The send buffer was being leaked; fix it.
Signed-off-by: NK. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: NHaiyang Zhang <haiyangz@microsoft.com>
Reported-by: NJason Wang <jasowang@redhat.com>
Acked-by: NJason Wang <jasowang@redhat.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

There is bug in the definition of struct dm_info_msg. This patch fixes
the definition of this structure and makes the corresponding adjustments.
Signed-off-by: NK. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: NHaiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Users of mei_amthif_read() expect it leaves dev->device_lock held,
while there is a path where mei_amthif_read() unlocks device_lock
and returns -ERESTARTSYS.

The patch move code locking device_lock back before the return.

Found by Linux Driver Verification project (linuxtesting.org).
Signed-off-by: NAlexey Khoroshilov <khoroshilov@ispras.ru>
Cc: Sedat Dilek <sedat.dilek@gmail.com>
Acked-by: NTomas Winkler <tomas.winkler@intel.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Pull LED fix from Bryan Wu.

* 'fixes-for-3.8' of git://git.kernel.org/pub/scm/linux/kernel/git/cooloney/linux-leds:
  leds: leds-gpio: set devm_gpio_request_one() flags param correctly

commit a99d76f9 leds: leds-gpio: use gpio_request_one

changed the leds-gpio driver to use gpio_request_one() instead
of gpio_request() + gpio_direction_output()

Unfortunately, it also made a semantic change that breaks the
leds-gpio driver.

The gpio_request_one() flags parameter was set to:

GPIOF_DIR_OUT | (led_dat->active_low ^ state)

Since GPIOF_DIR_OUT is 0, the final flags value will just be the
XOR'ed value of led_dat->active_low and state.

This value were used to distinguish between HIGH/LOW output initial
level and call gpio_direction_output() accordingly.

With this new semantic gpio_request_one() will take the flags value
of 1 as a configuration of input direction (GPIOF_DIR_IN) and will
call gpio_direction_input() instead of gpio_direction_output().

int gpio_request_one(unsigned gpio, unsigned long flags, const char *label)
{
..
	if (flags & GPIOF_DIR_IN)
		err = gpio_direction_input(gpio);
	else
		err = gpio_direction_output(gpio,
				(flags & GPIOF_INIT_HIGH) ? 1 : 0);
..
}

The right semantic is to evaluate led_dat->active_low ^ state and
set the output initial level explicitly.
Signed-off-by: NJavier Martinez Canillas <javier.martinez@collabora.co.uk>
Reported-by: NArnaud Patard <arnaud.patard@rtp-net.org>
Tested-by: NEzequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: NBryan Wu <cooloney@gmail.com>

Pull watchdog fixes from Wim Van Sebroeck:
 "This fixes some small errors in the new da9055 driver, eliminates a
  compiler warning and adds DT support for the twl4030_wdt driver (so
  that we can have multiple watchdogs with DT on the omap platforms)."

* git://www.linux-watchdog.org/linux-watchdog:
  watchdog: twl4030_wdt: add DT support
  watchdog: omap_wdt: eliminate unused variable and a compiler warning
  watchdog: da9055: Don't update wdt_dev->timeout in da9055_wdt_set_timeout error path
  watchdog: da9055: Fix invalid free of devm_ allocated data

Pull PCI updates from Bjorn Helgaas:
 "Some fixes for v3.8.  They include a fix for the new SR-IOV sysfs
  management support, an expanded quirk for Ricoh SD card readers, a
  Stratus DMI quirk fix, and a PME polling fix."

* tag '3.8-pci-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci:
  PCI: Reduce Ricoh 0xe822 SD card reader base clock frequency to 50MHz
  PCI/PM: Do not suspend port if any subordinate device needs PME polling
  PCI: Add PCIe Link Capability link speed and width names
  PCI: Work around Stratus ftServer broken PCIe hierarchy (fix DMI check)
  PCI: Remove spurious error for sriov_numvfs store and simplify flow

Commit 56c176c9 ("UAPI: strip the _UAPI prefix from header guards
during header installation") strips the _UAPI prefix from header guards,
but only if there's a single space between the cpp directive and the
label.

