PCI/portdrv: Don't disable AER reporting in get_port_device_capability()

[ Upstream commit 8795e182b02dc87e343c79e73af6b8b7f9c5e635 ]

AER reporting is currently disabled in the DevCtl registers of all non Root
Port PCIe devices on systems using pcie_ports_native || host->native_aer,
disabling AER completely in such systems. This is because 2bd50dd800b5
("PCI: PCIe: Disable PCIe port services during port initialization"), added
a call to pci_disable_pcie_error_reporting() *after* the AER setup was
completed for the PCIe device tree.

Here a longer analysis about the current status of AER enabling /
disabling upon bootup provided by Bjorn:

  pcie_portdrv_probe
    pcie_port_device_register
      get_port_device_capability
        pci_disable_pcie_error_reporting
          clear CERE NFERE FERE URRE               # <-- disable for RP USP DSP
      pcie_device_init
        device_register                            # new AER service device
          aer_probe
            aer_enable_rootport                    # RP only
              set_downstream_devices_error_reporting
                set_device_error_reporting         # self (RP)
                  if (RP || USP || DSP)
                    pci_enable_pcie_error_reporting
                      set CERE NFERE FERE URRE     # <-- enable for RP
                pci_walk_bus
                  set_device_error_reporting
                    if (RP || USP || DSP)
                      pci_enable_pcie_error_reporting
                        set CERE NFERE FERE URRE   # <-- enable for USP DSP

In a typical Root Port -> Endpoint hierarchy, the above:
  - Disables Error Reporting for the Root Port,
  - Enables Error Reporting for the Root Port,
  - Does NOT enable Error Reporting for the Endpoint because it is not a
    Root Port or Switch Port.

In a deeper Root Port -> Upstream Switch Port -> Downstream Switch
Port -> Endpoint hierarchy:
  - Disables Error Reporting for the Root Port,
  - Enables Error Reporting for the Root Port,
  - Enables Error Reporting for both Switch Ports,
  - Does NOT enable Error Reporting for the Endpoint because it is not a
    Root Port or Switch Port,
  - Disables Error Reporting for the Switch Ports when pcie_portdrv_probe()
    claims them.  AER does not re-enable it because these are not Root
    Ports.

Remove this call to pci_disable_pcie_error_reporting() from
get_port_device_capability(), leaving the already enabled AER configuration
intact. With this change, AER is enabled in the Root Port and the PCIe
switch upstream and downstream ports. Only the PCIe Endpoints don't have
AER enabled yet. A follow-up patch will take care of this Endpoint
enabling.

Fixes: 2bd50dd800b5 ("PCI: PCIe: Disable PCIe port services during port initialization")
Link: https://lore.kernel.org/r/20220125071820.2247260-3-sr@denx.de
Signed-off-by: Stefan Roese <sr@denx.de>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Pali Rohár <pali@kernel.org>
Cc: Rafael J. Wysocki <rjw@rjwysocki.net>
Cc: Bharat Kumar Gogada <bharat.kumar.gogada@xilinx.com>
Cc: Michal Simek <michal.simek@xilinx.com>
Cc: Yao Hongbo <yaohongbo@linux.alibaba.com>
Cc: Naveen Naidu <naveennaidu479@gmail.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

KVM: s390: pv: leak the topmost page table when destroy fails

[ Upstream commit faa2f72cb3569256480c5540d242c84e99965160 ]

Each secure guest must have a unique ASCE (address space control
element); we must avoid that new guests use the same page for their
ASCE, to avoid errors.

Since the ASCE mostly consists of the address of the topmost page table
(plus some flags), we must not return that memory to the pool unless
the ASCE is no longer in use.

Only a successful Destroy Secure Configuration UVC will make the ASCE
reusable again.

If the Destroy Configuration UVC fails, the ASCE cannot be reused for a
secure guest (either for the ASCE or for other memory areas). To avoid
a collision, it must not be used again. This is a permanent error and
the page becomes in practice unusable, so we set it aside and leak it.
On failure we already leak other memory that belongs to the ultravisor
(i.e. the variable and base storage for a guest) and not leaking the
topmost page table was an oversight.

This error (and thus the leakage) should not happen unless the hardware
is broken or KVM has some unknown serious bug.

Signed-off-by: Claudio Imbrenda <imbrenda@linux.ibm.com>
Fixes: 29b40f105ec8d55 ("KVM: s390: protvirt: Add initial vm and cpu lifecycle handling")
Reviewed-by: Janosch Frank <frankja@linux.ibm.com>
Link: https://lore.kernel.org/r/20220628135619.32410-2-imbrenda@linux.ibm.com
Message-Id: <20220628135619.32410-2-imbrenda@linux.ibm.com>
Signed-off-by: Janosch Frank <frankja@linux.ibm.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

mmc: block: Add single read for 4k sector cards

[ Upstream commit b3fa3e6dccc465969721b8bd2824213bd235efeb ]

Cards with 4k native sector size may only be read 4k-aligned,
accommodate for this in the single read recovery and use it.

Fixes: 81196976ed946 (mmc: block: Add blk-mq support)
Signed-off-by: Christian Loehle <cloehle@hyperstone.com>
Acked-by: Adrian Hunter <adrian.hunter@intel.com>
Reviewed-by: Avri Altman <avri.altman@wdc.com>
Link: https://lore.kernel.org/r/cf4f316274c5474586d0d99b17db4a4c@hyperstone.com
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

of: device: Fix missing of_node_put() in of_dma_set_restricted_buffer

[ Upstream commit d17e37c41b7ed38459957a5d2968ba61516fd5c2 ]

We should use of_node_put() for the reference 'node' returned by
of_parse_phandle() which will increase the refcount.

Fixes: fec9b625095f ("of: Add plumbing for restricted DMA pool")
Co-authored-by: Miaoqian Lin <linmq006@gmail.com>
Signed-off-by: Liang He <windhl@126.com>
Signed-off-by: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20220702014449.263772-1-windhl@126.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

mmc: sdhci-of-at91: fix set_uhs_signaling rewriting of MC1R

[ Upstream commit 5987e6ded29d52e42fc7b06aa575c60a25eee38e ]

In set_uhs_signaling, the DDR bit is being set by fully writing the MC1R
register.
This can lead to accidental erase of certain bits in this register.
Avoid this by doing a read-modify-write operation.

Fixes: d0918764c17b ("mmc: sdhci-of-at91: fix MMC_DDR_52 timing selection")
Signed-off-by: Eugen Hristev <eugen.hristev@microchip.com>
Tested-by: Karl Olsen <karl@micro-technic.com>
Acked-by: Adrian Hunter <adrian.hunter@intel.com>
Link: https://lore.kernel.org/r/20220630090926.15061-1-eugen.hristev@microchip.com
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

memstick/ms_block: Fix a memory leak

[ Upstream commit 54eb7a55be6779c4d0c25eaf5056498a28595049 ]

'erased_blocks_bitmap' is never freed. As it is allocated at the same time
as 'used_blocks_bitmap', it is likely that it should be freed also at the
same time.

Add the corresponding bitmap_free() in msb_data_clear().

Fixes: 0ab30494bc4f ("memstick: add support for legacy memorysticks")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Link: https://lore.kernel.org/r/b3b78926569445962ea5c3b6e9102418a9effb88.1656155715.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

memstick/ms_block: Fix some incorrect memory allocation

[ Upstream commit 2e531bc3e0d86362fcd8a577b3278d9ef3cc2ba0 ]

Some functions of the bitmap API take advantage of the fact that a bitmap
is an array of long.

So, to make sure this assertion is correct, allocate bitmaps with
bitmap_zalloc() instead of kzalloc()+hand-computed number of bytes.

While at it, also use bitmap_free() instead of kfree() to keep the
semantic.

Fixes: 0ab30494bc4f ("memstick: add support for legacy memorysticks")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Link: https://lore.kernel.org/r/dbf633c48c24ae6d95f852557e8d8b3bbdef65fe.1656155715.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

mmc: renesas_sdhi: Get the reset handle early in the probe

[ Upstream commit 0dac1e498f8130fdacfdd5289e3a7ac87ec1b9ad ]

In case of devm_reset_control_get_optional_exclusive() failure we returned
directly instead of jumping to the error path to roll back initialization.

This patch moves devm_reset_control_get_optional_exclusive() early in the
probe so that we have the reset handle prior to initialization of the
hardware.

Fixes: b4d86f37eacb7 ("mmc: renesas_sdhi: do hard reset if possible")
Reported-by: Pavel Machek <pavel@denx.de>
Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Link: https://lore.kernel.org/r/20220624181438.4355-2-prabhakar.mahadev-lad.rj@bp.renesas.com
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

mmc: mxcmmc: Silence a clang warning

[ Upstream commit 7dc65e3c0ef4b746a583b7c58f99873fddf5ccfa ]

Change the of_device_get_match_data() cast to (uintptr_t)
to silence the following clang warning:

drivers/mmc/host/mxcmmc.c:1028:18: warning: cast to smaller integer type 'enum mxcmci_type' from 'const void *' [-Wvoid-pointer-to-enum-cast]

Reported-by: kernel test robot <lkp@intel.com>
Fixes: 8223e885e74b ("mmc: mxc: Convert the driver to DT-only")
Signed-off-by: Fabio Estevam <festevam@gmail.com>
Link: https://lore.kernel.org/r/20220526010022.1163483-1-festevam@gmail.com
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

mmc: sdhci-of-esdhc: Fix refcount leak in esdhc_signal_voltage_switch

[ Upstream commit b5899a3e2f783a27b268e38d37f9b24c71bddf45 ]

of_find_matching_node() returns a node pointer with refcount
incremented, we should use of_node_put() on it when not need anymore.
Add missing of_node_put() to avoid refcount leak.
of_node_put() checks null pointer.

Fixes: ea35645a3c66 ("mmc: sdhci-of-esdhc: add support for signal voltage switch")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Link: https://lore.kernel.org/r/20220523144255.10310-1-linmq006@gmail.com
Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

staging: rtl8192u: Fix sleep in atomic context bug in dm_fsync_timer_callback

[ Upstream commit 6a0c054930d554ad8f8044ef1fc856d9da391c81 ]

There are sleep in atomic context bugs when dm_fsync_timer_callback is
executing. The root cause is that the memory allocation functions with
GFP_KERNEL or GFP_NOIO parameters are called in dm_fsync_timer_callback
which is a timer handler. The call paths that could trigger bugs are
shown below:

    (interrupt context)
dm_fsync_timer_callback
  write_nic_byte
    kzalloc(sizeof(data), GFP_KERNEL); //may sleep
    usb_control_msg
      kmalloc(.., GFP_NOIO); //may sleep
  write_nic_dword
    kzalloc(sizeof(data), GFP_KERNEL); //may sleep
    usb_control_msg
      kmalloc(.., GFP_NOIO); //may sleep

This patch uses delayed work to replace timer and moves the operations
that may sleep into the delayed work in order to mitigate bugs.

Fixes: 8fc8598e61f6 ("Staging: Added Realtek rtl8192u driver to staging")
Signed-off-by: Duoming Zhou <duoming@zju.edu.cn>
Link: https://lore.kernel.org/r/20220710103002.63283-1-duoming@zju.edu.cn
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

binder: fix redefinition of seq_file attributes

[ Upstream commit b7e241bbff24f9e106bf616408fd58bcedc44bae ]

The patchset in [1] exported some definitions to binder_internal.h in
order to make the debugfs entries such as 'stats' and 'transaction_log'
available in a binderfs instance. However, the DEFINE_SHOW_ATTRIBUTE
macro expands into a static function/variable pair, which in turn get
redefined each time a source file includes this internal header.

