secs_avail,
|
|
|
|
unsigned long secs_to_flush);
|
2018-10-09 11:11:51 +00:00
|
|
|
void pblk_down_rq(struct pblk *pblk, struct ppa_addr ppa,
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
unsigned long *lun_bitmap);
|
2018-10-09 11:11:51 +00:00
|
|
|
void pblk_down_chunk(struct pblk *pblk, struct ppa_addr ppa);
|
|
|
|
void pblk_up_chunk(struct pblk *pblk, struct ppa_addr ppa);
|
2018-10-09 11:11:48 +00:00
|
|
|
void pblk_up_rq(struct pblk *pblk, unsigned long *lun_bitmap);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
int pblk_bio_add_pages(struct pblk *pblk, struct bio *bio, gfp_t flags,
|
|
|
|
int nr_pages);
|
|
|
|
void pblk_bio_free_pages(struct pblk *pblk, struct bio *bio, int off,
|
|
|
|
int nr_pages);
|
|
|
|
void pblk_map_invalidate(struct pblk *pblk, struct ppa_addr ppa);
|
2017-06-26 09:57:19 +00:00
|
|
|
void __pblk_map_invalidate(struct pblk *pblk, struct pblk_line *line,
|
|
|
|
u64 paddr);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
void pblk_update_map(struct pblk *pblk, sector_t lba, struct ppa_addr ppa);
|
|
|
|
void pblk_update_map_cache(struct pblk *pblk, sector_t lba,
|
|
|
|
struct ppa_addr ppa);
|
|
|
|
void pblk_update_map_dev(struct pblk *pblk, sector_t lba,
|
|
|
|
struct ppa_addr ppa, struct ppa_addr entry_line);
|
|
|
|
int pblk_update_map_gc(struct pblk *pblk, sector_t lba, struct ppa_addr ppa,
|
2017-10-13 12:46:14 +00:00
|
|
|
struct pblk_line *gc_line, u64 paddr);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
void pblk_lookup_l2p_rand(struct pblk *pblk, struct ppa_addr *ppas,
|
|
|
|
u64 *lba_list, int nr_secs);
|
|
|
|
void pblk_lookup_l2p_seq(struct pblk *pblk, struct ppa_addr *ppas,
|
|
|
|
sector_t blba, int nr_secs);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* pblk user I/O write path
|
|
|
|
*/
|
|
|
|
int pblk_write_to_cache(struct pblk *pblk, struct bio *bio,
|
|
|
|
unsigned long flags);
|
2017-10-13 12:46:14 +00:00
|
|
|
int pblk_write_gc_to_cache(struct pblk *pblk, struct pblk_gc_rq *gc_rq);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* pblk map
|
|
|
|
*/
|
|
|
|
void pblk_map_erase_rq(struct pblk *pblk, struct nvm_rq *rqd,
|
|
|
|
unsigned int sentry, unsigned long *lun_bitmap,
|
|
|
|
unsigned int valid_secs, struct ppa_addr *erase_ppa);
|
|
|
|
void pblk_map_rq(struct pblk *pblk, struct nvm_rq *rqd, unsigned int sentry,
|
|
|
|
unsigned long *lun_bitmap, unsigned int valid_secs,
|
|
|
|
unsigned int off);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* pblk write thread
|
|
|
|
*/
|
|
|
|
int pblk_write_ts(void *data);
|
2017-10-18 04:10:19 +00:00
|
|
|
void pblk_write_timer_fn(struct timer_list *t);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
void pblk_write_should_kick(struct pblk *pblk);
|
2018-06-01 14:41:13 +00:00
|
|
|
void pblk_write_kick(struct pblk *pblk);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* pblk read path
|
|
|
|
*/
|
2018-05-20 22:25:50 +00:00
|
|
|
extern struct bio_set pblk_bio_set;
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
int pblk_submit_read(struct pblk *pblk, struct bio *bio);
|
2017-10-13 12:46:14 +00:00
|
|
|
int pblk_submit_read_gc(struct pblk *pblk, struct pblk_gc_rq *gc_rq);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
/*
|
|
|
|
* pblk recovery
|
|
|
|
*/
|
|
|
|
struct pblk_line *pblk_recov_l2p(struct pblk *pblk);
|
2017-06-26 09:57:29 +00:00
|
|
|
int pblk_recov_pad(struct pblk *pblk);
|
2018-01-05 13:16:08 +00:00
|
|
|
int pblk_recov_check_emeta(struct pblk *pblk, struct line_emeta *emeta);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* pblk gc
|
|
|
|
*/
|
lightnvm: pblk: redesign GC algorithm
At the moment, in order to get enough read parallelism, we have recycled
several lines at the same time. This approach has proven not to work
well when reaching capacity, since we end up mixing valid data from all
lines, thus not maintaining a sustainable free/recycled line ratio.