Make it more flexible and able to handle tabs and multiple white space
characters.
Signed-off-by: NDavid Howells <dhowell@redhat.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Empty files can get deleted by the patch program, so remove empty Kbuild
files and their links from the parent Kbuilds.
Signed-off-by: NDavid Howells <dhowells@redhat.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Merge tag 'ecryptfs-3.8-rc2-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tyhicks/ecryptfs

Pull ecryptfs fixes from Tyler Hicks:
 "Two self-explanatory fixes and a third patch which improves
  performance: when overwriting a full page in the eCryptfs page cache,
  skip reading in and decrypting the corresponding lower page."

* tag 'ecryptfs-3.8-rc2-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tyhicks/ecryptfs:
  fs/ecryptfs/crypto.c: make ecryptfs_encode_for_filename() static
  eCryptfs: fix to use list_for_each_entry_safe() when delete items
  eCryptfs: Avoid unnecessary disk read and data decryption during writing

Pull Ceph fixes from Sage Weil:
 "Two of Alex's patches deal with a race when reseting server
  connections for open RBD images, one demotes some non-fatal BUGs to
  WARNs, and my patch fixes a protocol feature bit failure path."

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client:
  libceph: fix protocol feature mismatch failure path
  libceph: WARN, don't BUG on unexpected connection states
  libceph: always reset osds when kicking
  libceph: move linger requests sooner in kick_requests()

Sasha was fuzzing with trinity and reported the following problem:

  BUG: sleeping function called from invalid context at kernel/mutex.c:269
  in_atomic(): 1, irqs_disabled(): 0, pid: 6361, name: trinity-main
  2 locks held by trinity-main/6361:
   #0:  (&mm->mmap_sem){++++++}, at: [<ffffffff810aa314>] __do_page_fault+0x1e4/0x4f0
   #1:  (&(&mm->page_table_lock)->rlock){+.+...}, at: [<ffffffff8122f017>] handle_pte_fault+0x3f7/0x6a0
  Pid: 6361, comm: trinity-main Tainted: G        W
  3.7.0-rc2-next-20121024-sasha-00001-gd95ef01-dirty #74
  Call Trace:
    __might_sleep+0x1c3/0x1e0
    mutex_lock_nested+0x29/0x50
    mpol_shared_policy_lookup+0x2e/0x90
    shmem_get_policy+0x2e/0x30
    get_vma_policy+0x5a/0xa0
    mpol_misplaced+0x41/0x1d0
    handle_pte_fault+0x465/0x6a0

This was triggered by a different version of automatic NUMA balancing
but in theory the current version is vunerable to the same problem.

do_numa_page
  -> numa_migrate_prep
    -> mpol_misplaced
      -> get_vma_policy
        -> shmem_get_policy

It's very unlikely this will happen as shared pages are not marked
pte_numa -- see the page_mapcount() check in change_pte_range() -- but
it is possible.

To address this, this patch restores sp->lock as originally implemented
by Kosaki Motohiro.  In the path where get_vma_policy() is called, it
should not be calling sp_alloc() so it is not necessary to treat the PTL
specially.
Signed-off-by: NKOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Tested-by: NKOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: NMel Gorman <mgorman@suse.de>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Pull ext4 bug fixes from Ted Ts'o:
 "Various bug fixes for ext4.  Perhaps the most serious bug fixed is one
  which could cause file system corruptions when performing file punch
  operations."

* tag 'ext4_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
  ext4: avoid hang when mounting non-journal filesystems with orphan list
  ext4: lock i_mutex when truncating orphan inodes
  ext4: do not try to write superblock on ro remount w/o journal
  ext4: include journal blocks in df overhead calcs
  ext4: remove unaligned AIO warning printk
  ext4: fix an incorrect comment about i_mutex
  ext4: fix deadlock in journal_unmap_buffer()
  ext4: split off ext4_journalled_invalidatepage()
  jbd2: fix assertion failure in jbd2_journal_flush()
  ext4: check dioread_nolock on remount
  ext4: fix extent tree corruption caused by hole punch

Remove the unused argument (formerly no_context) from mpol_parse_str()
and from mpol_to_str().
Signed-off-by: NHugh Dickins <hughd@google.com>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Recently I suggested using "mount -o remount,mpol=local /tmp" in NUMA
mempolicy testing.  Very nasty.  Reading /proc/mounts, /proc/pid/mounts
or /proc/pid/mountinfo may then corrupt one bit of kernel memory, often
in a page table (causing "Bad swap" or "Bad page map" warning or "Bad
pagetable" oops), sometimes in a vm_area_struct or rbnode or somewhere
worse.  "mpol=prefer" and "mpol=prefer:Node" are equally toxic.