This problem was made evident after a report from the kernel test robot
<lkp@intel.com> where several W=1 build warnings are seen in downstream
kernels. See the following example:

  include/../drivers/android/binder_internal.h:111:23: warning: 'binder_stats_fops' defined but not used [-Wunused-const-variable=]
     111 | DEFINE_SHOW_ATTRIBUTE(binder_stats);
         |                       ^~~~~~~~~~~~
  include/linux/seq_file.h:174:37: note: in definition of macro 'DEFINE_SHOW_ATTRIBUTE'
     174 | static const struct file_operations __name ## _fops = {                 \
         |                                     ^~~~~~

This patch fixes the above issues by moving back the definitions into
binder.c and instead creates an array of the debugfs entries which is
more convenient to share with binderfs and iterate through.

  [1] https://lore.kernel.org/all/20190903161655.107408-1-hridya@google.com/

Fixes: 0e13e452dafc ("binder: Add stats, state and transactions files")
Fixes: 03e2e07e3814 ("binder: Make transaction_log available in binderfs")
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Link: https://lore.kernel.org/r/20220701182041.2134313-1-cmllamas@google.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

intel_th: msu: Fix vmalloced buffers

[ Upstream commit ac12ad3ccf6d386e64a9d6a890595a2509d24edd ]

After commit f5ff79fddf0e ("dma-mapping: remove CONFIG_DMA_REMAP") there's
a chance of DMA buffer getting allocated via vmalloc(), which messes up
the mmapping code:

> RIP: msc_mmap_fault [intel_th_msu]
> Call Trace:
>  <TASK>
>  __do_fault
>  do_fault
...

Fix this by accounting for vmalloc possibility.

Fixes: ba39bd830605 ("intel_th: msu: Switch over to scatterlist")
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Link: https://lore.kernel.org/r/20220705082637.59979-4-alexander.shishkin@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

intel_th: msu-sink: Potential dereference of null pointer

[ Upstream commit 82f76a4a720791d889de775b5f7541d601efc8bd ]

The return value of dma_alloc_coherent() needs to be checked.
To avoid use of null pointer in sg_set_buf() in case of the failure of
alloc.

Fixes: f220df66f676 ("intel_th: msu-sink: An example msu buffer "sink"")
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Jiasheng Jiang <jiasheng@iscas.ac.cn>
Signed-off-by: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Link: https://lore.kernel.org/r/20220705082637.59979-3-alexander.shishkin@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

intel_th: Fix a resource leak in an error handling path

[ Upstream commit 086c28ab7c5699256aced0049aae9c42f1410313 ]

If an error occurs after calling 'pci_alloc_irq_vectors()',
'pci_free_irq_vectors()' must be called as already done in the remove
function.

Fixes: 7b7036d47c35 ("intel_th: pci: Use MSI interrupt signalling")
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Link: https://lore.kernel.org/r/20220705082637.59979-2-alexander.shishkin@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: qla2xxx: Check correct variable in qla24xx_async_gffid()

[ Upstream commit 7c33e477bd883f79cccec418980cb8f7f2d50347 ]

There is a copy and paste bug here.  It should check ".rsp" instead of
".req".  The error message is copy and pasted as well so update that too.

Link: https://lore.kernel.org/r/YrK1A/t3L6HKnswO@kili
Fixes: 9c40c36e75ff ("scsi: qla2xxx: edif: Reduce Initiator-Initiator thrashing")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

PCI: endpoint: Don't stop controller when unbinding endpoint function

[ Upstream commit 1bc2b7bfba6e2f64edf5e246f3af2967261f6c3d ]

Unbinding an endpoint function from the endpoint controller shouldn't stop
the controller.  This is especially a problem for multi-function endpoints
where other endpoints may still be active.

Don't stop the controller when unbinding one of its endpoints.  Normally
the controller is stopped via configfs.

Fixes: 349e7a85b25f ("PCI: endpoint: functions: Add an EP function to test PCI")
Link: https://lore.kernel.org/r/20220622040924.113279-1-mie@igel.co.jp
Signed-off-by: Shunsuke Mie <mie@igel.co.jp>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Kishon Vijay Abraham I <kishon@ti.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

dmaengine: sf-pdma: Add multithread support for a DMA channel

[ Upstream commit b2cc5c465c2cb8ab697c3fd6583c614e3f6cfbcc ]

When we get a DMA channel and try to use it in multiple threads it
will cause oops and hanging the system.

% echo 64 > /sys/module/dmatest/parameters/threads_per_chan
% echo 10000 > /sys/module/dmatest/parameters/iterations
% echo 1 > /sys/module/dmatest/parameters/run
[   89.480664] Unable to handle kernel NULL pointer dereference at virtual
               address 00000000000000a0
[   89.488725] Oops [#1]
[   89.494708] CPU: 2 PID: 1008 Comm: dma0chan0-copy0 Not tainted
               5.17.0-rc5
[   89.509385] epc : vchan_find_desc+0x32/0x46
[   89.513553]  ra : sf_pdma_tx_status+0xca/0xd6

This happens because of data race. Each thread rewrite channels's
descriptor as soon as device_prep_dma_memcpy() is called. It leads to the
situation when the driver thinks that it uses right descriptor that
actually is freed or substituted for other one.

With current fixes a descriptor changes its value only when it has
been used. A new descriptor is acquired from vc->desc_issued queue that
is already filled with descriptors that are ready to be sent. Threads
have no direct access to DMA channel descriptor. Now it is just possible
to queue a descriptor for further processing.

Fixes: 6973886ad58e ("dmaengine: sf-pdma: add platform DMA support for HiFive Unleashed A00")
Signed-off-by: Viacheslav Mitrofanov <v.v.mitrofanov@yadro.com>
Link: https://lore.kernel.org/r/20220701082942.12835-1-v.v.mitrofanov@yadro.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

KVM: arm64: Don't return from void function

[ Upstream commit 1c3ace2b8b3995d3213c5e2d2aca01a0577a3b0f ]

Although harmless, the return statement in kvm_unexpected_el2_exception
is rather confusing as the function itself has a void return type. The
C standard is also pretty clear that "A return statement with an
expression shall not appear in a function whose return type is void".
Given that this return statement does not seem to add any actual value,
let's not pointlessly violate the standard.

Build-tested with GCC 10 and CLANG 13 for good measure, the disassembled
code is identical with or without the return statement.

Fixes: e9ee186bb735 ("KVM: arm64: Add kvm_extable for vaxorcism code")
Signed-off-by: Quentin Perret <qperret@google.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20220705142310.3847918-1-qperret@google.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

soundwire: revisit driver bind/unbind and callbacks

[ Upstream commit bd29c00edd0a5dac8b6e7332bb470cd50f92e893 ]

In the SoundWire probe, we store a pointer from the driver ops into
the 'slave' structure. This can lead to kernel oopses when unbinding
codec drivers, e.g. with the following sequence to remove machine
driver and codec driver.

/sbin/modprobe -r snd_soc_sof_sdw
/sbin/modprobe -r snd_soc_rt711

The full details can be found in the BugLink below, for reference the
two following examples show different cases of driver ops/callbacks
being invoked after the driver .remove().

kernel: BUG: kernel NULL pointer dereference, address: 0000000000000150
kernel: Workqueue: events cdns_update_slave_status_work [soundwire_cadence]
kernel: RIP: 0010:mutex_lock+0x19/0x30
kernel: Call Trace:
kernel:  ? sdw_handle_slave_status+0x426/0xe00 [soundwire_bus 94ff184bf398570c3f8ff7efe9e32529f532e4ae]
kernel:  ? newidle_balance+0x26a/0x400
kernel:  ? cdns_update_slave_status_work+0x1e9/0x200 [soundwire_cadence 1bcf98eebe5ba9833cd433323769ac923c9c6f82]

kernel: BUG: unable to handle page fault for address: ffffffffc07654c8
kernel: Workqueue: pm pm_runtime_work
kernel: RIP: 0010:sdw_bus_prep_clk_stop+0x6f/0x160 [soundwire_bus]
kernel: Call Trace:
kernel:  <TASK>
kernel:  sdw_cdns_clock_stop+0xb5/0x1b0 [soundwire_cadence 1bcf98eebe5ba9833cd433323769ac923c9c6f82]
kernel:  intel_suspend_runtime+0x5f/0x120 [soundwire_intel aca858f7c87048d3152a4a41bb68abb9b663a1dd]
kernel:  ? dpm_sysfs_remove+0x60/0x60

This was not detected earlier in Intel tests since the tests first
remove the parent PCI device and shut down the bus. The sequence
above is a corner case which keeps the bus operational but without a
driver bound.

While trying to solve this kernel oopses, it became clear that the
existing SoundWire bus does not deal well with the unbind case.

Commit 528be501b7d4a ("soundwire: sdw_slave: add probe_complete structure and new fields")
added a 'probed' status variable and a 'probe_complete'
struct completion. This status is however not reset on remove and
likewise the 'probe complete' is not re-initialized, so the
bind/unbind/bind test cases would fail. The timeout used before the
'update_status' callback was also a bad idea in hindsight, there
should really be no timing assumption as to if and when a driver is
bound to a device.

An initial draft was based on device_lock() and device_unlock() was
tested. This proved too complicated, with deadlocks created during the
suspend-resume sequences, which also use the same device_lock/unlock()
as the bind/unbind sequences. On a CometLake device, a bad DSDT/BIOS
caused spurious resumes and the use of device_lock() caused hangs
during suspend. After multiple weeks or testing and painful
reverse-engineering of deadlocks on different devices, we looked for
alternatives that did not interfere with the device core.

A bus notifier was used successfully to keep track of DRIVER_BOUND and
DRIVER_UNBIND events. This solved the bind-unbind-bind case in tests,
but it can still be defeated with a theoretical corner case where the
memory is freed by a .remove while the callback is in use. The
notifier only helps make sure the driver callbacks are valid, but not
that the memory allocated in probe remains valid while the callbacks
are invoked.

This patch suggests the introduction of a new 'sdw_dev_lock' mutex
protecting probe/remove and all driver callbacks. Since this mutex is
'local' to SoundWire only, it does not interfere with existing locks
and does not create deadlocks. In addition, this patch removes the
'probe_complete' completion, instead we directly invoke the
'update_status' from the probe routine. That removes any sort of
timing dependency and a much better support for the device/driver
model, the driver could be bound before the bus started, or eons after
the bus started and the hardware would be properly initialized in all
cases.

BugLink: https://github.com/thesofproject/linux/issues/3531
Fixes: 56d4fe31af77 ("soundwire: Add MIPI DisCo property helpers")
Fixes: 528be501b7d4a ("soundwire: sdw_slave: add probe_complete structure and new fields")
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Rander Wang <rander.wang@intel.com>
Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Reviewed-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Reviewed-by: Péter Ujfalusi <peter.ujfalusi@linux.intel.com>
Link: https://lore.kernel.org/r/20220621225641.221170-2-pierre-louis.bossart@linux.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

soundwire: bus_type: fix remove and shutdown support

[ Upstream commit df6407782964dc7e35ad84230abb38f46314b245 ]

The bus sdw_drv_remove() and sdw_drv_shutdown() helpers are used
conditionally, if the driver provides these routines.

These helpers already test if the driver provides a .remove or
.shutdown callback, so there's no harm in invoking the
sdw_drv_remove() and sdw_drv_shutdown() unconditionally.

