Fix OF node leaks by calling of_node_put in
for_each_available_child_of_node when the cycle returns.

Generated by: scripts/coccinelle/iterators/for_each_child.cocci
Signed-off-by: Junlin Yang <yangjunlin@yulong.com>
Link: https://lore.kernel.org/r/20210215110425.1012-1-angkery@163.comSigned-off-by: Mark Brown <broonie@kernel.org>

The DMA cap mask is no longer used since:
commit 7758e390 ("spi: atmel: remove compat for non DT board when requesting dma chan")
Drop it now.
Signed-off-by: Tudor Ambarus <tudor.ambarus@microchip.com>
Link: https://lore.kernel.org/r/20210218132840.131898-1-tudor.ambarus@microchip.comSigned-off-by: Mark Brown <broonie@kernel.org>

Building this file with clang leads to a an unreachable code path
causing a warning from objtool:

drivers/spi/spi-rockchip.o: warning: objtool: rockchip_spi_transfer_one()+0x2e0: sibling call from callable instruction with modified stack frame

Change the unreachable() into an error return that can be
handled if it ever happens, rather than silently crashing
the kernel.

Fixes: 65498c6a ("spi: rockchip: support 4bit words")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Pratyush Yadav <p.yadav@ti.com>
Link: https://lore.kernel.org/r/20210226140109.3477093-1-arnd@kernel.orgSigned-off-by: Mark Brown <broonie@kernel.org>

Set the (optional) idle pinctrl state during runtime suspend. This is the
same schema used in PL022 driver and can help with HW designs sharing
the SPI lines for different purposes.
Signed-off-by: Alexander Sverdlin <alexander.sverdlin@gmail.com>
Link: https://lore.kernel.org/r/20210222023243.491432-1-alexander.sverdlin@gmail.comSigned-off-by: Mark Brown <broonie@kernel.org>

This removes the assignment of setup and cleanup functions for the ath79
target. Assigning the setup-method will lead to 'setup_transfer' not
being assigned in spi_bitbang_init. Because of this, performing any
TX/RX operation will lead to a kernel oops.

Also drop the redundant cleanup assignment, as it's also assigned in
spi_bitbang_init.
Signed-off-by: David Bauer <mail@david-bauer.net>
Link: https://lore.kernel.org/r/20210303160837.165771-2-mail@david-bauer.netSigned-off-by: Mark Brown <broonie@kernel.org>

spi-bitbang has to call the chipselect function on the ath79 SPI driver
in order to communicate with the SPI slave device, as the ath79 SPI
driver has three dedicated chipselect lines but can also be used with
GPIOs for the CS lines.

Fixes commit 4a07b8bc ("spi: bitbang: Make chipselect callback optional")
Signed-off-by: David Bauer <mail@david-bauer.net>
Link: https://lore.kernel.org/r/20210303160837.165771-1-mail@david-bauer.netSigned-off-by: Mark Brown <broonie@kernel.org>

The driver sets auto_runtime_pm to true, but it doesn't call
pm_runtime_enable(), which results in "Failed to power device" when PM support
is enabled.
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
Link: https://lore.kernel.org/r/20210223151851.4110-3-noltari@gmail.comSigned-off-by: Mark Brown <broonie@kernel.org>

The driver sets auto_runtime_pm to true, but it doesn't call
pm_runtime_enable(), which results in "Failed to power device" when PM support
is enabled.
Signed-off-by: Álvaro Fernández Rojas <noltari@gmail.com>
Link: https://lore.kernel.org/r/20210223151851.4110-2-noltari@gmail.comSigned-off-by: Mark Brown <broonie@kernel.org>

Release the RX channel and free the dma coherent memory when
devm_spi_register_master() fails.

Fixes: 5720ec0a ("spi: spi-ti-qspi: Add DMA support for QSPI mmap read")
Cc: stable@vger.kernel.org
Signed-off-by: Tudor Ambarus <tudor.ambarus@microchip.com>
Link: https://lore.kernel.org/r/20210218130950.90155-1-tudor.ambarus@microchip.comSigned-off-by: Mark Brown <broonie@kernel.org>

Pull rdma fixes from Jason Gunthorpe:
 "Nothing special here, though Bob's regression fixes for rxe would have
  made it before the rc cycle had there not been such strong winter
  weather!

