rtc: mt6397: check return value after calling platform_get_resource()

[ Upstream commit d3b43eb505bffb8e4cdf6800c15660c001553fe6 ]

It will cause null-ptr-deref if platform_get_resource() returns NULL,
we need check the return value.

Fixes: fc2979118f3f ("rtc: mediatek: Add MT6397 RTC driver")
Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Link: https://lore.kernel.org/r/20220505125043.1594771-1-yangyingliang@huawei.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

ARM: dts: aspeed: ast2600-evb: Enable RX delay for MAC0/MAC1

[ Upstream commit 4d338ee40ba89e508c5d3e1b4af956af7cb5e12e ]

Since mac0/1 and mac2/3 are physically located on different die,
they have different properties by nature, which is mac0/1 has smaller delay step.

The property 'phy-mode' on ast2600 mac0 and mac1 is recommended to set to 'rgmii-rxid'
which enables the RX interface delay from the PHY chip.
Refer page 45 of SDK User Guide v08.00
https://github.com/AspeedTech-BMC/openbmc/releases/download/v08.00/SDK_User_Guide_v08.00.pdf

Fixes: 2ca5646b5c2f ("ARM: dts: aspeed: Add AST2600 and EVB")
Signed-off-by: Howard Chiu <howard_chiu@aspeedtech.com>
Link: https://lore.kernel.org/r/SG2PR06MB23152A548AAE81140B57DD69E6E09@SG2PR06MB2315.apcprd06.prod.outlook.com
Signed-off-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

clocksource/drivers/riscv: Events are stopped during CPU suspend

[ Upstream commit 232ccac1bd9b5bfe73895f527c08623e7fa0752d ]

Some implementations of the SBI time extension depend on hart-local
state (for example, CSRs) that are lost or hardware that is powered
down when a CPU is suspended. To be safe, the clockevents driver
cannot assume that timer IRQs will be received during CPU suspend.

Fixes: 62b019436814 ("clocksource: new RISC-V SBI timer driver")
Signed-off-by: Samuel Holland <samuel@sholland.org>
Reviewed-by: Anup Patel <anup@brainfault.org>
Link: https://lore.kernel.org/r/20220509012121.40031-1-samuel@sholland.org
Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

soc: rockchip: Fix refcount leak in rockchip_grf_init

[ Upstream commit 9b59588d8be91c96bfb0371e912ceb4f16315dbf ]

of_find_matching_node_and_match 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: 4c58063d4258 ("soc: rockchip: add driver handling grf setup")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Link: https://lore.kernel.org/r/20220516072013.19731-1-linmq006@gmail.com
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

extcon: ptn5150: Add queue work sync before driver release

[ Upstream commit 782cd939cbe0f569197cd1c9b0477ee213167f04 ]

Add device managed action to sync pending queue work, otherwise
the queued work may run after the work is destroyed.

Fixes: 4ed754de2d66 ("extcon: Add support for ptn5150 extcon driver")
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Li Jun <jun.li@nxp.com>
Signed-off-by: Chanwoo Choi <cw00.choi@samsung.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ksmbd: fix reference count leak in smb_check_perm_dacl()

[ Upstream commit d21a580dafc69aa04f46e6099616146a536b0724 ]

The issue happens in a specific path in smb_check_perm_dacl(). When
"id" and "uid" have the same value, the function simply jumps out of
the loop without decrementing the reference count of the object
"posix_acls", which is increased by get_acl() earlier. This may
result in memory leaks.

Fix it by decreasing the reference count of "posix_acls" before
jumping to label "check_access_bits".

Fixes: 777cad1604d6 ("ksmbd: remove select FS_POSIX_ACL in Kconfig")
Signed-off-by: Xin Xiong <xiongx18@fudan.edu.cn>
Signed-off-by: Xin Tan <tanxin.ctf@gmail.com>
Acked-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

coresight: cpu-debug: Replace mutex with mutex_trylock on panic notifier

[ Upstream commit 1adff542d67a2ed1120955cb219bfff8a9c53f59 ]

The panic notifier infrastructure executes registered callbacks when
a panic event happens - such callbacks are executed in atomic context,
with interrupts and preemption disabled in the running CPU and all other
CPUs disabled. That said, mutexes in such context are not a good idea.

This patch replaces a regular mutex with a mutex_trylock safer approach;
given the nature of the mutex used in the driver, it should be pretty
uncommon being unable to acquire such mutex in the panic path, hence
no functional change should be observed (and if it is, that would be
likely a deadlock with the regular mutex).

Fixes: 2227b7c74634 ("coresight: add support for CPU debug module")
Cc: Leo Yan <leo.yan@linaro.org>
Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Guilherme G. Piccoli <gpiccoli@igalia.com>
Reviewed-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Link: https://lore.kernel.org/r/20220427224924.592546-10-gpiccoli@igalia.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

soundwire: intel: prevent pm_runtime resume prior to system suspend

[ Upstream commit 6d9f2dadba698114fed97b224578c5338a36b0d9 ]

commit e38f9ff63e6d ("ACPI: scan: Do not add device IDs from _CID if _HID is not valid")
exposes a race condition on a TGL RVP device leading to a timeout.

The detailed analysis shows the RT711 codec driver scheduling a jack
detection workqueue while attaching during a spurious pm_runtime
resume, and the work function happens to be scheduled after the
manager device is suspended.

The direct link between this ACPI patch and a spurious pm_runtime
resume is not obvious; the most likely explanation is that a change in
the ACPI device linked list management modifies the order in which the
pm_runtime device status is checked and exposes a race condition that
was probably present for a very long time, but was not identified.

We already have a check in the .prepare stage, where we will resume to
full power from specific clock-stop modes. In all other cases, we
don't need to resume to full power by default. Adding the
SMART_SUSPEND flag prevents the spurious resume from happening.

BugLink: https://github.com/thesofproject/linux/issues/3459
Fixes: 029bfd1cd53cd ("soundwire: intel: conditionally exit clock stop mode on system suspend")
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Ranjani Sridharan <ranjani.sridharan@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/20220420023241.14335-2-yung-chuan.liao@linux.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

export: fix string handling of namespace in EXPORT_SYMBOL_NS

[ Upstream commit d143b9db8069f0e2a0fa34484e806a55a0dd4855 ]

Commit c3a6cf19e695 ("export: avoid code duplication in
include/linux/export.h") broke the ability for a defined string to be
used as a namespace value.  Fix this up by using stringify to properly
encode the namespace name.

Fixes: c3a6cf19e695 ("export: avoid code duplication in include/linux/export.h")
Cc: Miroslav Benes <mbenes@suse.cz>
Cc: Emil Velikov <emil.l.velikov@gmail.com>
Cc: Jessica Yu <jeyu@kernel.org>
Cc: Quentin Perret <qperret@google.com>
Cc: Matthias Maennich <maennich@google.com>
Reviewed-by: Masahiro Yamada <masahiroy@kernel.org>
Link: https://lore.kernel.org/r/20220427090442.2105905-1-gregkh@linuxfoundation.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

serial: sifive: Report actual baud base rather than fixed 115200

[ Upstream commit 0a7ff843d507ce2cca2c3b7e169ee56e28133530 ]

The base baud value reported is supposed to be the highest baud rate
that can be set for a serial port.  The SiFive FU740-C000 SOC's on-chip
UART supports baud rates of up to 1/16 of the input clock rate, which is
the bus clock `tlclk'[1], often at 130MHz in the case of the HiFive
Unmatched board.

However the sifive UART driver reports a fixed value of 115200 instead:

10010000.serial: ttySIF0 at MMIO 0x10010000 (irq = 1, base_baud = 115200) is a SiFive UART v0
10011000.serial: ttySIF1 at MMIO 0x10011000 (irq = 2, base_baud = 115200) is a SiFive UART v0

even though we already support setting higher baud rates, e.g.:

$ tty
/dev/ttySIF1
$ stty speed
230400

The baud base value is computed by the serial core by dividing the UART
clock recorded in `struct uart_port' by 16, which is also the minimum
value of the clock divider supported, so correct the baud base value
reported by setting the UART clock recorded to the input clock rate
rather than 115200:

10010000.serial: ttySIF0 at MMIO 0x10010000 (irq = 1, base_baud = 8125000) is a SiFive UART v0
10011000.serial: ttySIF1 at MMIO 0x10011000 (irq = 2, base_baud = 8125000) is a SiFive UART v0

References:

[1] "SiFive FU740-C000 Manual", v1p3, SiFive, Inc., August 13, 2021,
    Section 16.9 "Baud Rate Divisor Register (div)", pp.143-144

Signed-off-by: Maciej W. Rozycki <macro@orcam.me.uk>
Fixes: 1f1496a923b6 ("riscv: Fix sifive serial driver")
Link: https://lore.kernel.org/r/alpine.DEB.2.21.2204291656280.9383@angie.orcam.me.uk
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

power: supply: axp288_fuel_gauge: Drop BIOS version check from "T3 MRD" DMI quirk

[ Upstream commit f61509a6f0b70f5bedea34efaf8065621689bd7a ]

Some "T3 MRD" mini-PCs / HDMI-sticks without a battery use a different
value then "5.11" for their DMI BIOS version field.

Drop the BIOS version check so that the no-battery "T3 MRD" DMI quirk
applies to these too.

Fixes: 3a06b912a5ce ("power: supply: axp288_fuel_gauge: Make "T3 MRD" no_battery_list DMI entry more generic")
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Sebastian Reichel <sebastian.reichel@collabora.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

phy: qcom-qmp: fix pipe-clock imbalance on power-on failure

[ Upstream commit 5e73b2d9867998278479ccc065a8a8227a5513ef ]

Make sure to disable the pipe clock also if ufs-reset deassertion fails
during power on.

Note that the ufs-reset is asserted in qcom_qmp_phy_com_exit().

Fixes: c9b589791fc1 ("phy: qcom: Utilize UFS reset controller")
Cc: Evan Green <evgreen@chromium.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20220502133130.4125-2-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

misc/pvpanic: Convert regular spinlock into trylock on panic path

[ Upstream commit e918c10265ef2bc82ce8a6fed6d8123d09ec1db3 ]

The pvpanic driver relies on panic notifiers to execute a callback
on panic event. Such function is executed in atomic context - the
panic function disables local IRQs, preemption and all other CPUs
that aren't running the panic code.

With that said, it's dangerous to use regular spinlocks in such path,
as introduced by commit b3c0f8774668 ("misc/pvpanic: probe multiple instances").
This patch fixes that by replacing regular spinlocks with the trylock
safer approach.

