xfs: updates for 4.7-rc1

Hi Linus,

Can you please pull the XFS update from the tag below? It's a pretty average collection of fixes, cleanups and improvements in this request, details are in the tag below. A couple of the branches were only pushed out last week (yes, I know this is generally considered bad form) but I had been testing them for a couple of weeks prior and had planned to push them out before the window opened. However, an unplanned high speed impact with a concrete wall a couple of days before the merge window opened forced me to change my plans a little, hence the late commit dates on the changes.

I've let them soak in linux-next for a week, and I did a merge with your tree this morning and a bunch of xfstests runs over the past 8 hours without any problems, so everything looks ok from my side. Just yell if this is not OK and I'll drop those branches for this merge and resend the pull request....

-Dave.

The following changes since commit f55532a0c0b8bb6148f4e07853b876ef73bc69ca:

Linux 4.6-rc1 (2016-03-26 16:03:24 -0700)

are available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/dgc/linux-xfs.git for-next

for you to fetch changes up to 555b67e4e729ca544bb4028ab12e532c68b70ddb:

Merge branch 'xfs-4.7-inode-reclaim' into for-next (2016-05-20 10:34:00 +1000)

----------------------------------------------------------------

xfs: update for 4.7-rc1

Changes in this update:
o fixes for mount line parsing, sparse warnings, read-only compat feature remount behaviour
o allow fast path symlink lookups for inline symlinks.
o attribute listing cleanups
o writeback goes direct to bios rather than indirecting through bufferheads
o transaction allocation cleanup
o optimised kmem_realloc
o added configurable error handling for metadata write errors, changed default error handling behaviour from "retry forever" to "retry until unmount then fail"
o fixed several inode cluster writeback lookup vs reclaim race conditions
o fixed inode cluster writeback checking wrong inode after lookup
o fixed bugs where struct xfs_inode freeing wasn't actually RCU safe
o cleaned up inode reclaim tagging

----------------------------------------------------------------
Alex Lyakas (1):
xfs: optimise xfs_iext_destroy

Brian Foster (1):

      xfs: buffer ->bi_end_io function requires irq-safe lock

Carlos Maiolino (8):
xfs: Add caller function output to xfs_log_force tracepoint
xfs: configurable error behavior via sysfs
xfs: introduce metadata IO error class
xfs: add configurable error support to metadata buffers
xfs: introduce table-based init for error behaviors
xfs: add configuration of error failure speed
xfs: add configuration handlers for specific errors
xfs: add "fail at unmount" error handling configuration

Christoph Hellwig (9):
xfs: factor out a helper to initialize a local format inode fork
xfs: set up inode operation vectors later
xfs: use ->readlink to implement the readlink_by_handle ioctl
xfs: optimize inline symlinks
xfs: optimize bio handling in the buffer writeback path
xfs: better xfs_trans_alloc interface
xfs: remove transaction types
xfs: improve kmem_realloc
xfs: fix warning in xfs_finish_page_writeback for non-debug builds

Dave Chinner (20):
xfs: Don't wrap growfs AGFL indexes
xfs: build bios directly in xfs_add_to_ioend
xfs: don't release bios on completion immediately
xfs: remove xfs_fs_evict_inode()
xfs: xfs_iflush_cluster fails to abort on error
xfs: fix inode validity check in xfs_iflush_cluster
xfs: skip stale inodes in xfs_iflush_cluster
xfs: xfs_inode_free() isn't RCU safe
xfs: mark reclaimed inodes invalid earlier
xfs: xfs_iflush_cluster has range issues
xfs: rename variables in xfs_iflush_cluster for clarity
xfs: simplify inode reclaim tagging interfaces
xfs: move reclaim tagging functions
Merge branch 'xfs-4.7-writeback-bio' into for-next Merge branch 'xfs-4.7-trans-type-cleanup' into for-next Merge branch 'xfs-4.7-optimise-inline-symlinks' into for-next Merge branch 'xfs-4.7-cleanup-attr-listent' into for-next Merge branch 'xfs-4.7-misc-fixes' into for-next Merge branch 'xfs-4.7-error-cfg' into for-next Merge branch 'xfs-4.7-inode-reclaim' into for-next

Eric Sandeen (5):

      xfs: disallow rw remount on fs with unknown ro-compat features
      xfs: only return -errno or success from attr ->put_listent
      xfs: don't pass value into attr ->put_listent
      xfs: remove put_value from attr ->put_listent context
      xfs: collapse cases in xfs_attr3_leaf_list_int

