2021-04-26 06:27:21 +00:00
|
|
|
/* SPDX-License-Identifier: GPL-2.0 */
|
|
|
|
|
|
|
|
#ifndef __BTRFS_ZONED_H__
|
|
|
|
#define __BTRFS_ZONED_H__
|
|
|
|
|
|
|
|
#include "kerncompat.h"
|
|
|
|
#include <stdbool.h>
|
btrfs-progs: zoned: implement log-structured superblock
Superblock (and its copies) is the only data structure in btrfs which has a
fixed location on a device. Since we cannot overwrite in a sequential write
required zone, we cannot place superblock in the zone. One easy solution
is limiting superblock and copies to be placed only in conventional zones.
However, this method has two downsides: one is reduced number of superblock
copies. The location of the second copy of superblock is 256GB, which is in
a sequential write required zone on typical devices in the market today.
So, the number of superblock and copies is limited to be two. Second
downside is that we cannot support devices which have no conventional zones
at all.
To solve these two problems, we employ superblock log writing. It uses two
adjacent zones as a circular buffer to write updated superblocks. Once the
first zone is filled up, start writing into the second one. Then, when
both zones are filled up and before starting to write to the first zone
again, reset the first zone.
We can determine the position of the latest superblock by reading write
pointer information from a device. One corner case is when both zones are
full. For this situation, we read out the last superblock of each zone, and
compare them to determine which zone is older.
The following zones are reserved as the circular buffer on ZONED btrfs.
- primary superblock: offset 0B (and the following zone)
- first copy: offset 512G (and the following zone)
- Second copy: offset 4T (4096G, and the following zone)
If these reserved zones are conventional, superblock is written fixed at
the start of the zone without logging.
Currently, superblock reading/writing is done by pread/pwrite. This
commit replace the call sites with sbread/sbwrite to wrap the functions.
For zoned btrfs, btrfs_sb_io which is called from sbread/sbwrite
reverses the IO position back to a mirror number, maps the mirror number
into the superblock logging position, and do the IO.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
Signed-off-by: David Sterba <dsterba@suse.com>
2021-04-26 06:27:26 +00:00
|
|
|
#include "kernel-shared/disk-io.h"
|
2021-04-26 06:27:27 +00:00
|
|
|
#include "kernel-shared/volumes.h"
|
2021-04-26 06:27:21 +00:00
|
|
|
|
|
|
|
#ifdef BTRFS_ZONED
|
|
|
|
#include <linux/blkzoned.h>
|
|
|
|
#else
|
|
|
|
|
|
|
|
struct blk_zone {
|
|
|
|
int dummy;
|
|
|
|
};
|
|
|
|
|
|
|
|
#endif /* BTRFS_ZONED */
|
|
|
|
|
btrfs-progs: zoned: implement log-structured superblock
Superblock (and its copies) is the only data structure in btrfs which has a
fixed location on a device. Since we cannot overwrite in a sequential write
required zone, we cannot place superblock in the zone. One easy solution
is limiting superblock and copies to be placed only in conventional zones.
However, this method has two downsides: one is reduced number of superblock
copies. The location of the second copy of superblock is 256GB, which is in
a sequential write required zone on typical devices in the market today.
So, the number of superblock and copies is limited to be two. Second
downside is that we cannot support devices which have no conventional zones
at all.
To solve these two problems, we employ superblock log writing. It uses two
adjacent zones as a circular buffer to write updated superblocks. Once the
first zone is filled up, start writing into the second one. Then, when
both zones are filled up and before starting to write to the first zone
again, reset the first zone.
We can determine the position of the latest superblock by reading write
pointer information from a device. One corner case is when both zones are
full. For this situation, we read out the last superblock of each zone, and
compare them to determine which zone is older.
The following zones are reserved as the circular buffer on ZONED btrfs.
- primary superblock: offset 0B (and the following zone)
- first copy: offset 512G (and the following zone)
- Second copy: offset 4T (4096G, and the following zone)
If these reserved zones are conventional, superblock is written fixed at
the start of the zone without logging.
Currently, superblock reading/writing is done by pread/pwrite. This
commit replace the call sites with sbread/sbwrite to wrap the functions.