   - Fix corner cases in the rxe reference counting cleanup that are
     causing regressions in blktests for SRP

   - Two kdoc fixes so W=1 is clean

   - Missing error return in error unwind for mlx5

   - Wrong lock type nesting in IB CM"

* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma:
  RDMA/rxe: Fix errant WARN_ONCE in rxe_completer()
  RDMA/rxe: Fix extra deref in rxe_rcv_mcast_pkt()
  RDMA/rxe: Fix missed IB reference counting in loopback
  RDMA/uverbs: Fix kernel-doc warning of _uverbs_alloc
  RDMA/mlx5: Set correct kernel-doc identifier
  IB/mlx5: Add missing error code
  RDMA/rxe: Fix missing kconfig dependency on CRYPTO
  RDMA/cm: Fix IRQ restore in ib_send_cm_sidr_rep

Pull gcc-plugins fixes from Kees Cook:
 "Tiny gcc-plugin fixes for v5.12-rc2. These issues are small but have
  been reported a couple times now by static analyzers, so best to get
  them fixed to reduce the noise. :)

   - Fix coding style issues (Jason Yan)"

* tag 'gcc-plugins-v5.12-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux:
  gcc-plugins: latent_entropy: remove unneeded semicolon
  gcc-plugins: structleak: remove unneeded variable 'ret'

Pull pstore fixes from Kees Cook:

 - Rate-limit ECC warnings (Dmitry Osipenko)

 - Fix error path check for NULL (Tetsuo Handa)

* tag 'pstore-v5.12-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux:
  pstore/ram: Rate-limit "uncorrectable error in header" message
  pstore: Fix warning in pstore_kill_sb()

Merge tag 'for-5.12/dm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm

Pull device mapper fixes from Mike Snitzer:
 "Fix DM verity target's optional Forward Error Correction (FEC) for
  Reed-Solomon roots that are unaligned to block size"

* tag 'for-5.12/dm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm:
  dm verity: fix FEC for RS roots unaligned to block size
  dm bufio: subtract the number of initial sectors in dm_bufio_get_device_size

Pull block fixes from Jens Axboe:

 - NVMe fixes:
      - more device quirks (Julian Einwag, Zoltán Böszörményi, Pascal
        Terjan)
      - fix a hwmon error return (Daniel Wagner)
      - fix the keep alive timeout initialization (Martin George)
      - ensure the model_number can't be changed on a used subsystem
        (Max Gurtovoy)

 - rsxx missing -EFAULT on copy_to_user() failure (Dan)

 - rsxx remove unused linux.h include (Tian)

 - kill unused RQF_SORTED (Jean)

 - updated outdated BFQ comments (Joseph)

 - revert work-around commit for bd_size_lock, since we removed the
   offending user in this merge window (Damien)

* tag 'block-5.12-2021-03-05' of git://git.kernel.dk/linux-block:
  nvmet: model_number must be immutable once set
  nvme-fabrics: fix kato initialization
  nvme-hwmon: Return error code when registration fails
  nvme-pci: add quirks for Lexar 256GB SSD
  nvme-pci: mark Kingston SKC2000 as not supporting the deepest power state
  nvme-pci: mark Seagate Nytro XM1440 as QUIRK_NO_NS_DESC_LIST.
  rsxx: Return -EFAULT if copy_to_user() fails
  block/bfq: update comments and default value in docs for fifo_expire
  rsxx: remove unused including <linux/version.h>
  block: Drop leftover references to RQF_SORTED
  block: revert "block: fix bd_size_lock use"

Pull io_uring fixes from Jens Axboe:
 "A bit of a mix between fallout from the worker change, cleanups and
  reductions now possible from that change, and fixes in general. In
  detail:

   - Fully serialize manager and worker creation, fixing races due to
     that.