It also fixes an old comment (about a long gone framebuffer code) and
the notifier priority - we should execute hypervisor notifiers early,
deferring this way the panic action to the hypervisor, as expected by
the users that are setting up pvpanic.

Fixes: b3c0f8774668 ("misc/pvpanic: probe multiple instances")
Cc: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Cc: Mihai Carabas <mihai.carabas@oracle.com>
Cc: Shile Zhang <shile.zhang@linux.alibaba.com>
Cc: Wang ShaoBo <bobo.shaobowang@huawei.com>
Cc: zhenwei pi <pizhenwei@bytedance.com>
Signed-off-by: Guilherme G. Piccoli <gpiccoli@igalia.com>
Link: https://lore.kernel.org/r/20220427224924.592546-6-gpiccoli@igalia.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

pvpanic: Fix typos in the comments

[ Upstream commit cc5b392d0f94f27743583140d819fa35a46899db ]

Fix a few spelling typos in the comments.

Reviewed-by: Mihai Carabas <mihai.carabas@oracle.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Link: https://lore.kernel.org/r/20210829124354.81653-2-andriy.shevchenko@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

rpmsg: qcom_smd: Fix returning 0 if irq_of_parse_and_map() fails

[ Upstream commit 59d6f72f6f9c92fec8757d9e29527da828e9281f ]

irq_of_parse_and_map() returns 0 on failure, so this should not be
passed further as error return code.

Fixes: 1a358d350664 ("rpmsg: qcom_smd: Fix irq_of_parse_and_map() return value")
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20220423093932.32136-1-krzysztof.kozlowski@linaro.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

iio: adc: sc27xx: Fine tune the scale calibration values

[ Upstream commit 5a7a184b11c6910f47600ff5cbbee34168f701a8 ]

Small adjustment the scale calibration value for the sc2731,
use new name sc2731_[big|small]_scale_graph_calib, and remove
the origin [big|small]_scale_graph_calib struct for unused.

Fixes: 8ba0dbfd07a35 (iio: adc: sc27xx: Add ADC scale calibration)
Signed-off-by: Cixi Geng <cixi.geng1@unisoc.com>
Link: https://lore.kernel.org/r/20220419142458.884933-4-gengcixi@gmail.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

iio: adc: sc27xx: fix read big scale voltage not right

[ Upstream commit ad930a75613282400179361e220e58b87386b8c7 ]

Fix wrong configuration value of SC27XX_ADC_SCALE_MASK and
SC27XX_ADC_SCALE_SHIFT by spec documetation.

Fixes: 5df362a6cf49c (iio: adc: Add Spreadtrum SC27XX PMICs ADC support)
Signed-off-by: Cixi Geng <cixi.geng1@unisoc.com>
Reviewed-by: Baolin Wang <baolin.wang7@gmail.com>
Link: https://lore.kernel.org/r/20220419142458.884933-3-gengcixi@gmail.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

iio: proximity: vl53l0x: Fix return value check of wait_for_completion_timeout

[ Upstream commit 50f2959113cb6756ffd73c4fedc712cf2661f711 ]

wait_for_completion_timeout() returns unsigned long not int.
It returns 0 if timed out, and positive if completed.
The check for <= 0 is ambiguous and should be == 0 here
indicating timeout which is the only error case.

Fixes: 3cef2e31b54b ("iio: proximity: vl53l0x: Add IRQ support")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Link: https://lore.kernel.org/r/20220412064210.10734-1-linmq006@gmail.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

iio: adc: stmpe-adc: Fix wait_for_completion_timeout return value check

[ Upstream commit d345b23200bcdbd2bd3582213d738c258b77718f ]

wait_for_completion_timeout() returns unsigned long not long.
it returns 0 if timed out, and positive if completed.
The check for <= 0 is ambiguous and should be == 0 here
indicating timeout which is the only error case

Fixes: e813dde6f833 ("iio: stmpe-adc: Use wait_for_completion_timeout")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Reviewed-by: Philippe Schenker <philippe.schenker@toradex.com>
Link: https://lore.kernel.org/r/20220412065150.14486-1-linmq006@gmail.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

rpmsg: virtio: Fix the unregistration of the device rpmsg_ctrl

[ Upstream commit df191796985922488e4e6b64f7bd79c3934412f2 ]

Unregister the rpmsg_ctrl device instead of just freeing the
the virtio_rpmsg_channel structure.
This will properly unregister the device and call
virtio_rpmsg_release_device() that frees the structure.

Fixes: c486682ae1e2 ("rpmsg: virtio: Register the rpmsg_char device")
Signed-off-by: Arnaud Pouliquen <arnaud.pouliquen@foss.st.com>
Reviewed-by: Hangyu Hua <hbh25y@gmail.com>
Link: https://lore.kernel.org/r/20220426060536.15594-4-hbh25y@gmail.com
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

rpmsg: virtio: Fix possible double free in rpmsg_virtio_add_ctrl_dev()

[ Upstream commit 1680939e9ecf7764fba8689cfb3429c2fe2bb23c ]

vch will be free in virtio_rpmsg_release_device() when
rpmsg_ctrldev_register_device() fails. There is no need to call
kfree() again.

Fixes: c486682ae1e2 ("rpmsg: virtio: Register the rpmsg_char device")
Signed-off-by: Hangyu Hua <hbh25y@gmail.com>
Tested-by: Arnaud Pouliquen <arnaud.pouliquen@foss.st.com>
Link: https://lore.kernel.org/r/20220426060536.15594-3-hbh25y@gmail.com
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

rpmsg: virtio: Fix possible double free in rpmsg_probe()

[ Upstream commit c2eecefec5df1306eafce28ccdf1ca159a552ecc ]

vch will be free in virtio_rpmsg_release_device() when
rpmsg_ns_register_device() fails. There is no need to call kfree() again.

Fix this by changing error path from free_vch to free_ctrldev.

Fixes: c486682ae1e2 ("rpmsg: virtio: Register the rpmsg_char device")
Signed-off-by: Hangyu Hua <hbh25y@gmail.com>
Tested-by: Arnaud Pouliquen <arnaud.pouliquen@foss.st.com>
Link: https://lore.kernel.org/r/20220426060536.15594-2-hbh25y@gmail.com
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

usb: typec: mux: Check dev_set_name() return value

[ Upstream commit b9fa0292490db39d6542f514117333d366ec0011 ]

It's possible that dev_set_name() returns -ENOMEM, catch and handle this.

Fixes: 3370db35193b ("usb: typec: Registering real device entries for the muxes")
Reported-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Acked-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20220422222351.1297276-4-bjorn.andersson@linaro.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

firmware: stratix10-svc: fix a missing check on list iterator

[ Upstream commit 5a0793ac66ac0e254d292f129a4d6c526f9f2aff ]

The bug is here:
pmem->vaddr = NULL;

The list iterator 'pmem' will point to a bogus position containing
HEAD if the list is empty or no element is found. This case must
be checked before any use of the iterator, otherwise it will
lead to a invalid memory access.

To fix this bug, just gen_pool_free/set NULL/list_del() and return
when found, otherwise list_del HEAD and return;

Fixes: 7ca5ce896524f ("firmware: add Intel Stratix10 service layer driver")
Signed-off-by: Xiaomeng Tong <xiam0nd.tong@gmail.com>
Link: https://lore.kernel.org/r/20220414035609.2239-1-xiam0nd.tong@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

misc: fastrpc: fix an incorrect NULL check on list iterator

[ Upstream commit 5ac11fe03a0a83042d1a040dbce4fa2fb5521e23 ]

The bug is here:
if (!buf) {

The list iterator value 'buf' will *always* be set and non-NULL
by list_for_each_entry(), so it is incorrect to assume that the
iterator value will be NULL if the list is empty (in this case, the
check 'if (!buf) {' will always be false and never exit expectly).

To fix the bug, use a new variable 'iter' as the list iterator,
while use the original variable 'buf' as a dedicated pointer to
point to the found element.

Fixes: 2419e55e532de ("misc: fastrpc: add mmap/unmap support")
Signed-off-by: Xiaomeng Tong <xiam0nd.tong@gmail.com>
Link: https://lore.kernel.org/r/20220327062202.5720-1-xiam0nd.tong@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

usb: dwc3: pci: Fix pm_runtime_get_sync() error checking

[ Upstream commit a03e2ddab8e735e2cc315609b297b300e9cc60d2 ]

If the device is already in a runtime PM enabled state
pm_runtime_get_sync() will return 1, so a test for negative
value should be used to check for errors.

Fixes: 8eed00b237a28 ("usb: dwc3: pci: Runtime resume child device from wq")
Signed-off-by: Zheng Yongjun <zhengyongjun3@huawei.com>
Link: https://lore.kernel.org/r/20220422062652.10575-1-zhengyongjun3@huawei.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

usb: dwc3: gadget: Replace list_for_each_entry_safe() if using giveback

[ Upstream commit bf594d1d0c1d7b895954018043536ffd327844f9 ]

The list_for_each_entry_safe() macro saves the current item (n) and
the item after (n+1), so that n can be safely removed without
corrupting the list.  However, when traversing the list and removing
items using gadget giveback, the DWC3 lock is briefly released,
allowing other routines to execute.  There is a situation where, while
items are being removed from the cancelled_list using
dwc3_gadget_ep_cleanup_cancelled_requests(), the pullup disable
routine is running in parallel (due to UDC unbind).  As the cleanup
routine removes n, and the pullup disable removes n+1, once the
cleanup retakes the DWC3 lock, it references a request who was already
removed/handled.  With list debug enabled, this leads to a panic.
Ensure all instances of the macro are replaced where gadget giveback
is used.

Example call stack:

Thread#1:
__dwc3_gadget_ep_set_halt() - CLEAR HALT
  -> dwc3_gadget_ep_cleanup_cancelled_requests()
    ->list_for_each_entry_safe()
    ->dwc3_gadget_giveback(n)
      ->dwc3_gadget_del_and_unmap_request()- n deleted[cancelled_list]
      ->spin_unlock
      ->Thread#2 executes
      ...
    ->dwc3_gadget_giveback(n+1)
      ->Already removed!

Thread#2:
dwc3_gadget_pullup()
  ->waiting for dwc3 spin_lock
  ...
  ->Thread#1 released lock
  ->dwc3_stop_active_transfers()
    ->dwc3_remove_requests()
      ->fetches n+1 item from cancelled_list (n removed by Thread#1)
      ->dwc3_gadget_giveback()
        ->dwc3_gadget_del_and_unmap_request()- n+1 deleted[cancelled_list]
        ->spin_unlock

Fixes: d4f1afe5e896 ("usb: dwc3: gadget: move requests to cancelled_list")
Signed-off-by: Wesley Cheng <quic_wcheng@quicinc.com>
Link: https://lore.kernel.org/r/20220414183521.23451-1-quic_wcheng@quicinc.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

rpmsg: qcom_smd: Fix irq_of_parse_and_map() return value

[ Upstream commit 1a358d35066487d228a68303d808bc4721c6b1b9 ]

The irq_of_parse_and_map() returns 0 on failure, not a negative ERRNO.

