optimize COW end I/O remapping v2

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hi all,

this series optimizes how we handle unmapping the old data fork
extents when finishing COW I/O.  To get there I also did a bunch
of cleanups for helpers called by this code.

Changes since v1:
 - remove the racy if_bytes in xfs_reflink_end_cow_extent enirely
 - pass an unsigned value to xfs_quota_unreserve_blkres
 - rename xfs_iext_count_upgrade to xfs_iext_count_ensure
 - only check for XFS_ERRTAG_REDUCE_MAX_IEXTENTS once in
   xfs_iext_count_ensure
 - roll the transaction after 128 EFIs
 - use a different name for the irec of the remapped extent in
   xfs_reflink_end_cow_extent

Diffstat:
 libxfs/xfs_attr.c       |    5 -
 libxfs/xfs_bmap.c       |   21 +-----
 libxfs/xfs_bmap.h       |    2 
 libxfs/xfs_inode_fork.c |   57 +++++++----------
 libxfs/xfs_inode_fork.h |    6 -
 libxfs/xfs_trans_resv.h |    7 ++
 scrub/newbt.c           |    2 
 scrub/reap.c            |    2 
 scrub/repair.h          |    8 --
 xfs_aops.c              |    6 -
 xfs_bmap_item.c         |    4 -
 xfs_bmap_util.c         |   33 ++-------
 xfs_bmap_util.h         |    2 
 xfs_dquot.c             |    5 -
 xfs_iomap.c             |   13 +--
 xfs_quota.h             |   23 ++----
 xfs_reflink.c           |  160 ++++++++++++++++++++++++++----------------------
 xfs_rtalloc.c           |    5 -
 18 files changed, 159 insertions(+), 202 deletions(-)




[Index of Archives]     [XFS Filesystem Development (older mail)]     [Linux Filesystem Development]     [Linux Audio Users]     [Yosemite Trails]     [Linux Kernel]     [Linux RAID]     [Linux SCSI]


  Powered by Linux