The IB/core provides address resolution service and invokes callback
handler when address resolve request completes of requester in worker
thread context.

Such caller might allocate or free memory in callback handler
depending on the completion status to make further progress or to
terminate a connection. Most ULPs resolve route which involves
allocating route entry and path record elements in callback event handler.

It has been noticed that WQ_MEM_RECLAIM flag should not be used for
workers that tend to allocate memory in this [1] thread discussion.

In order to mitigate this situation, WQ_MEM_RECLAIM flag was dropped for
other such WQs in this [2] patch.

Similar problem might arise with address resolution path, though its not
yet noticed. The ib_addr workqueue is not memory reclaim path due to its
nature of invoking callback that might allocate memory or don't free any
memory under memory pressure.

[1] https://www.spinics.net/lists/linux-rdma/msg53239.html
[2] https://www.spinics.net/lists/linux-rdma/msg53416.html

Fixes: f5481626 ("IB/addr: Remove deprecated create_singlethread_workqueue")
Fixes: 5fff41e1 ("IB/core: Fix race condition in resolving IP to MAC")
Signed-off-by: Parav Pandit <parav@mellanox.com>
Reviewed-by: Daniel Jurgens <danielj@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>

This patch fixes the case where 'lifespan' entry of the hw_counters
is not writable. Currently write callback is not exposed for for
the hw_counters sysfs operation. Due to this, modifying lifespan
value results into permission denied error in below example.

echo 10 > /sys/class/infiniband/mlx5_0/ports/1/hw_counters/lifespan
-bash: /sys/class/infiniband/mlx5_0/ports/1/hw_counters/lifespan:
Permission denied

This patch adds the hook to modify any attribute which implements
store() operation.

Fixes: b40f4757 ("IB/core: Make device counter infrastructure dynamic")
Signed-off-by: Parav Pandit <parav@mellanox.com>
Reviewed-by: Mark Bloch <markb@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>

Since IB/core resolves the destination mac address for user and kernel
consumers, avoid resolving in multiple provider drivers.

Only ib_core resolves DMAC now, therefore resolve_eth_dmac is removed as
exported symbol.
Signed-off-by: Parav Pandit <parav@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>

Introduce rdma_create_user_ah API which allows passing udata to
provider driver and additionally which resolves DMAC for RoCE.

ib_resolve_eth_dmac() resolves destination mac address for unicast,
multicast, link local ipv4 mapped ipv6 and ipv6 destination gid entry.
This allows all RoCE provider drivers to avoid duplicating such code.

Such change brings consistency where IB core always resolves dmac and pass
it to RoCE provider drivers for user and kernel consumers, with this
ah_attr->roce.dmac is always an input field for provider drivers.

This uniformity avoids exporting ib_resolve_eth_dmac symbol to providers
or other modules. Therefore its removed as exported symbol at later in
the patch series.

Now uverbs and umad both makes use of rdma_create_user_ah API which
fixes the issue where umad has invalid DMAC for address.
Signed-off-by: Parav Pandit <parav@mellanox.com>
Reviewed-by: Daniel Jurgens <danielj@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>

Merge tag 'mlx5-updates-2017-10-11' of git://git.kernel.org/pub/scm/linux/kernel/git/mellanox/linux into mlnx-shared

mlx5-updates-2017-10-11: IPoIB Muli Pkey support

This series provides the support for IPoIB Multi Pkey.  InfiniBand Pkeys
are the equivalent of Ethernet vlans.  Currently IPoIB device driver
supports only default Pkey and IPoIB Pkey child interfaces are not
supported with IPoIB offloads mode, this series will add the support for
that by allowing creating mlx5 multiple IPoIB netdevices with a
non-default Pkey.

mlx5 IPoIB Pkey child interface is smaller version of mlx5i IPoIB
interfaces and shares most of its resources with the parent IPoIB
interface, namely RX steering and ring queue resources.

The only mlx5 resources a child Pkey interface will be creating are the
TX rings, since they should be assigned to a specific Pkey.

mlx5i Pkey netdev is implemented via new mlx5e netdev profile
implemented in mlx5/core/ipoib/ipoib_vlan.c.

The series starts with a refactoring of mlx5e PTP and mlx5 clock
implementation to move the code to be part of mlx5 core rather than
mlx5e netdevice, in order to make mlx5 clock and PTP registration part
of the core to be shared with mlx5e master Ethernet netdev/IPoIB parent
netdev and mlx5_ib in the near future.