Fixes: 53e2822e56c7 ("rpmsg: Introduce Qualcomm SMD backend")
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20220422105326.78713-1-krzysztof.kozlowski@linaro.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

pwm: raspberrypi-poe: Fix endianness in firmware struct

[ Upstream commit 09f688f0718f57f9cf68ee1aa94490f641e759ba ]

The reg member of struct raspberrypi_pwm_prop is a little endian 32 bit
quantity. Explicitly convert the (native endian) value to little endian
on assignment as is already done in raspberrypi_pwm_set_property().

This fixes the following sparse warning:

drivers/pwm/pwm-raspberrypi-poe.c:69:24: warning: incorrect type in initializer (different base types)
drivers/pwm/pwm-raspberrypi-poe.c:69:24:    expected restricted __le32 [usertype] reg
drivers/pwm/pwm-raspberrypi-poe.c:69:24:    got unsigned int [usertype] reg

Fixes: 79caa362eab6 ("pwm: Add Raspberry Pi Firmware based PWM bus")
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

pwm: lp3943: Fix duty calculation in case period was clamped

[ Upstream commit 5e3b07ca5cc78cd4a987e78446849e41288d87cb ]

The hardware only supports periods <= 1.6 ms and if a bigger period is
requested it is clamped to 1.6 ms. In this case duty_cycle might be bigger
than 1.6 ms and then the duty cycle register is written with a value
bigger than LP3943_MAX_DUTY. So clamp duty_cycle accordingly.

Fixes: af66b3c0934e ("pwm: Add LP3943 PWM driver")
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

staging: fieldbus: Fix the error handling path in anybuss_host_common_probe()

[ Upstream commit 7079b3483a17be2cfba64cbd4feb1b7ae07f1ea7 ]

If device_register() fails, device_unregister() should not be called
because it will free some resources that are not allocated.
put_device() should be used instead.

Fixes: 308ee87a2f1e ("staging: fieldbus: anybus-s: support HMS Anybus-S bus")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Link: https://lore.kernel.org/r/5401a519608d6e1a4e7435c20f4f20b0c5c36c23.1650610082.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

usb: musb: Fix missing of_node_put() in omap2430_probe

[ Upstream commit 424bef51fa530389b0b9008c9e144e40c10e8458 ]

The device_node pointer is returned by of_parse_phandle() with refcount
incremented. We should use of_node_put() on it when done.

Fixes: 8934d3e4d0e7 ("usb: musb: omap2430: Don't use omap_get_control_dev()")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Link: https://lore.kernel.org/r/20220309111033.24487-1-linmq006@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

USB: storage: karma: fix rio_karma_init return

[ Upstream commit b92ffb1eddd9a66a90defc556dcbf65a43c196c7 ]

The function rio_karam_init() should return -ENOMEM instead of
value 0 (USB_STOR_TRANSPORT_GOOD) when allocation fails.

Similarly, it should return -EIO when rio_karma_send_command() fails.

Fixes: dfe0d3ba20e8 ("USB Storage: add rio karma eject support")
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Lin Ma <linma@zju.edu.cn>
Link: https://lore.kernel.org/r/20220412144359.28447-1-linma@zju.edu.cn
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

usb: usbip: add missing device lock on tweak configuration cmd

[ Upstream commit d088fabace2ca337b275d1d4b36db4fe7771e44f ]

The function documentation of usb_set_configuration says that its
callers should hold the device lock. This lock is held for all
callsites except tweak_set_configuration_cmd. The code path can be
executed for example when attaching a remote USB device.
The solution is to surround the call by the device lock.

This bug was found using my experimental own-developed static analysis
tool, which reported the missing lock on v5.17.2. I manually verified
this bug report by doing code review as well. I runtime checked that
the required lock is not held. I compiled and runtime tested this on
x86_64 with a USB mouse. After applying this patch, my analyser no
longer reports this potential bug.

Fixes: 2c8c98158946 ("staging: usbip: let client choose device configuration")
Reviewed-by: Shuah Khan <skhan@linuxfoundation.org>
Signed-off-by: Niels Dossche <dossche.niels@gmail.com>
Link: https://lore.kernel.org/r/20220412165055.257113-1-dossche.niels@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

usb: usbip: fix a refcount leak in stub_probe()

[ Upstream commit 9ec4cbf1cc55d126759051acfe328d489c5d6e60 ]

usb_get_dev() is called in stub_device_alloc(). When stub_probe() fails
after that, usb_put_dev() needs to be called to release the reference.

Fix this by moving usb_put_dev() to sdev_free error path handling.

Find this by code review.

Fixes: 3ff67445750a ("usbip: fix error handling in stub_probe()")
Reviewed-by: Shuah Khan <skhan@linuxfoundation.org>
Signed-off-by: Hangyu Hua <hbh25y@gmail.com>
Link: https://lore.kernel.org/r/20220412020257.9767-1-hbh25y@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

remoteproc: imx_rproc: Ignore create mem entry for resource table

[ Upstream commit 58b7c856519fe946620ee68dd0c37bd3c695484a ]

Resource table is used by Linux to get information published by
remote processor. It should be not be used for memory allocation, so
not create rproc mem entry.

Fixes: b29b4249f8f0 ("remoteproc: imx_rproc: add i.MX specific parse fw hook")
Signed-off-by: Peng Fan <peng.fan@nxp.com>
Link: https://lore.kernel.org/r/20220415025737.1561976-1-peng.fan@oss.nxp.com
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

tty: serial: fsl_lpuart: fix potential bug when using both of_alias_get_id and ida_simple_get

[ Upstream commit f398e0aa325c61fa20903833a5b534ecb8e6e418 ]

Now fsl_lpuart driver use both of_alias_get_id() and ida_simple_get() in
.probe(), which has the potential bug. For example, when remove the
lpuart7 alias in dts, of_alias_get_id() will return error, then call
ida_simple_get() to allocate the id 0 for lpuart7, this may confilct
with the lpuart4 which has alias 0.

    aliases {
...
        serial0 = &lpuart4;
        serial1 = &lpuart5;
        serial2 = &lpuart6;
        serial3 = &lpuart7;
    }

So remove the ida_simple_get() in .probe(), return an error directly
when calling of_alias_get_id() fails, which is consistent with other
uart drivers behavior.

Fixes: 3bc3206e1c0f ("serial: fsl_lpuart: Remove the alias node dependence")
Signed-off-by: Sherry Sun <sherry.sun@nxp.com>
Link: https://lore.kernel.org/r/20220321112211.8895-1-sherry.sun@nxp.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

serial: 8250_aspeed_vuart: Fix potential NULL dereference in aspeed_vuart_probe

[ Upstream commit 0e0fd55719fa081de6f9e5d9e6cef48efb04d34a ]

platform_get_resource() may fail and return NULL, so we should
better check it's return value to avoid a NULL pointer dereference.

Fixes: 54da3e381c2b ("serial: 8250_aspeed_vuart: use UPF_IOREMAP to set up register mapping")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Link: https://lore.kernel.org/r/20220404143842.16960-1-linmq006@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

tty: n_tty: Restore EOF push handling behavior

[ Upstream commit 65a8b287023da68c4550deab5c764e6891cf1caf ]

TTYs in ICANON mode have a special case that allows "pushing" a line
without a regular EOL character (like newline), by using EOF (the EOT
character - ASCII 0x4) as a pseudo-EOL. It is silently discarded, so
the reader of the PTS will receive the line *without* EOF or any other
terminating character.

This special case has an edge case: What happens if the readers buffer
is the same size as the line (without EOF)? Will they be able to tell
if the whole line is received, i.e. if the next read() will return more
of the same line or the next line?

There are two possibilities,  that both have (dis)advantages:

1. The next read() returns 0. FreeBSD (13.0) and OSX (10.11) do this.
   Advantage: The reader can interpret this as "the line is over".
   Disadvantage: read() returning 0 means EOF, the reader could also
   interpret it as "there's no more data" and stop reading or even
   close the PT.

2. The next read() returns the next line, the EOF is silently discarded.
   Solaris (or at least OpenIndiana 2021.10) does this, Linux has done
   do this since commit 40d5e0905a03 ("n_tty: Fix EOF push handling");
   this behavior was recently broken by commit 359303076163 ("tty:
   n_tty: do not look ahead for EOL character past the end of the buffer").
   Advantage: read() won't return 0 (EOF), reader less likely to be
   confused (and things like `while(read(..)>0)` don't break)
   Disadvantage: The reader can't really know if the read() continues
   the last line (that filled the whole read buffer) or starts a
   new line.

As both options are defensible (and are used by other Unix-likes), it's
best to stick to the "old" behavior since "n_tty: Fix EOF push handling"
of 2013, i.e. silently discard that EOF.

This patch - that I actually got from Linus for testing and only
modified slightly - restores that behavior by skipping an EOF
character if it's the next character after reading is done.

Based on a patch from Linus Torvalds.

Link: https://bugzilla.kernel.org/show_bug.cgi?id=215611
Fixes: 359303076163 ("tty: n_tty: do not look ahead for EOL character past the end of the buffer")
Cc: Peter Hurley <peter@hurleysoftware.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Jiri Slaby <jirislaby@kernel.org>
Reviewed-and-tested-by: Daniel Gibson <daniel@gibson.sh>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Daniel Gibson <daniel@gibson.sh>
Link: https://lore.kernel.org/r/20220329235810.452513-2-daniel@gibson.sh
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

tty: serial: owl: Fix missing clk_disable_unprepare() in owl_uart_probe

[ Upstream commit bcea0f547ec1a2ee44d429aaf0334633e386e67c ]

Fix the missing clk_disable_unprepare() before return
from owl_uart_probe() in the error handling case.

Fixes: abf42d2f333b ("tty: serial: owl: add "much needed" clk_prepare_enable()")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Link: https://lore.kernel.org/r/20220307105135.11698-1-linmq006@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

tty: goldfish: Use tty_port_destroy() to destroy port

[ Upstream commit 507b05063d1b7a1fcb9f7d7c47586fc4f3508f98 ]

In goldfish_tty_probe(), the port initialized through tty_port_init()
should be destroyed in error paths.In goldfish_tty_remove(), qtty->port
also should be destroyed or else might leak resources.

