Skip to contentSkip to conversation

xfs: add support for zoned devices

GIT PULL Applied Christoph Hellwig 2 participants 2 messages Mar 3, 2025 – Mar 10, 2025

lore
Share & tools
lore
#0 lore raw

The following changes since commit 0a1fd78080c8c9a5582e82100bd91b87ae5ac57c:

Merge branch 'vfs-6.15.iomap' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs into xfs-6.15-merge (2025-03-03 13:01:06 +0100)

are available in the Git repository at:

git://git.infradead.org/users/hch/xfs.git tags/xfs-zoned-allocator-2025-03-03

for you to fetch changes up to 9c477912b2f58da71751f244aceecf5f8cc549ed:

xfs: export max_open_zones in sysfs (2025-03-03 08:17:10 -0700)

----------------------------------------------------------------
xfs: add support for zoned devices

Add support for the new zoned space allocator and thus for zoned devices:

https://zonedstorage.io/docs/introduction/zoned-storage

to XFS. This has been developed for and tested on both SMR hard drives, which are the oldest and most common class of zoned devices:

https://zonedstorage.io/docs/introduction/smr

and ZNS SSDs:

https://zonedstorage.io/docs/introduction/zns

It has not been tested with zoned UFS devices, as their current capacity points and performance characteristics aren't too interesting for XFS use cases (but never say never).

Sequential write only zones are only supported for data using a new allocator for the RT device, which maps each zone to a rtgroup which is written sequentially. All metadata and (for now) the log require using randomly writable space. This means a realtime device is required to support zoned storage, but for the common case of SMR hard drives that contain random writable zones and sequential write required zones on the same block device, the concept of an internal RT device is added which means using XFS on a SMR HDD is as simple as:

$ mkfs.xfs /dev/sda
$ mount /dev/sda /mnt

When using NVMe ZNS SSDs that do not support conventional zones, the traditional multi-device RT configuration is required. E.g. for an SSD with a conventional namespace 1 and a zoned namespace 2:

$ mkfs.xfs /dev/nvme0n1 -o rtdev=/dev/nvme0n2
$ mount -o rtdev=/dev/nvme0n2 /dev/nvme0n1 /mnt

The zoned allocator can also be used on conventional block devices, or on conventional zones (e.g. when using an SMR HDD as the external RT device). For example using zoned XFS on normal SSDs shows very nice performance advantages and write amplification reduction for intelligent workloads like RocksDB.

Some work is still in progress or planned, but should not affect the integration with the rest of XFS or the on-disk format:

- support for quotas
- support for reflinks

Note that the I/O path already supports reflink, but garbage collection isn't refcount aware yet and would unshare shared blocks, thus rendering the feature useless.

----------------------------------------------------------------
Christoph Hellwig (42):
xfs: reflow xfs_dec_freecounter
xfs: generalize the freespace and reserved blocks handling
xfs: support reserved blocks for the rt extent counter
xfs: trace in-memory freecounter reservations
xfs: fixup the metabtree reservation in xrep_reap_metadir_fsblocks
xfs: make metabtree reservations global
xfs: reduce metafile reservations
xfs: factor out a xfs_rt_check_size helper
xfs: add a rtg_blocks helper
xfs: move xfs_bmapi_reserve_delalloc to xfs_iomap.c
xfs: skip always_cow inodes in xfs_reflink_trim_around_shared
xfs: refine the unaligned check for always COW inodes in xfs_file_dio_write
xfs: support XFS_BMAPI_REMAP in xfs_bmap_del_extent_delay
xfs: add a xfs_rtrmap_highest_rgbno helper
xfs: define the zoned on-disk format
xfs: allow internal RT devices for zoned mode
xfs: export zoned geometry via XFS_FSOP_GEOM
xfs: disable sb_frextents for zoned file systems
xfs: disable FITRIM for zoned RT devices
xfs: don't call xfs_can_free_eofblocks from ->release for zoned inodes
xfs: skip zoned RT inodes in xfs_inodegc_want_queue_rt_file
xfs: parse and validate hardware zone information
xfs: add the zoned space allocator
xfs: add support for zoned space reservations
xfs: implement zoned garbage collection
xfs: implement buffered writes to zoned RT devices
xfs: implement direct writes to zoned RT devices
xfs: wire up zoned block freeing in xfs_rtextent_free_finish_item
xfs: hide reserved RT blocks from statfs
xfs: support growfs on zoned file systems
xfs: allow COW forks on zoned file systems in xchk_bmap
xfs: support xchk_xref_is_used_rt_space on zoned file systems
xfs: support xrep_require_rtext_inuse on zoned file systems
xfs: enable fsmap reporting for internal RT devices
xfs: disable reflink for zoned file systems
xfs: disable rt quotas for zoned file systems
xfs: enable the zoned RT device feature
xfs: support zone gaps
xfs: add a max_open_zones mount option
xfs: wire up the show_stats super operation
xfs: contain more sysfs code in xfs_sysfs.c
xfs: export max_open_zones in sysfs