The new design, relies on a two level workqueue mechanism. In the first
level, we read the metadata for a number of lines based on the GC list
they reside on (this is governed by the number of valid sectors in each
line). In the second level, we recycle a single line at a time. Here, we
issue reads in parallel, while a single GC write thread places data in
the write buffer. This design allows to (i) only move data from one line
at a time, thus maintaining a sane free/recycled ration and (ii)
maintain the GC writer busy with recycled data.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2017-06-26 09:57:27 +00:00
|
|
|
#define PBLK_GC_MAX_READERS 8 /* Max number of outstanding GC reader jobs */
|
2017-10-13 12:46:11 +00:00
|
|
|
#define PBLK_GC_RQ_QD 128 /* Queue depth for inflight GC requests */
|
lightnvm: pblk: redesign GC algorithm
At the moment, in order to get enough read parallelism, we have recycled
several lines at the same time. This approach has proven not to work
well when reaching capacity, since we end up mixing valid data from all
lines, thus not maintaining a sustainable free/recycled line ratio.
The new design, relies on a two level workqueue mechanism. In the first
level, we read the metadata for a number of lines based on the GC list
they reside on (this is governed by the number of valid sectors in each
line). In the second level, we recycle a single line at a time. Here, we
issue reads in parallel, while a single GC write thread places data in
the write buffer. This design allows to (i) only move data from one line
at a time, thus maintaining a sane free/recycled ration and (ii)
maintain the GC writer busy with recycled data.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2017-06-26 09:57:27 +00:00
|
|
|
#define PBLK_GC_L_QD 4 /* Queue depth for inflight GC lines */
|
|
|
|
#define PBLK_GC_RSV_LINE 1 /* Reserved lines for GC */
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
|
|
|
int pblk_gc_init(struct pblk *pblk);
|
2018-06-01 13:04:24 +00:00
|
|
|
void pblk_gc_exit(struct pblk *pblk, bool graceful);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
void pblk_gc_should_start(struct pblk *pblk);
|
|
|
|
void pblk_gc_should_stop(struct pblk *pblk);
|
2017-10-13 12:46:37 +00:00
|
|
|
void pblk_gc_should_kick(struct pblk *pblk);
|
2017-10-13 12:46:36 +00:00
|
|
|
void pblk_gc_free_full_lines(struct pblk *pblk);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
void pblk_gc_sysfs_state_show(struct pblk *pblk, int *gc_enabled,
|
|
|
|
int *gc_active);
|
lightnvm: pblk: redesign GC algorithm
At the moment, in order to get enough read parallelism, we have recycled
several lines at the same time. This approach has proven not to work
well when reaching capacity, since we end up mixing valid data from all
lines, thus not maintaining a sustainable free/recycled line ratio.
The new design, relies on a two level workqueue mechanism. In the first
level, we read the metadata for a number of lines based on the GC list
they reside on (this is governed by the number of valid sectors in each
line). In the second level, we recycle a single line at a time. Here, we
issue reads in parallel, while a single GC write thread places data in
the write buffer. This design allows to (i) only move data from one line
at a time, thus maintaining a sane free/recycled ration and (ii)
maintain the GC writer busy with recycled data.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2017-06-26 09:57:27 +00:00
|
|
|
int pblk_gc_sysfs_force(struct pblk *pblk, int force);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* pblk rate limiter
|
|
|
|
*/
|
|
|
|
void pblk_rl_init(struct pblk_rl *rl, int budget);
|
|
|
|
void pblk_rl_free(struct pblk_rl *rl);
|
2017-10-13 12:46:37 +00:00
|
|
|
void pblk_rl_update_rates(struct pblk_rl *rl);
|
lightnvm: pblk: redesign GC algorithm
At the moment, in order to get enough read parallelism, we have recycled
several lines at the same time. This approach has proven not to work
well when reaching capacity, since we end up mixing valid data from all
lines, thus not maintaining a sustainable free/recycled line ratio.