In addition, the current code is imbalanced with
dev_pm_domain_attach() called from sdw_drv_probe(), but
dev_pm_domain_detach() called from sdw_drv_remove() only if the driver
provides a .remove callback.

Fixes: 9251345dca24b ("soundwire: Add SoundWire bus type")
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Rander Wang <rander.wang@intel.com>
Signed-off-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Link: https://lore.kernel.org/r/20220610015105.25987-1-yung-chuan.liao@linux.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

PCI: dwc: Always enable CDM check if "snps,enable-cdm-check" exists

[ Upstream commit ec7b952f453ce7eabe7e1bea584626934d44f668 ]

If the "snps,enable-cdm-check" property exists, we should enable the CDM
check.  But previously dw_pcie_setup() could exit before doing so if the
"num-lanes" property was absent or invalid.

Move the CDM enable earlier so we do it regardless of whether "num-lanes"
is present.

[bhelgaas: commit log]
Fixes: 07f123def73e ("PCI: dwc: Add support to enable CDM register check")
Link: https://lore.kernel.org/r/20220624143428.8334-7-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Vidya Sagar <vidyas@nvidia.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

PCI: dwc: Deallocate EPC memory on dw_pcie_ep_init() errors

[ Upstream commit 8161e9626b50892eaedbd8070ecb1586ecedb109 ]

If dw_pcie_ep_init() fails to perform any action after the EPC memory is
initialized and the MSI memory region is allocated, the latter parts won't
be undone thus causing a memory leak.  Add a cleanup-on-error path to fix
these leaks.

[bhelgaas: commit log]
Fixes: 2fd0c9d966cc ("PCI: designware-ep: Pre-allocate memory for MSI in dw_pcie_ep_init")
Link: https://lore.kernel.org/r/20220624143428.8334-6-Sergey.Semin@baikalelectronics.ru
Tested-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

PCI: dwc: Set INCREASE_REGION_SIZE flag based on limit address

[ Upstream commit 777e7c3ab73036105e6fc4a67ed950179dbffbab ]

We program the 64-bit ATU limit address (in PCIE_ATU_LIMIT/
PCIE_ATU_UPPER_LIMIT or PCIE_ATU_UNR_LOWER_LIMIT/PCIE_ATU_UNR_UPPER_LIMIT),
but in addition, the PCIE_ATU_INCREASE_REGION_SIZE bit must be set if the
upper 32 bits of the limit address differ from the upper 32 bits of the
base address (see [1,2]).

5b4cf0f65324 ("PCI: dwc: Add upper limit address for outbound iATU") set
PCIE_ATU_INCREASE_REGION_SIZE, but only when the *size* was greater than
4GB.  It did not set it when a smaller region crossed a 4GB boundary, e.g.,
[mem 0x0_f0000000-0x1_0fffffff].

Set PCIE_ATU_INCREASE_REGION_SIZE whenever PCIE_ATU_UPPER_LIMIT is
greater than PCIE_ATU_UPPER_BASE.

[1] DesignWare Cores PCI Express Controller Databook - DWC PCIe Root Port,
    v5.40a, March 2019, fig.3-36, p.175
[2] DesignWare Cores PCI Express Controller Databook - DWC PCIe Root Port,
    v5.40a, March 2019, fig.3-37, p.176

[bhelgaas: commit log]
Fixes: 5b4cf0f65324 ("PCI: dwc: Add upper limit address for outbound iATU")
Link: https://lore.kernel.org/r/20220624143428.8334-5-Sergey.Semin@baikalelectronics.ru
Tested-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

PCI: dwc: Disable outbound windows only for controllers using iATU

[ Upstream commit d60a2e281e9de2b2f67343b2e39417ca0f4fd54e ]

Some DWC-based controllers (e.g., pcie-al.c and pci-keystone.c, identified
by the fact that they override the default dw_child_pcie_ops) use their own
address translation approach instead of the DWC internal ATU (iATU).  For
those controllers, skip disabling the iATU outbound windows.

[bhelgaas: commit log, update multiple window comment]
Fixes: 458ad06c4cdd ("PCI: dwc: Ensure all outbound ATU windows are reset")
Link: https://lore.kernel.org/r/20220624143428.8334-4-Sergey.Semin@baikalelectronics.ru
Tested-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

PCI: dwc: Add unroll iATU space support to dw_pcie_disable_atu()

[ Upstream commit d1cf738f2b65a5640234e1da90a68d3523fbed83 ]

dw_pcie_disable_atu() was introduced by f8aed6ec624f ("PCI: dwc:
designware: Add EP mode support") and supported only the viewport version
of the iATU CSRs.

DW PCIe IP cores v4.80a and newer also support unrolled iATU/eDMA space.
Callers of dw_pcie_disable_atu(), including pci_epc_ops.clear_bar(),
pci_epc_ops.unmap_addr(), and dw_pcie_setup_rc(), don't work correctly when
it is enabled.

Add dw_pcie_disable_atu() support for controllers with unrolled iATU CSRs
enabled.

[bhelgaas: commit log]
Fixes: f8aed6ec624f ("PCI: dwc: designware: Add EP mode support")
Link: https://lore.kernel.org/r/20220624143428.8334-3-Sergey.Semin@baikalelectronics.ru
Tested-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

PCI: dwc: Stop link on host_init errors and de-initialization

[ Upstream commit 113fa857b74c947137d845e7e635afcf6a59c43a ]

It's logically correct to undo everything that was done when an error is
discovered or in the corresponding cleanup counterpart. Otherwise the host
controller will be left in an undetermined state. Since the link is set up
in the host_init method, deactivate it there in the cleanup-on-error block
and stop the link in the antagonistic routine - dw_pcie_host_deinit(). Link
deactivation is platform-specific and should be implemented in
dw_pcie_ops.stop_link().

Fixes: 886a9c134755 ("PCI: dwc: Move link handling into common code")
Link: https://lore.kernel.org/r/20220624143428.8334-2-Sergey.Semin@baikalelectronics.ru
Tested-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

mm/mempolicy: fix get_nodes out of bound access

[ Upstream commit 000eca5d044d1ee23b4ca311793cf3fc528da6c6 ]

When user specified more nodes than supported, get_nodes will access nmask
array out of bounds.

Link: https://lkml.kernel.org/r/20220601093211.2970565-1-tianyu.li@arm.com
Fixes: e130242dc351 ("mm: simplify compat numa syscalls")
Signed-off-by: Tianyu Li <tianyu.li@arm.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: qcom: clk-rcg2: Make sure to not write d=0 to the NMD register

[ Upstream commit d0696770cef35a1fd16ea2167e2198c18aa6fbfe ]

Sometimes calculation of d value may result in 0 because of the
rounding after integer division. This causes the following error:

[  113.969689] camss_gp1_clk_src: rcg didn't update its configuration.
[  113.969754] WARNING: CPU: 3 PID: 35 at drivers/clk/qcom/clk-rcg2.c:122 update_config+0xc8/0xdc

Make sure that D value is never zero.

Fixes: 7f891faf596e ("clk: qcom: clk-rcg2: Add support for duty-cycle for RCG")
Signed-off-by: Nikita Travkin <nikita@trvn.ru>
Reviewed-by: Stephen Boyd <sboyd@kernel.org>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20220612145955.385787-3-nikita@trvn.ru
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: qcom: clk-rcg2: Fail Duty-Cycle configuration if MND divider is not enabled.

[ Upstream commit bdafb609c3bb848d710ad9cd4debd2ee9d6a4049 ]

In cases when MND is not enabled (e.g. when only Half Integer Divider is
used), setting D registers makes no effect.

Fail instead of making ineffective write.

Fixes: 7f891faf596e ("clk: qcom: clk-rcg2: Add support for duty-cycle for RCG")
Signed-off-by: Nikita Travkin <nikita@trvn.ru>
Reviewed-by: Stephen Boyd <sboyd@kernel.org>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20220612145955.385787-2-nikita@trvn.ru
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: qcom: camcc-sm8250: Fix topology around titan_top power domain

[ Upstream commit f8acf01a6a4f84baf05181e24bd48def4ba23f5b ]

On SM8250 two found VFE GDSC power domains shall not be operated, if
titan top is turned off, thus the former power domains will be set as
subdomains by a GDSC registration routine.

Fixes: 5d66ca79b58c ("clk: qcom: Add camera clock controller driver for SM8250")
Signed-off-by: Vladimir Zapolskiy <vladimir.zapolskiy@linaro.org>
Reviewed-by: Robert Foss <robert.foss@linaro.org>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20220519214133.1728979-3-vladimir.zapolskiy@linaro.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: qcom: camcc-sdm845: Fix topology around titan_top power domain

[ Upstream commit 103dd2338bbff567bce7acd00fc5a09c806b38ec ]

On SDM845 two found VFE GDSC power domains shall not be operated, if
titan top is turned off, thus the former power domains will be set as
subdomains by a GDSC registration routine.

Fixes: 78412c262004 ("clk: qcom: Add camera clock controller driver for SDM845")
Signed-off-by: Vladimir Zapolskiy <vladimir.zapolskiy@linaro.org>
Reviewed-by: Robert Foss <robert.foss@linaro.org>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20220519214133.1728979-2-vladimir.zapolskiy@linaro.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: qcom: ipq8074: set BRANCH_HALT_DELAY flag for UBI clocks

[ Upstream commit 2bd357e698207e2e65db03007e4be65bf9d6a7b3 ]

Currently, attempting to enable the UBI clocks will cause the stuck at
off warning to be printed and clk_enable will fail.

[   14.936694] gcc_ubi1_ahb_clk status stuck at 'off'

Downstream 5.4 QCA kernel has fixed this by seting the BRANCH_HALT_DELAY
flag on UBI clocks, so lets do the same.

Fixes: 5736294aef83 ("clk: qcom: ipq8074: add NSS clocks")
Signed-off-by: Robert Marko <robimarko@gmail.com>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20220515210048.483898-6-robimarko@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: qcom: ipq8074: fix NSS port frequency tables

[ Upstream commit 0e9e61a2815b5cd34f1b495b2d72e8127ce9b794 ]

NSS port 5 and 6 frequency tables are currently broken and are causing a
wide ranges of issue like 1G not working at all on port 6 or port 5 being
clocked with 312 instead of 125 MHz as UNIPHY1 gets selected.

So, update the frequency tables with the ones from the downstream QCA 5.4
based kernel which has already fixed this.

Fixes: 7117a51ed303 ("clk: qcom: ipq8074: add NSS ethernet port clocks")
Signed-off-by: Robert Marko <robimarko@gmail.com>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20220515210048.483898-3-robimarko@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: qcom: ipq8074: SW workaround for UBI32 PLL lock

[ Upstream commit 3401ea2856ef84f39b75f0dc5ebcaeda81cb90ec ]

UBI32 Huayra PLL fails to lock in 5 us in some SoC silicon and thus it
will cause the wait_for_pll() to timeout and thus return the error
indicating that the PLL failed to lock.

This is bug in Huayra PLL HW for which SW workaround
is to set bit 26 of TEST_CTL register.

This is ported from the QCA 5.4 based downstream kernel.

Fixes: b8e7e519625f ("clk: qcom: ipq8074: add remaining PLL’s")
Signed-off-by: Robert Marko <robimarko@gmail.com>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20220515210048.483898-2-robimarko@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: qcom: ipq8074: fix NSS core PLL-s

[ Upstream commit ca41ec1b30434636c56c5600b24a8d964d359d9c ]

Like in IPQ6018 the NSS related Alpha PLL-s require initial configuration
to work.