Recent NUMA enhancements are not to blame: this dates back to 2.6.35,
when commit e17f74af "mempolicy: don't call mpol_set_nodemask() when
no_context" skipped mpol_parse_str()'s call to mpol_set_nodemask(),
which used to initialize v.preferred_node, or set MPOL_F_LOCAL in flags.
With slab poisoning, you can then rely on mpol_to_str() to set the bit
for node 0x6b6b, probably in the next page above the caller's stack.

mpol_parse_str() is only called from shmem_parse_options(): no_context
is always true, so call it unused for now, and remove !no_context code.
Set v.nodes or v.preferred_node or MPOL_F_LOCAL as mpol_to_str() might
expect.  Then mpol_to_str() can ignore its no_context argument also,
the mpol being appropriately initialized whether contextualized or not.
Rename its no_context unused too, and let subsequent patch remove them
(that's not needed for stable backporting, which would involve rejects).

I don't understand why MPOL_LOCAL is described as a pseudo-policy:
it's a reasonable policy which suffers from a confusing implementation
in terms of MPOL_PREFERRED with MPOL_F_LOCAL.  I believe this would be
much more robust if MPOL_LOCAL were recognized in switch statements
throughout, MPOL_F_LOCAL deleted, and MPOL_PREFERRED use the (possibly
empty) nodes mask like everyone else, instead of its preferred_node
variant (I presume an optimization from the days before MPOL_LOCAL).
But that would take me too long to get right and fully tested.
Signed-off-by: NHugh Dickins <hughd@google.com>
Cc: stable@vger.kernel.org
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

EPOLL_CTL_MOD sets the interest mask before calling f_op->poll() to
ensure events are not missed.  Since the modifications to the interest
mask are not protected by the same lock as ep_poll_callback, we need to
ensure the change is visible to other CPUs calling ep_poll_callback.

We also need to ensure f_op->poll() has an up-to-date view of past
events which occured before we modified the interest mask.  So this
barrier also pairs with the barrier in wq_has_sleeper().

This should guarantee either ep_poll_callback or f_op->poll() (or both)
will notice the readiness of a recently-ready/modified item.

This issue was encountered by Andreas Voellmy and Junchang(Jason) Wang in:
http://thread.gmane.org/gmane.linux.kernel/1408782/Signed-off-by: NEric Wong <normalperson@yhbt.net>
Cc: Hans Verkuil <hans.verkuil@cisco.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davide Libenzi <davidel@xmailserver.org>
Cc: Hans de Goede <hdegoede@redhat.com>
Cc: Mauro Carvalho Chehab <mchehab@infradead.org>
Cc: David Miller <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Andreas Voellmy <andreas.voellmy@yale.edu>
Tested-by: N"Junchang(Jason) Wang" <junchang.wang@yale.edu>
Cc: netdev@vger.kernel.org
Cc: linux-fsdevel@vger.kernel.org
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Add DT support for twl4030_wdt. This is needed to get twl4030_wdt to
probe when booting with DT.
Signed-off-by: NAaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

We forgot to delete this in the commit 4f4753d9 (watchdog: omap_wdt:
convert to devm_ functions), and as a result the following compilation
warning was introduced:

drivers/watchdog/omap_wdt.c: In function 'omap_wdt_remove':
drivers/watchdog/omap_wdt.c:299:19: warning: unused variable 'res' [-Wunused-variable]
Signed-off-by: NAaro Koskinen <aaro.koskinen@iki.fi>
Reviewed-by: NPaul Walmsley <paul@pwsan.com>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

Otherwise, WDIOC_GETTIMEOUT returns wrong value if set_timeout fails.
This patch also removes unnecessary ret variable in da9055_wdt_ping function.
Signed-off-by: NAxel Lin <axel.lin@ingics.com>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

It is not required to free devm_ allocated data. Since kref_put
needs a valid release function, da9055_wdt_release_resources()
is not deleted.
Signed-off-by: NAxel Lin <axel.lin@ingics.com>
Signed-off-by: NWim Van Sebroeck <wim@iguana.be>

Pull DRM update from Dave Airlie:
 "This is a bit larger due to me not bothering to do anything since
  before Xmas, and other people working too hard after I had clearly
  given up.

