The following bit in ceph_msg_revoke_incoming() is unsafe:

    struct ceph_connection *con = msg->con;
    if (!con)
            return;
    mutex_lock(&con->mutex);
    <more msg->con use>

There is nothing preventing con from getting destroyed right after
msg->con test.  One easy way to reproduce this is to disable message
signing only on the server side and try to map an image.  The system
will go into a

    libceph: read_partial_message ffff880073f0ab68 signature check failed
    libceph: osd0 192.168.255.155:6801 bad crc/signature
    libceph: read_partial_message ffff880073f0ab68 signature check failed
    libceph: osd0 192.168.255.155:6801 bad crc/signature

loop which has to be interrupted with Ctrl-C.  Hit Ctrl-C and you are
likely to end up with a random GP fault if the reset handler executes
"within" ceph_msg_revoke_incoming():

                     <yet another reply w/o a signature>
                                   ...
          <Ctrl-C>
    rbd_obj_request_end
      ceph_osdc_cancel_request
        __unregister_request
          ceph_osdc_put_request
            ceph_msg_revoke_incoming
                                   ...
                                osd_reset
                                  __kick_osd_requests
                                    __reset_osd
                                      remove_osd
                                        ceph_con_close
                                          reset_connection
                                            <clear con->in_msg->con>
                                            <put con ref>
                                              put_osd
                                                <free osd/con>
              <msg->con use> <-- !!!

If ceph_msg_revoke_incoming() executes "before" the reset handler,
osd/con will be leaked because ceph_msg_revoke_incoming() clears
con->in_msg but doesn't put con ref, while reset_connection() only puts
con ref if con->in_msg != NULL.

The current msg->con scheme was introduced by commits 38941f80
("libceph: have messages point to their connection") and 92ce034b
("libceph: have messages take a connection reference"), which defined
when messages get associated with a connection and when that
association goes away.  Part of the problem is that this association is
supposed to go away in much too many places; closing this race entirely
requires either a rework of the existing or an addition of a new layer
of synchronization.

In lieu of that, we can make it *much* less likely to hit by
disassociating messages only on their destruction and resend through
a different connection.  This makes the code simpler and is probably
a good thing to do regardless - this patch adds a msg_con_set() helper
which is is called from only three places: ceph_con_send() and
ceph_con_in_msg_alloc() to set msg->con and ceph_msg_release() to clear
it.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

Support for message signing was merged into 3.19, along with
nocephx_require_signatures option.  But, all that option does is allow
the kernel client to talk to clusters that don't support MSG_AUTH
feature bit.  That's pretty useless, given that it's been supported
since bobtail.

Meanwhile, if one disables message signing on the server side with
"cephx sign messages = false", it becomes impossible to use the kernel
client since it expects messages to be signed if MSG_AUTH was
negotiated.  Add nocephx_sign_messages option to support this use case.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

supported_features and required_features serve no purpose at all, while
nocrc and tcp_nodelay belong to ceph_options::flags.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

I don't see a way for auth->authorizer to be NULL in
ceph_x_sign_message() or ceph_x_check_message_signature().
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

We can use msg->con instead - at the point we sign an outgoing message
or check the signature on the incoming one, msg->con is always set.  We
wouldn't know how to sign a message without an associated session (i.e.
msg->con == NULL) and being able to sign a message using an explicitly
provided authorizer is of no use.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

This patch changes the osd_req_op_data() macro to not evaluate
arguments more than once in order to follow the kernel coding style.
Signed-off-by: Ioana Ciornei <ciorneiioana@gmail.com>
Reviewed-by: Alex Elder <elder@linaro.org>
[idryomov@gmail.com: changelog, formatting]
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

If we get a unsafe reply for request that created/modified inode,
add the unsafe request to a list in the newly created/modified
inode. So we can make fsync() wait these unsafe requests.
Signed-off-by: Yan, Zheng <zyan@redhat.com>

Previously we add request to i_unsafe_dirops when registering
request. So ceph_fsync() also waits for imcomplete requests.
This is unnecessary, ceph_fsync() only needs to wait unsafe
requests.
Signed-off-by: Yan, Zheng <zyan@redhat.com>

