This patch fixes a cluster coherency problem that occurs when one
node creates a file, does several writes, then a different node
tries to write to the same file. When the inode's glock is demoted,
the inode wasn't synced to the media properly because the gl_object
wasn't set. Later, the flush daemon noticed the uncommitted data
and tried to flush it, only to discover the glock was no longer locked
properly in exclusive mode. That caused an assert withdraw.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

This patch adds a return code check after attempting to allocate
a new inode during dinode creation.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

This patch changes the block allocation trace so that it references
the rgd's glock rather than the inode's glock. Now that the order
of inode creation is switched, this prevents a reference to the
glock which may not be set yet.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

The lksb struct already contains a pointer to the lvb,
so another directly from the glock struct is not needed.

Signed-off-by: David Teigland <teigland@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

Save the effort of allocating, reading and writing
the lvb for most glocks that do not use it.

Signed-off-by: David Teigland <teigland@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

When unmounting, gfs2 does a full dlm_unlock operation on every
cached lock.  This can create a very large amount of work and can
take a long time to complete.  However, the vast majority of these
dlm unlock operations are unnecessary because after all the unlocks
are done, gfs2 leaves the dlm lockspace, which automatically clears
the locks of the leaving node, without unlocking each one individually.
So, gfs2 can skip explicit dlm unlocks, and use dlm_release_lockspace to
remove the locks implicitly.  The one exception is when the lock's lvb is
being used.  In this case, dlm_unlock is called because it may update the
lvb of the resource.

Signed-off-by: David Teigland <teigland@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

For filesystems with only a single resource group, we need to be careful
that the allocation loop will not land up with a NULL resource group. This
fixes a bug in a previous patch where the gfs2_rgrpd_get_next() function
was being used instead of gfs2_rgrpd_get_first()

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

Since we now have a dirty_inode that takes care of manipulating the
inode buffer and writing from the inode to the buffer, we can
eliminate some unnecessary buffer manipulations in gfs2_unlink_inode
that are now redundant.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

This patch changes the gfs2_dir_add function so that it uses
the dirty_inode function (via mark_inode_dirty) rather than manually
updating the dinode.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

This patch fixes an issue relating to not having enough revokes
available when truncating journaled data files. In order to ensure
that we do no run out, the truncation is broken into separate pieces
if it is large enough.

Tested using fsx on a journaled data file.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

Just like ext3, this works on the root directory and any directory
with the +T flag set. Also, just like ext3, any subdirectory created
in one of the just mentioned cases will be allocated to a random
resource group (GFS2 equivalent of a block group).

If you are creating a set of directories, each of which will contain a
job running on a different node, then by setting +T on the parent
directory before creating the subdirectories, each will land up in a
different resource group, and thus resource group contention between
nodes will be kept to a minimum.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

Rather than using the parent directory's allocation context, this
patch allocated the new inode earlier in the process and then uses
it to contain all the information required. As a result, we can now
use the new inode's own allocation context to allocate it rather
than having to use the parent directory's context. This give us a
lot more flexibility in where the inode is placed on disk.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

This patch uses information gathered by the recent glock statistics
patch in order to derrive a boolean verdict on the congestion
status of a resource group. This is then used when making decisions
on which resource group to choose during block allocation.

The aim is to avoid resource groups which are heavily contended
by other nodes, while still ensuring locality of access wherever
possible.

Once a reservation has been made in a particular resource group
we continue to use that resource group until a new reservation is
required. This should help to ensure that we do not change resource
groups too often.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

[Editorial: This is a nit, but has been a minor irritation for a long time:]

This patch renames glops structure item for go_xmote_th to go_sync.
The functionality is unchanged; it's just for readability.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

This patch is a rewrite of function gfs2_rbm_from_block. Rather than
looping to find the right bitmap, the code now does a few simple
math calculations.