  It's got the 3 main x86 driver fixes pulls, and a bunch of tegra
  fixes, doesn't fix the Ironlake bug yet, but that does seem to be
  getting closer.

   - radeon: gpu reset fixes and userspace packet support
   - i915: watermark fixes, workarounds, i830/845 fix,
   - nouveau: nvd9/kepler microcode fixes, accel is now enabled and
     working, gk106 support
   - tegra: misc fixes."

* 'drm-next' of git://people.freedesktop.org/~airlied/linux: (34 commits)
  Revert "drm: tegra: protect DC register access with mutex"
  drm: tegra: program only one window during modeset
  drm: tegra: clean out old gem prototypes
  drm: tegra: remove redundant tegra2_tmds_config entry
  drm: tegra: protect DC register access with mutex
  drm: tegra: don't leave clients host1x member uninitialized
  drm: tegra: fix front_porch <-> back_porch mixup
  drm/nve0/graph: fix fuc, and enable acceleration on all known chipsets
  drm/nvc0/graph: fix fuc, and enable acceleration on GF119
  drm/nouveau/bios: cache ramcfg strap on later chipsets
  drm/nouveau/mxm: silence output if no bios data
  drm/nouveau/bios: parse/display extra version component
  drm/nouveau/bios: implement opcode 0xa9
  drm/nouveau/bios: update gpio parsing apis to match current design
  drm/nouveau: initial support for GK106
  drm/radeon: add WAIT_UNTIL to evergreen VM safe reg list
  drm/i915: disable shrinker lock stealing for create_mmap_offset
  drm/i915: optionally disable shrinker lock stealing
  drm/i915: fix flags in dma buf exporting
  drm/radeon: add support for MEM_WRITE packet
  ...

Pull late ARM cleanups for omap from Olof Johansson:
 "From Tony Lindgren:

  Here are few more patches to finish the omap changes for multiplatform
  conversion that are not strictly fixes, but were too complex to do
  with the dependencies during the merge window.  Those are to move of
  serial-omap.h to platform_data, and the removal of remaining
  cpu_is_omap macro usage outside mach-omap2.

  Then there are several trivial fixes for typos and few minimal
  omap2plus_defconfig updates."

* tag 'omap-late-cleanups' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc:
  arch/arm/mach-omap2/dpll3xxx.c: drop if around WARN_ON
  OMAP2: Fix a typo - replace regist with register.
  ARM/omap: use module_platform_driver macro
  ARM: OMAP2+: PMU: Remove unused header
  ARM: OMAP4: remove duplicated include from omap_hwmod_44xx_data.c
  ARM: OMAP2+: omap2plus_defconfig: enable twl4030 SoC audio
  ARM: OMAP2+: omap2plus_defconfig: Add tps65217 support
  ARM: OMAP2+: enable devtmpfs and devtmpfs automount
  ARM: OMAP2+: omap_twl: Change TWL4030_MODULE_PM_RECEIVER to TWL_MODULE_PM_RECEIVER
  ARM: OMAP2+: Drop plat/cpu.h for omap2plus
  ARM: OMAP: Split fb.c to remove last remaining cpu_is_omap usage
  MAINTAINERS: Add an entry for omap related .dts files

Pull ARM SoC fixes from Olof Johansson:
 "It's been quiet over the holidays, but we have had a couple of trivial
  fixes coming in for the newly introduced sunxi platform; one to add it
  to the multiplatform defconfig for build coverage, and one fixup for
  device tree strings."

* tag 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc:
  sunxi: Change the machine compatible string.
  ARM: multi_v7_defconfig: Add ARCH_SUNXI

This reverts commit 83c0bcb6.