Add the support for attaching multiple underlay QPs for the different
Pkeys in mlx5 core RX steering.

Add Pkey index to rdma_netdev to add the ability to set PKEY index to
lower IPoIB offload netdev.

Use hash-table to map between DQPN (Destination QP number) to child
netdev for the IPoIB parent netdev to forward RX packets to the
corresponding child Pkey netdev, since the RX rings are shared.

The reset of the series adds the ipoib child Pkey: mlx5e netdev profile,
netdev nods implementation and minimal set of ethtool callbacks.

Thanks,
Saeed.
Signed-off-by: Doug Ledford <dledford@redhat.com>

Resources such as FT, QPN HT and mdev resources should be allocated
only by parent netdev. Shared resources are allocated and freed by the
parent interface since the parent is always present and created
before the IPoIB PKEY sub-interface.
Signed-off-by: Alex Vesker <valex@mellanox.com>
Reviewed-by: Erez Shitrit <erezsh@mellanox.com>

Similar to VLAN interfaces child interfaces have limited ethtool
support. In current code the main limitation that does not
allow child interface ethtool configuration is due to shared
resources which are managed by the parent.
Signed-off-by: Alex Vesker <valex@mellanox.com>
Reviewed-by: Erez Shitrit <erezsh@mellanox.com>

Child interface ndos will be called to support child interface
specific behaviour.

ndo_init flow:
-Acquire shared QPN to net-device HT from parent
-Continue with the same flow as parent interface

ndo_open flow:
-Initialize child underlay QP and connect to shared FT
-Create child send TIS
-Open child send channels
Signed-off-by: Alex Vesker <valex@mellanox.com>
Reviewed-by: Erez Shitrit <erezsh@mellanox.com>

Child interface profile will be called to support child interface
specific behaviour. The child code is sparse compared to the parent
since the RX channels are shared between the interfaces.
Creating a septate profile for child and parent will make a smother
code with a better ability for future expansion.
The profile stuct is exposed to the parent using a getter function.
Signed-off-by: Alex Vesker <valex@mellanox.com>
Reviewed-by: Erez Shitrit <erezsh@mellanox.com>

This change is needed for PKEY support, since the RQs are shared
between the child interface and the parent. The parent is responsible
for NAPI and the precessing of RX completions. Using the dqpn in the
completion descriptor we set the corresponding child IPoIB netdevice
on the SKB.
The mapping between the dqpn and the netdevice is done using a HT,
each mlx5 IPoIB interface registers its mapping on creation.
Signed-off-by: Alex Vesker <valex@mellanox.com>
Reviewed-by: Erez Shitrit <erezsh@mellanox.com>

Added a function to set PKEY index to IPoIB device driver using the
already present set_id function. PKEY index is attached to the QP
during state modification.
Signed-off-by: Alex Vesker <valex@mellanox.com>
Reviewed-by: Erez Shitrit <erezsh@mellanox.com>

To support passing child interfaces to the lower device a new
rdma_netdev function was used, set_id. This will allow us to
attach the PKEY index lower device resources such as TIS/QP.
For devices that do not support offloads in IPoIB same logic
will be used, setting the PKEY index to priv struct.
Signed-off-by: Alex Vesker <valex@mellanox.com>
Reviewed-by: Erez Shitrit <erezsh@mellanox.com>

When ndo_open and ndo_stop are called RTNL lock should be held.
In this specific case ipoib_ib_dev_open calls the offloaded ndo_open
which re-sets the number of TX queue assuming RTNL lock is held.
Since RTNL lock is not held, RTNL assert will fail.
Signed-off-by: Alex Vesker <valex@mellanox.com>

Previous support allowed connecting only a single QPN to the FT.
Now using a linked list multiple QPNs can be attached to the same FT.

Supporting attaching multiple underlay QPs is required for PKEY
support in which child and parent share the same FT.

The actual attaching/detaching FW commands will be called inside the
function symmetrically.

This change requires a change in IPoIB open and close functions, the
attaching/detaching to/from the FT is done each time we open/close.
Signed-off-by: Alex Vesker <valex@mellanox.com>
Reviewed-by: Maor Gottlieb <maorg@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>

During the creation of the underlay QP the PKEY index is unknown, the
PKEY index is known only when calling ndo_open.
PKEY index attached to the QP during state modification.