I compared the performance of both algorithms side by side and the new
algorithm is noticeably faster. Sample instrumentation output from a
"fast" machine:

5 million calls: millisec spent: Orig: 166 New: 113
5 million calls: millisec spent: Orig: 189 New: 114

In addition, I ran postmark (on a somewhat slowr CPU) before the after
the new algorithm was put in place and postmark showed a decent
improvement:

Before the new algorithm:
-------------------------
Time:
	645 seconds total
	584 seconds of transactions (171 per second)

Files:
	150087 created (232 per second)
		Creation alone: 100000 files (2083 per second)
		Mixed with transactions: 50087 files (85 per second)
	49995 read (85 per second)
	49991 appended (85 per second)
	150087 deleted (232 per second)
		Deletion alone: 100174 files (7705 per second)
		Mixed with transactions: 49913 files (85 per second)

Data:
	273.42 megabytes read (434.08 kilobytes per second)
	852.13 megabytes written (1.32 megabytes per second)

With the new algorithm:
-----------------------
Time:
	599 seconds total
	530 seconds of transactions (188 per second)

Files:
	150087 created (250 per second)
		Creation alone: 100000 files (1886 per second)
		Mixed with transactions: 50087 files (94 per second)
	49995 read (94 per second)
	49991 appended (94 per second)
	150087 deleted (250 per second)
		Deletion alone: 100174 files (6260 per second)
		Mixed with transactions: 49913 files (94 per second)

Data:
	273.42 megabytes read (467.42 kilobytes per second)
	852.13 megabytes written (1.42 megabytes per second)

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

Two of the bug traps here could really be warnings. The others are
converted from BUG() to GLOCK_BUG_ON() since we'll most likely
need to know the glock state in order to debug any issues which
arise. As a result of this, __dump_glock has to be renamed and
is no longer static.

Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

Pull gfs2 fixes from Steven Whitehouse:
 "Here are a number of GFS2 bug fixes.  There are three from Andy Price
  which fix various issues spotted by automated code analysis.  There
  are two from Lukas Czerner fixing my mistaken assumptions as to how
  FITRIM should work.  Finally Ben Marzinski has fixed a bug relating to
  mmap and atime and also a bug relating to a locking issue in the
  transaction code."

* git://git.kernel.org/pub/scm/linux/kernel/git/steve/gfs2-3.0-fixes:
  GFS2: Test bufdata with buffer locked and gfs2_log_lock held
  GFS2: Don't call file_accessed() with a shared glock
  GFS2: Fix FITRIM argument handling
  GFS2: Require user to provide argument for FITRIM
  GFS2: Clean up some unused assignments
  GFS2: Fix possible null pointer deref in gfs2_rs_alloc
  GFS2: Fix an unchecked error from gfs2_rs_alloc

Pull hwmon fixes from Jean Delvare.

* 'hwmon-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jdelvare/staging:
  hwmon: Fix chip feature table headers
  hwmon: (w83627ehf) Force initial bank selection

In gfs2_trans_add_bh(), gfs2 was testing if a there was a bd attached to the
buffer without having the gfs2_log_lock held. It was then assuming it would
stay attached for the rest of the function. However, without either the log
lock being held of the buffer locked, __gfs2_ail_flush() could detach bd at any
time.  This patch moves the locking before the test.  If there isn't a bd
already attached, gfs2 can safely allocate one and attach it before locking.
There is no way that the newly allocated bd could be on the ail list,
and thus no way for __gfs2_ail_flush() to detach it.

Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

file_accessed() was being called by gfs2_mmap() with a shared glock. If it
needed to update the atime, it was crashing because it dirtied the inode in
gfs2_dirty_inode() without holding an exclusive lock. gfs2_dirty_inode()
checked if the caller was already holding a glock, but it didn't make sure that
the glock was in the exclusive state. Now, instead of calling file_accessed()
while holding the shared lock in gfs2_mmap(), file_accessed() is called after
grabbing and releasing the glock to update the inode.  If file_accessed() needs
to update the atime, it will grab an exclusive lock in gfs2_dirty_inode().

gfs2_dirty_inode() now also checks to make sure that if the calling process has
already locked the glock, it has an exclusive lock.

Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

Currently implementation in gfs2 uses FITRIM arguments as it were in
file system blocks units which is wrong. The FITRIM arguments
(fstrim_range.start, fstrim_range.len and fstrim_range.minlen) are
actually in bytes.

Moreover, check for start argument beyond the end of file system, len
argument being smaller than file system block and minlen argument being
bigger than biggest resource group were missing.

This commit converts the code to convert FITRIM argument to file system
blocks and also adds appropriate checks mentioned above.

All the problems were recognised by xfstests 251 and 260.

Signed-off-by: Lukas Czerner <lczerner@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

When the fstrim_range argument is not provided by user in FITRIM ioctl
we should just return EFAULT and not promoting bad behaviour by filling
the structure in kernel. Let the user deal with it.

Signed-off-by: Lukas Czerner <lczerner@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

Cleans up two cases where variables were assigned values but then never
used again.

Signed-off-by: Andrew Price <anprice@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

Despite the return value from kmem_cache_zalloc() being checked, the
error wasn't being returned until after a possible null pointer
dereference. This patch returns the error immediately, allowing the
removal of the error variable.

Signed-off-by: Andrew Price <anprice@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

Check the return value of gfs2_rs_alloc(ip) and avoid a possible null
pointer dereference.

Signed-off-by: Andrew Price <anprice@redhat.com>
Signed-off-by: Steven Whitehouse <swhiteho@redhat.com>

Pull drm fixes from Dave Airlie:
 "A single radeon typo fix for a regressions and two fixes for a
  regression in the open helper address space stuff."

* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux:
  drm/radeon: fix typo in evergreen_mc_resume()
  drm: set dev_mapping before calling drm_open_helper
  drm: restore open_count if drm_setup fails

Pull arm fixes from Russell King:
 "Not much here again.

  The two most notable things here are the sched_clock() fix, which was
  causing problems with the scheduling of threaded IRQs after a suspend
  event, and the vfp fix, which afaik has only been seen on some older
  OMAP boards.  Nevertheless, both are fairly important fixes."

* 'fixes' of git://git.linaro.org/people/rmk/linux-arm:
  ARM: 7569/1: mm: uninitialized warning corrections
  ARM: 7567/1: io: avoid GCC's offsettable addressing modes for halfword accesses
  ARM: 7566/1: vfp: fix save and restore when running on pre-VFPv3 and CONFIG_VFPv3 set
  ARM: 7565/1: sched: stop sched_clock() during suspend

Add missing index that may have led us to enabling
more crtcs than necessary.

May also fix:
https://bugs.freedesktop.org/show_bug.cgi?id=56139



Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

Some drivers (specifically vmwgfx) look at dev_mapping
in their open hook, so we have to set dev->dev_mapping
earlier in the process.

Reference:
http://lists.freedesktop.org/archives/dri-devel/2012-October/029420.html



Signed-off-by: Ilija Hadzic <ihadzic@research.bell-labs.com>
Reported-by: Thomas Hellstrom <thellstrom@vmware.com>
Cc: stable@vger.kernel.org
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

If drm_setup (called at first open) fails, the whole
open call has failed, so we should not keep the
open_count incremented.

Signed-off-by: Ilija Hadzic <ihadzic@research.bell-labs.com>
Cc: stable@vger.kernel.org
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>

These got broken by recent patches fixing checkpatch warnings in these
drivers. The trick is that the patches themselves looked good, but the
source files after applying them do not. That's why I am not a big fan
of using tabs inside comments.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Acked-by: Guenter Roeck <linux@roeck-us.net>

Don't assume bank 0 is selected at device probe time. This may not be
the case. Force bank selection at first register access to guarantee
that we read the right registers upon driver loading.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Reviewed-by: Guenter Roeck <linux@roeck-us.net>
Cc: stable@vger.kernel.org