Eryu Guan (2):
xfs: add missing break in xfs_parseargs()
xfs: mute some sparse warnings

 fs/namei.c                     |    1 -
 fs/xfs/kmem.c                  |   26 +--
 fs/xfs/kmem.h                  |    2 +-
 fs/xfs/libxfs/xfs_attr.c       |   58 ++-----
 fs/xfs/libxfs/xfs_bmap.c       |   22 +--
 fs/xfs/libxfs/xfs_dir2_sf.c    |    9 +-
 fs/xfs/libxfs/xfs_inode_fork.c |   99 +++++++----
 fs/xfs/libxfs/xfs_inode_fork.h |    1 +
 fs/xfs/libxfs/xfs_log_format.h |    5 +
 fs/xfs/libxfs/xfs_sb.c         |    8 +-
 fs/xfs/libxfs/xfs_shared.h     |  102 +-----------
 fs/xfs/xfs_aops.c              |  353 ++++++++++++++++++++--------------------
 fs/xfs/xfs_aops.h              |   15 +-
 fs/xfs/xfs_attr.h              |    4 +-
 fs/xfs/xfs_attr_inactive.c     |   16 +-
 fs/xfs/xfs_attr_list.c         |   85 +++-------
 fs/xfs/xfs_bmap_util.c         |   45 ++---
 fs/xfs/xfs_buf.c               |   12 +-
 fs/xfs/xfs_buf.h               |   20 +++
 fs/xfs/xfs_buf_item.c          |  121 +++++++++-----
 fs/xfs/xfs_dquot.c             |    9 +-
 fs/xfs/xfs_file.c              |    8 +-
 fs/xfs/xfs_fsops.c             |   14 +-
 fs/xfs/xfs_icache.c            |  290 ++++++++++++++++++---------------
 fs/xfs/xfs_inode.c             |  167 ++++++++++---------
 fs/xfs/xfs_inode.h             |    5 +-
 fs/xfs/xfs_inode_item.c        |    6 +-
 fs/xfs/xfs_ioctl.c             |   31 +---
 fs/xfs/xfs_iomap.c             |   53 ++----
 fs/xfs/xfs_iops.c              |  117 ++++++++-----
 fs/xfs/xfs_log.c               |   62 +------
 fs/xfs/xfs_log.h               |    3 +-
 fs/xfs/xfs_log_cil.c           |    1 -
 fs/xfs/xfs_log_priv.h          |    1 -
 fs/xfs/xfs_log_recover.c       |   12 +-
 fs/xfs/xfs_mount.c             |   23 ++-
 fs/xfs/xfs_mount.h             |   34 ++++
 fs/xfs/xfs_pnfs.c              |    7 +-
 fs/xfs/xfs_qm.c                |    9 +-
 fs/xfs/xfs_qm_syscalls.c       |   26 +--
 fs/xfs/xfs_rtalloc.c           |   21 +--
 fs/xfs/xfs_super.c             |   65 +++-----
 fs/xfs/xfs_symlink.c           |   37 ++---
 fs/xfs/xfs_sysfs.c             |  291 ++++++++++++++++++++++++++++++++-
 fs/xfs/xfs_sysfs.h             |    3 +
 fs/xfs/xfs_trace.h             |   16 +-
 fs/xfs/xfs_trans.c             |   88 +++++-----
 fs/xfs/xfs_trans.h             |    8 +-
 fs/xfs/xfs_xattr.c             |   17 +-
 49 files changed, 1293 insertions(+), 1135 deletions(-)

Dave Chinner
[email protected]

i'm ok with the late branches, it's not like xfs has been a problem spot.

However:

Your pull request mentions the 'for-next' branch, but I think you *meant* to send me the "xfs-for-linus-4.7-rc1" tag which points to the same commit and has your summary in it.

Please try to make sure your pull requests actually contain what you intend for them to do,

Linus

Still, I'll try to avoid them because it reduces testing time.

<sigh>

Mea culpa. I ran this command after creating the tag:

$ git request-pull v4.6 git://git.kernel.org/pub/scm/linux/kernel/git/dgc/linux-xfs.git tags/xfs-for-linus-4.7-rc1 > ~/tmp/t.txt

And I didn't check the output closely enough. I forgot to push the tag to the upstream repo before running request-pull.

Git often makes it very easy to make mistakes whilst simultaneously making it hard to notice you've made a mistake. :(

Cheers,

Dave.

Dave Chinner
[email protected]

Oh, 100% agreed. I'm just saying that you get a pass because you've historically been very reliable.

Actually, that one got fixed in git quite some time ago.

You probably have a fairly old version of git, which had the "helpful" logic to pick another remote location if it couldn't find the exact one you had specified, but it could find another one that matched in content

It was a misguided attempt at automatically just "doing the right thing" and being easier to use, but yes, it means that it does something subtly different from what you ask for if you haven't pushed out the remote you talked about yet.

Bu that was all fixed about two years ago, and I think made v2.0.

Do you perhaps happen to run something like debian-stable or similar that never upgrades major versions, so you're still at git-1.x?

Building your own git is trivial, and a "make install" will just install in your ~/bin directory so you don't even have to uninstall the system one (assuming you have your own ~/bin before /usr/bin in the path, of course).

Linus

Ah, unlike my workstation which is running 2.8.3, the server I host my all my internal network services (including staging trees) is running....

... debian stable.

Given the only thing I do manually in those staging trees is tag commits, push branches to kernel.org, and run "git request-pull" I hadn't noticed that I was still on an old git.

*nod*

But it's far easier just to do 'apt-get install git/testing' and have 2.8.1 installed 15s later.... :P

Cheers,

Dave.

Dave Chinner
[email protected]