Splitting the functions will also make the code symmetric and more
readable. This split is also required for later PKEY support to be
called with the PKEY index during ndo_open.
Signed-off-by: Alex Vesker <valex@mellanox.com>
Reviewed-by: Erez Shitrit <erezsh@mellanox.com>
Signed-off-by: Leon Romanovsky <leon@kernel.org>

PTP code is moved to core section of mlx5 driver in order to share
it between ethernet and infiniband. This movement involves the following
changes:
- Change mlx5e_ prefix to be mlx5_
- Add clock structs to Core
- Add clock object to mlx5_core_dev
- Call Init/Uninit clock from core init/cleanup
- Rename mlx5e_tstamp to be mlx5_clock
Signed-off-by: Feras Daoud <ferasda@mellanox.com>
Signed-off-by: Eitan Rabin <rabin@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>

en_clock.c renamed clock.c and moved to lib/ as first step
towards relocating code to core part of the driver to allow
sharing between Ethernet and Infiniband.
Signed-off-by: Feras Daoud <ferasda@mellanox.com>
Signed-off-by: Eitan Rabin <rabin@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>

Pull SCSI fixes from James Bottomley:

 - a couple of serious fixes: use after free and blacklist for WRITE
   SAME

 - one error leg fix: write_pending failure

 - one user experience problem: do not override max_sectors_kb

 - one minor unused function removal

* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  scsi: ibmvscsis: Fix write_pending failure path
  scsi: libiscsi: Remove iscsi_destroy_session
  scsi: libiscsi: Fix use-after-free race during iscsi_session_teardown
  scsi: sd: Do not override max_sectors_kb sysfs setting
  scsi: sd: Implement blacklist option for WRITE SAME w/ UNMAP

Pull i2c fixes from Wolfram Sang:
 "I2C has three driver fixes for the newly introduced drivers and one ID
  addition for the i801 driver"

* 'i2c/for-current-4.14' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
  i2c: i2c-stm32f7: make structure stm32f7_setup static const
  i2c: ensure termination of *_device_id tables
  i2c: i801: Add support for Intel Cedar Fork
  i2c: stm32f7: fix setup structure

Pull MMC fixes from Ulf Hansson:
 "MMC core:

   - Fix driver strength selection when selecting hs400es

   - Delete bounce buffer handling:

     This change fixes a problem related to how bounce buffers are being
     allocated. However, instead of trying to fix that, let's just
     remove the mmc bounce buffer code altogether, as it has practically
     no use.

  MMC host:

   - meson-gx: A couple of fixes related to clock/phase/tuning

   - sdhci-xenon: Fix clock resource by adding an optional bus clock"

* tag 'mmc-v4.14-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/ulfh/mmc:
  mmc: sdhci-xenon: Fix clock resource by adding an optional bus clock
  mmc: meson-gx: include tx phase in the tuning process
  mmc: meson-gx: fix rx phase reset
  mmc: meson-gx: make sure the clock is rounded down
  mmc: Delete bounce buffer handling
  mmc: core: add driver strength selection when selecting hs400es

Merge tag 'hwmon-for-linus-v4.14-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging

Pull hwmon fix from Guenter Roeck:
 "Fix up error path in xgene driver"

* tag 'hwmon-for-linus-v4.14-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging:
  hwmon: (xgene) Fix up error handling path mixup in 'xgene_hwmon_probe()'

Pull clk fixes from Stephen Boyd:

 - build fix to export the clk_bulk_prepare() symbol

 - suspend fix for Samsung Exynos SoCs where we need to keep clks on
   across suspend

 - two critical clk markings for clks that shouldn't ever turn off on
   Rockchip SoCs

 - a fix for a copy-paste mistake on Rockchip rk3128 causing some clks
   to touch the same bit and trample over one another

* tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux:
  clk: samsung: exynos4: Enable VPLL and EPLL clocks for suspend/resume cycle
  clk: Export clk_bulk_prepare()
  clk: rockchip: add sclk_timer5 as critical clock on rk3128
  clk: rockchip: fix up rk3128 pvtm and mipi_24m gate regs error
  clk: rockchip: add pclk_pmu as critical clock on rk3128

Pull ARC udpates from Vineet Gupta:

 - updates for various platforms

 - boot log updates for upcoming HS48 family of cores (dual issue)