So, obtain the regmap that is required for the Alpha PLL configuration
and thus utilize the qcom_cc_really_probe() as we already have the regmap.
Then utilize the Alpha PLL configs from the downstream QCA 5.4 based
kernel to configure them.

This fixes the UBI32 and NSS crypto PLL-s failing to get enabled by the
kernel.

Fixes: b8e7e519625f ("clk: qcom: ipq8074: add remaining PLL’s")
Signed-off-by: Robert Marko <robimarko@gmail.com>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20220515210048.483898-1-robimarko@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/rxe: Fix deadlock in rxe_do_local_ops()

[ Upstream commit 7cb33d1bc1ac8e51fd88928f96674d392f8e07c4 ]

When a local operation (invalidate mr, reg mr, bind mw) is finished there
will be no ack packet coming from a responder to cause the wqe to be
completed. This may happen anyway if a subsequent wqe performs
IO. Currently if the wqe is signalled the completer tasklet is scheduled
immediately but not otherwise.

This leads to a deadlock if the next wqe has the fence bit set in send
flags and the operation is not signalled. This patch removes the condition
that the wqe must be signalled in order to schedule the completer tasklet
which is the simplest fix for this deadlock and is fairly low cost. This
is the analog for local operations of always setting the ackreq bit in all
last or only request packets even if the operation is not signalled.

Link: https://lore.kernel.org/r/20220523223251.15350-1-rpearsonhpe@gmail.com
Reported-by: Jenny Hack <jhack@hpe.com>
Fixes: c1a411268a4b ("RDMA/rxe: Move local ops to subroutine")
Signed-off-by: Bob Pearson <rpearsonhpe@gmail.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

usb: host: xhci: use snprintf() in xhci_decode_trb()

[ Upstream commit 1ce69c35b86038dd11d3a6115a04501c5b89a940 ]

Commit cbf286e8ef83 ("xhci: fix unsafe memory usage in xhci tracing")
apparently missed one sprintf() call in xhci_decode_trb() -- replace
it with the snprintf() call as well...

Found by Linux Verification Center (linuxtesting.org) with the SVACE static
analysis tool.

Fixes: cbf286e8ef83 ("xhci: fix unsafe memory usage in xhci tracing")
Signed-off-by: Sergey Shtylyov <s.shtylyov@omp.ru>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Link: https://lore.kernel.org/r/20220630124645.1805902-2-mathias.nyman@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: qcom: gcc-msm8939: Point MM peripherals to system_mm_noc clock

[ Upstream commit 05eed0990927aa9634682fec58660e30f7b7ae30 ]

Qcom docs indciate the following peripherals operating from System NOC
MM not from System NOC clocks.

- MDP
- VFE
- JPEGe
- Venus

Switch over the relevant parent pointers.

Fixes: 1664014e4679 ("clk: qcom: gcc-msm8939: Add MSM8939 Generic Clock Controller")
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20220504163835.40130-5-bryan.odonoghue@linaro.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: qcom: gcc-msm8939: Add missing system_mm_noc_bfdcd_clk_src

[ Upstream commit dd363e2f7196278e7a30f509a0e8a841cb763b14 ]

The msm8939 has an additional higher operating point for the multi-media
peripherals. The higher throughput MM componets operate off of the
system-mm noc not the system noc.

system_mm_noc_bfdcd_clk_src is the source clock for the higher frequency
capable system noc mm.

Maximum frequency for the MM SNOC is 400 MHz.

Fixes: 1664014e4679 ("clk: qcom: gcc-msm8939: Add MSM8939 Generic Clock Controller")
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20220504163835.40130-4-bryan.odonoghue@linaro.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: qcom: gcc-msm8939: Fix bimc_ddr_clk_src rcgr base address

[ Upstream commit 63d42708320d6d2ca9ed505123d50ff4a542c36f ]

Reviewing qcom docs for the 8939 we can see the command rcgr is pointing to
the wrong address. bimc_ddr_clk_src_rcgr is @ 0x01832024 not 0x01832004.

Fixes: 1664014e4679 ("clk: qcom: gcc-msm8939: Add MSM8939 Generic Clock Controller")
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20220504163835.40130-3-bryan.odonoghue@linaro.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: qcom: gcc-msm8939: Add missing SYSTEM_MM_NOC_BFDCD_CLK_SRC

[ Upstream commit 07e7fcf1714c5f9930ad27613fea940aedba68da ]

When adding in the indexes for this clock-controller we missed
SYSTEM_MM_NOC_BFDCD_CLK_SRC.

Add it in now.

Fixes: 4c71d6abc4fc ("clk: qcom: Add DT bindings for MSM8939 GCC")
Cc: Rob Herring <robh+dt@kernel.org>
Cc: Krzysztof Kozlowski <krzysztof.kozlowski+dt@linaro.org>
Cc: devicetree@vger.kernel.org
Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Acked-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20220504163835.40130-2-bryan.odonoghue@linaro.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: qcom: clk-krait: unlock spin after mux completion

[ Upstream commit df83d2c9e72910416f650ade1e07cc314ff02731 ]

Unlock spinlock after the mux switch is completed to prevent any corner
case of mux request while the switch still needs to be done.

Fixes: 4d7dc77babfe ("clk: qcom: Add support for Krait clocks")
Signed-off-by: Ansuel Smith <ansuelsmth@gmail.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20220430054458.31321-3-ansuelsmth@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

driver core: fix potential deadlock in __driver_attach

[ Upstream commit 70fe758352cafdee72a7b13bf9db065f9613ced8 ]

In __driver_attach function, There are also AA deadlock problem,
like the commit b232b02bf3c2 ("driver core: fix deadlock in
__device_attach").

stack like commit b232b02bf3c2 ("driver core: fix deadlock in
__device_attach").
list below:
    In __driver_attach function, The lock holding logic is as follows:
    ...
    __driver_attach
    if (driver_allows_async_probing(drv))
      device_lock(dev)      // get lock dev
        async_schedule_dev(__driver_attach_async_helper, dev); // func
          async_schedule_node
            async_schedule_node_domain(func)
              entry = kzalloc(sizeof(struct async_entry), GFP_ATOMIC);
              /* when fail or work limit, sync to execute func, but
                 __driver_attach_async_helper will get lock dev as
                 will, which will lead to A-A deadlock.  */
              if (!entry || atomic_read(&entry_count) > MAX_WORK) {
                func;
              else
                queue_work_node(node, system_unbound_wq, &entry->work)
      device_unlock(dev)

    As above show, when it is allowed to do async probes, because of
    out of memory or work limit, async work is not be allowed, to do
    sync execute instead. it will lead to A-A deadlock because of
    __driver_attach_async_helper getting lock dev.

Reproduce:
and it can be reproduce by make the condition
(if (!entry || atomic_read(&entry_count) > MAX_WORK)) untenable, like
below:

[  370.785650] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables
this message.
[  370.787154] task:swapper/0       state:D stack:    0 pid:    1 ppid:
0 flags:0x00004000
[  370.788865] Call Trace:
[  370.789374]  <TASK>
[  370.789841]  __schedule+0x482/0x1050
[  370.790613]  schedule+0x92/0x1a0
[  370.791290]  schedule_preempt_disabled+0x2c/0x50
[  370.792256]  __mutex_lock.isra.0+0x757/0xec0
[  370.793158]  __mutex_lock_slowpath+0x1f/0x30
[  370.794079]  mutex_lock+0x50/0x60
[  370.794795]  __device_driver_lock+0x2f/0x70
[  370.795677]  ? driver_probe_device+0xd0/0xd0
[  370.796576]  __driver_attach_async_helper+0x1d/0xd0
[  370.797318]  ? driver_probe_device+0xd0/0xd0
[  370.797957]  async_schedule_node_domain+0xa5/0xc0
[  370.798652]  async_schedule_node+0x19/0x30
[  370.799243]  __driver_attach+0x246/0x290
[  370.799828]  ? driver_allows_async_probing+0xa0/0xa0
[  370.800548]  bus_for_each_dev+0x9d/0x130
[  370.801132]  driver_attach+0x22/0x30
[  370.801666]  bus_add_driver+0x290/0x340
[  370.802246]  driver_register+0x88/0x140
[  370.802817]  ? virtio_scsi_init+0x116/0x116
[  370.803425]  scsi_register_driver+0x1a/0x30
[  370.804057]  init_sd+0x184/0x226
[  370.804533]  do_one_initcall+0x71/0x3a0
[  370.805107]  kernel_init_freeable+0x39a/0x43a
[  370.805759]  ? rest_init+0x150/0x150
[  370.806283]  kernel_init+0x26/0x230
[  370.806799]  ret_from_fork+0x1f/0x30

To fix the deadlock, move the async_schedule_dev outside device_lock,
as we can see, in async_schedule_node_domain, the parameter of
queue_work_node is system_unbound_wq, so it can accept concurrent
operations. which will also not change the code logic, and will
not lead to deadlock.

Fixes: ef0ff68351be ("driver core: Probe devices asynchronously instead of the driver")
Signed-off-by: Zhang Wensheng <zhangwensheng5@huawei.com>
Link: https://lore.kernel.org/r/20220622074327.497102-1-zhangwensheng5@huawei.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

misc: rtsx: Fix an error handling path in rtsx_pci_probe()

[ Upstream commit 44fd1917314e9d4f53dd95dd65df1c152f503d3a ]

If an error occurs after a successful idr_alloc() call, the corresponding
resource must be released with idr_remove() as already done in the .remove
function.

Update the error handling path to add the missing idr_remove() call.

Fixes: ada8a8a13b13 ("mfd: Add realtek pcie card reader driver")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Link: https://lore.kernel.org/r/e8dc41716cbf52fb37a12e70d8972848e69df6d6.1655271216.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: qcom: camcc-sm8250: Fix halt on boot by reducing driver's init level

[ Upstream commit c4f40351901a10cd662ac2c081396d8fb04f584d ]

Access to I/O of SM8250 camera clock controller IP depends on enabled
GCC_CAMERA_AHB_CLK clock supplied by global clock controller, the latter
one is inited on subsys level, so, to satisfy the dependency, it would
make sense to deprive the init level of camcc-sm8250 driver.

If both drivers are compiled as built-in, there is a change that a board
won't boot up due to a race, which happens on the same init level.

Fixes: 5d66ca79b58c ("clk: qcom: Add camera clock controller driver for SM8250")
Signed-off-by: Vladimir Zapolskiy <vladimir.zapolskiy@linaro.org>
Reviewed-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Tested-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20220518103554.949511-1-vladimir.zapolskiy@linaro.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

mtd: dataflash: Add SPI ID table

[ Upstream commit ac4f83482afbfd927d0fe118151b747cf175e724 ]

Currently autoloading for SPI devices does not use the DT ID table, it uses
SPI modalises. Supporting OF modalises is going to be difficult if not
impractical, an attempt was made but has been reverted, so ensure that
module autoloading works for this driver by adding an id_table listing the
SPI IDs for everything.

Fixes: 96c8395e2166 ("spi: Revert modalias changes")
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Link: https://lore.kernel.org/linux-mtd/20220620152313.708768-1-broonie@kernel.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

dmaengine: dw-edma: Fix eDMA Rd/Wr-channels and DMA-direction semantics

[ Upstream commit c1e33979171da63cf47e56243ccb8ba82363c7d3 ]

In accordance with [1, 2] the DW eDMA controller has been created to be
part of the DW PCIe Root Port and DW PCIe End-point controllers and to
offload the transferring of large blocks of data between application and
remote PCIe domains leaving the system CPU free for other tasks. In the
first case (eDMA being part of DW PCIe Root Port) the eDMA controller is
always accessible via the CPU DBI interface and never over the PCIe wire.