   - Clean up some naming that had gone stale.

   - SQPOLL fixes.

   - Fix race condition around task_work rework that went into this
     merge window.

   - Implement unshare. Used for when the original task does unshare(2)
     or setuid/seteuid and friends, drops the original workers and forks
     new ones.

   - Drop the only remaining piece of state shuffling we had left, which
     was cred. Move it into issue instead, and we can drop all of that
     code too.

   - Kill f_op->flush() usage. That was such a nasty hack that we had
     out of necessity, we no longer need it.

   - Following from ->flush() removal, we can also drop various bits of
     ctx state related to SQPOLL and cancelations.

   - Fix an issue with IOPOLL retry, which originally was fallout from a
     filemap change (removing iov_iter_revert()), but uncovered an issue
     with iovec re-import too late.

   - Fix an issue with system suspend.

   - Use xchg() for fallback work, instead of cmpxchg().

   - Properly destroy io-wq on exec.

   - Add create_io_thread() core helper, and use that in io-wq and
     io_uring. This allows us to remove various silly completion events
     related to thread setup.

   - A few error handling fixes.

  This should be the grunt of fixes necessary for the new workers, next
  week should be quieter. We've got a pending series from Pavel on
  cancelations, and how tasks and rings are indexed. Outside of that,
  should just be minor fixes. Even with these fixes, we're still killing
  a net ~80 lines"

* tag 'io_uring-5.12-2021-03-05' of git://git.kernel.dk/linux-block: (41 commits)
  io_uring: don't restrict issue_flags for io_openat
  io_uring: make SQPOLL thread parking saner
  io-wq: kill hashed waitqueue before manager exits
  io_uring: clear IOCB_WAITQ for non -EIOCBQUEUED return
  io_uring: don't keep looping for more events if we can't flush overflow
  io_uring: move to using create_io_thread()
  kernel: provide create_io_thread() helper
  io_uring: reliably cancel linked timeouts
  io_uring: cancel-match based on flags
  io-wq: ensure all pending work is canceled on exit
  io_uring: ensure that threads freeze on suspend
  io_uring: remove extra in_idle wake up
  io_uring: inline __io_queue_async_work()
  io_uring: inline io_req_clean_work()
  io_uring: choose right tctx->io_wq for try cancel
  io_uring: fix -EAGAIN retry with IOPOLL
  io-wq: fix error path leak of buffered write hash map
  io_uring: remove sqo_task
  io_uring: kill sqo_dead and sqo submission halting
  io_uring: ignore double poll add on the same waitqueue head
  ...

Pull power management fixes from Rafael Wysocki:
 "These fix the usage of device links in the runtime PM core code and
  update the DTPM (Dynamic Thermal Power Management) feature added
  recently.

  Specifics:

   - Make the runtime PM core code avoid attempting to suspend supplier
     devices before updating the PM-runtime status of a consumer to
     'suspended' (Rafael Wysocki).

   - Fix DTPM (Dynamic Thermal Power Management) root node
     initialization and label that feature as EXPERIMENTAL in Kconfig
     (Daniel Lezcano)"

* tag 'pm-5.12-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  powercap/drivers/dtpm: Add the experimental label to the option description
  powercap/drivers/dtpm: Fix root node initialization
  PM: runtime: Update device status before letting suppliers suspend

Pull ACPI fix from Rafael Wysocki:
 "Make the empty stubs of some helper functions used when CONFIG_ACPI is
  not set actually match those functions (Andy Shevchenko)"

* tag 'acpi-5.12-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  ACPI: bus: Constify is_acpi_node() and friends (part 2)

Pull iommu fixes from Joerg Roedel:

 - Fix a sleeping-while-atomic issue in the AMD IOMMU code

 - Disable lazy IOTLB flush for untrusted devices in the Intel VT-d
   driver

 - Fix status code definitions for Intel VT-d

 - Fix IO Page Fault issue in Tegra IOMMU driver

* tag 'iommu-fixes-v5.12-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu:
  iommu/vt-d: Fix status code for Allocate/Free PASID command
  iommu: Don't use lazy flush for untrusted device
  iommu/tegra-smmu: Fix mc errors on tegra124-nyan
  iommu/amd: Fix sleeping in atomic in increase_address_space()

Pull btrfs fixes from David Sterba:
 "More regression fixes and stabilization.