* tag 'arc-4.14-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/vgupta/arc:
  ARC: [plat-hsdk]: Add reset controller node to manage ethernet reset
  ARC: [plat-hsdk]: Temporary fix to set CPU frequency to 1GHz
  ARC: fix allnoconfig build warning
  ARCv2: boot log: identify HS48 cores (dual issue)
  ARC: boot log: decontaminate ARCv2 ISA_CONFIG register
  arc: remove redundant UTS_MACHINE define in arch/arc/Makefile
  ARC: [plat-eznps] Update platform maintainer as Noam left
  ARC: [plat-hsdk] use actual clk driver to manage cpu clk
  ARC: [*defconfig] Reenable soft lock-up detector
  ARC: [plat-axs10x] sdio: Temporary fix of sdio ciu frequency
  ARC: [plat-hsdk] sdio: Temporary fix of sdio ciu frequency
  ARC: [plat-axs103] Add temporary quirk to reset ethernet IP

Pull xfs fixes from Darrick Wong:

 - fix a race between overlapping copy on write aio

 - fix cow fork swapping when we defragment reflinked files

* tag 'xfs-4.14-fixes-4' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux:
  xfs: handle racy AIO in xfs_reflink_end_cow
  xfs: always swap the cow forks when swapping extents

Pull block fixes from Jens Axboe:
 "A collection of fixes for this series. This contains:

   - NVMe pull request from Christoph, one uuid attribute fix, and one
     fix for the controller memory buffer address for remapped BARs.

   - use-after-free fix for bsg, from Benjamin Block.

   - bcache race/use-after-free fix for a list traversal, fixing a
     regression in this merge window. From Coly Li.

   - null_blk change configfs dependency change from a 'depends' to a
     'select'. This is a change from this merge window as well. From me.

   - nbd signal fix from Josef, fixing a regression introduced with the
     status code changes.

   - nbd MAINTAINERS mailing list entry update.

   - blk-throttle stall fix from Joseph Qi.

   - blk-mq-debugfs fix from Omar, fixing an issue where we don't
     register the IO scheduler debugfs directory, if the driver is
     loaded with it. Only shows up if you switch through the sysfs
     interface"

* 'for-linus' of git://git.kernel.dk/linux-block:
  bsg-lib: fix use-after-free under memory-pressure
  nvme-pci: Use PCI bus address for data/queues in CMB
  blk-mq-debugfs: fix device sched directory for default scheduler
  null_blk: change configfs dependency to select
  blk-throttle: fix possible io stall when upgrade to max
  MAINTAINERS: update list for NBD
  nbd: fix -ERESTARTSYS handling
  nvme: fix visibility of "uuid" ns attribute
  bcache: use llist_for_each_entry_safe() in __closure_wake_up()

Pull PCI fixes from Bjorn Helgaas:
 "Fix legacy IDE probe issues exposed by recent PCI core IRQ mapping
  changes (Bartlomiej Zolnierkiewicz, Lorenzo Pieralisi)"

* tag 'pci-v4.14-fixes-4' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci:
  ide: fix IRQ assignment for PCI bus order probing
  ide: pci: free PCI BARs on initialization failure
  ide: free hwif->portdev on hwif_init() failure

Pull arm64 fixes from Catalin Marinas:

 - Bring initialisation of user space undefined instruction handling
   early (core_initcall) since late_initcall() happens after modprobe in
   initramfs is invoked. Similar fix for fpsimd initialisation

 - Increase the kernel stack when KASAN is enabled

 - Bring the PCI ACS enabling earlier via the
   iort_init_platform_devices()

 - Fix misleading data abort address printing (decimal vs hex)

* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
  arm64: Ensure fpsimd support is ready before userspace is active
  arm64: Ensure the instruction emulation is ready for userspace
  arm64: Use larger stacks when KASAN is selected
  ACPI/IORT: Fix PCI ACS enablement
  arm64: fix misleading data abort decoding

Pull KVM fixes from Radim Krčmář:

 - fix PPC XIVE interrupt delivery

 - fix x86 RCU breakage from asynchronous page faults when built without
   PREEMPT_COUNT

 - fix x86 build with -frecord-gcc-switches

 - fix x86 build without X86_LOCAL_APIC

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  KVM: add X86_LOCAL_APIC dependency
  x86/kvm: Move kvm_fastop_exception to .fixup section
  kvm/x86: Avoid async PF preempting the kernel incorrectly
  KVM: PPC: Book3S: Fix server always zero from kvmppc_xive_get_xive()

Pull rdma fixes from Doug Ledford:
 "This is a pretty small pull request. Only 6 patches in total. There
  are no outstanding -rc patches on the mailing list after this pull
  request, so only if some new issues are discovered in the remainder of
  the rc cycles will you hear from me again.