The latter case is more complex. Depending on the DW PCIe End-Point IP-core
synthesize parameters it's possible to have the eDMA registers accessible
not only from the application CPU side, but also via mapping the eDMA CSRs
over a dedicated endpoint BAR. So based on the specifics denoted above the
eDMA driver is supposed to support two types of the DMA controller setups:

  1) eDMA embedded into the DW PCIe Root Port/End-point and accessible over
     the local CPU from the application side.

  2) eDMA embedded into the DW PCIe End-point and accessible via the PCIe
     wire with MWr/MRd TLPs generated by the CPU PCIe host controller.

Since the CPU memory resides different sides in these cases the semantics
of the MEM_TO_DEV and DEV_TO_MEM operations is flipped with respect to the
Tx and Rx DMA channels. So MEM_TO_DEV/DEV_TO_MEM corresponds to the Tx/Rx
channels in setup 1) and to the Rx/Tx channels in case of setup 2).

The DW eDMA driver has supported the case 2) since e63d79d1ffcd
("dmaengine: Add Synopsys eDMA IP core driver") in the framework of the
drivers/dma/dw-edma/dw-edma-pcie.c driver.

The case 1) support was added later by bd96f1b2f43a ("dmaengine: dw-edma:
support local dma device transfer semantics").  Afterwards the driver was
supposed to cover the both possible eDMA setups, but the latter commit
turned out to be not fully correct.

The problem was that the commit together with the new functionality support
also changed the channel direction semantics so the eDMA Read-channel
(corresponding to the DMA_DEV_TO_MEM direction for case 1) now uses the
sgl/cyclic base addresses as the Source addresses of the DMA transfers and
dma_slave_config.dst_addr as the Destination address of the DMA transfers.

Similarly the eDMA Write-channel (corresponding to the DMA_MEM_TO_DEV
direction for case 1) now uses dma_slave_config.src_addr as a source
address of the DMA transfers and sgl/cyclic base address as the Destination
address of the DMA transfers. This contradicts the logic of the
DMA-interface, which implies that DEV side is supposed to belong to the
PCIe device memory and MEM - to the CPU/Application memory. Indeed it seems
irrational to have the SG-list defined in the PCIe bus space, while
expecting a contiguous buffer allocated in the CPU memory. Moreover the
passed SG-list and cyclic DMA buffers are supposed to be mapped in a way so
to be seen by the DW eDMA Application (CPU) interface.

So in order to have the correct DW eDMA interface we need to invert the
eDMA Rd/Wr-channels and DMA-slave directions semantics by selecting the
src/dst addresses based on the DMA transfer direction instead of using the
channel direction capability.

[1] DesignWare Cores PCI Express Controller Databook - DWC PCIe Root Port,
    v.5.40a, March 2019, p.1092
[2] DesignWare Cores PCI Express Controller Databook - DWC PCIe Endpoint,
    v.5.40a, March 2019, p.1189

Co-developed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Fixes: bd96f1b2f43a ("dmaengine: dw-edma: support local dma device transfer semantics")
Link: https://lore.kernel.org/r/20220524152159.2370739-7-Frank.Li@nxp.com
Tested-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Signed-off-by: Frank Li <Frank.Li@nxp.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-By: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: iscsi: Fix session removal on shutdown

[ Upstream commit 31500e902759322ba3c64b60dabae2704e738df8 ]

When the system is shutting down, iscsid is not running so we will not get
a response to the ISCSI_ERR_INVALID_HOST error event. The system shutdown
will then hang waiting on userspace to remove the session.

This has libiscsi force the destruction of the session from the kernel when
iscsi_host_remove() is called from a driver's shutdown callout.

This fixes a regression added in qedi boot with commit d1f2ce77638d ("scsi:
qedi: Fix host removal with running sessions") which made qedi use the
common session removal function that waits on userspace instead of rolling
its own kernel based removal.

Link: https://lore.kernel.org/r/20220616222738.5722-7-michael.christie@oracle.com
Fixes: d1f2ce77638d ("scsi: qedi: Fix host removal with running sessions")
Tested-by: Nilesh Javali <njavali@marvell.com>
Reviewed-by: Lee Duncan <lduncan@suse.com>
Reviewed-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Mike Christie <michael.christie@oracle.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: iscsi: Add helper to remove a session from the kernel

[ Upstream commit bb42856bfd54fda1cbc7c470fcf5db1596938f4f ]

During qedi shutdown we need to stop the iSCSI layer from sending new nops
as pings and from responding to target ones and make sure there is no
running connection cleanups. Commit d1f2ce77638d ("scsi: qedi: Fix host
removal with running sessions") converted the driver to use the libicsi
helper to drive session removal, so the above issues could be handled. The
problem is that during system shutdown iscsid will not be running so when
we try to remove the root session we will hang waiting for userspace to
reply.

Add a helper that will drive the destruction of sessions like these during
system shutdown.

Link: https://lore.kernel.org/r/20220616222738.5722-5-michael.christie@oracle.com
Tested-by: Nilesh Javali <njavali@marvell.com>
Reviewed-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Mike Christie <michael.christie@oracle.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: iscsi: Allow iscsi_if_stop_conn() to be called from kernel

[ Upstream commit 3328333b47f4163504267440ec0a36087a407a5f ]

iscsi_if_stop_conn() is only called from the userspace interface but in a
subsequent commit we will want to call it from the kernel interface to
allow drivers like qedi to remove sessions from inside the kernel during
shutdown. This removes the iscsi_uevent code from iscsi_if_stop_conn() so we
can call it in a new helper.

Link: https://lore.kernel.org/r/20220616222738.5722-3-michael.christie@oracle.com
Tested-by: Nilesh Javali <njavali@marvell.com>
Reviewed-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Mike Christie <michael.christie@oracle.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

mwifiex: fix sleep in atomic context bugs caused by dev_coredumpv

[ Upstream commit a52ed4866d2b90dd5e4ae9dabd453f3ed8fa3cbc ]

There are sleep in atomic context bugs when uploading device dump
data in mwifiex. The root cause is that dev_coredumpv could not
be used in atomic contexts, because it calls dev_set_name which
include operations that may sleep. The call tree shows execution
paths that could lead to bugs:

   (Interrupt context)
fw_dump_timer_fn
  mwifiex_upload_device_dump
    dev_coredumpv(..., GFP_KERNEL)
      dev_coredumpm()
        kzalloc(sizeof(*devcd), gfp); //may sleep
        dev_set_name
          kobject_set_name_vargs
            kvasprintf_const(GFP_KERNEL, ...); //may sleep
            kstrdup(s, GFP_KERNEL); //may sleep

The corresponding fail log is shown below:

[  135.275938] usb 1-1: == mwifiex dump information to /sys/class/devcoredump start
[  135.281029] BUG: sleeping function called from invalid context at include/linux/sched/mm.h:265
...
[  135.293613] Call Trace:
[  135.293613]  <IRQ>
[  135.293613]  dump_stack_lvl+0x57/0x7d
[  135.293613]  __might_resched.cold+0x138/0x173
[  135.293613]  ? dev_coredumpm+0xca/0x2e0
[  135.293613]  kmem_cache_alloc_trace+0x189/0x1f0
[  135.293613]  ? devcd_match_failing+0x30/0x30
[  135.293613]  dev_coredumpm+0xca/0x2e0
[  135.293613]  ? devcd_freev+0x10/0x10
[  135.293613]  dev_coredumpv+0x1c/0x20
[  135.293613]  ? devcd_match_failing+0x30/0x30
[  135.293613]  mwifiex_upload_device_dump+0x65/0xb0
[  135.293613]  ? mwifiex_dnld_fw+0x1b0/0x1b0
[  135.293613]  call_timer_fn+0x122/0x3d0
[  135.293613]  ? msleep_interruptible+0xb0/0xb0
[  135.293613]  ? lock_downgrade+0x3c0/0x3c0
[  135.293613]  ? __next_timer_interrupt+0x13c/0x160
[  135.293613]  ? lockdep_hardirqs_on_prepare+0xe/0x220
[  135.293613]  ? mwifiex_dnld_fw+0x1b0/0x1b0
[  135.293613]  __run_timers.part.0+0x3f8/0x540
[  135.293613]  ? call_timer_fn+0x3d0/0x3d0
[  135.293613]  ? arch_restore_msi_irqs+0x10/0x10
[  135.293613]  ? lapic_next_event+0x31/0x40
[  135.293613]  run_timer_softirq+0x4f/0xb0
[  135.293613]  __do_softirq+0x1c2/0x651
...
[  135.293613] RIP: 0010:default_idle+0xb/0x10
[  135.293613] RSP: 0018:ffff888006317e68 EFLAGS: 00000246
[  135.293613] RAX: ffffffff82ad8d10 RBX: ffff888006301cc0 RCX: ffffffff82ac90e1
[  135.293613] RDX: ffffed100d9ff1b4 RSI: ffffffff831ad140 RDI: ffffffff82ad8f20
[  135.293613] RBP: 0000000000000003 R08: 0000000000000000 R09: ffff88806cff8d9b
[  135.293613] R10: ffffed100d9ff1b3 R11: 0000000000000001 R12: ffffffff84593410
[  135.293613] R13: 0000000000000000 R14: 0000000000000000 R15: 1ffff11000c62fd2
...
[  135.389205] usb 1-1: == mwifiex dump information to /sys/class/devcoredump end

This patch uses delayed work to replace timer and moves the operations
that may sleep into a delayed work in order to mitigate bugs, it was
tested on Marvell 88W8801 chip whose port is usb and the firmware is
usb8801_uapsta.bin. The following is the result after using delayed
work to replace timer.

[  134.936453] usb 1-1: == mwifiex dump information to /sys/class/devcoredump start
[  135.043344] usb 1-1: == mwifiex dump information to /sys/class/devcoredump end

As we can see, there is no bug now.

Fixes: f5ecd02a8b20 ("mwifiex: device dump support for usb interface")
Reviewed-by: Brian Norris <briannorris@chromium.org>
Signed-off-by: Duoming Zhou <duoming@zju.edu.cn>
Link: https://lore.kernel.org/r/b63b77fc84ed3e8a6bef02378e17c7c71a0bc3be.1654569290.git.duoming@zju.edu.cn
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

mwifiex: Ignore BTCOEX events from the 88W8897 firmware

[ Upstream commit 84d94e16efa268e4f2887d858cd67ee37b870f25 ]

The firmware of the 88W8897 PCIe+USB card sends those events very
unreliably, sometimes bluetooth together with 2.4ghz-wifi is used and no
COEX event comes in, and sometimes bluetooth is disabled but the
coexistance mode doesn't get disabled.

This means we sometimes end up capping the rx/tx window size while
bluetooth is not enabled anymore, artifically limiting wifi speeds even
though bluetooth is not being used.

Since we can't fix the firmware, let's just ignore those events on the
88W8897 device. From some Wireshark capture sessions it seems that the
Windows driver also doesn't change the rx/tx window sizes when bluetooth
gets enabled or disabled, so this is fairly consistent with the Windows
driver.