Commit ae385eaf ("libceph: store session key in cephx authorizer")
introduced ceph_x_authorizer::session_key, but didn't update all the
exit/error paths.  Introduce ceph_x_authorizer_cleanup() to encapsulate
ceph_x_authorizer cleanup and switch to it.  This fixes ceph_x_destroy(),
which currently always leaks key and ceph_x_build_authorizer() error
paths.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Yan, Zheng <zyan@redhat.com>

ceph_check_caps() invalidate page cache when inode is not used
by any open file. This behaviour is not friendly for workload
that repeatly read files.
Signed-off-by: Yan, Zheng <zyan@redhat.com>

Commit d1cf5788 ("rbd: set mapping info earlier") defined
rbd_dev_mapping_clear(), but, just a few days after, commit
f35a4dee ("rbd: set the mapping size and features later") moved
rbd_dev_mapping_set() calls and added another rbd_dev_mapping_clear()
call instead of moving the old one.  Around the same time, another
duplicate was introduced in rbd_dev_device_release() - kill both.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

No point in providing an empty device_type::release callback and then
setting device::release for each rbd_dev dynamically.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

struct rbd_device has struct device embedded in it, which means it's
part of kobject universe and has an unpredictable life cycle.  Freeing
its memory outside of the release callback is flawed, yet commits
200a6a8b ("rbd: don't destroy rbd_dev in device release function")
and 8ad42cd0 ("rbd: don't have device release destroy rbd_dev")
moved rbd_dev_destroy() out to rbd_dev_image_release().

This commit reverts most of that, the key points are:

- rbd_dev->dev is initialized in rbd_dev_create(), making it possible
  to use rbd_dev_destroy() - which is just a put_device() - both before
  we register with device core and after.

- rbd_dev_release() (the release callback) is the only place we
  kfree(rbd_dev).  It's also where we do module_put(), keeping the
  module unload race window as small as possible.

- We pin the module in rbd_dev_create(), but only for mapping
  rbd_dev-s.  Moving image related stuff out of struct rbd_device into
  another struct which isn't tied with sysfs and device core is long
  overdue, but until that happens, this will keep rbd module refcount
  (which users can observe with lsmod) sane.

Fixes: http://tracker.ceph.com/issues/12697

Cc: Alex Elder <elder@linaro.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

Returning pool id (i.e. >= 0) from a sysfs ->store() callback makes
userspace think it needs to retry the write.  Fix it - it's a leftover
from the times when the equivalent of rbd_dev_create() was the first
action in rbd_add().
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

Use local variable cursor in place of &msg->cursor in
read_partial_msg_data() and write_partial_msg_data().
Signed-off-by: Shraddha Barke <shraddha.6596@gmail.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

Since handle_reply() does not use its con argument, remove it.
Signed-off-by: Shraddha Barke <shraddha.6596@gmail.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

Both ceph_sync_direct_write and ceph_sync_read iterate iovec elements
one by one, send one OSD request for each iovec. This is sub-optimal,
We can combine serveral iovec into one page vector, and send an OSD
request for the whole page vector.
Signed-off-by: Zhu, Caifeng <zhucaifeng@unissoft-nj.com>
Signed-off-by: Yan, Zheng <zyan@redhat.com>

create_request_message() computes the maximum length of a message,
but uses the wrong type for the time stamp: sizeof(struct timespec)
may be 8 or 16 depending on the architecture, while sizeof(struct
ceph_timespec) is always 8, and that is what gets put into the
message.

Found while auditing the uses of timespec for y2038 problems.

Fixes: b8e69066 ("ceph: include time stamp in every MDS request")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Yan, Zheng <zyan@redhat.com>

Signed-off-by: Geliang Tang <geliangtang@163.com>
Signed-off-by: Yan, Zheng <zyan@redhat.com>

Remove unneeded NULL test.

The semantic patch that makes this change is as follows:
(http://coccinelle.lip6.fr/)

// <smpl>
@@ expression x; @@
-if (x != NULL) {
  \(kmem_cache_destroy\|mempool_destroy\|dma_pool_destroy\)(x);
  x = NULL;
-}
// </smpl>
Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

Pull memremap fix from Dan Williams:
 "The new memremap() api introduced in the 4.3 cycle to unify/replace
  ioremap_cache() and ioremap_wt() is mishandling the highmem case.
  This patch has received a build success notification from a
  0day-kbuild-robot run and has received an ack from Ard"

From the commit message:
 "The impact of this bug is low for now since the pmem driver is the
  only user of memremap(), but this is important to fix before more
  conversions to memremap arrive in 4.4"

* 'libnvdimm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/nvdimm/nvdimm:
  memremap: fix highmem support