The variables here are really not used uninitialized.

arch/arm/mm/alignment.c: In function 'do_alignment':
arch/arm/mm/alignment.c:327:15: warning: 'offset.un' may be used uninitialized in this function [-Wmaybe-uninitialized]
arch/arm/mm/alignment.c:748:21: note: 'offset.un' was declared here

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

Pull NFS client bugfixes from Trond Myklebust:

 - Fix a bunch of deadlock situations:
   * State recovery can deadlock if we fail to release sequence ids
     before scheduling the recovery thread.
   * Calling deactivate_super() from an RPC workqueue thread can
     deadlock because of the call to rpc_shutdown_client.

 - Display the device name correctly in /proc/*/mounts

 - Fix a number of incorrect error return values:
   * When NFSv3 mounts fail due to a timeout.
   * On NFSv4.1 backchannel setup failure
   * On NFSv4 open access checks

 - pnfs_find_alloc_layout() must check the layout pointer for NULL

 - Fix a regression in the legacy DNS resolved

* tag 'nfs-for-3.7-4' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
  NFS4: nfs4_opendata_access should return errno
  NFSv4: Initialise the NFSv4.1 slot table highest_used_slotid correctly
  SUNRPC: return proper errno from backchannel_rqst
  NFS: add nfs_sb_deactive_async to avoid deadlock
  nfs: Show original device name verbatim in /proc/*/mount{s,info}
  nfsv3: Make v3 mounts fail with ETIMEDOUTs instead EIO on mountd timeouts
  nfs: Check whether a layout pointer is NULL before free it
  NFS: fix bug in legacy DNS resolver.
  NFSv4: nfs4_locku_done must release the sequence id
  NFSv4.1: We must release the sequence id when we fail to get a session slot
  NFS: Wait for session recovery to finish before returning

Pull thermal management & ACPI update from Zhang Rui,

Ho humm.  Normally these things go through Len.  But it's just three
small fixes, I guess I can pull directly too.

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/rzhang/linux:
  exynos4_tmu_driver_ids should be exynos_tmu_driver_ids.
  ACPI video: Ignore errors after _DOD evaluation.
  thermal: solve compilation errors in rcar_thermal

Pull i2c embedded fixes from Wolfram Sang:
 "Two patches are usual stuff.

  The bigger patch is needed to correct a wrong decision made in this
  merge window.  We hoped to get the PIOQUEUE mode in the mxs driver
  working with DMA, but it turned out to be too broken (leading to data
  loss), so we now think it is best to remove it entirely and work only
  with DMA now.  The patch should be in 3.7.  IMO, so users never get
  the chance to use both modes in parallel."

* 'i2c-embedded/for-current' of git://git.pengutronix.de/git/wsa/linux:
  i2c: tegra: set irq name as device name
  i2c-nomadik: Fixup clock handling
  i2c: mxs: remove broken PIOQUEUE support

Pull drm fixes from Dave Airlie:
 "Scattered selection of fixes:

   - radeon: load detect fixes from SuSE/AMD
   - intel: misc i830, sdvo regression, vesafb kickoff ums fix
   - exynos: maintainers entry update + fixes
   - udl: fix stride scanout issue

  it's slightly bigger than I'd probably like, but nothing looked
  dangerous enough to hold off on."

* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux:
  drm/udl: fix stride issues scanning out stride != width*bpp
  drm/radeon: add load detection support for ext DAC on R200 (v2)
  DRM/radeon: For single CRTC GPUs move handling of CRTC_CRT_ON to crtc_dpms().
  DRM/Radeon: Fix TV DAC Load Detection for single CRTC chips.
  DRM/Radeon: Clean up code in TV DAC load detection.
  drm/radeon: fix ATPX function documentation
  drivers/gpu/drm/radeon/evergreen_cs.c: Remove unnecessary semicolon
  DRM/Radeon: On DVI-I use Load Detection when EDID is bogus.
  DRM/Radeon: Fix primary DAC Load Detection for RV100 chips.
  DRM/Radeon: Fix Load Detection on legacy primary DAC.
  drm: exynos: removed warning due to missing typecast for mixer driver data
  drm/exynos: add support for ARCH_MULTIPLATFORM
  MAINTAINERS: Add git repository for Exynos DRM
  drm/exynos: fix display on issue
  drm/i915: Only kick out vesafb if we takeover the fbcon with KMS
  drm/i915: be less verbose about inability to provide vendor backlight
  drm/i915: clear the entire sdvo infoframe buffer
  drm/i915: VGA needs to be on pipe A on i830M
  drm/i915: fix overlay on i830M

Pull networking fixes from David Miller:
 "First post-Sandy pull request"

 1) Fix antenna gain handling and initialization of chan->max_reg_power
    in wireless, from Felix Fietkau.

 2) Fix nexthop handling in H.232 conntrack helper, from Julian
    Anastasov.

 3) Only process 80211 mesh config header in certain kinds of frames,
    from Javier Cardona.

 4) 80211 management frame header length needs to be validated, from
    Johannes Berg.

 5) Don't access free'd SKBs in ath9k driver, from Felix Fietkay.

 6) Test for permanent state correctly in VXLAN driver, from Stephen
    Hemminger.

 7) BNX2X bug fixes from Yaniv Rosner and Dmitry Kravkov.

 8) Fix off by one errors in bonding, from Nikolay ALeksandrov.

 9) Fix divide by zero in TCP-Illinois congestion control.  From Jesper
    Dangaard Brouer.

10) TCP metrics code says "Yo dawg, I heard you like sizeof, so I did a
    sizeof of a sizeof, so you can size your size" Fix from Julian
    Anastasov.

11) Several drivers do mdiobus_free without first doing an
    mdiobus_unregister leading to stray pointer references.  Fix from
    Peter Senna Tschudin.

12) Fix OOPS in l2tp_eth_create() error path, it's another danling
    pointer kinda situation.  Fix from Tom Parkin.

13) Hardware driven by the vmxnet driver can't handle larger than 16K
    fragments, so split them up when necessary.  From Eric Dumazet.

14) Handle zero length data length in tcp_send_rcvq() properly.  Fix
    from Pavel Emelyanov.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (38 commits)
  tcp-repair: Handle zero-length data put in rcv queue
  vmxnet3: must split too big fragments
  l2tp: fix oops in l2tp_eth_create() error path
  cxgb4: Fix unable to get UP event from the LLD
  drivers/net/phy/mdio-bitbang.c: Call mdiobus_unregister before mdiobus_free
  drivers/net/ethernet/nxp/lpc_eth.c: Call mdiobus_unregister before mdiobus_free
  bnx2x: fix HW initialization using fw 7.8.x
  tcp: Fix double sizeof in new tcp_metrics code
  net: fix divide by zero in tcp algorithm illinois
  net: sctp: Fix typo in net/sctp
  bonding: fix second off-by-one error
  bonding: fix off-by-one error
  bnx2x: Disable FCoE for 57840 since not yet supported by FW
  bnx2x: Fix no link on 577xx 10G-baseT
  bnx2x: Fix unrecognized SFP+ module after driver is loaded
  bnx2x: Fix potential incorrect link speed provision
  bnx2x: Restore global registers back to default.
  bnx2x: Fix link down in 57712 following LFA
  bnx2x: Fix 57810 1G-KR link against certain switches.
  ixgbe: PTP get_ts_info missing software support
  ...

When sending data into a tcp socket in repair state we should check
for the amount of data being 0 explicitly. Otherwise we'll have an skb
with seq == end_seq in rcv queue, but tcp doesn't expect this to happen
(in particular a warn_on in tcp_recvmsg shoots).

Signed-off-by: Pavel Emelyanov <xemul@parallels.com>
Reported-by: Giorgos Mavrikas <gmavrikas@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>