Hans Holmberg (2):
xfs: support write life time based data placement
xfs: export zone stats in /proc/*/mountstats

 fs/xfs/Makefile                  |    7 +-
 fs/xfs/libxfs/xfs_bmap.c         |  316 +---------
 fs/xfs/libxfs/xfs_bmap.h         |    7 +-
 fs/xfs/libxfs/xfs_format.h       |   20 +-
 fs/xfs/libxfs/xfs_fs.h           |   14 +-
 fs/xfs/libxfs/xfs_group.h        |   31 +-
 fs/xfs/libxfs/xfs_ialloc.c       |    2 +-
 fs/xfs/libxfs/xfs_inode_buf.c    |   21 +-
 fs/xfs/libxfs/xfs_inode_util.c   |    1 +
 fs/xfs/libxfs/xfs_log_format.h   |    7 +-
 fs/xfs/libxfs/xfs_metafile.c     |  167 ++++--
 fs/xfs/libxfs/xfs_metafile.h     |    6 +-
 fs/xfs/libxfs/xfs_ondisk.h       |    6 +-
 fs/xfs/libxfs/xfs_rtbitmap.c     |   11 +
 fs/xfs/libxfs/xfs_rtgroup.c      |   39 +-
 fs/xfs/libxfs/xfs_rtgroup.h      |   50 +-
 fs/xfs/libxfs/xfs_rtrmap_btree.c |   19 +
 fs/xfs/libxfs/xfs_rtrmap_btree.h |    2 +
 fs/xfs/libxfs/xfs_sb.c           |   82 ++-
 fs/xfs/libxfs/xfs_types.h        |   28 +
 fs/xfs/libxfs/xfs_zones.c        |  186 ++++++
 fs/xfs/libxfs/xfs_zones.h        |   35 ++
 fs/xfs/scrub/agheader.c          |    2 +
 fs/xfs/scrub/bmap.c              |    4 +-
 fs/xfs/scrub/fscounters.c        |   22 +-
 fs/xfs/scrub/fscounters_repair.c |   12 +-
 fs/xfs/scrub/inode.c             |    7 +
 fs/xfs/scrub/inode_repair.c      |    4 +-
 fs/xfs/scrub/newbt.c             |    2 +-
 fs/xfs/scrub/reap.c              |    9 +-
 fs/xfs/scrub/repair.c            |   37 +-
 fs/xfs/scrub/rtbitmap.c          |   11 +-
 fs/xfs/scrub/rtrefcount_repair.c |   34 +-
 fs/xfs/scrub/rtrmap_repair.c     |   29 +-
 fs/xfs/scrub/scrub.c             |    2 +
 fs/xfs/xfs_aops.c                |  171 +++++-
 fs/xfs/xfs_aops.h                |    3 +-
 fs/xfs/xfs_bmap_util.c           |   32 +-
 fs/xfs/xfs_bmap_util.h           |   12 +-
 fs/xfs/xfs_discard.c             |    3 +-
 fs/xfs/xfs_extent_busy.c         |    2 +-
 fs/xfs/xfs_extfree_item.c        |   35 +-
 fs/xfs/xfs_file.c                |  347 +++++++++--
 fs/xfs/xfs_fsmap.c               |   86 ++-
 fs/xfs/xfs_fsops.c               |   50 +-
 fs/xfs/xfs_fsops.h               |    3 +-
 fs/xfs/xfs_icache.c              |    6 +-
 fs/xfs/xfs_inode.c               |    3 +-