Signed-off-by: Jonas Dreßler <verdre@v0yd.nl>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20211103205827.14559-1-verdre@v0yd.nl
Signed-off-by: Sasha Levin <sashal@kernel.org>

KVM: Don't set Accessed/Dirty bits for ZERO_PAGE

[ Upstream commit a1040b0d42acf69bb4f6dbdc54c2dcd78eea1de5 ]

Don't set Accessed/Dirty bits for a struct page with PG_reserved set,
i.e. don't set A/D bits for the ZERO_PAGE.  The ZERO_PAGE (or pages
depending on the architecture) should obviously never be written, and
similarly there's no point in marking it accessed as the page will never
be swapped out or reclaimed.  The comment in page-flags.h is quite clear
that PG_reserved pages should be managed only by their owner, and
strictly following that mandate also simplifies KVM's logic.

Fixes: 7df003c85218 ("KVM: fix overflow of zero page refcount with ksm running")
Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20220429010416.2788472-4-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

mm/memremap: fix memunmap_pages() race with get_dev_pagemap()

[ Upstream commit 1e57ffb6e3fd9583268c6462c4e3853575b21701 ]

Think about the below scene:

 CPU1 CPU2
 memunmap_pages
   percpu_ref_exit
     __percpu_ref_exit
       free_percpu(percpu_count);
         /* percpu_count is freed here! */
 get_dev_pagemap
   xa_load(&pgmap_array, PHYS_PFN(phys))
     /* pgmap still in the pgmap_array */
   percpu_ref_tryget_live(&pgmap->ref)
     if __ref_is_percpu
       /* __PERCPU_REF_ATOMIC_DEAD not set yet */
       this_cpu_inc(*percpu_count)
         /* access freed percpu_count here! */
      ref->percpu_count_ptr = __PERCPU_REF_ATOMIC_DEAD;
        /* too late... */
   pageunmap_range

To fix the issue, do percpu_ref_exit() after pgmap_array is emptied. So
we won't do percpu_ref_tryget_live() against a being freed percpu_ref.

Link: https://lkml.kernel.org/r/20220609121305.2508-1-linmiaohe@huawei.com
Fixes: b7b3c01b1915 ("mm/memremap_pages: support multiple ranges per invocation")
Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

memremap: remove support for external pgmap refcounts

[ Upstream commit b80892ca022e9eb484771a66eb68e12364695a2a ]

No driver is left using the external pgmap refcount, so remove the
code to support it.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Link: https://lore.kernel.org/r/20211028151017.50234-1-hch@lst.de
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

lib/test_hmm: avoid accessing uninitialized pages

[ Upstream commit ed913b055a74b723976f8e885a3395162a0371e6 ]

If make_device_exclusive_range() fails or returns pages marked for
exclusive access less than required, remaining fields of pages will left
uninitialized.  So dmirror_atomic_map() will access those yet
uninitialized fields of pages.  To fix it, do dmirror_atomic_map() iff all
pages are marked for exclusive access (we will break if mapped is less
than required anyway) so we won't access those uninitialized fields of
pages.

Link: https://lkml.kernel.org/r/20220609130835.35110-1-linmiaohe@huawei.com
Fixes: b659baea7546 ("mm: selftests for exclusive device memory")
Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Cc: Jerome Glisse <jglisse@redhat.com>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: Ralph Campbell <rcampbell@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: mediatek: reset: Fix written reset bit offset

[ Upstream commit edabcf71d100fd433a0fc2d0c97057c446c33b2a ]

Original assert/deassert bit is BIT(0), but it's more resonable to modify
them to BIT(id % 32) which is based on id.

This patch will not influence any previous driver because the reset is
only used for thermal. The id (MT8183_INFRACFG_AO_THERM_SW_RST) is 0.

Fixes: 64ebb57a3df6 ("clk: reset: Modify reset-controller driver")
Signed-off-by: Rex-BC Chen <rex-bc.chen@mediatek.com>
Reviewed-by: Chen-Yu Tsai <wenst@chromium.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Reviewed-by: Nícolas F. R. A. Prado <nfraprado@collabora.com>
Tested-by: Nícolas F. R. A. Prado <nfraprado@collabora.com>
Link: https://lore.kernel.org/r/20220523093346.28493-3-rex-bc.chen@mediatek.com
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

iio: accel: bma400: Reordering of header files

[ Upstream commit 1bd2dc6ea863690aee5c45ebf09c9194c7a42c0d ]

Reordering of header files and removing the iio/sysfs.h since
custom attributes are not being used in the driver.

Signed-off-by: Jagath Jog J <jagathjog1996@gmail.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Link: https://lore.kernel.org/r/20220505133021.22362-3-jagathjog1996@gmail.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

platform/chrome: cros_ec: Always expose last resume result

[ Upstream commit 74bb746407bf0d7c7d126c7731dbcd66d467619b ]

The last resume result exposing logic in cros_ec_sleep_event()
incorrectly requires S0ix support, which doesn't work on ARM based
systems where S0ix doesn't exist. That's because cros_ec_sleep_event()
only reports the last resume result when the EC indicates the last sleep
event was an S0ix resume. On ARM systems, the last sleep event is always
S3 resume, but the EC can still detect sleep hang events in case some
other part of the AP is blocking sleep.

Always expose the last resume result if the EC supports it so that this
works on all devices regardless of S0ix support. This fixes sleep hang
detection on ARM based chromebooks like Trogdor.

Cc: Rajat Jain <rajatja@chromium.org>
Cc: Matthias Kaehlcke <mka@chromium.org>
Cc: Hsin-Yi Wang <hsinyi@chromium.org>
Cc: Tzung-Bi Shih <tzungbi@kernel.org>
Reviewed-by: Guenter Roeck <groeck@chromium.org>
Reviewed-by: Evan Green <evgreen@chromium.org>
Fixes: 7235560ac77a ("platform/chrome: Add support for v1 of host sleep event")
Signed-off-by: Stephen Boyd <swboyd@chromium.org>
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Link: https://lore.kernel.org/r/20220614075726.2729987-1-swboyd@chromium.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

iio: accel: bma400: Fix the scale min and max macro values

[ Upstream commit 747c7cf1592e226d40543231b26502b332d0ea2f ]

Changing the scale macro values to match the bma400 sensitivity
for 1 LSB of all the available ranges.

Fixes: 465c811f1f20 ("iio: accel: Add driver for the BMA400")
Signed-off-by: Jagath Jog J <jagathjog1996@gmail.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Link: https://lore.kernel.org/r/20220505133021.22362-2-jagathjog1996@gmail.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: qla2xxx: edif: Fix no logout on delete for N2N

[ Upstream commit ec538eb838f334453b10e7e9b260f0c358018a37 ]

The driver failed to send implicit logout on session delete. For edif, this
failed to flush any lingering SA index in FW.

Set a flag to turn on implicit logout early in the session recovery to make
sure the logout will go out in case of error.

Link: https://lore.kernel.org/r/20220608115849.16693-8-njavali@marvell.com
Fixes: 4de067e5df12 ("scsi: qla2xxx: edif: Add N2N support for EDIF")
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: qla2xxx: edif: Fix session thrash

[ Upstream commit a8fdfb0b39c2b31722c70bdf2272b949d5af4b7b ]

Current code prematurely sends out PRLI before authentication application
has given the OK to do so. This causes PRLI failure and session teardown.

Prevents PRLI from going out before authentication app gives the OK.

Link: https://lore.kernel.org/r/20220608115849.16693-7-njavali@marvell.com
Fixes: 91f6f5fbe87b ("scsi: qla2xxx: edif: Reduce connection thrash")
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: qla2xxx: edif: Tear down session if keys have been removed

[ Upstream commit d7e2e4a68fc047a025afcd200e6b7e1fbc8b1999 ]

If all keys for a session have been deleted, trigger a session teardown.

Link: https://lore.kernel.org/r/20220608115849.16693-6-njavali@marvell.com
Fixes: dd30706e73b7 ("scsi: qla2xxx: edif: Add key update")
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: qla2xxx: edif: Fix no login after app start

[ Upstream commit 24c796098f5395477f7f7ebf8e24f3f08a139f71 ]

The scenario is this: User loaded driver but has not started authentication
app. All sessions to secure device will exhaust all login attempts, fail,
and in stay in deleted state. Then some time later the app is started. The
driver will replenish the login retry count, trigger delete to prepare for
secure login. After deletion, relogin is triggered.

For the session that is already deleted, the delete trigger is a no-op. If
none of the sessions trigger a relogin, no progress is made.

Add a relogin trigger.

Link: https://lore.kernel.org/r/20220608115849.16693-5-njavali@marvell.com
Fixes: 7ebb336e45ef ("scsi: qla2xxx: edif: Add start + stop bsgs")
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: qla2xxx: edif: Reduce disruption due to multiple app start

[ Upstream commit 0dbfce5255fe8d069a1a3b712a25b263264cfa58 ]

Multiple app start can trigger a session bounce. Make driver skip over
session teardown if app start is seen more than once.

Link: https://lore.kernel.org/r/20220608115849.16693-4-njavali@marvell.com
Fixes: 7ebb336e45ef ("scsi: qla2xxx: edif: Add start + stop bsgs")
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: qla2xxx: edif: Send LOGO for unexpected IKE message

[ Upstream commit 2b659ed67a12f39f56d8dcad9b5d5a74d67c01b3 ]

If the session is down and the local port continues to receive AUTH ELS
messages, the driver needs to send back LOGO so that the remote device
knows to tear down its session. Terminate and clean up the AUTH ELS
exchange followed by a passthrough LOGO.

Link: https://lore.kernel.org/r/20220608115849.16693-3-njavali@marvell.com
Fixes: 225479296c4f ("scsi: qla2xxx: edif: Reject AUTH ELS on session down")
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

netfilter: xtables: Bring SPDX identifier back

[ Upstream commit 20646f5b1e798bcc20044ae90ac3702f177bf254 ]

Commit e2be04c7f995 ("License cleanup: add SPDX license identifier to
uapi header files with a license") added the correct SPDX identifier to
include/uapi/linux/netfilter/xt_IDLETIMER.h.

A subsequent commit removed it for no reason and reintroduced the UAPI
license incorrectness as the file is now missing the UAPI exception
again.

Add it back and remove the GPLv2 boilerplate while at it.

Fixes: 68983a354a65 ("netfilter: xtables: Add snapshot of hardidletimer target")
Cc: Manoj Basapathi <manojbm@codeaurora.org>
Cc: Subash Abhinov Kasiviswanathan <subashab@codeaurora.org>
Cc: Pablo Neira Ayuso <pablo@netfilter.org>
Cc: netfilter-devel@vger.kernel.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

usb: xhci: tegra: Fix error check

[ Upstream commit 18fc7c435be3f17ea26a21b2e2312fcb9088e01f ]

In the function tegra_xusb_powerdomain_init(),
dev_pm_domain_attach_by_name() may return NULL in some cases,
so IS_ERR() doesn't meet the requirements. Thus fix it.

Fixes: 6494a9ad86de ("usb: xhci: tegra: Add genpd support")
Signed-off-by: Tang Bin <tangbin@cmss.chinamobile.com>
Link: https://lore.kernel.org/r/20220524121404.18376-1-tangbin@cmss.chinamobile.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

usb: gadget: tegra-xudc: Fix error check in tegra_xudc_powerdomain_init()

[ Upstream commit f08aa7c80dac27ee00fa6827f447597d2fba5465 ]

dev_pm_domain_attach_by_name() may return NULL in some cases,
so IS_ERR() doesn't meet the requirements. Thus fix it.