Lucas pointed out this was a mistake, and I missed the discussion,
so just revert it out to save a rebase.
Signed-off-by: NDave Airlie <airlied@redhat.com>

The intention is to program exactly WIN_A, not WIN_A and possibly
others.
Signed-off-by: NLucas Stach <dev@lynxeye.de>
Signed-off-by: NDave Airlie <airlied@redhat.com>

There is no gem.c anymore, those functions are implemented by the
drm_cma_helpers now.
Signed-off-by: NLucas Stach <dev@lynxeye.de>
Signed-off-by: NDave Airlie <airlied@redhat.com>

The 720p and 1080p entries are completely redundant, as we are matching
the table entries against <=pclk.
Also generalize the comment, as we are using those table entries even
when driving other modes than the standard TV ones.
Signed-off-by: NLucas Stach <dev@lynxeye.de>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Window properties are programmed through a shared aperture and have to
happen atomically. Also we do the read-update-write dance on some of the
shared regs.
To make sure that different functions don't stumble over each other
protect the register access with a mutex.
Signed-off-by: NLucas Stach <dev@lynxeye.de>
Signed-off-by: NDave Airlie <airlied@redhat.com>

No real problem for now, as nothing is using this, but leaving it
unitialized is asking for trouble later on.
Signed-off-by: NLucas Stach <dev@lynxeye.de>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Fixes wrong picture offset observed when using HDMI output with a
Technisat HD TV.
Signed-off-by: NLucas Stach <dev@lynxeye.de>
Acked-by: NMark Zhang <markz@nvidia.com>
Tested-by: NMark Zhang <markz@nvidia.com>
Signed-off-by: NDave Airlie <airlied@redhat.com>

Some fixes for 3.8:
- Watermark fixups from Chris Wilson (4 pieces).
- 2 snb workarounds, seem to be recently added to our internal DB.
- workaround for the infamous i830/i845 hang, seems now finally solid!
  Based on Chris' fix for SNA, now also for UXA/mesa&old SNA.
- Some more fixlets for shrinker-pulls-the-rug issues (Chris&me).
- Fix dma-buf flags when exporting (you).
- Disable the VGA plane if it's enabled on lid open - similar fix in
  spirit to the one I've sent you last weeek, BIOS' really like to mess
  with the display when closing the lid (awesome debug work from Krzysztof
  Mazur).

* 'drm-intel-fixes' of git://people.freedesktop.org/~danvet/drm-intel:
  drm/i915: disable shrinker lock stealing for create_mmap_offset
  drm/i915: optionally disable shrinker lock stealing
  drm/i915: fix flags in dma buf exporting
  i915: ensure that VGA plane is disabled
  drm/i915: Preallocate the drm_mm_node prior to manipulating the GTT drm_mm manager
  drm: Export routines for inserting preallocated nodes into the mm manager
  drm/i915: don't disable disconnected outputs
  drm/i915: Implement workaround for broken CS tlb on i830/845
  drm/i915: Implement WaSetupGtModeTdRowDispatch
  drm/i915: Implement WaDisableHiZPlanesWhenMSAAEnabled
  drm/i915: Prefer CRTC 'active' rather than 'enabled' during WM computations
  drm/i915: Clear self-refresh watermarks when disabled
  drm/i915: Double the cursor self-refresh latency on Valleyview
  drm/i915: Fixup cursor latency used for IVB lp3 watermarks

Misc fixes for reset and new packets for userspace usage.

* 'drm-fixes-3.8' of git://people.freedesktop.org/~agd5f/linux:
  drm/radeon: add WAIT_UNTIL to evergreen VM safe reg list
  drm/radeon: add support for MEM_WRITE packet
  drm/radeon: restore modeset late in GPU reset path
  drm/radeon: avoid deadlock in pm path when waiting for fence
  drm/radeon: don't leave fence blocked process on failed GPU reset

Merge branch 'drm-nouveau-fixes-3.8' of git://anongit.freedesktop.org/git/nouveau/linux-2.6 into drm-next

Fixes the accel support for nvd9 + kepler chipsets, also fixes GK106 support.

* 'drm-nouveau-fixes-3.8' of git://anongit.freedesktop.org/git/nouveau/linux-2.6:
  drm/nve0/graph: fix fuc, and enable acceleration on all known chipsets
  drm/nvc0/graph: fix fuc, and enable acceleration on GF119
  drm/nouveau/bios: cache ramcfg strap on later chipsets
  drm/nouveau/mxm: silence output if no bios data
  drm/nouveau/bios: parse/display extra version component
  drm/nouveau/bios: implement opcode 0xa9
  drm/nouveau/bios: update gpio parsing apis to match current design
  drm/nouveau: initial support for GK106

An unintended consequence of commit 4ae0a48b ("mm: modify
pgdat_balanced() so that it also handles order-0") is that
wait_iff_congested() can now be called with NULL 'struct zone *'
producing kernel oops like this:

  BUG: unable to handle kernel NULL pointer dereference
  IP: [<ffffffff811542d9>] wait_iff_congested+0x59/0x140

This trivial patch fixes it.
Reported-by: NZhouping Liu <zliu@redhat.com>
Reported-and-tested-by: NSedat Dilek <sedat.dilek@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: NZlatko Calusic <zlatko.calusic@iskon.hr>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

From Maxime Ripard:
Fixes for the sunxi core to be merged in 3.8-rc2

* tag 'sunxi-fixes-for-3.8-rc2' of git://github.com/mripard/linux:
  sunxi: Change the machine compatible string.
  ARM: multi_v7_defconfig: Add ARCH_SUNXI

We should not set con->state to CLOSED here; that happens in
ceph_fault() in the caller, where it first asserts that the state
is not yet CLOSED.  Avoids a BUG when the features don't match.

Since the fail_protocol() has become a trivial wrapper, replace
calls to it with direct calls to reset_connection().
Signed-off-by: NSage Weil <sage@inktank.com>
Reviewed-by: NAlex Elder <elder@inktank.com>

A number of assertions in the ceph messenger are implemented with
BUG_ON(), killing the system if connection's state doesn't match
what's expected.  At this point our state model is (evidently) not
well understood enough for these assertions to trigger a BUG().
Convert all BUG_ON(con->state...) calls to be WARN_ON(con->state...)
so we learn about these issues without killing the machine.

We now recognize that a connection fault can occur due to a socket
closure at any time, regardless of the state of the connection.  So
there is really nothing we can assert about the state of the
connection at that point so eliminate that assertion.
Reported-by: NUgis <ugis22@gmail.com>
Tested-by: NUgis <ugis22@gmail.com>
Signed-off-by: NAlex Elder <elder@inktank.com>
Reviewed-by: NSage Weil <sage@inktank.com>

When ceph_osdc_handle_map() is called to process a new osd map,
kick_requests() is called to ensure all affected requests are
updated if necessary to reflect changes in the osd map.  This
happens in two cases:  whenever an incremental map update is
processed; and when a full map update (or the last one if there is
more than one) gets processed.

In the former case, the kick_requests() call is followed immediately
by a call to reset_changed_osds() to ensure any connections to osds
affected by the map change are reset.  But for full map updates
this isn't done.

Both cases should be doing this osd reset.

Rather than duplicating the reset_changed_osds() call, move it into
the end of kick_requests().
Signed-off-by: NAlex Elder <elder@inktank.com>
Reviewed-by: NSage Weil <sage@inktank.com>

The kick_requests() function is called by ceph_osdc_handle_map()
when an osd map change has been indicated.  Its purpose is to
re-queue any request whose target osd is different from what it
was when it was originally sent.

It is structured as two loops, one for incomplete but registered
requests, and a second for handling completed linger requests.
As a special case, in the first loop if a request marked to linger
has not yet completed, it is moved from the request list to the
linger list.  This is as a quick and dirty way to have the second
loop handle sending the request along with all the other linger
requests.

Because of the way it's done now, however, this quick and dirty
solution can result in these incomplete linger requests never
getting re-sent as desired.  The problem lies in the fact that
the second loop only arranges for a linger request to be sent
if it appears its target osd has changed.  This is the proper
handling for *completed* linger requests (it avoids issuing
the same linger request twice to the same osd).

But although the linger requests added to the list in the first loop
may have been sent, they have not yet completed, so they need to be
re-sent regardless of whether their target osd has changed.

The first required fix is we need to avoid calling __map_request()
on any incomplete linger request.  Otherwise the subsequent
__map_request() call in the second loop will find the target osd
has not changed and will therefore not re-send the request.

Second, we need to be sure that a sent but incomplete linger request
gets re-sent.  If the target osd is the same with the new osd map as
it was when the request was originally sent, this won't happen.
This can be fixed through careful handling when we move these
requests from the request list to the linger list, by unregistering
the request *before* it is registered as a linger request.  This
works because a side-effect of unregistering the request is to make
the request's r_osd pointer be NULL, and *that* will ensure the
second loop actually re-sends the linger request.

Processing of such a request is done at that point, so continue with
the next one once it's been moved.
Signed-off-by: NAlex Elder <elder@inktank.com>
Reviewed-by: NSage Weil <sage@inktank.com>