For zoned btrfs, btrfs_sb_io which is called from sbread/sbwrite
reverses the IO position back to a mirror number, maps the mirror number
into the superblock logging position, and do the IO.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
Signed-off-by: David Sterba <dsterba@suse.com>
2021-04-26 06:27:26 +00:00
|
|
|
/* Number of superblock log zones */
|
|
|
|
#define BTRFS_NR_SB_LOG_ZONES 2
|
|
|
|
|
2021-04-26 06:27:21 +00:00
|
|
|
/*
|
|
|
|
* Zoned block device models
|
|
|
|
*/
|
|
|
|
enum btrfs_zoned_model {
|
|
|
|
ZONED_NONE,
|
|
|
|
ZONED_HOST_AWARE,
|
|
|
|
ZONED_HOST_MANAGED,
|
|
|
|
};
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Zone information for a zoned block device.
|
|
|
|
*/
|
|
|
|
struct btrfs_zoned_device_info {
|
|
|
|
enum btrfs_zoned_model model;
|
|
|
|
u64 zone_size;
|
2021-04-26 06:27:23 +00:00
|
|
|
u64 max_zone_append_size;
|
2021-04-26 06:27:21 +00:00
|
|
|
u32 nr_zones;
|
|
|
|
struct blk_zone *zones;
|
|
|
|
};
|
|
|
|
|
|
|
|
enum btrfs_zoned_model zoned_model(const char *file);
|
|
|
|
u64 zone_size(const char *file);
|
|
|
|
int btrfs_get_zone_info(int fd, const char *file,
|
|
|
|
struct btrfs_zoned_device_info **zinfo);
|
|
|
|
int btrfs_get_dev_zone_info_all_devices(struct btrfs_fs_info *fs_info);
|
2021-04-26 06:27:22 +00:00
|
|
|
int btrfs_check_zoned_mode(struct btrfs_fs_info *fs_info);
|
2021-04-26 06:27:21 +00:00
|
|
|
|
btrfs-progs: zoned: implement log-structured superblock
Superblock (and its copies) is the only data structure in btrfs which has a
fixed location on a device. Since we cannot overwrite in a sequential write
required zone, we cannot place superblock in the zone. One easy solution
is limiting superblock and copies to be placed only in conventional zones.
However, this method has two downsides: one is reduced number of superblock
copies. The location of the second copy of superblock is 256GB, which is in
a sequential write required zone on typical devices in the market today.
So, the number of superblock and copies is limited to be two. Second
downside is that we cannot support devices which have no conventional zones
at all.
To solve these two problems, we employ superblock log writing. It uses two
adjacent zones as a circular buffer to write updated superblocks. Once the
first zone is filled up, start writing into the second one. Then, when
both zones are filled up and before starting to write to the first zone
again, reset the first zone.
We can determine the position of the latest superblock by reading write
pointer information from a device. One corner case is when both zones are
full. For this situation, we read out the last superblock of each zone, and
compare them to determine which zone is older.
The following zones are reserved as the circular buffer on ZONED btrfs.
- primary superblock: offset 0B (and the following zone)
- first copy: offset 512G (and the following zone)
- Second copy: offset 4T (4096G, and the following zone)
If these reserved zones are conventional, superblock is written fixed at
the start of the zone without logging.
Currently, superblock reading/writing is done by pread/pwrite. This
commit replace the call sites with sbread/sbwrite to wrap the functions.