The new design, relies on a two level workqueue mechanism. In the first
level, we read the metadata for a number of lines based on the GC list
they reside on (this is governed by the number of valid sectors in each
line). In the second level, we recycle a single line at a time. Here, we
issue reads in parallel, while a single GC write thread places data in
the write buffer. This design allows to (i) only move data from one line
at a time, thus maintaining a sane free/recycled ration and (ii)
maintain the GC writer busy with recycled data.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2017-06-26 09:57:27 +00:00
|
|
|
int pblk_rl_high_thrs(struct pblk_rl *rl);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
unsigned long pblk_rl_nr_free_blks(struct pblk_rl *rl);
|
2018-01-05 13:16:13 +00:00
|
|
|
unsigned long pblk_rl_nr_user_free_blks(struct pblk_rl *rl);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
int pblk_rl_user_may_insert(struct pblk_rl *rl, int nr_entries);
|
2017-06-26 09:57:29 +00:00
|
|
|
void pblk_rl_inserted(struct pblk_rl *rl, int nr_entries);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
void pblk_rl_user_in(struct pblk_rl *rl, int nr_entries);
|
|
|
|
int pblk_rl_gc_may_insert(struct pblk_rl *rl, int nr_entries);
|
|
|
|
void pblk_rl_gc_in(struct pblk_rl *rl, int nr_entries);
|
|
|
|
void pblk_rl_out(struct pblk_rl *rl, int nr_user, int nr_gc);
|
2017-10-13 12:46:05 +00:00
|
|
|
int pblk_rl_max_io(struct pblk_rl *rl);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
void pblk_rl_free_lines_inc(struct pblk_rl *rl, struct pblk_line *line);
|
2018-01-05 13:16:13 +00:00
|
|
|
void pblk_rl_free_lines_dec(struct pblk_rl *rl, struct pblk_line *line,
|
|
|
|
bool used);
|
2017-06-26 09:57:29 +00:00
|
|
|
int pblk_rl_is_limit(struct pblk_rl *rl);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
2018-06-01 14:41:06 +00:00
|
|
|
void pblk_rl_werr_line_in(struct pblk_rl *rl);
|
|
|
|
void pblk_rl_werr_line_out(struct pblk_rl *rl);
|
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
/*
|
|
|
|
* pblk sysfs
|
|
|
|
*/
|
|
|
|
int pblk_sysfs_init(struct gendisk *tdisk);
|
|
|
|
void pblk_sysfs_exit(struct gendisk *tdisk);
|
|
|
|
|
|
|
|
static inline void *pblk_malloc(size_t size, int type, gfp_t flags)
|
|
|
|
{
|
|
|
|
if (type == PBLK_KMALLOC_META)
|
|
|
|
return kmalloc(size, flags);
|
|
|
|
return vmalloc(size);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void pblk_mfree(void *ptr, int type)
|
|
|
|
{
|
|
|
|
if (type == PBLK_KMALLOC_META)
|
|
|
|
kfree(ptr);
|
|
|
|
else
|
|
|
|
vfree(ptr);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline struct nvm_rq *nvm_rq_from_c_ctx(void *c_ctx)
|
|
|
|
{
|
|
|
|
return c_ctx - sizeof(struct nvm_rq);
|
|
|
|
}
|
|
|
|
|
2017-06-26 09:57:17 +00:00
|
|
|
static inline void *emeta_to_bb(struct line_emeta *emeta)
|
|
|
|
{
|
|
|
|
return emeta->bb_bitmap;
|
|
|
|
}
|
|
|
|
|
2018-03-29 22:04:52 +00:00
|
|
|
static inline void *emeta_to_wa(struct pblk_line_meta *lm,
|
|
|
|
struct line_emeta *emeta)
|
|
|
|
{
|
|
|
|
return emeta->bb_bitmap + lm->blk_bitmap_len;
|
|
|
|
}
|
|
|
|
|
2017-06-26 09:57:17 +00:00
|
|
|
static inline void *emeta_to_lbas(struct pblk *pblk, struct line_emeta *emeta)
|
|
|
|
{
|
|
|
|
return ((void *)emeta + pblk->lm.emeta_len[1]);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void *emeta_to_vsc(struct pblk *pblk, struct line_emeta *emeta)
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
{
|
2017-06-26 09:57:17 +00:00
|
|
|
return (emeta_to_lbas(pblk, emeta) + pblk->lm.emeta_len[2]);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
}
|
|
|
|
|
lightnvm: pblk: redesign GC algorithm
At the moment, in order to get enough read parallelism, we have recycled
several lines at the same time. This approach has proven not to work
well when reaching capacity, since we end up mixing valid data from all
lines, thus not maintaining a sustainable free/recycled line ratio.
The new design, relies on a two level workqueue mechanism. In the first
level, we read the metadata for a number of lines based on the GC list
they reside on (this is governed by the number of valid sectors in each
line). In the second level, we recycle a single line at a time. Here, we
issue reads in parallel, while a single GC write thread places data in
the write buffer. This design allows to (i) only move data from one line
at a time, thus maintaining a sane free/recycled ration and (ii)
maintain the GC writer busy with recycled data.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2017-06-26 09:57:27 +00:00
|
|
|
static inline int pblk_line_vsc(struct pblk_line *line)
|
|
|
|
{
|
2017-10-13 12:46:14 +00:00
|
|
|
return le32_to_cpu(*line->vsc);
|
lightnvm: pblk: redesign GC algorithm
At the moment, in order to get enough read parallelism, we have recycled
several lines at the same time. This approach has proven not to work
well when reaching capacity, since we end up mixing valid data from all
lines, thus not maintaining a sustainable free/recycled line ratio.