Fixes: 49db427232fe ("usb: gadget: Add UDC driver for tegra XUSB device mode controller")
Signed-off-by: Tang Bin <tangbin@cmss.chinamobile.com>
Link: https://lore.kernel.org/r/20220525135332.23144-1-tangbin@cmss.chinamobile.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

usb: ohci-nxp: Fix refcount leak in ohci_hcd_nxp_probe

[ Upstream commit 302970b4cad3ebfda2c05ce06c322ccdc447d17e ]

of_parse_phandle() returns a node pointer with refcount
incremented, we should use of_node_put() on it when not need anymore.
Add missing of_node_put() to avoid refcount leak.

Fixes: 73108aa90cbf ("USB: ohci-nxp: Use isp1301 driver")
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Link: https://lore.kernel.org/r/20220603141231.979-1-linmq006@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

usb: host: Fix refcount leak in ehci_hcd_ppc_of_probe

[ Upstream commit b5c5b13cb45e2c88181308186b0001992cb41954 ]

of_find_compatible_node() returns a node pointer with refcount
incremented, we should use of_node_put() on it when done.
Add missing of_node_put() to avoid refcount leak.

Fixes: 796bcae7361c ("USB: powerpc: Workaround for the PPC440EPX USBH_23 errata [take 3]")
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Link: https://lore.kernel.org/r/20220602110849.58549-1-linmq006@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

fpga: altera-pr-ip: fix unsigned comparison with less than zero

[ Upstream commit 2df84a757d87fd62869fc401119d429735377ec5 ]

Fix the "comparison with less than zero" warning reported by
cppcheck for the unsigned (size_t) parameter count of the
alt_pr_fpga_write() function.

Fixes: d201cc17a8a3 ("fpga pr ip: Core driver support for Altera Partial Reconfiguration IP")
Reviewed-by: Tom Rix <trix@redhat.com>
Acked-by: Xu Yilun <yilun.xu@intel.com>
Signed-off-by: Marco Pagani <marpagan@redhat.com>
Link: https://lore.kernel.org/r/20220609140520.42662-1-marpagan@redhat.com
Signed-off-by: Xu Yilun <yilun.xu@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

PCI: mediatek-gen3: Fix refcount leak in mtk_pcie_init_irq_domains()

[ Upstream commit bf038503d5fe90189743124233fe7aeb0984e961 ]

of_get_child_by_name() returns a node pointer with refcount incremented, so
we should use of_node_put() on it when we don't need it anymore.

Add missing of_node_put() to avoid refcount leak.

Fixes: 814cceebba9b ("PCI: mediatek-gen3: Add INTx support")
Link: https://lore.kernel.org/r/20220601041259.56185-1-linmq006@gmail.com
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Miles Chen <miles.chen@mediatek.com>
Acked-by: Jianjun Wang <jianjun.wang@mediatek.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

mtd: st_spi_fsm: Add a clk_disable_unprepare() in .probe()'s error path

[ Upstream commit 28607b426c3d050714f250d0faeb99d2e9106e90 ]

For all but one error path clk_disable_unprepare() is already there. Add
it to the one location where it's missing.

Fixes: 481815a6193b ("mtd: st_spi_fsm: Handle clk_prepare_enable/clk_disable_unprepare.")
Fixes: 69d5af8d016c ("mtd: st_spi_fsm: Obtain and use EMI clock")
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Link: https://lore.kernel.org/linux-mtd/20220607152458.232847-2-u.kleine-koenig@pengutronix.de
Signed-off-by: Sasha Levin <sashal@kernel.org>

mtd: parsers: ofpart: Fix refcount leak in bcm4908_partitions_fw_offset

[ Upstream commit e607879b0da18c451de5e91daf239cc2f2f8ff2d ]

of_find_node_by_path() returns a node pointer with refcount incremented,
we should use of_node_put() on it when not need anymore.
Add missing of_node_put() to avoid refcount leak.

Fixes: bb17230c61a6 ("mtd: parsers: ofpart: support BCM4908 fixed partitions")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Link: https://lore.kernel.org/linux-mtd/20220605070726.5979-1-linmq006@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

mtd: partitions: Fix refcount leak in parse_redboot_of

[ Upstream commit 9f7e62815cf3cbbcb1b8cb21649fb4dfdb3aa016 ]

of_get_child_by_name() returns a node pointer with refcount
incremented, we should use of_node_put() on it when not need anymore.
Add missing of_node_put() to avoid refcount leak.

Fixes: 237960880960 ("mtd: partitions: redboot: seek fis-index-block in the right node")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Link: https://lore.kernel.org/linux-mtd/20220526110652.64849-1-linmq006@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

mtd: sm_ftl: Fix deadlock caused by cancel_work_sync in sm_release

[ Upstream commit a61528d997619a518ee8c51cf0ef0513021afaff ]

There is a deadlock between sm_release and sm_cache_flush_work
which is a work item. The cancel_work_sync in sm_release will
not return until sm_cache_flush_work is finished. If we hold
mutex_lock and use cancel_work_sync to wait the work item to
finish, the work item also requires mutex_lock. As a result,
the sm_release will be blocked forever. The race condition is
shown below:

    (Thread 1)             |   (Thread 2)
sm_release                 |
  mutex_lock(&ftl->mutex)  | sm_cache_flush_work
                           |   mutex_lock(&ftl->mutex)
  cancel_work_sync         |   ...

This patch moves del_timer_sync and cancel_work_sync out of
mutex_lock in order to mitigate deadlock.

Fixes: 7d17c02a01a1 ("mtd: Add new SmartMedia/xD FTL")
Signed-off-by: Duoming Zhou <duoming@zju.edu.cn>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Link: https://lore.kernel.org/linux-mtd/20220524044841.10517-1-duoming@zju.edu.cn
Signed-off-by: Sasha Levin <sashal@kernel.org>

HID: cp2112: prevent a buffer overflow in cp2112_xfer()

[ Upstream commit 381583845d19cb4bd21c8193449385f3fefa9caf ]

Smatch warnings:
drivers/hid/hid-cp2112.c:793 cp2112_xfer() error: __memcpy()
'data->block[1]' too small (33 vs 255)
drivers/hid/hid-cp2112.c:793 cp2112_xfer() error: __memcpy() 'buf' too
small (64 vs 255)

The 'read_length' variable is provided by 'data->block[0]' which comes
from user and it(read_length) can take a value between 0-255. Add an
upper bound to 'read_length' variable to prevent a buffer overflow in
memcpy().

Fixes: 542134c0375b ("HID: cp2112: Fix I2C_BLOCK_DATA transactions")
Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Sasha Levin <sashal@kernel.org>

PCI: tegra194: Fix PM error handling in tegra_pcie_config_ep()

[ Upstream commit e8fbd344a5ea62663554b8546b6bf9f88b93785a ]

pm_runtime_enable() will increase power disable depth.  If
dw_pcie_ep_init() fails, we should use pm_runtime_disable() to balance it
with pm_runtime_enable().

Add missing pm_runtime_disable() for tegra_pcie_config_ep().

Fixes: c57247f940e8 ("PCI: tegra: Add support for PCIe endpoint mode in Tegra194")
Link: https://lore.kernel.org/r/20220602031910.55859-1-linmq006@gmail.com
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Vidya Sagar <vidyas@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

PCI: microchip: Fix refcount leak in mc_pcie_init_irq_domains()

[ Upstream commit f030304fdeb87ec8f1b518c73703214aec6cc24a ]

of_get_next_child() returns a node pointer with refcount incremented, so we
should use of_node_put() on it when we don't need it anymore.

mc_pcie_init_irq_domains() only calls of_node_put() in the normal path,
missing it in some error paths.  Add missing of_node_put() to avoid
refcount leak.

Fixes: 6f15a9c9f941 ("PCI: microchip: Add Microchip PolarFire PCIe controller driver")
Link: https://lore.kernel.org/r/20220605055123.59127-1-linmq006@gmail.com
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

phy: samsung: exynosautov9-ufs: correct TSRV register configurations

[ Upstream commit f7fdc4db071f7ee7d408ea3f083222a060c76623 ]

For exynos auto v9's UFS MPHY, We should use 0x50 offset of TSRV register
configurations. So, it must be

s/PHY_TRSV_REG_CFG/PHY_TRSV_REG_CFG_AUTOV9/g

Fixes: d64519249e1d ("phy: samsung-ufs: support exynosauto ufs phy driver")
Signed-off-by: Chanho Park <chanho61.park@samsung.com>
Acked-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://lore.kernel.org/r/20220603050536.61957-1-chanho61.park@samsung.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

KVM: SVM: Stuff next_rip on emulated INT3 injection if NRIPS is supported

[ Upstream commit 3741aec4c38fa4123ab08ae552f05366d4fd05d8 ]

If NRIPS is supported in hardware but disabled in KVM, set next_rip to
the next RIP when advancing RIP as part of emulating INT3 injection.
There is no flag to tell the CPU that KVM isn't using next_rip, and so
leaving next_rip is left as is will result in the CPU pushing garbage
onto the stack when vectoring the injected event.

Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com>
Fixes: 66b7138f9136 ("KVM: SVM: Emulate nRIP feature when reinjecting INT3")
Signed-off-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Maciej S. Szmigiero <maciej.szmigiero@oracle.com>
Message-Id: <cd328309a3b88604daa2359ad56f36cb565ce2d4.1651440202.git.maciej.szmigiero@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

KVM: SVM: Unwind "speculative" RIP advancement if INTn injection "fails"

[ Upstream commit cd9e6da8048c5b40315ee2d929b6230ce1252c3c ]

Unwind the RIP advancement done by svm_queue_exception() when injecting
an INT3 ultimately "fails" due to the CPU encountering a VM-Exit while
vectoring the injected event, even if the exception reported by the CPU
isn't the same event that was injected.  If vectoring INT3 encounters an
exception, e.g. #NP, and vectoring the #NP encounters an intercepted
exception, e.g. #PF when KVM is using shadow paging, then the #NP will
be reported as the event that was in-progress.

Note, this is still imperfect, as it will get a false positive if the
INT3 is cleanly injected, no VM-Exit occurs before the IRET from the INT3
handler in the guest, the instruction following the INT3 generates an
exception (directly or indirectly), _and_ vectoring that exception
encounters an exception that is intercepted by KVM.  The false positives
could theoretically be solved by further analyzing the vectoring event,
e.g. by comparing the error code against the expected error code were an
exception to occur when vectoring the original injected exception, but
SVM without NRIPS is a complete disaster, trying to make it 100% correct
is a waste of time.

Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com>
Fixes: 66b7138f9136 ("KVM: SVM: Emulate nRIP feature when reinjecting INT3")
Signed-off-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Maciej S. Szmigiero <maciej.szmigiero@oracle.com>
Message-Id: <450133cf0a026cb9825a2ff55d02cb136a1cb111.1651440202.git.maciej.szmigiero@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: qla2xxx: edif: Fix n2n login retry for secure device

[ Upstream commit aec55325ddec975216119da000092cb8664a3399 ]

After initiator has burned up all login retries, target authentication
application begins to run. This triggers a link bounce on target side.
Initiator will attempt another login. Due to N2N, the PRLI [nvme | fcp] can
fail because of the mode mismatch with target. This patch add a few more
login retries to revive the connection.

Link: https://lore.kernel.org/r/20220607044627.19563-11-njavali@marvell.com
Fixes: 4de067e5df12 ("scsi: qla2xxx: edif: Add N2N support for EDIF")
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: qla2xxx: edif: Fix n2n discovery issue with secure target

[ Upstream commit 789d54a4178634850e441f60c0326124138e7269 ]

User failed to see disk via n2n topology. Driver used up all login retries
before authentication application started. When authentication application
started, driver did not have enough login retries to connect securely. On
app_start, driver will reset the login retry attempt count.