  Regressions:

   - zoned mode
      - count zone sizes in wider int types
      - fix space accounting for read-only block groups

   - subpage: fix page tail zeroing

  Fixes:

   - fix spurious warning when remounting with free space tree

   - fix warning when creating a directory with smack enabled

   - ioctl checks for qgroup inheritance when creating a snapshot

   - qgroup
      - fix missing unlock on error path in zero range
      - fix amount of released reservation on error
      - fix flushing from unsafe context with open transaction,
        potentially deadlocking

   - minor build warning fixes"

* tag 'for-5.12-rc1-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
  btrfs: zoned: do not account freed region of read-only block group as zone_unusable
  btrfs: zoned: use sector_t for zone sectors
  btrfs: subpage: fix the false data csum mismatch error
  btrfs: fix warning when creating a directory with smack enabled
  btrfs: don't flush from btrfs_delayed_inode_reserve_metadata
  btrfs: export and rename qgroup_reserve_meta
  btrfs: free correct amount of space in btrfs_delayed_inode_reserve_metadata
  btrfs: fix spurious free_space_tree remount warning
  btrfs: validate qgroup inherit for SNAP_CREATE_V2 ioctl
  btrfs: unlock extents in btrfs_zero_range in case of quota reservation errors
  btrfs: ref-verify: use 'inline void' keyword ordering

Pull devicetree fixes from Rob Herring:

 - Another batch of graph and video-interfaces schema conversions

 - Drop DT header symlink for dropped C6X arch

 - Fix bcm2711-hdmi schema error

* tag 'devicetree-fixes-for-5.12-1' of git://git.kernel.org/pub/scm/linux/kernel/git/robh/linux:
  dt-bindings: media: Use graph and video-interfaces schemas, round 2
  dts: drop dangling c6x symlink
  dt-bindings: bcm2711-hdmi: Fix broken schema

Pull tracing fixes from Steven Rostedt:
 "Functional fixes:

   - Fix big endian conversion for arm64 in recordmcount processing

   - Fix timestamp corruption in ring buffer on discarding events

   - Fix memory leak in __create_synth_event()

   - Skip selftests if tracing is disabled as it will cause them to
     fail.

  Non-functional fixes:

   - Fix help text in Kconfig

   - Remove duplicate prototype for trace_empty()

   - Fix stale comment about the trace_event_call flags.

  Self test update:

   - Add more information to the validation output of when a corrupt
     timestamp is found in the ring buffer, and also trigger a warning
     to make sure that tests catch it"

* tag 'trace-v5.12-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace:
  tracing: Fix comment about the trace_event_call flags
  tracing: Skip selftests if tracing is disabled
  tracing: Fix memory leak in __create_synth_event()
  ring-buffer: Add a little more information and a WARN when time stamp going backwards is detected
  ring-buffer: Force before_stamp and write_stamp to be different on discard
  tracing: Fix help text of TRACEPOINT_BENCHMARK in Kconfig
  tracing: Remove duplicate declaration from trace.h
  ftrace: Have recordmcount use w8 to read relp->r_info in arm64_is_fake_mcount

In rxe_comp.c in rxe_completer() the function free_pkt() did not clear skb
which triggered a warning at 'done:' and could possibly at 'exit:'. The
WARN_ONCE() calls are not actually needed.  The call to free_pkt() is
moved to the end to clearly show that all skbs are freed.

Fixes: 899aba89 ("RDMA/rxe: Fix FIXME in rxe_udp_encap_recv()")
Link: https://lore.kernel.org/r/20210304192048.2958-1-rpearson@hpe.comSigned-off-by: Bob Pearson <rpearsonhpe@gmail.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>

rxe_rcv_mcast_pkt() dropped a reference to ib_device when no error
occurred causing an underflow on the reference counter.  This code is
cleaned up to be clearer and easier to read.