Fix the above by calling tty_port_destroy().

Fixes: 666b7793d4bf ("goldfish: tty driver")
Reviewed-by: Jiri Slaby <jirislaby@kernel.org>
Signed-off-by: Wang Weiyang <wangweiyang2@huawei.com>
Link: https://lore.kernel.org/r/20220328115844.86032-1-wangweiyang2@huawei.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

lkdtm/bugs: Don't expect thread termination without CONFIG_UBSAN_TRAP

[ Upstream commit 8bfdbddd68249e0d8598777cca8249619ee51df0 ]

When you don't select CONFIG_UBSAN_TRAP, you get:

  # echo ARRAY_BOUNDS > /sys/kernel/debug/provoke-crash/DIRECT
[  102.265827] ================================================================================
[  102.278433] UBSAN: array-index-out-of-bounds in drivers/misc/lkdtm/bugs.c:342:16
[  102.287207] index 8 is out of range for type 'char [8]'
[  102.298722] ================================================================================
[  102.313712] lkdtm: FAIL: survived array bounds overflow!
[  102.318770] lkdtm: Unexpected! This kernel (5.16.0-rc1-s3k-dev-01884-g720dcf79314a ppc) was built with CONFIG_UBSAN_BOUNDS=y

It is not correct because when CONFIG_UBSAN_TRAP is not selected
you can't expect array bounds overflow to kill the thread.

Modify the logic so that when the kernel is built with
CONFIG_UBSAN_BOUNDS but without CONFIG_UBSAN_TRAP, you get a warning
about CONFIG_UBSAN_TRAP not been selected instead.

This also require a fix of pr_expected_config(), otherwise the
following error is encountered.

  CC      drivers/misc/lkdtm/bugs.o
drivers/misc/lkdtm/bugs.c: In function 'lkdtm_ARRAY_BOUNDS':
drivers/misc/lkdtm/bugs.c:351:2: error: 'else' without a previous 'if'
  351 |  else
      |  ^~~~

Fixes: c75be56e35b2 ("lkdtm/bugs: Add ARRAY_BOUNDS to selftests")
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Kees Cook <keescook@chromium.org>
Link: https://lore.kernel.org/r/363b58690e907c677252467a94fe49444c80ea76.1649704381.git.christophe.leroy@csgroup.eu
Signed-off-by: Sasha Levin <sashal@kernel.org>

lkdtm/bugs: Check for the NULL pointer after calling kmalloc

[ Upstream commit 4a9800c81d2f34afb66b4b42e0330ae8298019a2 ]

As the possible failure of the kmalloc(), the not_checked and checked
could be NULL pointer.
Therefore, it should be better to check it in order to avoid the
dereference of the NULL pointer.
Also, we need to kfree the 'not_checked' and 'checked' to avoid
the memory leak if fails.
And since it is just a test, it may directly return without error
number.

Fixes: ae2e1aad3e48 ("drivers/misc/lkdtm/bugs.c: add arithmetic overflow and array bounds checks")
Signed-off-by: Jiasheng Jiang <jiasheng@iscas.ac.cn>
Acked-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Link: https://lore.kernel.org/r/20220120092936.1874264-1-jiasheng@iscas.ac.cn
Signed-off-by: Sasha Levin <sashal@kernel.org>

iio: adc: ad7124: Remove shift from scan_type

[ Upstream commit fe78ccf79b0e29fd6d8dc2e2c3b0dbeda4ce3ad8 ]

The 24 bits data is stored in 32 bits in BE. There
is no need to shift it. This confuses user-space apps.

Fixes: b3af341bbd966 ("iio: adc: Add ad7124 support")
Signed-off-by: Alexandru Tachici <alexandru.tachici@analog.com>
Link: https://lore.kernel.org/r/20220322105029.86389-2-alexandru.tachici@analog.com
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

staging: greybus: codecs: fix type confusion of list iterator variable

[ Upstream commit 84ef256550196bc06e6849a34224c998b45bd557 ]

If the list does not exit early then data == NULL and 'module' does not
point to a valid list element.
Using 'module' in such a case is not valid and was therefore removed.

Fixes: 6dd67645f22c ("greybus: audio: Use single codec driver registration")
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Vaibhav Agarwal <vaibhav.sr@gmail.com>
Reviewed-by: Mark Greer <mgreer@animalcreek.com>
Signed-off-by: Jakob Koschel <jakobkoschel@gmail.com>
Link: https://lore.kernel.org/r/20220321123626.3068639-1-jakobkoschel@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

pcmcia: db1xxx_ss: restrict to MIPS_DB1XXX boards

[ Upstream commit 3928cf08334ed895a31458cbebd8d4ec6d84c080 ]

When the MIPS_ALCHEMY board selection is MIPS_XXS1500 instead of
MIPS_DB1XXX, the PCMCIA driver 'db1xxx_ss' has build errors due
to missing DB1XXX symbols. The PCMCIA driver should be restricted
to MIPS_DB1XXX instead of MIPS_ALCHEMY to fix this build error.

ERROR: modpost: "bcsr_read" [drivers/pcmcia/db1xxx_ss.ko] undefined!
ERROR: modpost: "bcsr_mod" [drivers/pcmcia/db1xxx_ss.ko] undefined!

Fixes: 42a4f17dc356 ("MIPS: Alchemy: remove SOC_AU1X00 in favor of MIPS_ALCHEMY")
Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Reported-by: kernel test robot <lkp@intel.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Kees Cook <keescook@chromium.org>
Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Cc: linux-mips@vger.kernel.org
Acked-by: Manuel Lauss <manuel.lauss@gmail.com>
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

Linux 5.15.46

Link: https://lore.kernel.org/r/20220607164934.766888869@linuxfoundation.org
Tested-by: Fox Chen <foxhlchen@gmail.com>
Tested-by: Shuah Khan <skhan@linuxfoundation.org>
Tested-by: Sudip Mukherjee <sudip.mukherjee@codethink.co.uk>
Tested-by: Bagas Sanjaya <bagasdotme@gmail.com>
Tested-by: Linux Kernel Functional Testing <lkft@linaro.org>
Tested-by: Guenter Roeck <linux@roeck-us.net>
Tested-by: Ron Economos <re@w6rz.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

block: fix bio_clone_blkg_association() to associate with proper blkcg_gq

commit 22b106e5355d6e7a9c3b5cb5ed4ef22ae585ea94 upstream.

Commit d92c370a16cb ("block: really clone the block cgroup in
bio_clone_blkg_association") changed bio_clone_blkg_association() to
just clone bio->bi_blkg reference from source to destination bio. This
is however wrong if the source and destination bios are against
different block devices because struct blkcg_gq is different for each
bdev-blkcg pair. This will result in IOs being accounted (and throttled
as a result) multiple times against the same device (src bdev) while
throttling of the other device (dst bdev) is ignored. In case of BFQ the
inconsistency can even result in crashes in bfq_bic_update_cgroup().
Fix the problem by looking up correct blkcg_gq for the cloned bio.

Reported-by: Logan Gunthorpe <logang@deltatee.com>
Reported-and-tested-by: Donald Buczek <buczek@molgen.mpg.de>
Fixes: d92c370a16cb ("block: really clone the block cgroup in bio_clone_blkg_association")
CC: stable@vger.kernel.org
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20220602081242.7731-1-jack@suse.cz
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

pinctrl/rockchip: support setting input-enable param

commit 42d90a1e5caf73138ddde42da5a9fe2b543f9a2c upstream.

Handle the PIN_CONFIG_INPUT_ENABLE param for configuring GPIOs as input.

Signed-off-by: Caleb Connolly <kc@postmarketos.org>
Link: https://lore.kernel.org/r/20220328005005.72492-3-kc@postmarketos.org
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

md: bcache: check the return value of kzalloc() in detached_dev_do_request()

commit 40f567bbb3b0639d2ec7d1c6ad4b1b018f80cf19 upstream.

The function kzalloc() in detached_dev_do_request() can fail, so its
return value should be checked.

Fixes: bc082a55d25c ("bcache: fix inaccurate io state for detached bcache devices")
Reported-by: TOTE Robot <oslab@tsinghua.edu.cn>
Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Signed-off-by: Coly Li <colyli@suse.de>
Link: https://lore.kernel.org/r/20220527152818.27545-4-colyli@suse.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

md: fix double free of io_acct_set bioset

commit 42b805af102471f53e3c7867b8c2b502ea4eef7e upstream.

Now io_acct_set is alloc and free in personality. Remove the codes that
free io_acct_set in md_free and md_stop.

Fixes: 0c031fd37f69 (md: Move alloc/free acct bioset in to personality)
Signed-off-by: Xiao Ni <xni@redhat.com>
Signed-off-by: Song Liu <song@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

md: Don't set mddev private to NULL in raid0 pers->free

commit 0f2571ad7a30ff6b33cde142439f9378669f8b4f upstream.

In normal stop process, it does like this:
   do_md_stop
      |
   __md_stop (pers->free(); mddev->private=NULL)
      |
   md_free (free mddev)
__md_stop sets mddev->private to NULL after pers->free. The raid device
will be stopped and mddev memory is free. But in reshape, it doesn't
free the mddev and mddev will still be used in new raid.

In reshape, it first sets mddev->private to new_pers and then runs
old_pers->free(). Now raid0 sets mddev->private to NULL in raid0_free.
The new raid can't work anymore. It will panic when dereference
mddev->private because of NULL pointer dereference.

It can panic like this:
[63010.814972] kernel BUG at drivers/md/raid10.c:928!
[63010.819778] invalid opcode: 0000 [#1] PREEMPT SMP NOPTI
[63010.825011] CPU: 3 PID: 44437 Comm: md0_resync Kdump: loaded Not tainted 5.14.0-86.el9.x86_64 #1
[63010.833789] Hardware name: Dell Inc. PowerEdge R6415/07YXFK, BIOS 1.15.0 09/11/2020
[63010.841440] RIP: 0010:raise_barrier+0x161/0x170 [raid10]
[63010.865508] RSP: 0018:ffffc312408bbc10 EFLAGS: 00010246
[63010.870734] RAX: 0000000000000000 RBX: ffffa00bf7d39800 RCX: 0000000000000000
[63010.877866] RDX: 0000000000000000 RSI: 0000000000000001 RDI: ffffa00bf7d39800
[63010.884999] RBP: 0000000000000000 R08: fffffa4945e74400 R09: 0000000000000000
[63010.892132] R10: ffffa00eed02f798 R11: 0000000000000000 R12: ffffa00bbc435200
[63010.899266] R13: ffffa00bf7d39800 R14: 0000000000000400 R15: 0000000000000003
[63010.906399] FS:  0000000000000000(0000) GS:ffffa00eed000000(0000) knlGS:0000000000000000
[63010.914485] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[63010.920229] CR2: 00007f5cfbe99828 CR3: 0000000105efe000 CR4: 00000000003506e0
[63010.927363] Call Trace:
[63010.929822]  ? bio_reset+0xe/0x40
[63010.933144]  ? raid10_alloc_init_r10buf+0x60/0xa0 [raid10]
[63010.938629]  raid10_sync_request+0x756/0x1610 [raid10]
[63010.943770]  md_do_sync.cold+0x3e4/0x94c
[63010.947698]  md_thread+0xab/0x160
[63010.951024]  ? md_write_inc+0x50/0x50
[63010.954688]  kthread+0x149/0x170
[63010.957923]  ? set_kthread_struct+0x40/0x40
[63010.962107]  ret_from_fork+0x22/0x30

Removing the code that sets mddev->private to NULL in raid0 can fix
problem.