  Summary:
   - a fix for iwpm netlink usage
   - a fix for error unwinding in mlx5
   - two fixes to vlan handling in qedr
   - a couple small i40iw fixes"

* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dledford/rdma:
  i40iw: Fix port number for query QP
  i40iw: Add missing memory barriers
  RDMA/qedr: Parse vlan priority as sl
  RDMA/qedr: Parse VLAN ID correctly and ignore the value of zero
  IB/mlx5: Fix label order in error path handling
  RDMA/iwpm: Properly mark end of NL messages

Pull btrfs fixes from David Sterba:
 "Two more fixes for bugs introduced in 4.13.

  The sector_t problem with 32bit architecture and !LBDAF config seems
  serious but the number of affected deployments is hopefully low.

  The clashing status bits could lead to a confusing in-memory state of
  the whole-filesystem operations if used with the quota override sysfs
  knob"

* 'for-4.14-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
  Btrfs: fix overlap of fs_info::flags values
  btrfs: avoid overflow when sector_t is 32 bit

Pull ceph fixes from Ilya Dryomov:
 "Two fixups for CephFS snapshot-handling patches in -rc1"

* tag 'ceph-for-4.14-rc4' of git://github.com/ceph/ceph-client:
  ceph: fix __choose_mds() for LSSNAP request
  ceph: properly queue cap snap for newly created snap realm

DW ethernet controller on HSDK hangs sometimes after SW reset, so
add reset node to make possible to reset DW ethernet controller HW.
Signed-off-by: Eugeniy Paltsev <Eugeniy.Paltsev@synopsys.com>
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>

Pull overlayfs fixes from Miklos Szeredi:
 "Fix a regression in 4.14 and one in 4.13. The latter is a case when
  Docker is doing something it really shouldn't and gets away with it.
  We now print a warning instead of erroring out.

  There are also fixes to several error paths"

* 'overlayfs-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/vfs:
  ovl: fix regression caused by exclusive upper/work dir protection
  ovl: fix missing unlock_rename() in ovl_do_copy_up()
  ovl: fix dentry leak in ovl_indexdir_cleanup()
  ovl: fix dput() of ERR_PTR in ovl_cleanup_index()
  ovl: fix error value printed in ovl_lookup_index()
  ovl: fix may_write_real() for overlayfs directories

Pull powerpc fixes from Michael Ellerman:
 "Nine small fixes, really nothing that stands out.

  A work-around for a spurious MCE on Power9. A CXL fault handling fix,
  some fixes to the new XIVE code, and a fix to the new 32-bit
  STRICT_KERNEL_RWX code.

  Fixes for old code/stable: an fix to an incorrect TLB flush on boot
  but not on any current machines, a compile error on 4xx and a fix to
  memory hotplug when using radix (Power9).

  Thanks to: Anton Blanchard, Cédric Le Goater, Christian Lamparter,
  Christophe Leroy, Christophe Lombard, Guenter Roeck, Jeremy Kerr,
  Michael Neuling, Nicholas Piggin"

* tag 'powerpc-4.14-4' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
  powerpc/powernv: Increase memory block size to 1GB on radix
  powerpc/mm: Call flush_tlb_kernel_range with interrupts enabled
  powerpc/xive: Clear XIVE internal structures when a CPU is removed
  powerpc/xive: Fix IPI reset
  powerpc/4xx: Fix compile error with 64K pages on 40x, 44x
  powerpc: Fix action argument for cpufeatures-based TLB flush
  cxl: Fix memory page not handled
  powerpc: Fix workaround for spurious MCE on POWER9
  powerpc: Handle MCE on POWER9 with only DSISR bit 30 set

Pull drm fixes from Dave Airlie:
 "Some i915 fixes from the last two weeks (as they were on a strange
  base and I just waited for rc3), also a single sun4i hdmi fix"

* tag 'drm-fixes-for-v4.14-rc4' of git://people.freedesktop.org/~airlied/linux:
  drm/i915/glk: Fix DMC/DC state idleness calculation
  drm/i915/cnl: Reprogram DMC firmware after S3/S4 resume
  drm/i915: Fix DDI PHY init if it was already on
  drm/sun4i: hdmi: Disable clks in bind function error path and unbind function
  drm/i915/bios: ignore HDMI on port A
  drm/i915: remove redundant variable hw_check
  drm/i915: always update ELD connector type after get modes

Pull watchddog clean-up and fixes from Thomas Gleixner:
 "The watchdog (hard/softlockup detector) code is pretty much broken in
  its current state. The patch series addresses this by removing all
  duct tape and refactoring it into a workable state.