The new design, relies on a two level workqueue mechanism. In the first
level, we read the metadata for a number of lines based on the GC list
they reside on (this is governed by the number of valid sectors in each
line). In the second level, we recycle a single line at a time. Here, we
issue reads in parallel, while a single GC write thread places data in
the write buffer. This design allows to (i) only move data from one line
at a time, thus maintaining a sane free/recycled ration and (ii)
maintain the GC writer busy with recycled data.
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
2017-06-26 09:57:27 +00:00
|
|
|
}
|
|
|
|
|
2018-10-09 11:11:42 +00:00
|
|
|
static inline int pblk_ppa_to_line_id(struct ppa_addr p)
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
{
|
2018-03-29 22:05:15 +00:00
|
|
|
return p.a.blk;
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
}
|
|
|
|
|
2018-10-09 11:11:42 +00:00
|
|
|
static inline struct pblk_line *pblk_ppa_to_line(struct pblk *pblk,
|
|
|
|
struct ppa_addr p)
|
|
|
|
{
|
|
|
|
return &pblk->lines[pblk_ppa_to_line_id(p)];
|
|
|
|
}
|
|
|
|
|
2018-01-05 13:16:06 +00:00
|
|
|
static inline int pblk_ppa_to_pos(struct nvm_geo *geo, struct ppa_addr p)
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
{
|
2018-03-29 22:05:15 +00:00
|
|
|
return p.a.lun * geo->num_ch + p.a.ch;
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
}
|
|
|
|
|
2018-01-05 13:16:06 +00:00
|
|
|
static inline struct ppa_addr addr_to_gen_ppa(struct pblk *pblk, u64 paddr,
|
|
|
|
u64 line_id)
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
{
|
2018-03-29 22:05:21 +00:00
|
|
|
struct nvm_tgt_dev *dev = pblk->dev;
|
|
|
|
struct nvm_geo *geo = &dev->geo;
|
2018-01-05 13:16:06 +00:00
|
|
|
struct ppa_addr ppa;
|
|
|
|
|
2018-03-29 22:05:21 +00:00
|
|
|
if (geo->version == NVM_OCSSD_SPEC_12) {
|
|
|
|
struct nvm_addrf_12 *ppaf = (struct nvm_addrf_12 *)&pblk->addrf;
|
|
|
|
|
|
|
|
ppa.ppa = 0;
|
|
|
|
ppa.g.blk = line_id;
|
|
|
|
ppa.g.pg = (paddr & ppaf->pg_mask) >> ppaf->pg_offset;
|
|
|
|
ppa.g.lun = (paddr & ppaf->lun_mask) >> ppaf->lun_offset;
|
|
|
|
ppa.g.ch = (paddr & ppaf->ch_mask) >> ppaf->ch_offset;
|
|
|
|
ppa.g.pl = (paddr & ppaf->pln_mask) >> ppaf->pln_offset;
|
|
|
|
ppa.g.sec = (paddr & ppaf->sec_mask) >> ppaf->sec_offset;
|
|
|
|
} else {
|
|
|
|
struct pblk_addrf *uaddrf = &pblk->uaddrf;
|
|
|
|
int secs, chnls, luns;
|
|
|
|
|
|
|
|
ppa.ppa = 0;
|
|
|
|
|
|
|
|
ppa.m.chk = line_id;
|
|
|
|
|
|
|
|
paddr = div_u64_rem(paddr, uaddrf->sec_stripe, &secs);
|
|
|
|
ppa.m.sec = secs;
|
|
|
|
|
|
|
|
paddr = div_u64_rem(paddr, uaddrf->ch_stripe, &chnls);
|
|
|
|
ppa.m.grp = chnls;
|
|
|
|
|
|
|
|
paddr = div_u64_rem(paddr, uaddrf->lun_stripe, &luns);
|
|
|
|
ppa.m.pu = luns;
|
|
|
|
|
|
|
|
ppa.m.sec += uaddrf->sec_stripe * paddr;
|
|
|
|
}
|
2018-01-05 13:16:06 +00:00
|
|
|
|
|
|
|
return ppa;
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
}
|
|
|
|
|
2018-10-09 11:11:41 +00:00
|
|
|
static inline struct nvm_chk_meta *pblk_dev_ppa_to_chunk(struct pblk *pblk,
|
|
|
|
struct ppa_addr p)
|
|
|
|
{
|
|
|
|
struct nvm_tgt_dev *dev = pblk->dev;
|
|
|
|
struct nvm_geo *geo = &dev->geo;
|
2018-10-09 11:11:42 +00:00
|
|
|
struct pblk_line *line = pblk_ppa_to_line(pblk, p);
|
2018-10-09 11:11:41 +00:00
|
|
|
int pos = pblk_ppa_to_pos(geo, p);
|
|
|
|
|
|
|
|
return &line->chks[pos];
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline u64 pblk_dev_ppa_to_chunk_addr(struct pblk *pblk,