Fixes: 0c031fd37f69 (md: Move alloc/free acct bioset in to personality)
Reported-by: Fine Fan <ffan@redhat.com>
Signed-off-by: Xiao Ni <xni@redhat.com>
Signed-off-by: Song Liu <song@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

fs/ntfs3: Fix invalid free in log_replay

commit f26967b9f7a830e228bb13fb41bd516ddd9d789d upstream.

log_read_rst() returns ENOMEM error when there is not enough memory.
In this case, if info is returned without initialization,
it attempts to kfree the uninitialized info->r_page pointer. This patch
moves the memset initialization code to before log_read_rst() is called.

Reported-by: Gerald Lee <sundaywind2004@gmail.com>
Signed-off-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Konstantin Komarov <almaz.alexandrovich@paragon-software.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

exportfs: support idmapped mounts

commit 3a761d72fa62eec8913e45d29375344f61706541 upstream.

Make the two locations where exportfs helpers check permission to lookup
a given inode idmapped mount aware by switching it to the lookup_one()
helper. This is a bugfix for the open_by_handle_at() system call which
doesn't take idmapped mounts into account currently. It's not tied to a
specific commit so we'll just Cc stable.

In addition this is required to support idmapped base layers in overlay.
The overlay filesystem uses exportfs to encode and decode file handles
for its index=on mount option and when nfs_export=on.

Cc: <stable@vger.kernel.org>
Cc: <linux-fsdevel@vger.kernel.org>
Tested-by: Giuseppe Scrivano <gscrivan@redhat.com>
Reviewed-by: Amir Goldstein <amir73il@gmail.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

fs: add two trivial lookup helpers

commit 00675017e0aeba5305665c52ded4ddce6a4c0231 upstream.

Similar to the addition of lookup_one() add a version of
lookup_one_unlocked() and lookup_one_positive_unlocked() that take
idmapped mounts into account. This is required to port overlay to
support idmapped base layers.

Cc: <linux-fsdevel@vger.kernel.org>
Tested-by: Giuseppe Scrivano <gscrivan@redhat.com>
Reviewed-by: Amir Goldstein <amir73il@gmail.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

interconnect: qcom: icc-rpmh: Add BCMs to commit list in pre_aggregate

commit b95b668eaaa2574e8ee72f143c52075e9955177e upstream.

We're only adding BCMs to the commit list in aggregate(), but there are
cases where pre_aggregate() is called without subsequently calling
aggregate(). In particular, in icc_sync_state() when a node with initial
BW has zero requests. Since BCMs aren't added to the commit list in
these cases, we don't actually send the zero BW request to HW. So the
resources remain on unnecessarily.

Add BCMs to the commit list in pre_aggregate() instead, which is always
called even when there are no requests.

Signed-off-by: Mike Tipton <mdtipton@codeaurora.org>
[georgi: remove icc_sync_state for platforms with incomplete support]
Link: https://lore.kernel.org/r/20211125174751.25317-1-djakov@kernel.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
Signed-off-by: Stephen Boyd <swboyd@chromium.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

interconnect: qcom: sc7180: Drop IP0 interconnects

commit 2f3724930eb4bba74f7d10bc3bef5bb22dd323df upstream.

The IPA BCM resource ("IP0") on sc7180 was moved to the clk-rpmh driver
in commit bcd63d222b60 ("clk: qcom: rpmh: Add IPA clock for SC7180") and
modeled as a clk, but this interconnect driver still had it modeled as
an interconnect. This was mostly OK because nobody used the interconnect
definition, until the interconnect framework started dropping bandwidth
requests on interconnects that aren't used via the sync_state callback
in commit 7d3b0b0d8184 ("interconnect: qcom: Use icc_sync_state"). Once
that patch was applied the IP0 resource was going to be controlled from
two places, the clk framework and the interconnect framework.

Even then, things were probably going to be OK, because commit
b95b668eaaa2 ("interconnect: qcom: icc-rpmh: Add BCMs to commit list in
pre_aggregate") was needed to actually drop bandwidth requests on unused
interconnects, of which the IPA was one of the interconnect that wasn't
getting dropped to zero. Combining the three commits together leads to
bad behavior where the interconnect framework is disabling the IP0
resource because it has no users while the clk framework thinks the IP0
resource is on because the only user, the IPA driver, has turned it on
via clk_prepare_enable(). Depending on when sync_state is called, we can
get into a situation like below:

  IPA driver probes
  IPA driver gets notified modem started
   runtime PM get()
    IPA clk enabled -> IP0 resource is ON
  sync_state runs
   interconnect zeroes out the IP0 resource -> IP0 resource is off
  IPA driver tries to access a register and blows up

The crash is an unclocked access that manifest as an SError.

 SError Interrupt on CPU0, code 0xbe000011 -- SError
 CPU: 0 PID: 3595 Comm: mmdata_mgr Not tainted 5.17.1+ #166
 Hardware name: Google Lazor (rev1 - 2) with LTE (DT)
 pstate: 60400009 (nZCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
 pc : mutex_lock+0x4c/0x80
 lr : mutex_lock+0x30/0x80
 sp : ffffffc00da9b9c0
 x29: ffffffc00da9b9c0 x28: 0000000000000000 x27: 0000000000000000
 x26: ffffffc00da9bc90 x25: ffffff80c2024010 x24: ffffff80c2024000
 x23: ffffff8083100000 x22: ffffff80831000d0 x21: ffffff80831000a8
 x20: ffffff80831000a8 x19: ffffff8083100070 x18: 00000000ffff0a00
 x17: 000000002f7254f1 x16: 0000000000000100 x15: 0000000000000000
 x14: 0000000000000000 x13: 0000000000000000 x12: 0000000000000000
 x11: 000000000001f0b8 x10: ffffffc00931f0b8 x9 : 0000000000000000
 x8 : 0000000000000000 x7 : fefefefefeff2f60 x6 : 0000808080808080
 x5 : 0000000000000000 x4 : 8080808080800000 x3 : ffffff80d2d4ee28
 x2 : ffffff808c1d6e40 x1 : 0000000000000000 x0 : ffffff8083100070
 Kernel panic - not syncing: Asynchronous SError Interrupt
 CPU: 0 PID: 3595 Comm: mmdata_mgr Not tainted 5.17.1+ #166
 Hardware name: Google Lazor (rev1 - 2) with LTE (DT)
 Call trace:
  dump_backtrace+0xf4/0x114
  show_stack+0x24/0x30
  dump_stack_lvl+0x64/0x7c
  dump_stack+0x18/0x38
  panic+0x150/0x38c
  nmi_panic+0x88/0xa0
  arm64_serror_panic+0x74/0x80
  do_serror+0x0/0x80
  do_serror+0x58/0x80
  el1h_64_error_handler+0x34/0x4c
  el1h_64_error+0x78/0x7c
  mutex_lock+0x4c/0x80
  __gsi_channel_start+0x50/0x17c
  gsi_channel_start+0x54/0x90
  ipa_endpoint_enable_one+0x34/0xc0
  ipa_open+0x4c/0x120

Remove all IP0 resource management from the interconnect driver so that
clk-rpmh is the sole owner. This fixes the issue by preventing the
interconnect driver from overwriting the IP0 resource data that the
clk-rpmh driver wrote.

Cc: Alex Elder <elder@linaro.org>
Cc: Bjorn Andersson <bjorn.andersson@linaro.org>
Cc: Taniya Das <quic_tdas@quicinc.com>
Cc: Mike Tipton <quic_mdtipton@quicinc.com>
Fixes: b95b668eaaa2 ("interconnect: qcom: icc-rpmh: Add BCMs to commit list in pre_aggregate")
Fixes: bcd63d222b60 ("clk: qcom: rpmh: Add IPA clock for SC7180")
Fixes: 7d3b0b0d8184 ("interconnect: qcom: Use icc_sync_state")
Signed-off-by: Stephen Boyd <swboyd@chromium.org>
Tested-by: Alex Elder <elder@linaro.org>
Reviewed-by: Alex Elder <elder@linaro.org>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20220412220033.1273607-2-swboyd@chromium.org
Signed-off-by: Georgi Djakov <djakov@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ext4: only allow test_dummy_encryption when supported

commit 5f41fdaea63ddf96d921ab36b2af4a90ccdb5744 upstream.

Make the test_dummy_encryption mount option require that the encrypt
feature flag be already enabled on the filesystem, rather than
automatically enabling it.  Practically, this means that "-O encrypt"
will need to be included in MKFS_OPTIONS when running xfstests with the
test_dummy_encryption mount option.  (ext4/053 also needs an update.)

Moreover, as long as the preconditions for test_dummy_encryption are
being tightened anyway, take the opportunity to start rejecting it when
!CONFIG_FS_ENCRYPTION rather than ignoring it.

The motivation for requiring the encrypt feature flag is that:

- Having the filesystem auto-enable feature flags is problematic, as it
  bypasses the usual sanity checks.  The specific issue which came up
  recently is that in kernel versions where ext4 supports casefold but
  not encrypt+casefold (v5.1 through v5.10), the kernel will happily add
  the encrypt flag to a filesystem that has the casefold flag, making it
  unmountable -- but only for subsequent mounts, not the initial one.
  This confused the casefold support detection in xfstests, causing
  generic/556 to fail rather than be skipped.

- The xfstests-bld test runners (kvm-xfstests et al.) already use the
  required mkfs flag, so they will not be affected by this change.  Only
  users of test_dummy_encryption alone will be affected.  But, this
  option has always been for testing only, so it should be fine to
  require that the few users of this option update their test scripts.

- f2fs already requires it (for its equivalent feature flag).