Pull x86 fixes from Thomas Gleixner:
 "This set of updates contains:

   - Another bugfix for the pathologic vm86 machinery.  Clear
     thread.vm86 on fork to prevent corrupting the parent state.  This
     comes along with an update to the vm86 selftest case

   - Fix another corner case in the ioapic setup code which causes a
     boot crash on some oddball systems

   - Fix the fallout from the dma allocation consolidation work, which
     leads to a NULL pointer dereference when the allocation code is
     called with a NULL device"

* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/vm86: Set thread.vm86 to NULL on fork/clone
  selftests/x86: Add a fork() to entry_from_vm86 to catch fork bugs
  x86/ioapic: Prevent NULL pointer dereference in setup_ioapic_dest()
  x86/dma-mapping: Fix arch_dma_alloc_attrs() oops with NULL dev

Pull timer fixes from Thomas Gleixner:
 "The last round of minimalistic fixes for clocksource drivers:

   - Prevent multiple shutdown of the sh_mtu2 clocksource

   - Annotate a bunch of clocksource/schedclock functions with notrace
     to prevent an annoying ftrace recursion issue"

* 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  clocksource/drivers/sh_mtu2: Fix multiple shutdown call issue
  clocksource/drivers/digicolor: Prevent ftrace recursion
  clocksource/drivers/fsl_ftm_timer: Prevent ftrace recursion
  clocksource/drivers/vf_pit_timer: Prevent ftrace recursion
  clocksource/drivers/prima2: Prevent ftrace recursion
  clocksource/drivers/samsung_pwm_timer: Prevent ftrace recursion
  clocksource/drivers/pistachio: Prevent ftrace recursion
  clocksource/drivers/arm_global_timer: Prevent ftrace recursion

Pull irq fixes from Thomas Gleixner:
 "The last two one-liners for 4.3 from the irqchip space:

   - Regression fix for armada SoC which addresses the fallout from the
     set_irq_flags() cleanup

   - Add the missing propagation of the irq_set_type() callback to the
     parent interrupt controller of the tegra interrupt chip"

* 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  irqchip/tegra: Propagate IRQ type setting to parent
  irqchip/armada-370-xp: Fix regression by clearing IRQ_NOAUTOEN

Pull ARM SoC fixes from Olof Johansson:
 "This should be our final batch of fixes for 4.3:

   - A patch from Sudeep Holla that fixes annotation of wakeup sources
     properly, old unused format seems to have spread through copying.

   - Two patches from Tony for OMAP.  One dealing with MUSB setup
     problems due to runtime PM being enabled too early on the parent
     device.  The other fixes IRQ numbering for OMAP1"

* tag 'armsoc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc:
  usb: musb: omap2430: Fix regression caused by driver core change
  ARM: OMAP1: fix incorrect INT_DMA_LCD
  ARM: dts: fix gpio-keys wakeup-source property

Pull SCSI fixes from James Bottomley:
 "This is three essential bug fixes for various SCSI parts.

  The only affected users are SCSI multi-path via device handler
  (basically all the enterprise) and mvsas users.  The dh bugs are an
  async entanglement in boot resulting in a serious WARN_ON trip and a
  use after free on remove leading to a crash with strict memory
  accounting.  The mvsas bug manifests as a null deref oops but only on
  abort sequences; however, these can commonly occur with SATA attached
  devices, hence the fix"

* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  scsi_dh: don't try to load a device handler during async probing
  scsi_dh: fix use-after-free when removing scsi device
  mvsas: Fix NULL pointer dereference in mvs_slot_task_free

Pull md bug fixes from Neil Brown:
 "Two more bug fixes for md.

  One bugfix for a list corruption in raid5 because of incorrect
  locking.

  Other for possible data corruption when a recovering device is failed,
  removed, and re-added.

  Both tagged for -stable"

* tag 'md/4.3-rc7-fixes' of git://neil.brown.name/md:
  Revert "md: allow a partially recovered device to be hot-added to an array."
  md/raid5: fix locking in handle_stripe_clean_event()

Pull drm fixes from Dave Airlie:
 "Two drm atomic core fixes.