Fixes: 899aba89 ("RDMA/rxe: Fix FIXME in rxe_udp_encap_recv()")
Link: https://lore.kernel.org/r/20210304192048.2958-1-rpearson@hpe.comSigned-off-by: Bob Pearson <rpearsonhpe@gmail.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>

When the noted patch below extending the reference taken by
rxe_get_dev_from_net() in rxe_udp_encap_recv() until each skb is freed it
was not matched by a reference in the loopback path resulting in
underflows.

Fixes: 899aba89 ("RDMA/rxe: Fix FIXME in rxe_udp_encap_recv()")
Link: https://lore.kernel.org/r/20210304192048.2958-1-rpearson@hpe.comSigned-off-by: Bob Pearson <rpearsonhpe@gmail.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>

45d189c6 ("io_uring: replace force_nonblock with flags") did
something strange for io_openat() slicing all issue_flags but
IO_URING_F_NONBLOCK. Not a bug for now, but better to just forward the
flags.
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

Pull NVMe fixes from Christoph:

"nvme fixes for 5.12:

 - more device quirks (Julian Einwag, Zoltán Böszörményi, Pascal Terjan)
 - fix a hwmon error return (Daniel Wagner)
 - fix the keep alive timeout initialization (Martin George)
 - ensure the model_number can't be changed on a used subsystem
   (Max Gurtovoy)"

* tag 'nvme-5.12-2021-03-05' of git://git.infradead.org/nvme:
  nvmet: model_number must be immutable once set
  nvme-fabrics: fix kato initialization
  nvme-hwmon: Return error code when registration fails
  nvme-pci: add quirks for Lexar 256GB SSD
  nvme-pci: mark Kingston SKC2000 as not supporting the deepest power state
  nvme-pci: mark Seagate Nytro XM1440 as QUIRK_NO_NS_DESC_LIST.

We have this weird true/false return from parking, and then some of the
callers decide to look at that. It can lead to unbalanced parks and
sqd locking. Have the callers check the thread status once it's parked.
We know we have the lock at that point, so it's either valid or it's NULL.

Fix race with parking on thread exit. We need to be careful here with
ordering of the sdq->lock and the IO_SQ_THREAD_SHOULD_PARK bit.

Rename sqd->completion to sqd->parked to reflect that this is the only
thing this completion event doesn.
Signed-off-by: Jens Axboe <axboe@kernel.dk>

If we race with shutting down the io-wq context and someone queueing
a hashed entry, then we can exit the manager with it armed. If it then
triggers after the manager has exited, we can have a use-after-free where
io_wqe_hash_wake() attempts to wake a now gone manager process.

Move the killing of the hashed write queue into the manager itself, so
that we know we've killed it before the task exits.

Fixes: e941894e ("io-wq: make buffered file write hashed work map per-ctx")
Signed-off-by: Jens Axboe <axboe@kernel.dk>

The callback can only be armed, if we get -EIOCBQUEUED returned. It's
important that we clear the WAITQ bit for other cases, otherwise we can
queue for async retry and filemap will assume that we're armed and
return -EAGAIN instead of just blocking for the IO.

Cc: stable@vger.kernel.org # 5.9+
Signed-off-by: Jens Axboe <axboe@kernel.dk>

It doesn't make sense to wait for more events to come in, if we can't
even flush the overflow we already have to the ring. Return -EBUSY for
that condition, just like we do for attempts to submit with overflow
pending.