Signed-off-by: Eric Biggers <ebiggers@google.com>
Reviewed-by: Gabriel Krisman Bertazi <krisman@collabora.com>
Link: https://lore.kernel.org/r/20220519204437.61645-1-ebiggers@kernel.org
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

MIPS: IP30: Remove incorrect `cpu_has_fpu' override

commit f44b3e74c33fe04defeff24ebcae98c3bcc5b285 upstream.

Remove unsupported forcing of `cpu_has_fpu' to 1, which makes the `nofpu'
kernel parameter non-functional, and also causes a link error:

ld: arch/mips/kernel/traps.o: in function `trap_init':
./arch/mips/include/asm/msa.h:(.init.text+0x348): undefined reference to `handle_fpe'
ld: ./arch/mips/include/asm/msa.h:(.init.text+0x354): undefined reference to `handle_fpe'
ld: ./arch/mips/include/asm/msa.h:(.init.text+0x360): undefined reference to `handle_fpe'

where the CONFIG_MIPS_FP_SUPPORT configuration option has been disabled.

Signed-off-by: Maciej W. Rozycki <macro@orcam.me.uk>
Reported-by: Stephen Zhang <starzhangzsd@gmail.com>
Fixes: 7505576d1c1a ("MIPS: add support for SGI Octane (IP30)")
Cc: stable@vger.kernel.org # v5.5+
Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

MIPS: IP27: Remove incorrect `cpu_has_fpu' override

commit 424c3781dd1cb401857585331eaaa425a13f2429 upstream.

Remove unsupported forcing of `cpu_has_fpu' to 1, which makes the `nofpu'
kernel parameter non-functional, and also causes a link error:

ld: arch/mips/kernel/traps.o: in function `trap_init':
./arch/mips/include/asm/msa.h:(.init.text+0x348): undefined reference to `handle_fpe'
ld: ./arch/mips/include/asm/msa.h:(.init.text+0x354): undefined reference to `handle_fpe'
ld: ./arch/mips/include/asm/msa.h:(.init.text+0x360): undefined reference to `handle_fpe'

where the CONFIG_MIPS_FP_SUPPORT configuration option has been disabled.

Signed-off-by: Maciej W. Rozycki <macro@orcam.me.uk>
Reported-by: Stephen Zhang <starzhangzsd@gmail.com>
Fixes: 0ebb2f4159af ("MIPS: IP27: Update/restructure CPU overrides")
Cc: stable@vger.kernel.org # v4.2+
Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

RDMA/rxe: Generate a completion for unsupported/invalid opcode

commit 2f917af777011c88e977b9b9a5d00b280d3a59ce upstream.

Current rxe_requester() doesn't generate a completion when processing an
unsupported/invalid opcode. If rxe driver doesn't support a new opcode
(e.g. RDMA Atomic Write) and RDMA library supports it, an application
using the new opcode can reproduce this issue. Fix the issue by calling
"goto err;".

Fixes: 8700e3e7c485 ("Soft RoCE driver")
Link: https://lore.kernel.org/r/20220410113513.27537-1-yangx.jy@fujitsu.com
Signed-off-by: Xiao Yang <yangx.jy@fujitsu.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

RDMA/hns: Remove the num_cqc_timer variable

commit db5dfbf5b201df65c1f5332c4d9d5e7c2f42396b upstream.

The bt number of cqc_timer of HIP09 increases compared with that of HIP08.
Therefore, cqc_timer_bt_num and num_cqc_timer do not match. As a result,
the driver may fail to allocate cqc_timer. So the driver needs to uniquely
uses cqc_timer_bt_num to represent the bt number of cqc_timer.

Fixes: 0e40dc2f70cd ("RDMA/hns: Add timer allocation support for hip08")
Link: https://lore.kernel.org/r/20220429093545.58070-1-liangwenpeng@huawei.com
Signed-off-by: Yixing Liu <liuyixing1@huawei.com>
Signed-off-by: Wenpeng Liang <liangwenpeng@huawei.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: r8188eu: delete rtw_wx_read/write32()

commit 4d0cc9e0e53e9946d7b8dc58279c62dfa7a2191b upstream.

These debugging tools let you call:

status = usb_control_msg_recv/send(udev, 0, REALTEK_USB_VENQT_CMD_REQ,
      REALTEK_USB_VENQT_READ/WRITE, value,
      REALTEK_USB_VENQT_CMD_IDX, io_buf,
      size, RTW_USB_CONTROL_MSG_TIMEOUT,
      GFP_KERNEL);

with a user controlled "value" in the 0-0xffff range.  It's not a valid
API.

Fixes: 2b42bd58b321 ("staging: r8188eu: introduce new os_dep dir for RTL8188eu driver")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Link: https://lore.kernel.org/r/YoXS4OaD1oauPvmj@kili
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Revert "random: use static branch for crng_ready()"

This reverts upstream commit f5bda35fba615ace70a656d4700423fa6c9bebee
from stable. It's not essential and will take some time during 5.19 to
work out properly.

Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>

list: test: Add a test for list_is_head()

commit 37dc573c0a547e1aed0c9abb480fab797bd3833f upstream.

list_is_head() was added recently[1], and didn't have a KUnit test. The
implementation is trivial, so it's not a particularly exciting test, but
it'd be nice to get back to full coverage of the list functions.

[1]: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/include/linux/list.h?id=0425473037db40d9e322631f2d4dc6ef51f97e88

Signed-off-by: David Gow <davidgow@google.com>
Acked-by: Daniel Latypov <dlatypov@google.com>
Acked-by: Brendan Higgins <brendanhiggins@google.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Shuah Khan <skhan@linuxfoundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

kseltest/cgroup: Make test_stress.sh work if run interactively

commit 213adc63dfbcdff9a0c19ec1f2681fda9c05cf6d upstream.

Commit 54de76c01239 ("kselftest/cgroup: fix test_stress.sh to use OUTPUT
dir") changes the test_core command path from . to $OUTPUT. However,
variable OUTPUT may not be defined if the command is run interactively.
Fix that by using ${OUTPUT:-.} to cover both cases.

Signed-off-by: Waiman Long <longman@redhat.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net: ipa: fix page free in ipa_endpoint_replenish_one()

commit 70132763d5d2e94cd185e3aa92ac6a3ba89068fa upstream.

Currently the (possibly compound) pages used for receive buffers are
freed using __free_pages().  But according to this comment above the
definition of that function, that's wrong:
    If you want to use the page's reference count to decide
    when to free the allocation, you should allocate a compound
    page, and use put_page() instead of __free_pages().

Convert the call to __free_pages() in ipa_endpoint_replenish_one()
to use put_page() instead.

Fixes: 6a606b90153b8 ("net: ipa: allocate transaction in replenish loop")
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net: ipa: fix page free in ipa_endpoint_trans_release()

commit 155c0c90bca918de6e4327275dfc1d97fd604115 upstream.

Currently the (possibly compound) page used for receive buffers are
freed using __free_pages().  But according to this comment above the
definition of that function, that's wrong:
    If you want to use the page's reference count to decide when
    to free the allocation, you should allocate a compound page,
    and use put_page() instead of __free_pages().

Convert the call to __free_pages() in ipa_endpoint_trans_release()
to use put_page() instead.

Fixes: ed23f02680caa ("net: ipa: define per-endpoint receive buffer size")
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

phy: qcom-qmp: fix reset-controller leak on probe errors

commit 4d2900f20edfe541f75756a00deeb2ffe7c66bc1 upstream.

Make sure to release the lane reset controller in case of a late probe
error (e.g. probe deferral).

Note that due to the reset controller being defined in devicetree in
"lane" child nodes, devm_reset_control_get_exclusive() cannot be used
directly.

Fixes: e78f3d15e115 ("phy: qcom-qmp: new qmp phy driver for qcom-chipsets")
Cc: stable@vger.kernel.org # 4.12
Cc: Vivek Gautam <vivek.gautam@codeaurora.org>
Reviewed-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/20220427063243.32576-3-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

coresight: core: Fix coresight device probe failure issue

commit 8c1d3f79d9ca48e406b78e90e94cf09a8c076bf2 upstream.

It is possibe that probe failure issue happens when the device
and its child_device's probe happens at the same time.
In coresight_make_links, has_conns_grp is true for parent, but
has_conns_grp is false for child device as has_conns_grp is set
to true in coresight_create_conns_sysfs_group. The probe of parent
device will fail at this condition. Add has_conns_grp check for
child device before make the links and make the process from
device_register to connection_create be atomic to avoid this
probe failure issue.

Cc: stable@vger.kernel.org
Suggested-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Suggested-by: Mike Leach <mike.leach@linaro.org>
Signed-off-by: Mao Jinlong <quic_jinlmao@quicinc.com>
Link: https://lore.kernel.org/r/20220309142206.15632-1-quic_jinlmao@quicinc.com
[ Added Cc stable ]
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

blk-iolatency: Fix inflight count imbalances and IO hangs on offline

commit 8a177a36da6c54c98b8685d4f914cb3637d53c0d upstream.

iolatency needs to track the number of inflight IOs per cgroup. As this
tracking can be expensive, it is disabled when no cgroup has iolatency
configured for the device. To ensure that the inflight counters stay
balanced, iolatency_set_limit() freezes the request_queue while manipulating
the enabled counter, which ensures that no IO is in flight and thus all
counters are zero.

Unfortunately, iolatency_set_limit() isn't the only place where the enabled
counter is manipulated. iolatency_pd_offline() can also dec the counter and
trigger disabling. As this disabling happens without freezing the q, this
can easily happen while some IOs are in flight and thus leak the counts.

This can be easily demonstrated by turning on iolatency on an one empty
cgroup while IOs are in flight in other cgroups and then removing the
cgroup. Note that iolatency shouldn't have been enabled elsewhere in the
system to ensure that removing the cgroup disables iolatency for the whole
device.

