[PATCH 0/6 v5] xfs: get rid of xfs_cluster_write()

[Dave Chinner <david@xxxxxxxxxxxxx>]

Hi folks,

Version 5 of the patchset, hopefully addressing all the remaining
review comments. I dropped the two RFC patches from this posting -
they are not necessary at this point.

Cheers,

Dave.

Version 5:
- return errors currectly when transaction allocation fails (patch
  2, patch 3, patch 6)
- Added XFS_IO_INVALID to the iotype enum, and initialised the
  writepage context with it. (patch 3)
- s/fail/status/ in xfs_submit_ioend (patch 6)
- pass submit_list into xfs_add_to_ioend() rather than returning the
  ioend to add to the submit list (patch 6)
- fixed bug where dirty pages that are not added to an ioend would
  remain dirty rather than being marked clean and transitioning
  through the writeback state before returning (patch 6)