  The reasons why I ask for inclusion that late in the cycle are:

   1) The code causes lockdep splats vs. hotplug locking which get
      reported over and over. Unfortunately there is no easy fix.

   2) The risk of breakage is minimal because it's already broken

   3) As 4.14 is a long term stable kernel, I prefer to have working
      watchdog code in that and the lockdep issues resolved. I wouldn't
      ask you to pull if 4.14 wouldn't be a LTS kernel or if the
      solution would be easy to backport.

   4) The series was around before the merge window opened, but then got
      delayed due to the UP failure caused by the for_each_cpu()
      surprise which we discussed recently.

  Changes vs. V1:

   - Addressed your review points

   - Addressed the warning in the powerpc code which was discovered late

   - Changed two function names which made sense up to a certain point
     in the series. Now they match what they do in the end.

   - Fixed a 'unused variable' warning, which got not detected by the
     intel robot. I triggered it when trying all possible related config
     combinations manually. Randconfig testing seems not random enough.

  The changes have been tested by and reviewed by Don Zickus and tested
  and acked by Micheal Ellerman for powerpc"

* 'core-watchdog-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (36 commits)
  watchdog/core: Put softlockup_threads_initialized under ifdef guard
  watchdog/core: Rename some softlockup_* functions
  powerpc/watchdog: Make use of watchdog_nmi_probe()
  watchdog/core, powerpc: Lock cpus across reconfiguration
  watchdog/core, powerpc: Replace watchdog_nmi_reconfigure()
  watchdog/hardlockup/perf: Fix spelling mistake: "permanetely" -> "permanently"
  watchdog/hardlockup/perf: Cure UP damage
  watchdog/hardlockup: Clean up hotplug locking mess
  watchdog/hardlockup/perf: Simplify deferred event destroy
  watchdog/hardlockup/perf: Use new perf CPU enable mechanism
  watchdog/hardlockup/perf: Implement CPU enable replacement
  watchdog/hardlockup/perf: Implement init time detection of perf
  watchdog/hardlockup/perf: Implement init time perf validation
  watchdog/core: Get rid of the racy update loop
  watchdog/core, powerpc: Make watchdog_nmi_reconfigure() two stage
  watchdog/sysctl: Clean up sysctl variable name space
  watchdog/sysctl: Get rid of the #ifdeffery
  watchdog/core: Clean up header mess
  watchdog/core: Further simplify sysctl handling
  watchdog/core: Get rid of the thread teardown/setup dance
  ...

We register the pm/hotplug callbacks for FPSIMD as late_initcall,
which happens after the userspace is active (from initramfs via
populate_rootfs, a rootfs_initcall). Make sure we are ready even
before the userspace could potentially use it, by promoting to
a core_initcall.

Cc: Will Deacon <will.deacon@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Dave Martin <dave.martin@arm.com>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>

We trap and emulate some instructions (e.g, mrs, deprecated instructions)
for the userspace. However the handlers for these are registered as
late_initcalls and the userspace could be up and running from the initramfs
by that time (with populate_rootfs, which is a rootfs_initcall()). This
could cause problems for the early applications ending up in failure
like :

[   11.152061] modprobe[93]: undefined instruction: pc=0000ffff8ca48ff4

This patch promotes the specific calls to core_initcalls, which are
guaranteed to be completed before we hit userspace.

Cc: stable@vger.kernel.org
Cc: Dave Martin <dave.martin@arm.com>
Cc: Matthias Brugger <mbrugger@suse.com>
Cc: James Morse <james.morse@arm.com>
Reported-by: Matwey V. Kornilov <matwey.kornilov@gmail.com>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>

Memory hot unplug on PowerNV radix hosts is broken. Our memory block
size is 256MB but since we map the linear region with very large
pages, each pte we tear down maps 1GB.

A hot unplug of one 256MB memory block results in 768MB of memory
getting unintentionally unmapped. At this point we are likely to oops.

Fix this by increasing our memory block size to 1GB on PowerNV radix
hosts.

Fixes: 4b5d62ca ("powerpc/mm: add radix__remove_section_mapping()")
Cc: stable@vger.kernel.org # v4.11+
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>