The following keeps flipping on and off iolatency on sda:

  echo +io > /sys/fs/cgroup/cgroup.subtree_control
  while true; do
      mkdir -p /sys/fs/cgroup/test
      echo '8:0 target=100000' > /sys/fs/cgroup/test/io.latency
      sleep 1
      rmdir /sys/fs/cgroup/test
      sleep 1
  done

and there's concurrent fio generating direct rand reads:

  fio --name test --filename=/dev/sda --direct=1 --rw=randread \
      --runtime=600 --time_based --iodepth=256 --numjobs=4 --bs=4k

while monitoring with the following drgn script:

  while True:
    for css in css_for_each_descendant_pre(prog['blkcg_root'].css.address_of_()):
        for pos in hlist_for_each(container_of(css, 'struct blkcg', 'css').blkg_list):
            blkg = container_of(pos, 'struct blkcg_gq', 'blkcg_node')
            pd = blkg.pd[prog['blkcg_policy_iolatency'].plid]
            if pd.value_() == 0:
                continue
            iolat = container_of(pd, 'struct iolatency_grp', 'pd')
            inflight = iolat.rq_wait.inflight.counter.value_()
            if inflight:
                print(f'inflight={inflight} {disk_name(blkg.q.disk).decode("utf-8")} '
                      f'{cgroup_path(css.cgroup).decode("utf-8")}')
    time.sleep(1)

The monitoring output looks like the following:

  inflight=1 sda /user.slice
  inflight=1 sda /user.slice
  ...
  inflight=14 sda /user.slice
  inflight=13 sda /user.slice
  inflight=17 sda /user.slice
  inflight=15 sda /user.slice
  inflight=18 sda /user.slice
  inflight=17 sda /user.slice
  inflight=20 sda /user.slice
  inflight=19 sda /user.slice <- fio stopped, inflight stuck at 19
  inflight=19 sda /user.slice
  inflight=19 sda /user.slice

If a cgroup with stuck inflight ends up getting throttled, the throttled IOs
will never get issued as there's no completion event to wake it up leading
to an indefinite hang.

This patch fixes the bug by unifying enable handling into a work item which
is automatically kicked off from iolatency_set_min_lat_nsec() which is
called from both iolatency_set_limit() and iolatency_pd_offline() paths.
Punting to a work item is necessary as iolatency_pd_offline() is called
under spinlocks while freezing a request_queue requires a sleepable context.

This also simplifies the code reducing LOC sans the comments and avoids the
unnecessary freezes which were happening whenever a cgroup's latency target
is newly set or cleared.

Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Josef Bacik <josef@toxicpanda.com>
Cc: Liu Bo <bo.liu@linux.alibaba.com>
Fixes: 8c772a9bfc7c ("blk-iolatency: fix IO hang due to negative inflight counter")
Cc: stable@vger.kernel.org # v5.0+
Link: https://lore.kernel.org/r/Yn9ScX6Nx2qIiQQi@slm.duckdns.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

vdpasim: allow to enable a vq repeatedly

commit 242436973831aa97e8ce19533c6c912ea8def31b upstream.

Code must be resilient to enable a queue many times.

At the moment the queue is resetting so it's definitely not the expected
behavior.

v2: set vq->ready = 0 at disable.

Fixes: 2c53d0f64c06 ("vdpasim: vDPA device simulator")
Cc: stable@vger.kernel.org
Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Message-Id: <20220519145919.772896-1-eperezma@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

dt-bindings: gpio: altera: correct interrupt-cells

commit 3a21c3ac93aff7b4522b152399df8f6a041df56d upstream.

update documentation to correctly state the interrupt-cells to be 2.

Cc: stable@vger.kernel.org
Fixes: 4fd9bbc6e071 ("drivers/gpio: Altera soft IP GPIO driver devicetree binding")
Signed-off-by: Dinh Nguyen <dinguyen@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

docs/conf.py: Cope with removal of language=None in Sphinx 5.0.0

commit 627f01eab93d8671d4e4afee9b148f9998d20e7c upstream.

One of the changes in Sphinx 5.0.0 [1] says [sic]:

    5.0.0 final

     - #10474: language does not accept None as it value.
       The default value of language becomes to 'en' now.

[1]: https://www.sphinx-doc.org/en/master/changes.html#release-5-0-0-released-may-30-2022

It results in a new warning from Sphinx 5.0.0 [sic]:

    WARNING: Invalid configuration value found: 'language = None'.
    Update your configuration to a valid langauge code. Falling
    back to 'en' (English).

Silence the warning by using 'en'.
It works with all the Sphinx versions required for building
kernel documentation (1.7.9 or later).

Signed-off-by: Akira Yokosawa <akiyks@gmail.com>
Link: https://lore.kernel.org/r/bd0c2ddc-2401-03cb-4526-79ca664e1cbe@gmail.com
Cc: stable@vger.kernel.org
Signed-off-by: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

SMB3: EBADF/EIO errors in rename/open caused by race condition in smb2_compound_op

commit 0a55cf74ffb5d004b93647e4389096880ce37d6b upstream.

There is  a race condition in smb2_compound_op:

after_close:
num_rqst++;

if (cfile) {
cifsFileInfo_put(cfile); // sends SMB2_CLOSE to the server
cfile = NULL;

This is triggered by smb2_query_path_info operation that happens during
revalidate_dentry. In smb2_query_path_info, get_readable_path is called to
load the cfile, increasing the reference counter. If in the meantime, this
reference becomes the very last, this call to cifsFileInfo_put(cfile) will
trigger a SMB2_CLOSE request sent to the server just before sending this compound
request – and so then the compound request fails either with EBADF/EIO depending
on the timing at the server, because the handle is already closed.

In the first scenario, the race seems to be happening between smb2_query_path_info
triggered by the rename operation, and between “cleanup” of asynchronous writes – while
fsync(fd) likely waits for the asynchronous writes to complete, releasing the writeback
structures can happen after the close(fd) call. So the EBADF/EIO errors will pop up if
the timing is such that:
1) There are still outstanding references after close(fd) in the writeback structures
2) smb2_query_path_info successfully fetches the cfile, increasing the refcounter by 1
3) All writeback structures release the same cfile, reducing refcounter to 1
4) smb2_compound_op is called with that cfile

In the second scenario, the race seems to be similar – here open triggers the
smb2_query_path_info operation, and if all other threads in the meantime decrease the
refcounter to 1 similarly to the first scenario, again SMB2_CLOSE will be sent to the
server just before issuing the compound request. This case is harder to reproduce.

See https://bugzilla.samba.org/show_bug.cgi?id=15051

Cc: stable@vger.kernel.org
Fixes: 8de9e86c67ba ("cifs: create a helper to find a writeable handle by path name")
Signed-off-by: Ondrej Hubsch <ohubsch@purestorage.com>
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
Reviewed-by: Paulo Alcantara (SUSE) <pc@cjr.nz>
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ARM: pxa: maybe fix gpio lookup tables

commit 2672a4bff6c03a20d5ae460a091f67ee782c3eff upstream.

From inspection I found a couple of GPIO lookups that are
listed with device "gpio-pxa", but actually have a number
from a different gpio controller.

Try to rectify that here, with a guess of what the actual
device name is.

Acked-by: Robert Jarzmik <robert.jarzmik@free.fr>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Cc: stable@vger.kernel.org
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ARM: dts: s5pv210: Remove spi-cs-high on panel in Aries

commit 096f58507374e1293a9e9cff8a1ccd5f37780a20 upstream.

Since commit 766c6b63aa04 ("spi: fix client driver breakages when using
GPIO descriptors"), the panel has been blank due to an inverted CS GPIO.
In order to correct this, drop the spi-cs-high from the panel SPI device.

Fixes: 766c6b63aa04 ("spi: fix client driver breakages when using GPIO descriptors")
Cc: <stable@vger.kernel.org>
Signed-off-by: Jonathan Bakker <xc-racer2@live.ca>
Link: https://lore.kernel.org/r/CY4PR04MB05670C771062570E911AF3B4CB1C9@CY4PR04MB0567.namprd04.prod.outlook.com
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

phy: qcom-qmp: fix struct clk leak on probe errors

commit f0a4bc38a12f5a0cc5ad68670d9480e91e6a94df upstream.

Make sure to release the pipe clock reference in case of a late probe
error (e.g. probe deferral).

Fixes: e78f3d15e115 ("phy: qcom-qmp: new qmp phy driver for qcom-chipsets")
Cc: stable@vger.kernel.org # 4.12
Cc: Vivek Gautam <vivek.gautam@codeaurora.org>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20220427063243.32576-2-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

clk: tegra: Add missing reset deassertion

commit 23a43cc437e747473d5f8f98b4fe189fb5c433b7 upstream.

Commit 4782c0a5dd88 ("clk: tegra: Don't deassert reset on enabling
clocks") removed deassertion of reset lines when enabling peripheral
clocks. This breaks the initialization of the DFLL driver which relied
on this behaviour.

Fix this problem by adding explicit deassert/assert requests to the
driver. Tested on Google Pixel C.

Cc: stable@vger.kernel.org
Fixes: 4782c0a5dd88 ("clk: tegra: Don't deassert reset on enabling clocks")
Signed-off-by: Diogo Ivo <diogo.ivo@tecnico.ulisboa.pt>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

arm64: tegra: Add missing DFLL reset on Tegra210

commit 0017f2c856e21bb900be88469e15dac4f41f4065 upstream.

Commit 4782c0a5dd88 ("clk: tegra: Don't deassert reset on enabling
clocks") removed deassertion of reset lines when enabling peripheral
clocks. This breaks the initialization of the DFLL driver which relied
on this behaviour.

In order to be able to fix this, add the corresponding reset to the DT.
Tested on Google Pixel C.

Cc: stable@vger.kernel.org
Fixes: 4782c0a5dd88 ("clk: tegra: Don't deassert reset on enabling clocks")
Signed-off-by: Diogo Ivo <diogo.ivo@tecnico.ulisboa.pt>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

arm64: dts: qcom: ipq8074: fix the sleep clock frequency

commit f607dd767f5d6800ffbdce5b99ba81763b023781 upstream.

Sleep clock frequency should be 32768Hz. Lets fix it.

Cc: stable@vger.kernel.org
Fixes: 41dac73e243d ("arm64: dts: Add ipq8074 SoC and HK01 board support")
Link: https://lore.kernel.org/all/e2a447f8-6024-0369-f698-2027b6edcf9e@codeaurora.org/
Signed-off-by: Kathiravan T <quic_kathirav@quicinc.com>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Link: https://lore.kernel.org/r/1644581655-11568-1-git-send-email-quic_kathirav@quicinc.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

gma500: fix an incorrect NULL check on list iterator

commit bdef417d84536715145f6dc9cc3275c46f26295a upstream.

The bug is here:
return crtc;

The list iterator value 'crtc' will *always* be set and non-NULL by
list_for_each_entry(), so it is incorrect to assume that the iterator
value will be NULL if the list is empty or no element is found.

To fix the bug, return 'crtc' when found, otherwise return NULL.

Cc: stable@vger.kernel.org
fixes: 89c78134cc54d ("gma500: Add Poulsbo support")
Signed-off-by: Xiaomeng Tong <xiam0nd.tong@gmail.com>
Signed-off-by: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220327052028.2013-1-xiam0nd.tong@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tilcdc: tilcdc_external: fix an incorrect NULL check on list iterator

commit 8b917cbe38e9b0d002492477a9fc2bfee2412ce4 upstream.