 fs/xfs/xfs_inode.h               |   28 +-
 fs/xfs/xfs_inode_item.c          |    1 +
 fs/xfs/xfs_inode_item_recover.c  |    1 +
 fs/xfs/xfs_ioctl.c               |   12 +-
 fs/xfs/xfs_iomap.c               |  528 ++++++++++++++++-
 fs/xfs/xfs_iomap.h               |    7 +-
 fs/xfs/xfs_iops.c                |   31 +-
 fs/xfs/xfs_log.c                 |    4 +
 fs/xfs/xfs_message.c             |    4 +
 fs/xfs/xfs_message.h             |    1 +
 fs/xfs/xfs_mount.c               |  206 ++++---
 fs/xfs/xfs_mount.h               |  131 ++++-
 fs/xfs/xfs_qm.c                  |    3 +-
 fs/xfs/xfs_reflink.c             |   18 +-
 fs/xfs/xfs_rtalloc.c             |  237 +++++---
 fs/xfs/xfs_rtalloc.h             |    5 -
 fs/xfs/xfs_super.c               |  165 ++++--
 fs/xfs/xfs_sysfs.c               |   75 ++-
 fs/xfs/xfs_sysfs.h               |    5 +-
 fs/xfs/xfs_trace.c               |    2 +
 fs/xfs/xfs_trace.h               |  214 ++++++-
 fs/xfs/xfs_zone_alloc.c          | 1211 ++++++++++++++++++++++++++++++++++++++
 fs/xfs/xfs_zone_alloc.h          |   70 +++
 fs/xfs/xfs_zone_gc.c             | 1165 ++++++++++++++++++++++++++++++++++++
 fs/xfs/xfs_zone_info.c           |  105 ++++
 fs/xfs/xfs_zone_priv.h           |  119 ++++
 fs/xfs/xfs_zone_space_resv.c     |  253 ++++++++
 75 files changed, 5649 insertions(+), 937 deletions(-)
 create mode 100644 fs/xfs/libxfs/xfs_zones.c
 create mode 100644 fs/xfs/libxfs/xfs_zones.h
 create mode 100644 fs/xfs/xfs_zone_alloc.c
 create mode 100644 fs/xfs/xfs_zone_alloc.h
 create mode 100644 fs/xfs/xfs_zone_gc.c
 create mode 100644 fs/xfs/xfs_zone_info.c
 create mode 100644 fs/xfs/xfs_zone_priv.h
 create mode 100644 fs/xfs/xfs_zone_space_resv.c
#1 lore raw
Quoted 9 lines · Christoph Hellwig

The following changes since commit 0a1fd78080c8c9a5582e82100bd91b87ae5ac57c:

Merge branch 'vfs-6.15.iomap' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs into xfs-6.15-merge (2025-03-03 13:01:06 +0100)

are available in the Git repository at:

git://git.infradead.org/users/hch/xfs.git tags/xfs-zoned-allocator-2025-03-03

[...]

Merged, thanks!

merge commit: 4c6283ec9284bb72906dba83bc7a809747e6331e

Best regards,

Signature · 1 line
Carlos Maiolino <[email protected]>