|
|
|
|
struct ppa_addr p)
|
|
|
|
{
|
|
|
|
struct nvm_tgt_dev *dev = pblk->dev;
|
|
|
|
|
|
|
|
return dev_to_chunk_addr(dev->parent, &pblk->addrf, p);
|
|
|
|
}
|
|
|
|
|
2018-01-05 13:16:06 +00:00
|
|
|
static inline u64 pblk_dev_ppa_to_line_addr(struct pblk *pblk,
|
|
|
|
struct ppa_addr p)
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
{
|
2018-03-29 22:05:21 +00:00
|
|
|
struct nvm_tgt_dev *dev = pblk->dev;
|
|
|
|
struct nvm_geo *geo = &dev->geo;
|
2018-01-05 13:16:06 +00:00
|
|
|
u64 paddr;
|
|
|
|
|
2018-03-29 22:05:21 +00:00
|
|
|
if (geo->version == NVM_OCSSD_SPEC_12) {
|
|
|
|
struct nvm_addrf_12 *ppaf = (struct nvm_addrf_12 *)&pblk->addrf;
|
|
|
|
|
|
|
|
paddr = (u64)p.g.ch << ppaf->ch_offset;
|
|
|
|
paddr |= (u64)p.g.lun << ppaf->lun_offset;
|
|
|
|
paddr |= (u64)p.g.pg << ppaf->pg_offset;
|
|
|
|
paddr |= (u64)p.g.pl << ppaf->pln_offset;
|
|
|
|
paddr |= (u64)p.g.sec << ppaf->sec_offset;
|
|
|
|
} else {
|
|
|
|
struct pblk_addrf *uaddrf = &pblk->uaddrf;
|
|
|
|
u64 secs = p.m.sec;
|
|
|
|
int sec_stripe;
|
|
|
|
|
|
|
|
paddr = (u64)p.m.grp * uaddrf->sec_stripe;
|
|
|
|
paddr += (u64)p.m.pu * uaddrf->sec_lun_stripe;
|
|
|
|
|
|
|
|
secs = div_u64_rem(secs, uaddrf->sec_stripe, &sec_stripe);
|
|
|
|
paddr += secs * uaddrf->sec_ws_stripe;
|
|
|
|
paddr += sec_stripe;
|
|
|
|
}
|
2018-01-05 13:16:06 +00:00
|
|
|
|
|
|
|
return paddr;
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static inline struct ppa_addr pblk_ppa32_to_ppa64(struct pblk *pblk, u32 ppa32)
|
|
|
|
{
|
2018-10-09 11:11:56 +00:00
|
|
|
struct nvm_tgt_dev *dev = pblk->dev;
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
2018-10-09 11:11:56 +00:00
|
|
|
return nvm_ppa32_to_ppa64(dev->parent, &pblk->addrf, ppa32);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static inline u32 pblk_ppa64_to_ppa32(struct pblk *pblk, struct ppa_addr ppa64)
|
|
|
|
{
|
2018-10-09 11:11:56 +00:00
|
|
|
struct nvm_tgt_dev *dev = pblk->dev;
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
2018-10-09 11:11:56 +00:00
|
|
|
return nvm_ppa64_to_ppa32(dev->parent, &pblk->addrf, ppa64);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
}
|
|
|
|
|
2018-01-05 13:16:06 +00:00
|
|
|
static inline struct ppa_addr pblk_trans_map_get(struct pblk *pblk,
|
|
|
|
sector_t lba)
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
{
|
2018-01-05 13:16:06 +00:00
|
|
|
struct ppa_addr ppa;
|
|
|
|
|
2018-03-29 22:05:19 +00:00
|
|
|
if (pblk->addrf_len < 32) {
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
u32 *map = (u32 *)pblk->trans_map;
|
|
|
|
|
2018-01-05 13:16:06 +00:00
|
|
|
ppa = pblk_ppa32_to_ppa64(pblk, map[lba]);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
} else {
|
2018-01-05 13:16:06 +00:00
|
|
|
struct ppa_addr *map = (struct ppa_addr *)pblk->trans_map;
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
2018-01-05 13:16:06 +00:00
|
|
|
ppa = map[lba];
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
}
|
2018-01-05 13:16:06 +00:00
|
|
|
|
|
|
|
return ppa;
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
}
|
|
|
|
|
2018-01-05 13:16:06 +00:00
|
|
|
static inline void pblk_trans_map_set(struct pblk *pblk, sector_t lba,
|
|
|
|
struct ppa_addr ppa)
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
{
|
2018-03-29 22:05:19 +00:00
|
|
|
if (pblk->addrf_len < 32) {
|
2018-01-05 13:16:06 +00:00
|
|
|
u32 *map = (u32 *)pblk->trans_map;
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