Link: https://lore.kernel.org/r/20220607044627.19563-10-njavali@marvell.com
Fixes: 4de067e5df12 ("scsi: qla2xxx: edif: Add N2N support for EDIF")
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: qla2xxx: edif: Add retry for ELS passthrough

[ Upstream commit 0b3f3143d473b489a7aa0779c43bcdb344bd3014 ]

Relating to EDIF, when sending IKE message, updating key or deleting key,
driver can encounter IOCB queue full. Add additional retries to reduce
higher level recovery.

Link: https://lore.kernel.org/r/20220607044627.19563-8-njavali@marvell.com
Fixes: dd30706e73b7 ("scsi: qla2xxx: edif: Add key update")
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: qla2xxx: edif: Synchronize NPIV deletion with authentication application

[ Upstream commit cf79716e6636400ae38c37bc8a652b1e522abbba ]

Notify authentication application of a NPIV deletion event is about to
occur. This allows app to perform cleanup.

Link: https://lore.kernel.org/r/20220607044627.19563-7-njavali@marvell.com
Fixes: 9efea843a906 ("scsi: qla2xxx: edif: Add detection of secure device")
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: qla2xxx: edif: Fix inconsistent check of db_flags

[ Upstream commit 36f468bfe98c7de7916ab3391ee5dd6fd2549979 ]

db_flags field is a bit field. Replace value check with bit flag check.

Link: https://lore.kernel.org/r/20211026115412.27691-12-njavali@marvell.com
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: qla2xxx: edif: Reduce connection thrash

[ Upstream commit 91f6f5fbe87ba834133fcc61d34881cb8ec9e518 ]

On ipsec start by remote port, target port may use RSCN to trigger
initiator to relogin. If driver is already in the process of a relogin,
then ignore the RSCN and allow the current relogin to continue. This
reduces thrashing of the connection.

Link: https://lore.kernel.org/r/20211026115412.27691-10-njavali@marvell.com
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: qla2xxx: edif: Fix potential stuck session in sa update

[ Upstream commit e0fb8ce2bb9e52c846e54ad2c58b5b7beb13eb09 ]

When a thread is in the process of reestablish a session, a flag is set to
prevent multiple threads/triggers from doing the same task. This flag was
left on, and any attempt to relogin was locked out. Clear this flag if the
attempt has failed.

Link: https://lore.kernel.org/r/20220607044627.19563-6-njavali@marvell.com
Fixes: dd30706e73b7 ("scsi: qla2xxx: edif: Add key update")
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: qla2xxx: edif: Reduce Initiator-Initiator thrashing

[ Upstream commit 9c40c36e75ffd49952cd4ead0672defc4b4dbdf7 ]

This patch uses GFFID switch command to scan whether remote device is
Target or Initiator mode.  Based on that info, driver will not pass up
Initiator info to authentication application. This helps reduce unnecessary
stress for authentication application to deal with unused connections.

Link: https://lore.kernel.org/r/20220607044627.19563-2-njavali@marvell.com
Fixes: 7ebb336e45ef ("scsi: qla2xxx: edif: Add start + stop bsgs")
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

of: check previous kernel's ima-kexec-buffer against memory bounds

[ Upstream commit cbf9c4b9617b6767886a913705ca14b7600c77db ]

Presently ima_get_kexec_buffer() doesn't check if the previous kernel's
ima-kexec-buffer lies outside the addressable memory range. This can result
in a kernel panic if the new kernel is booted with 'mem=X' arg and the
ima-kexec-buffer was allocated beyond that range by the previous kernel.
The panic is usually of the form below:

$ sudo kexec --initrd initrd vmlinux --append='mem=16G'

<snip>
 BUG: Unable to handle kernel data access on read at 0xc000c01fff7f0000
 Faulting instruction address: 0xc000000000837974
 Oops: Kernel access of bad area, sig: 11 [#1]
<snip>
 NIP [c000000000837974] ima_restore_measurement_list+0x94/0x6c0
 LR [c00000000083b55c] ima_load_kexec_buffer+0xac/0x160
 Call Trace:
 [c00000000371fa80] [c00000000083b55c] ima_load_kexec_buffer+0xac/0x160
 [c00000000371fb00] [c0000000020512c4] ima_init+0x80/0x108
 [c00000000371fb70] [c0000000020514dc] init_ima+0x4c/0x120
 [c00000000371fbf0] [c000000000012240] do_one_initcall+0x60/0x2c0
 [c00000000371fcc0] [c000000002004ad0] kernel_init_freeable+0x344/0x3ec
 [c00000000371fda0] [c0000000000128a4] kernel_init+0x34/0x1b0
 [c00000000371fe10] [c00000000000ce64] ret_from_kernel_thread+0x5c/0x64
 Instruction dump:
 f92100b8 f92100c0 90e10090 910100a0 4182050c 282a0017 3bc00000 40810330
 7c0802a6 fb610198 7c9b2378 f80101d0 <a1240000> 2c090001 40820614 e9240010
 ---[ end trace 0000000000000000 ]---

Fix this issue by checking returned PFN range of previous kernel's
ima-kexec-buffer with page_is_ram() to ensure correct memory bounds.

Fixes: 467d27824920 ("powerpc: ima: get the kexec buffer passed by the previous kernel")
Cc: Frank Rowand <frowand.list@gmail.com>
Cc: Prakhar Srivastava <prsriva@linux.microsoft.com>
Cc: Lakshmi Ramasubramanian <nramas@linux.microsoft.com>
Cc: Thiago Jung Bauermann <bauerman@linux.ibm.com>
Cc: Rob Herring <robh@kernel.org>
Cc: Ritesh Harjani <ritesh.list@gmail.com>
Cc: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Vaibhav Jain <vaibhav@linux.ibm.com>
Signed-off-by: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20220531041446.3334259-1-vaibhav@linux.ibm.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

mtd: rawnand: meson: Fix a potential double free issue

[ Upstream commit ec0da06337751b18f6dee06b6526e0f0d6e80369 ]

When meson_nfc_nand_chip_cleanup() is called, it will call:
meson_nfc_free_buffer(&meson_chip->nand);
nand_cleanup(&meson_chip->nand);

nand_cleanup() in turn will call nand_detach() which calls the
.detach_chip() which is here meson_nand_detach_chip().

meson_nand_detach_chip() already calls meson_nfc_free_buffer(), so we
could double free some memory.

Fix it by removing the unneeded explicit call to meson_nfc_free_buffer().

Fixes: 8fae856c5350 ("mtd: rawnand: meson: add support for Amlogic NAND flash controller")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Acked-by: Liang Yang <liang.yang@amlogic.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Link: https://lore.kernel.org/linux-mtd/ec15c358b8063f7c50ff4cd628cf0d2e14e43f49.1653064877.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Sasha Levin <sashal@kernel.org>

mtd: maps: Fix refcount leak in ap_flash_init

[ Upstream commit 77087a04c8fd554134bddcb8a9ff87b21f357926 ]

of_find_matching_node() returns a node pointer with refcount
incremented, we should use of_node_put() on it when not need anymore.
Add missing of_node_put() to avoid refcount leak.

Fixes: b0afd44bc192 ("mtd: physmap_of: add a hook for Versatile write protection")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Link: https://lore.kernel.org/linux-mtd/20220523143255.4376-1-linmq006@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

mtd: maps: Fix refcount leak in of_flash_probe_versatile

[ Upstream commit 33ec82a6d2b119938f26e5c8040ed5d92378eb54 ]

of_find_matching_node_and_match() returns a node pointer with refcount
incremented, we should use of_node_put() on it when not need anymore.
Add missing of_node_put() to avoid refcount leak.

Fixes: b0afd44bc192 ("mtd: physmap_of: add a hook for Versatile write protection")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Link: https://lore.kernel.org/linux-mtd/20220523140205.48625-1-linmq006@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: renesas: r9a06g032: Fix UART clkgrp bitsel

[ Upstream commit 2dee50ab9e72a3cae75b65e5934c8dd3e9bf01bc ]

There are two UART clock groups, each having a mux to select its
upstream clock source. The register/bit definitions for accessing these
two muxes appear to have been reversed since introduction. Correct them
so as to match the hardware manual.

Fixes: 4c3d88526eba ("clk: renesas: Renesas R9A06G032 clock driver")

Signed-off-by: Ralph Siemsen <ralph.siemsen@linaro.org>
Reviewed-by: Phil Edworthy <phil.edworthy@renesas.com>
Link: https://lore.kernel.org/r/20220518182527.1693156-1-ralph.siemsen@linaro.org
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Sasha Levin <sashal@kernel.org>

HID: amd_sfh: Don't show client init failed as error when discovery fails

[ Upstream commit e51d8d3ea3d773334d2c047c8d1623dba66f592a ]

When sensor discovery fails, this means that the system doesn't have
any sensors connected and a user should only be notified at most one time.
A message is already displayed at WARN level of "failed to discover,
sensors not enabled".  It's pointless to show that the client init failed
at ERR level for the same condition.

Check the return code and don't display this message in those conditions.

Fixes: b5d7f43e97da ("HID: amd_sfh: Add support for sensor discovery")
Reported-by: David Chang <David.Chang@amd.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Acked-by: Basavaraj Natikar <Basavaraj.Natikar@amd.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wireguard: allowedips: don't corrupt stack when detecting overflow

[ Upstream commit c31b14d86dfe7174361e8c6e5df6c2c3a4d5918c ]

In case push_rcu() and related functions are buggy, there's a
WARN_ON(len >= 128), which the selftest tries to hit by being tricky. In
case it is hit, we shouldn't corrupt the kernel's stack, though;
otherwise it may be hard to even receive the report that it's buggy. So
conditionalize the stack write based on that WARN_ON()'s return value.

Note that this never *actually* happens anyway. The WARN_ON() in the
first place is bounded by IS_ENABLED(DEBUG), and isn't expected to ever
actually hit. This is just a debugging sanity check.

Additionally, hoist the constant 128 into a named enum,
MAX_ALLOWEDIPS_BITS, so that it's clear why this value is chosen.

Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Link: https://lore.kernel.org/all/CAHk-=wjJZGA6w_DxA+k7Ejbqsq+uGK==koPai3sqdsfJqemvag@mail.gmail.com/
Fixes: e7096c131e51 ("net: WireGuard secure network tunnel")
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wireguard: ratelimiter: use hrtimer in selftest

[ Upstream commit 151c8e499f4705010780189377f85b57400ccbf5 ]

Using msleep() is problematic because it's compared against
ratelimiter.c's ktime_get_coarse_boottime_ns(), which means on systems
with slow jiffies (such as UML's forced HZ=100), the result is
inaccurate. So switch to using schedule_hrtimeout().

However, hrtimer gives us access only to the traditional posix timers,
and none of the _COARSE variants. So now, rather than being too
imprecise like jiffies, it's too precise.

One solution would be to give it a large "range" value, but this will
still fire early on a loaded system. A better solution is to align the
timeout to the actual coarse timer, and then round up to the nearest
tick, plus change.

So add the timeout to the current coarse time, and then
schedule_hrtimer() until the absolute computed time.

This should hopefully reduce flakes in CI as well. Note that we keep the
retry loop in case the entire function is running behind, because the
test could still be scheduled out, by either the kernel or by the
hypervisor's kernel, in which case restarting the test and hoping to not
be scheduled out still helps.

Fixes: e7096c131e51 ("net: WireGuard secure network tunnel")
Suggested-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>