  And two radeon patches needed to fix a backlight regression on some
  older hardware"

* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux:
  drm: Correct arguments to list_tail_add in create blob ioctl
  drm: crtc: integer overflow in drm_property_create_blob()
  drm/radeon: fix dpms when driver backlight control is disabled
  drm/radeon: move bl encoder assignment into bl init

Pull Ceph fix from Sage Weil:
 "This sets the stable pages flag on the RBD block device when we have
  CRCs enabled.  (This is necessary since the default assumption for
  block devices changed in 3.9)"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client:
  rbd: require stable pages if message data CRCs are enabled

Pull overlayfs bug fixes from Miklos Szeredi:
 "This contains fixes for bugs that appeared in earlier kernels (all are
  marked for -stable)"

* 'overlayfs-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/vfs:
  ovl: free lower_mnt array in ovl_put_super
  ovl: free stack of paths in ovl_fill_super
  ovl: fix open in stacked overlay
  ovl: fix dentry reference leak
  ovl: use O_LARGEFILE in ovl_copy_up()

Pull networking fixes from David Miller:

 1) Fix two regressions in ipv6 route lookups, particularly wrt output
    interface specifications in the lookup key.  From David Ahern.

 2) Fix checks in ipv6 IPSEC tunnel pre-encap fragmentation, from
    Herbert Xu.

 3) Fix mis-advertisement of 1000BASE-T on bcm63xx_enet, from Simon
    Arlott.

 4) Some smsc phys misbehave with energy detect mode enabled, so add a
    DT property and disable it on such switches.  From Heiko Schocher.

 5) Fix TSO corruption on TX in mv643xx_eth, from Philipp Kirchhofer.

 6) Fix regression added by removal of openvswitch vport stats, from
    James Morse.

 7) Vendor Kconfig options should be bool, not tristate, from Andreas
    Schwab.

 8) Use non-_BH() net stats bump in tcp_xmit_probe_skb(), otherwise we
    barf during TCP REPAIR operations.

 9) Fix various bugs in openvswitch conntrack support, from Joe
    Stringer.

10) Fix NETLINK_LIST_MEMBERSHIPS locking, from David Herrmann.

11) Don't have VSOCK do sock_put() in interrupt context, from Jorgen
    Hansen.

12) Fix skb_realloc_headroom() failures properly in ISDN, from Karsten
    Keil.

13) Add some device IDs to qmi_wwan, from Bjorn Mork.

14) Fix ovs egress tunnel information when using lwtunnel devices, from
    Pravin B Shelar.

15) Add missing NETIF_F_FRAGLIST to macvtab feature list, from Jason
    Wang.

16) Fix incorrect handling of throw routes when the result of the throw
    cannot find a match, from Xin Long.

17) Protect ipv6 MTU calculations from wrap-around, from Hannes Frederic
    Sowa.

18) Fix failed autonegotiation on KSZ9031 micrel PHYs, from Nathan
    Sullivan.

19) Add missing memory barries in descriptor accesses or xgbe driver,
    from Thomas Lendacky.

20) Fix release conditon test in pppoe_release(), from Guillaume Nault.

21) Fix gianfar bugs wrt filter configuration, from Claudiu Manoil.

22) Fix violations of RX buffer alignment in sh_eth driver, from Sergei
    Shtylyov.

23) Fixing missing of_node_put() calls in various places around the
    networking, from Julia Lawall.

24) Fix incorrect leaf now walking in ipv4 routing tree, from Alexander
    Duyck.

25) RDS doesn't check pskb_pull()/pskb_trim() return values, from
    Sowmini Varadhan.

26) Fix VLAN configuration in mlx4 driver, from Jack Morgenstein.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (79 commits)
  ipv6: protect mtu calculation of wrap-around and infinite loop by rounding issues
  Revert "Merge branch 'ipv6-overflow-arith'"
  net/mlx4: Copy/set only sizeof struct mlx4_eqe bytes
  net/mlx4_en: Explicitly set no vlan tags in WQE ctrl segment when no vlan is present
  vhost: fix performance on LE hosts
  bpf: sample: define aarch64 specific registers
  amd-xgbe: Fix race between access of desc and desc index
  RDS-TCP: Recover correctly from pskb_pull()/pksb_trim() failure in rds_tcp_data_recv
  forcedeth: fix unilateral interrupt disabling in netpoll path
  openvswitch: Fix skb leak using IPv6 defrag
  ipv6: Export nf_ct_frag6_consume_orig()
  openvswitch: Fix double-free on ip_defrag() errors
  fib_trie: leaf_walk_rcu should not compute key if key is less than pn->key
  net: mv643xx_eth: add missing of_node_put
  ath6kl: add missing of_node_put
  net: phy: mdio: add missing of_node_put
  netdev/phy: add missing of_node_put
  net: netcp: add missing of_node_put
  net: thunderx: add missing of_node_put
  ipv6: gre: support SIT encapsulation
  ...

thread.vm86 points to per-task information -- the pointer should not
be copied on clone.