Cc: stable@vger.kernel.org # 5.11
Signed-off-by: Jens Axboe <axboe@kernel.dk>

This allows us to do task creation and setup without needing to use
completions to try and synchronize with the starting thread. Get rid of
the old io_wq_fork_thread() wrapper, and the 'wq' and 'worker' startup
completion events - we can now do setup before the task is running.
Signed-off-by: Jens Axboe <axboe@kernel.dk>

* powercap:
  powercap/drivers/dtpm: Add the experimental label to the option description
  powercap/drivers/dtpm: Fix root node initialization

In case we have already established connection to nvmf target, it
shouldn't be allowed to change the model_number. E.g. if someone will
identify ctrl and get model_number of "my_model" later on will change
the model_numbel via configfs to "my_new_model" this will break the NVMe
specification for "Get Log Page – Persistent Event Log" that refers to
Model Number as: "This field contains the same value as reported in the
Model Number field of the Identify Controller data structure, bytes
63:24."

Although it doesn't mentioned explicitly that this field can't be
changed, we can assume it.

So allow setting this field only once: using configfs or in the first
identify ctrl operation.
Signed-off-by: Max Gurtovoy <mgurtovoy@nvidia.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

Currently kato is initialized to NVME_DEFAULT_KATO for both
discovery & i/o controllers. This is a problem specifically
for non-persistent discovery controllers since it always ends
up with a non-zero kato value. Fix this by initializing kato
to zero instead, and ensuring various controllers are assigned
appropriate kato values as follows:

non-persistent controllers  - kato set to zero
persistent controllers      - kato set to NVMF_DEV_DISC_TMO
                              (or any positive int via nvme-cli)
i/o controllers             - kato set to NVME_DEFAULT_KATO
                              (or any positive int via nvme-cli)
Signed-off-by: Martin George <marting@netapp.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

The hwmon pointer wont be NULL if the registration fails. Though the
exit code path will assign it to ctrl->hwmon_device. Later
nvme_hwmon_exit() will try to free the invalid pointer. Avoid this by
returning the error code from hwmon_device_register_with_info().

Fixes: ed7770f6 ("nvme/hwmon: rework to avoid devm allocation")
Signed-off-by: Daniel Wagner <dwagner@suse.de>
Signed-off-by: Christoph Hellwig <hch@lst.de>

Add the NVME_QUIRK_NO_NS_DESC_LIST and NVME_QUIRK_IGNORE_DEV_SUBNQN
quirks for this buggy device.

Reported and tested in https://bugs.mageia.org/show_bug.cgi?id=28417Signed-off-by: Pascal Terjan <pterjan@google.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

My 2TB SKC2000 showed the exact same symptoms that were provided
in 538e4a8c ("nvme-pci: avoid the deepest sleep state on
Kingston A2000 SSDs"), i.e. a complete NVME lockup that needed
cold boot to get it back.

According to some sources, the A2000 is simply a rebadged
SKC2000 with a slightly optimized firmware.

Adding the SKC2000 PCI ID to the quirk list with the same workaround
as the A2000 made my laptop survive a 5 hours long Yocto bootstrap
buildfest which reliably triggered the SSD lockup previously.
Signed-off-by: Zoltán Böszörményi <zboszor@gmail.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>

The kernel fails to fully detect these SSDs, only the character devices
are present:

[   10.785605] nvme nvme0: pci function 0000:04:00.0
[   10.876787] nvme nvme1: pci function 0000:81:00.0
[   13.198614] nvme nvme0: missing or invalid SUBNQN field.
[   13.198658] nvme nvme1: missing or invalid SUBNQN field.
[   13.206896] nvme nvme0: Shutdown timeout set to 20 seconds
[   13.215035] nvme nvme1: Shutdown timeout set to 20 seconds
[   13.225407] nvme nvme0: 16/0/0 default/read/poll queues
[   13.233602] nvme nvme1: 16/0/0 default/read/poll queues
[   13.239627] nvme nvme0: Identify Descriptors failed (8194)
[   13.246315] nvme nvme1: Identify Descriptors failed (8194)

Adding the NVME_QUIRK_NO_NS_DESC_LIST fixes this problem.

BugLink: https://bugzilla.kernel.org/show_bug.cgi?id=205679Signed-off-by: Julian Einwag <jeinwag-nvme@marcapo.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Keith Busch <kbusch@kernel.org>