2018-01-05 13:16:06 +00:00
|
|
|
map[lba] = pblk_ppa64_to_ppa32(pblk, ppa);
|
|
|
|
} else {
|
|
|
|
u64 *map = (u64 *)pblk->trans_map;
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
|
2018-01-05 13:16:06 +00:00
|
|
|
map[lba] = ppa.ppa;
|
|
|
|
}
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static inline int pblk_ppa_empty(struct ppa_addr ppa_addr)
|
|
|
|
{
|
|
|
|
return (ppa_addr.ppa == ADDR_EMPTY);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void pblk_ppa_set_empty(struct ppa_addr *ppa_addr)
|
|
|
|
{
|
|
|
|
ppa_addr->ppa = ADDR_EMPTY;
|
|
|
|
}
|
|
|
|
|
2017-06-30 15:56:42 +00:00
|
|
|
static inline bool pblk_ppa_comp(struct ppa_addr lppa, struct ppa_addr rppa)
|
|
|
|
{
|
2018-01-05 13:16:21 +00:00
|
|
|
return (lppa.ppa == rppa.ppa);
|
2017-06-30 15:56:42 +00:00
|
|
|
}
|
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
static inline int pblk_addr_in_cache(struct ppa_addr ppa)
|
|
|
|
{
|
|
|
|
return (ppa.ppa != ADDR_EMPTY && ppa.c.is_cached);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline int pblk_addr_to_cacheline(struct ppa_addr ppa)
|
|
|
|
{
|
|
|
|
return ppa.c.line;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline struct ppa_addr pblk_cacheline_to_addr(int addr)
|
|
|
|
{
|
|
|
|
struct ppa_addr p;
|
|
|
|
|
|
|
|
p.c.line = addr;
|
|
|
|
p.c.is_cached = 1;
|
|
|
|
|
|
|
|
return p;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline u32 pblk_calc_meta_header_crc(struct pblk *pblk,
|
2017-06-26 09:57:17 +00:00
|
|
|
struct line_header *header)
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
{
|
|
|
|
u32 crc = ~(u32)0;
|
|
|
|
|
2017-06-26 09:57:17 +00:00
|
|
|
crc = crc32_le(crc, (unsigned char *)header + sizeof(crc),
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
sizeof(struct line_header) - sizeof(crc));
|
|
|
|
|
|
|
|
return crc;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline u32 pblk_calc_smeta_crc(struct pblk *pblk,
|
|
|
|
struct line_smeta *smeta)
|
|
|
|
{
|
|
|
|
struct pblk_line_meta *lm = &pblk->lm;
|
|
|
|
u32 crc = ~(u32)0;
|
|
|
|
|
|
|
|
crc = crc32_le(crc, (unsigned char *)smeta +
|
|
|
|
sizeof(struct line_header) + sizeof(crc),
|
|
|
|
lm->smeta_len -
|
|
|
|
sizeof(struct line_header) - sizeof(crc));
|
|
|
|
|
|
|
|
return crc;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline u32 pblk_calc_emeta_crc(struct pblk *pblk,
|
|
|
|
struct line_emeta *emeta)
|
|
|
|
{
|
|
|
|
struct pblk_line_meta *lm = &pblk->lm;
|
|
|
|
u32 crc = ~(u32)0;
|
|
|
|
|
|
|
|
crc = crc32_le(crc, (unsigned char *)emeta +
|
|
|
|
sizeof(struct line_header) + sizeof(crc),
|
2017-06-26 09:57:17 +00:00
|
|
|
lm->emeta_len[0] -
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
sizeof(struct line_header) - sizeof(crc));
|
|
|
|
|
|
|
|
return crc;
|
|
|
|
}
|
|
|
|
|
2017-06-26 09:57:20 +00:00
|
|
|
static inline int pblk_io_aligned(struct pblk *pblk, int nr_secs)
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
{
|
2017-06-26 09:57:20 +00:00
|
|
|
return !(nr_secs % pblk->min_write_pgs);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
}
|
|
|
|
|
2018-07-13 08:48:37 +00:00
|
|
|
#ifdef CONFIG_NVM_PBLK_DEBUG
|
2018-07-13 08:48:42 +00:00
|
|
|
static inline void print_ppa(struct pblk *pblk, struct ppa_addr *p,
|
2018-03-29 22:05:21 +00:00
|
|
|
char *msg, int error)
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
{
|
2018-07-13 08:48:42 +00:00
|
|
|
struct nvm_geo *geo = &pblk->dev->geo;
|
