[v6,00/13] large atomic writes for xfs with CoW

Message

John Garry March 13, 2025, 5:12 p.m. UTC

Currently atomic write support for xfs is limited to writing a single
block as we have no way to guarantee alignment and that the write covers
a single extent.

This series introduces a method to issue atomic writes via a software
emulated method.

The software emulated method is used as a fallback for when attempting to
issue an atomic write over misaligned or multiple extents.

For XFS, this support is based on CoW.

The basic idea of this CoW method is to alloc a range in the CoW fork,
write the data, and atomically update the mapping.

Initial mysql performance testing has shown this method to perform ok.
However, there we are only using 16K atomic writes (and 4K block size),
so typically - and thankfully - this software fallback method won't be
used often.

For other FSes which want large atomics writes and don't support CoW, I
think that they can follow the example in [0].

Based on 4c6283ec9284 (xfs/xfs-6.15-zoned_devices, xfs/xfs-6.15-merge)
Merge tag 'xfs-zoned-allocator-2025-03-03' of
git://git.infradead.org/users/hch/xfs into xfs-6.15-zoned_devices

[0] https://lore.kernel.org/linux-xfs/[e-mail]/

Differences to v5:
- comment on atomic write checks in iomap_dio_bio_iter() (Christoph)
- inline iomap_dio_bio_opflags (Christoph)
- many comments added (Christoph, Dave)
- reorder patches
- rework iomap flags (Christoph, Dave)
- add xfs_get_atomic_write_{min,max}_attr (Christoph)
- add XFS_REFLINK_ALLOC_EXTSZALIGN
- change XFS_REFLINK_CONVERT -> XFS_REFLINK_CONVERT_UNWRITTEN (Christoph)

Differences to v4:
- Omit iomap patches which have already been queued
- Add () in xfs_bmap_compute_alignments() (Dave)
- Rename awu_max -> m_awu_max (Carlos)
- Add RFC to change IOMAP flag names
- Rebase

John Garry (13):
  iomap: inline iomap_dio_bio_opflags()
  iomap: comment on atomic write checks in iomap_dio_bio_iter()
  iomap: rework IOMAP atomic flags
  xfs: pass flags to xfs_reflink_allocate_cow()
  xfs: allow block allocator to take an alignment hint
  xfs: switch atomic write size check in xfs_file_write_iter()
  xfs: refactor xfs_reflink_end_cow_extent()
  xfs: reflink CoW-based atomic write support
  xfs: add XFS_REFLINK_ALLOC_EXTSZALIGN
  xfs: iomap COW-based atomic write support
  xfs: add xfs_file_dio_write_atomic()
  xfs: commit CoW-based atomic writes atomically
  xfs: update atomic write max size

 fs/ext4/inode.c          |   5 +-
 fs/iomap/direct-io.c     | 125 ++++++++++++++----------------
 fs/iomap/trace.h         |   2 +-
 fs/xfs/libxfs/xfs_bmap.c |   5 ++
 fs/xfs/libxfs/xfs_bmap.h |   6 +-
 fs/xfs/xfs_file.c        |  90 ++++++++++++++++++++--
 fs/xfs/xfs_iomap.c       | 141 +++++++++++++++++++++++++++++++++-
 fs/xfs/xfs_iomap.h       |   1 +
 fs/xfs/xfs_iops.c        |  40 ++++++++--
 fs/xfs/xfs_iops.h        |   2 +
 fs/xfs/xfs_mount.c       |  29 +++++++
 fs/xfs/xfs_mount.h       |   1 +
 fs/xfs/xfs_reflink.c     | 160 +++++++++++++++++++++++++++++----------
 fs/xfs/xfs_reflink.h     |  14 +++-
 include/linux/iomap.h    |  12 ++-
 15 files changed, 499 insertions(+), 134 deletions(-)