For zoned btrfs, btrfs_sb_io which is called from sbread/sbwrite
reverses the IO position back to a mirror number, maps the mirror number
into the superblock logging position, and do the IO.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
Signed-off-by: David Sterba <dsterba@suse.com>
2021-04-26 06:27:26 +00:00
|
|
|
#ifdef BTRFS_ZONED
|
|
|
|
size_t btrfs_sb_io(int fd, void *buf, off_t offset, int rw);
|
|
|
|
|
|
|
|
static inline size_t sbread(int fd, void *buf, off_t offset)
|
|
|
|
{
|
|
|
|
return btrfs_sb_io(fd, buf, offset, READ);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline size_t sbwrite(int fd, void *buf, off_t offset)
|
|
|
|
{
|
|
|
|
return btrfs_sb_io(fd, buf, offset, WRITE);
|
|
|
|
}
|
|
|
|
|
2021-04-26 06:27:27 +00:00
|
|
|
static inline bool zone_is_sequential(struct btrfs_zoned_device_info *zinfo,
|
|
|
|
u64 bytenr)
|
|
|
|
{
|
|
|
|
unsigned int zno;
|
|
|
|
|
|
|
|
if (!zinfo || zinfo->model == ZONED_NONE)
|
|
|
|
return false;
|
|
|
|
|
|
|
|
zno = bytenr / zinfo->zone_size;
|
|
|
|
return zinfo->zones[zno].type == BLK_ZONE_TYPE_SEQWRITE_REQ;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline bool btrfs_dev_is_empty_zone(struct btrfs_device *device, u64 pos)
|
|
|
|
{
|
|
|
|
struct btrfs_zoned_device_info *zinfo = device->zone_info;
|
|
|
|
unsigned int zno;
|
|
|
|
|
|
|
|
if (!zone_is_sequential(zinfo, pos))
|
|
|
|
return true;
|
|
|
|
|
|
|
|
zno = pos / zinfo->zone_size;
|
|
|
|
return zinfo->zones[zno].cond == BLK_ZONE_COND_EMPTY;
|
|
|
|
}
|
|
|
|
|
btrfs-progs: zoned: implement log-structured superblock
Superblock (and its copies) is the only data structure in btrfs which has a
fixed location on a device. Since we cannot overwrite in a sequential write
required zone, we cannot place superblock in the zone. One easy solution
is limiting superblock and copies to be placed only in conventional zones.
However, this method has two downsides: one is reduced number of superblock
copies. The location of the second copy of superblock is 256GB, which is in
a sequential write required zone on typical devices in the market today.
So, the number of superblock and copies is limited to be two. Second
downside is that we cannot support devices which have no conventional zones
at all.
To solve these two problems, we employ superblock log writing. It uses two
adjacent zones as a circular buffer to write updated superblocks. Once the
first zone is filled up, start writing into the second one. Then, when
both zones are filled up and before starting to write to the first zone
again, reset the first zone.
We can determine the position of the latest superblock by reading write
pointer information from a device. One corner case is when both zones are
full. For this situation, we read out the last superblock of each zone, and
compare them to determine which zone is older.
The following zones are reserved as the circular buffer on ZONED btrfs.
- primary superblock: offset 0B (and the following zone)
- first copy: offset 512G (and the following zone)
- Second copy: offset 4T (4096G, and the following zone)
If these reserved zones are conventional, superblock is written fixed at
the start of the zone without logging.
Currently, superblock reading/writing is done by pread/pwrite. This
commit replace the call sites with sbread/sbwrite to wrap the functions.
For zoned btrfs, btrfs_sb_io which is called from sbread/sbwrite
reverses the IO position back to a mirror number, maps the mirror number
into the superblock logging position, and do the IO.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
Signed-off-by: David Sterba <dsterba@suse.com>
2021-04-26 06:27:26 +00:00
|
|
|
int btrfs_reset_dev_zone(int fd, struct blk_zone *zone);
|
2021-04-26 06:27:27 +00:00
|
|
|
u64 btrfs_find_allocatable_zones(struct btrfs_device *device, u64 hole_start,
|
|
|
|
u64 hole_end, u64 num_bytes);
|
2021-04-26 06:27:28 +00:00
|
|
|
int btrfs_load_block_group_zone_info(struct btrfs_fs_info *fs_info,
|
|
|
|
struct btrfs_block_group *cache);
|
btrfs-progs: zoned: implement log-structured superblock
Superblock (and its copies) is the only data structure in btrfs which has a
fixed location on a device. Since we cannot overwrite in a sequential write
required zone, we cannot place superblock in the zone. One easy solution
is limiting superblock and copies to be placed only in conventional zones.
However, this method has two downsides: one is reduced number of superblock
copies. The location of the second copy of superblock is 256GB, which is in
a sequential write required zone on typical devices in the market today.
So, the number of superblock and copies is limited to be two. Second
downside is that we cannot support devices which have no conventional zones
at all.
To solve these two problems, we employ superblock log writing. It uses two
adjacent zones as a circular buffer to write updated superblocks. Once the
first zone is filled up, start writing into the second one. Then, when
both zones are filled up and before starting to write to the first zone
again, reset the first zone.
We can determine the position of the latest superblock by reading write
pointer information from a device. One corner case is when both zones are
full. For this situation, we read out the last superblock of each zone, and
compare them to determine which zone is older.