Fixes: d4ce0f26 ("x86/vm86: Move fields from 'struct kernel_vm86_struct' to 'struct vm86'")
Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Stas Sergeev <stsp@list.ru>
Link: http://lkml.kernel.org/r/71c5d6985d70ec8197c8d72f003823c81b7dcf99.1446270067.git.luto@kernel.orgSigned-off-by: Thomas Gleixner <tglx@linutronix.de>

Mere possession of vm86 state is strange.  Make sure that nothing
gets corrupted if we fork after calling vm86().
Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Stas Sergeev <stsp@list.ru>
Link: http://lkml.kernel.org/r/08f83295460a80e41dc5e3e81ec40d6844d316f5.1446270067.git.luto@kernel.orgSigned-off-by: Thomas Gleixner <tglx@linutronix.de>

Pull input layer fixes from Dmitry Torokhov:

 - a change to the ALPS driver where we had limit the quirk for
   trackstick handling from being active on all Dells to just a few
   models

 - a fix for a build dependency issue in the sur40 driver

 - a small clock handling fixup in the LPC32xx touchscreen driver

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
  Input: alps - only the Dell Latitude D420/430/620/630 have separate stick button bits
  Input: sur40 - add dependency on VIDEO_V4L2
  Input: lpc32xx_ts - fix warnings caused by enabling unprepared clock

Pull PCI fix from Bjorn Helgaas:
 "Sorry for this last-minute update; it's been in -next for quite a
  while, but I forgot about it until I started getting ready for the
  merge window.

  It's small and fixes a way a user could cause a panic via sysfs, so I
  think it's worth getting it in v4.3.

  NUMA:
    - Prevent out of bounds access in sysfs numa_node override (Sasha Levin)"

* tag 'pci-v4.3-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci:
  PCI: Prevent out of bounds access in numa_node override

Merge tag 'omap-for-v4.3/fixes-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap into fixes

Two omap regression fixes:

- Fix omap3 MUSB with DMA caused by driver core changes

- Fix LCD DMA interrupt number for omap1 that did not
  get changed for sparse IRQ changes

* tag 'omap-for-v4.3/fixes-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/tmlind/linux-omap:
  usb: musb: omap2430: Fix regression caused by driver core change
  ARM: OMAP1: fix incorrect INT_DMA_LCD
Signed-off-by: Olof Johansson <olof@lixom.net>

Arguments passed to list_add_tail were reversed resulting in deletion
of old blob property everytime the new one is added.

Fixes

commit e2f5d2ea
Author: Daniel Stone <daniels@collabora.com>
Date:   Fri May 22 13:34:51 2015 +0100

    drm/mode: Add user blob-creation ioctl
Signed-off-by: Maneet Singh <mmaneetsingh@nvidia.com>
[seanpaul tweaked commit subject a little]
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Cc: stable@kernel.org # v4.2
Reviewed-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Dave Airlie <airlied@gmail.com>

This reverts commit 7eb41885.

This commit is poorly justified, I can find not discusison in email,
and it clearly causes a problem.

If a device which is being recovered fails and is subsequently
re-added to an array, there could easily have been changes to the
array *before* the point where the recovery was up to.  So the
recovery must start again from the beginning.

If a spare is being recovered and fails, then when it is re-added we
really should do a bitmap-based recovery up to the recovery-offset,
and then a full recovery from there.  Before this reversion, we only
did the "full recovery from there" which is not corect.  After this
reversion with will do a full recovery from the start, which is safer
but not ideal.

It will be left to a future patch to arrange the two different styles
of recovery.
Reported-and-tested-by: Nate Dailey <nate.dailey@stratus.com>
Signed-off-by: NeilBrown <neilb@suse.com>
Cc: stable@vger.kernel.org (3.14+)
Fixes: 7eb41885 ("md: allow a partially recovered device to be hot-added to an array.")

The size here comes from the user via the ioctl, it is a number between
1-u32max so the addition here could overflow on 32 bit systems.

Fixes: f453ba04 ('DRM: add mode setting support')
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Cc: stable@kernel.org # v4.2
Signed-off-by: Dave Airlie <airlied@gmail.com>