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
if (p->c.is_cached) {
|
2018-07-13 08:48:42 +00:00
|
|
|
pblk_err(pblk, "ppa: (%s: %x) cache line: %llu\n",
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
msg, error, (u64)p->c.line);
|
2018-03-29 22:05:21 +00:00
|
|
|
} else if (geo->version == NVM_OCSSD_SPEC_12) {
|
2018-07-13 08:48:42 +00:00
|
|
|
pblk_err(pblk, "ppa: (%s: %x):ch:%d,lun:%d,blk:%d,pg:%d,pl:%d,sec:%d\n",
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
msg, error,
|
|
|
|
p->g.ch, p->g.lun, p->g.blk,
|
|
|
|
p->g.pg, p->g.pl, p->g.sec);
|
2018-03-29 22:05:21 +00:00
|
|
|
} else {
|
2018-07-13 08:48:42 +00:00
|
|
|
pblk_err(pblk, "ppa: (%s: %x):ch:%d,lun:%d,chk:%d,sec:%d\n",
|
2018-03-29 22:05:21 +00:00
|
|
|
msg, error,
|
|
|
|
p->m.grp, p->m.pu, p->m.chk, p->m.sec);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void pblk_print_failed_rqd(struct pblk *pblk, struct nvm_rq *rqd,
|
|
|
|
int error)
|
|
|
|
{
|
|
|
|
int bit = -1;
|
|
|
|
|
|
|
|
if (rqd->nr_ppas == 1) {
|
2018-07-13 08:48:42 +00:00
|
|
|
print_ppa(pblk, &rqd->ppa_addr, "rqd", error);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
while ((bit = find_next_bit((void *)&rqd->ppa_status, rqd->nr_ppas,
|
|
|
|
bit + 1)) < rqd->nr_ppas) {
|
2018-07-13 08:48:42 +00:00
|
|
|
print_ppa(pblk, &rqd->ppa_list[bit], "rqd", error);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
}
|
|
|
|
|
2018-07-13 08:48:42 +00:00
|
|
|
pblk_err(pblk, "error:%d, ppa_status:%llx\n", error, rqd->ppa_status);
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static inline int pblk_boundary_ppa_checks(struct nvm_tgt_dev *tgt_dev,
|
|
|
|
struct ppa_addr *ppas, int nr_ppas)
|
|
|
|
{
|
|
|
|
struct nvm_geo *geo = &tgt_dev->geo;
|
|
|
|
struct ppa_addr *ppa;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
for (i = 0; i < nr_ppas; i++) {
|
|
|
|
ppa = &ppas[i];
|
|
|
|
|
2018-03-29 22:05:21 +00:00
|
|
|
if (geo->version == NVM_OCSSD_SPEC_12) {
|
|
|
|
if (!ppa->c.is_cached &&
|
|
|
|
ppa->g.ch < geo->num_ch &&
|
|
|
|
ppa->g.lun < geo->num_lun &&
|
|
|
|
ppa->g.pl < geo->num_pln &&
|
|
|
|
ppa->g.blk < geo->num_chk &&
|
|
|
|
ppa->g.pg < geo->num_pg &&
|
|
|
|
ppa->g.sec < geo->ws_min)
|
|
|
|
continue;
|
|
|
|
} else {
|
|
|
|
if (!ppa->c.is_cached &&
|
|
|
|
ppa->m.grp < geo->num_ch &&
|
|
|
|
ppa->m.pu < geo->num_lun &&
|
|
|
|
ppa->m.chk < geo->num_chk &&
|
|
|
|
ppa->m.sec < geo->clba)
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
2018-07-13 08:48:42 +00:00
|
|
|
print_ppa(tgt_dev->q->queuedata, ppa, "boundary", i);
|
2017-10-13 12:46:47 +00:00
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2017-10-13 12:46:47 +00:00
|
|
|
static inline int pblk_check_io(struct pblk *pblk, struct nvm_rq *rqd)
|
|
|
|
{
|
|
|
|
struct nvm_tgt_dev *dev = pblk->dev;
|
2018-10-09 11:11:46 +00:00
|
|
|
struct ppa_addr *ppa_list = nvm_rq_to_ppa_list(rqd);
|
2017-10-13 12:46:47 +00:00
|
|
|
|
|
|
|
if (pblk_boundary_ppa_checks(dev, ppa_list, rqd->nr_ppas)) {
|
|
|
|
WARN_ON(1);
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (rqd->opcode == NVM_OP_PWRITE) {
|
|
|
|
struct pblk_line *line;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
for (i = 0; i < rqd->nr_ppas; i++) {
|
2018-10-09 11:11:42 +00:00
|
|
|
line = pblk_ppa_to_line(pblk, ppa_list[i]);
|
2017-10-13 12:46:47 +00:00
|
|
|
|
|
|
|
spin_lock(&line->lock);
|
|
|
|
if (line->state != PBLK_LINESTATE_OPEN) {
|
2018-07-13 08:48:42 +00:00
|
|
|
pblk_err(pblk, "bad ppa: line:%d,state:%d\n",