The following zones are reserved as the circular buffer on ZONED btrfs.
- primary superblock: offset 0B (and the following zone)
- first copy: offset 512G (and the following zone)
- Second copy: offset 4T (4096G, and the following zone)
If these reserved zones are conventional, superblock is written fixed at
the start of the zone without logging.
Currently, superblock reading/writing is done by pread/pwrite. This
commit replace the call sites with sbread/sbwrite to wrap the functions.
For zoned btrfs, btrfs_sb_io which is called from sbread/sbwrite
reverses the IO position back to a mirror number, maps the mirror number
into the superblock logging position, and do the IO.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
Signed-off-by: David Sterba <dsterba@suse.com>
2021-04-26 06:27:26 +00:00
|
|
|
|
|
|
|
#else
|
|
|
|
|
|
|
|
#define sbread(fd, buf, offset) \
|
|
|
|
pread64(fd, buf, BTRFS_SUPER_INFO_SIZE, offset)
|
|
|
|
#define sbwrite(fd, buf, offset) \
|
|
|
|
pwrite64(fd, buf, BTRFS_SUPER_INFO_SIZE, offset)
|
|
|
|
|
|
|
|
static inline int btrfs_reset_dev_zone(int fd, struct blk_zone *zone)
|
|
|
|
{
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2021-04-26 06:27:27 +00:00
|
|
|
static inline bool zone_is_sequential(struct btrfs_zoned_device_info *zinfo,
|
|
|
|
u64 bytenr)
|
|
|
|
{
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline u64 btrfs_find_allocatable_zones(struct btrfs_device *device,
|
|
|
|
u64 hole_start, u64 hole_end,
|
|
|
|
u64 num_bytes)
|
|
|
|
{
|
|
|
|
return hole_start;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline bool btrfs_dev_is_empty_zone(struct btrfs_device *device, u64 pos)
|
|
|
|
{
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2021-04-26 06:27:28 +00:00
|
|
|
static inline int btrfs_load_block_group_zone_info(
|
|
|
|
struct btrfs_fs_info *fs_info, struct btrfs_block_group *cache)
|
|
|
|
{
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
btrfs-progs: zoned: implement log-structured superblock
Superblock (and its copies) is the only data structure in btrfs which has a
fixed location on a device. Since we cannot overwrite in a sequential write
required zone, we cannot place superblock in the zone. One easy solution
is limiting superblock and copies to be placed only in conventional zones.
However, this method has two downsides: one is reduced number of superblock
copies. The location of the second copy of superblock is 256GB, which is in
a sequential write required zone on typical devices in the market today.
So, the number of superblock and copies is limited to be two. Second
downside is that we cannot support devices which have no conventional zones
at all.
To solve these two problems, we employ superblock log writing. It uses two
adjacent zones as a circular buffer to write updated superblocks. Once the
first zone is filled up, start writing into the second one. Then, when
both zones are filled up and before starting to write to the first zone
again, reset the first zone.
We can determine the position of the latest superblock by reading write
pointer information from a device. One corner case is when both zones are
full. For this situation, we read out the last superblock of each zone, and
compare them to determine which zone is older.
The following zones are reserved as the circular buffer on ZONED btrfs.
- primary superblock: offset 0B (and the following zone)
- first copy: offset 512G (and the following zone)
- Second copy: offset 4T (4096G, and the following zone)
If these reserved zones are conventional, superblock is written fixed at
the start of the zone without logging.
Currently, superblock reading/writing is done by pread/pwrite. This
commit replace the call sites with sbread/sbwrite to wrap the functions.
For zoned btrfs, btrfs_sb_io which is called from sbread/sbwrite
reverses the IO position back to a mirror number, maps the mirror number
into the superblock logging position, and do the IO.
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
Signed-off-by: David Sterba <dsterba@suse.com>
2021-04-26 06:27:26 +00:00
|
|
|
#endif /* BTRFS_ZONED */
|
|
|
|
|
2021-04-26 06:27:27 +00:00
|
|
|
static inline bool btrfs_dev_is_sequential(struct btrfs_device *device, u64 pos)
|
|
|
|
{
|
|
|
|
return zone_is_sequential(device->zone_info, pos);
|
|
|
|
}
|
|
|
|
|
2021-04-26 06:27:21 +00:00
|
|
|
#endif /* __BTRFS_ZONED_H__ */
|