|
2017-10-13 12:46:47 +00:00
|
|
|
line->id, line->state);
|
|
|
|
WARN_ON(1);
|
|
|
|
spin_unlock(&line->lock);
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
spin_unlock(&line->lock);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
static inline int pblk_boundary_paddr_checks(struct pblk *pblk, u64 paddr)
|
|
|
|
{
|
|
|
|
struct pblk_line_meta *lm = &pblk->lm;
|
|
|
|
|
|
|
|
if (paddr > lm->sec_per_line)
|
|
|
|
return 1;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline unsigned int pblk_get_bi_idx(struct bio *bio)
|
|
|
|
{
|
|
|
|
return bio->bi_iter.bi_idx;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline sector_t pblk_get_lba(struct bio *bio)
|
|
|
|
{
|
|
|
|
return bio->bi_iter.bi_sector / NR_PHY_IN_LOG;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline unsigned int pblk_get_secs(struct bio *bio)
|
|
|
|
{
|
|
|
|
return bio->bi_iter.bi_size / PBLK_EXPOSED_PAGE_SIZE;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void pblk_setup_uuid(struct pblk *pblk)
|
|
|
|
{
|
|
|
|
uuid_le uuid;
|
|
|
|
|
|
|
|
uuid_le_gen(&uuid);
|
|
|
|
memcpy(pblk->instance_uuid, uuid.b, 16);
|
|
|
|
}
|
2018-10-09 11:11:52 +00:00
|
|
|
|
|
|
|
static inline char *pblk_disk_name(struct pblk *pblk)
|
|
|
|
{
|
|
|
|
struct gendisk *disk = pblk->disk;
|
|
|
|
|
|
|
|
return disk->disk_name;
|
|
|
|
}
|
lightnvm: physical block device (pblk) target
This patch introduces pblk, a host-side translation layer for
Open-Channel SSDs to expose them like block devices. The translation
layer allows data placement decisions, and I/O scheduling to be
managed by the host, enabling users to optimize the SSD for their
specific workloads.
An open-channel SSD has a set of LUNs (parallel units) and a
collection of blocks. Each block can be read in any order, but
writes must be sequential. Writes may also fail, and if a block
requires it, must also be reset before new writes can be
applied.
To manage the constraints, pblk maintains a logical to
physical address (L2P) table, write cache, garbage
collection logic, recovery scheme, and logic to rate-limit
user I/Os versus garbage collection I/Os.
The L2P table is fully-associative and manages sectors at a
4KB granularity. Pblk stores the L2P table in two places, in
the out-of-band area of the media and on the last page of a
line. In the cause of a power failure, pblk will perform a
scan to recover the L2P table.
The user data is organized into lines. A line is data
striped across blocks and LUNs. The lines enable the host to
reduce the amount of metadata to maintain besides the user
data and makes it easier to implement RAID or erasure coding
in the future.
pblk implements multi-tenant support and can be instantiated
multiple times on the same drive. Each instance owns a
portion of the SSD - both regarding I/O bandwidth and
capacity - providing I/O isolation for each case.
Finally, pblk also exposes a sysfs interface that allows
user-space to peek into the internals of pblk. The interface
is available at /dev/block/*/pblk/ where * is the block
device name exposed.
This work also contains contributions from:
Matias Bjørling <matias@cnexlabs.com>
Simon A. F. Lund <slund@cnexlabs.com>
Young Tack Jin <youngtack.jin@gmail.com>
Huaicheng Li <huaicheng@cs.uchicago.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Signed-off-by: Matias Bjørling <matias@cnexlabs.com>
Signed-off-by: Jens Axboe <axboe@fb.com>
2017-04-15 18:55:50 +00:00
|
|
|
#endif /* PBLK_H_ */
|