The bug is here:
if (!encoder) {

The list iterator value 'encoder' will *always* be set and non-NULL
by list_for_each_entry(), so it is incorrect to assume that the
iterator value will be NULL if the list is empty or no element
is found.

To fix the bug, use a new variable 'iter' as the list iterator,
while use the original variable 'encoder' as a dedicated pointer
to point to the found element.

Cc: stable@vger.kernel.org
Fixes: ec9eab097a500 ("drm/tilcdc: Add drm bridge support for attaching drm bridge drivers")
Signed-off-by: Xiaomeng Tong <xiam0nd.tong@gmail.com>
Reviewed-by: Jyri Sarha <jyri.sarha@iki.fi>
Tested-by: Jyri Sarha <jyri.sarha@iki.fi>
Signed-off-by: Jyri Sarha <jyri.sarha@iki.fi>
Link: https://patchwork.freedesktop.org/patch/msgid/20220327061516.5076-1-xiam0nd.tong@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

serial: pch: don't overwrite xmit->buf[0] by x_char

commit d9f3af4fbb1d955bbaf872d9e76502f6e3e803cb upstream.

When x_char is to be sent, the TX path overwrites whatever is in the
circular buffer at offset 0 with x_char and sends it using
pch_uart_hal_write(). I don't understand how this was supposed to work
if xmit->buf[0] already contained some character. It must have been
lost.

Remove this whole pop_tx_x() concept and do the work directly in the
callers. (Without printing anything using dev_dbg().)

Cc: <stable@vger.kernel.org>
Fixes: 3c6a483275f4 (Serial: EG20T: add PCH_UART driver)
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Link: https://lore.kernel.org/r/20220503080808.28332-1-jslaby@suse.cz
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

bcache: avoid journal no-space deadlock by reserving 1 journal bucket

commit 32feee36c30ea06e38ccb8ae6e5c44c6eec790a6 upstream.

The journal no-space deadlock was reported time to time. Such deadlock
can happen in the following situation.

When all journal buckets are fully filled by active jset with heavy
write I/O load, the cache set registration (after a reboot) will load
all active jsets and inserting them into the btree again (which is
called journal replay). If a journaled bkey is inserted into a btree
node and results btree node split, new journal request might be
triggered. For example, the btree grows one more level after the node
split, then the root node record in cache device super block will be
upgrade by bch_journal_meta() from bch_btree_set_root(). But there is no
space in journal buckets, the journal replay has to wait for new journal
bucket to be reclaimed after at least one journal bucket replayed. This
is one example that how the journal no-space deadlock happens.

The solution to avoid the deadlock is to reserve 1 journal bucket in
run time, and only permit the reserved journal bucket to be used during
cache set registration procedure for things like journal replay. Then
the journal space will never be fully filled, there is no chance for
journal no-space deadlock to happen anymore.

This patch adds a new member "bool do_reserve" in struct journal, it is
inititalized to 0 (false) when struct journal is allocated, and set to
1 (true) by bch_journal_space_reserve() when all initialization done in
run_cache_set(). In the run time when journal_reclaim() tries to
allocate a new journal bucket, free_journal_buckets() is called to check
whether there are enough free journal buckets to use. If there is only
1 free journal bucket and journal->do_reserve is 1 (true), the last
bucket is reserved and free_journal_buckets() will return 0 to indicate
no free journal bucket. Then journal_reclaim() will give up, and try
next time to see whetheer there is free journal bucket to allocate. By
this method, there is always 1 jouranl bucket reserved in run time.

During the cache set registration, journal->do_reserve is 0 (false), so
the reserved journal bucket can be used to avoid the no-space deadlock.

Reported-by: Nikhil Kshirsagar <nkshirsagar@gmail.com>
Signed-off-by: Coly Li <colyli@suse.de>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20220524102336.10684-5-colyli@suse.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

bcache: remove incremental dirty sector counting for bch_sectors_dirty_init()

commit 80db4e4707e78cb22287da7d058d7274bd4cb370 upstream.

After making bch_sectors_dirty_init() being multithreaded, the existing
incremental dirty sector counting in bch_root_node_dirty_init() doesn't
release btree occupation after iterating 500000 (INIT_KEYS_EACH_TIME)
bkeys. Because a read lock is added on btree root node to prevent the
btree to be split during the dirty sectors counting, other I/O requester
has no chance to gain the write lock even restart bcache_btree().

That is to say, the incremental dirty sectors counting is incompatible
to the multhreaded bch_sectors_dirty_init(). We have to choose one and
drop another one.

In my testing, with 512 bytes random writes, I generate 1.2T dirty data
and a btree with 400K nodes. With single thread and incremental dirty
sectors counting, it takes 30+ minites to register the backing device.
And with multithreaded dirty sectors counting, the backing device
registration can be accomplished within 2 minutes.

The 30+ minutes V.S. 2- minutes difference makes me decide to keep
multithreaded bch_sectors_dirty_init() and drop the incremental dirty
sectors counting. This is what this patch does.

But INIT_KEYS_EACH_TIME is kept, in sectors_dirty_init_fn() the CPU
will be released by cond_resched() after every INIT_KEYS_EACH_TIME keys
iterated. This is to avoid the watchdog reports a bogus soft lockup
warning.

Fixes: b144e45fc576 ("bcache: make bch_sectors_dirty_init() to be multithreaded")
Signed-off-by: Coly Li <colyli@suse.de>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20220524102336.10684-4-colyli@suse.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

bcache: improve multithreaded bch_sectors_dirty_init()

commit 4dc34ae1b45fe26e772a44379f936c72623dd407 upstream.

Commit b144e45fc576 ("bcache: make bch_sectors_dirty_init() to be
multithreaded") makes bch_sectors_dirty_init() to be much faster
when counting dirty sectors by iterating all dirty keys in the btree.
But it isn't in ideal shape yet, still can be improved.

This patch does the following changes to improve current parallel dirty
keys iteration on the btree,
- Add read lock to root node when multiple threads iterating the btree,
  to prevent the root node gets split by I/Os from other registered
  bcache devices.
- Remove local variable "char name[32]" and generate kernel thread name
  string directly when calling kthread_run().
- Allocate "struct bch_dirty_init_state state" directly on stack and
  avoid the unnecessary dynamic memory allocation for it.
- Decrease BCH_DIRTY_INIT_THRD_MAX from 64 to 12 which is enough indeed.
- Increase &state->started to count created kernel thread after it
  succeeds to create.
- When wait for all dirty key counting threads to finish, use
  wait_event() to replace wait_event_interruptible().

With the above changes, the code is more clear, and some potential error
conditions are avoided.

Fixes: b144e45fc576 ("bcache: make bch_sectors_dirty_init() to be multithreaded")
Signed-off-by: Coly Li <colyli@suse.de>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20220524102336.10684-3-colyli@suse.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

bcache: improve multithreaded bch_btree_check()

commit 622536443b6731ec82c563aae7807165adbe9178 upstream.

Commit 8e7102273f59 ("bcache: make bch_btree_check() to be
multithreaded") makes bch_btree_check() to be much faster when checking
all btree nodes during cache device registration. But it isn't in ideal
shap yet, still can be improved.

This patch does the following thing to improve current parallel btree
nodes check by multiple threads in bch_btree_check(),
- Add read lock to root node while checking all the btree nodes with
  multiple threads. Although currently it is not mandatory but it is
  good to have a read lock in code logic.
- Remove local variable 'char name[32]', and generate kernel thread name
  string directly when calling kthread_run().
- Allocate local variable "struct btree_check_state check_state" on the
  stack and avoid unnecessary dynamic memory allocation for it.
- Reduce BCH_BTR_CHKTHREAD_MAX from 64 to 12 which is enough indeed.
- Increase check_state->started to count created kernel thread after it
  succeeds to create.
- When wait for all checking kernel threads to finish, use wait_event()
  to replace wait_event_interruptible().

With this change, the code is more clear, and some potential error
conditions are avoided.

Fixes: 8e7102273f59 ("bcache: make bch_btree_check() to be multithreaded")
Signed-off-by: Coly Li <colyli@suse.de>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20220524102336.10684-2-colyli@suse.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

stm: ltdc: fix two incorrect NULL checks on list iterator

commit 2e6c86be0e57079d1fb6c7c7e5423db096d0548a upstream.

The two bugs are here:
if (encoder) {
if (bridge && bridge->timings)

The list iterator value 'encoder/bridge' will *always* be set and
non-NULL by drm_for_each_encoder()/list_for_each_entry(), so it is
incorrect to assume that the iterator value will be NULL if the
list is empty or no element is found.

To fix the bug, use a new variable '*_iter' as the list iterator,
while use the old variable 'encoder/bridge' as a dedicated pointer
to point to the found element.

Cc: stable@vger.kernel.org
Fixes: 99e360442f223 ("drm/stm: Fix bus_flags handling")
Signed-off-by: Xiaomeng Tong <xiam0nd.tong@gmail.com>
Acked-by: Raphael Gallais-Pou <raphael.gallais-pou@foss.st.com>
Signed-off-by: Philippe Cornu <philippe.cornu@foss.st.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220327055355.3808-1-xiam0nd.tong@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

carl9170: tx: fix an incorrect use of list iterator

commit 54a6f29522da3c914da30e50721dedf51046449a upstream.

If the previous list_for_each_entry_continue_rcu() don't exit early
(no goto hit inside the loop), the iterator 'cvif' after the loop
will be a bogus pointer to an invalid structure object containing
the HEAD (&ar->vif_list). As a result, the use of 'cvif' after that
will lead to a invalid memory access (i.e., 'cvif->id': the invalid
pointer dereference when return back to/after the callsite in the
carl9170_update_beacon()).

The original intention should have been to return the valid 'cvif'
when found in list, NULL otherwise. So just return NULL when no
entry found, to fix this bug.

Cc: stable@vger.kernel.org
Fixes: 1f1d9654e183c ("carl9170: refactor carl9170_update_beacon")
Signed-off-by: Xiaomeng Tong <xiam0nd.tong@gmail.com>
Acked-by: Christian Lamparter <chunkeey@gmail.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220328122820.1004-1-xiam0nd.tong@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ASoC: rt5514: Fix event generation for "DSP Voice Wake Up" control

commit 4213ff556740bb45e2d9ff0f50d056c4e7dd0921 upstream.

The driver has a custom put function for "DSP Voice Wake Up" which does
not generate event notifications on change, instead returning 0. Since we
already exit early in the case that there is no change this can be fixed
by unconditionally returning 1 at the end of the function.

Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20220428162444.3883147-1-broonie@kernel.org
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

rtl818x: Prevent using not initialized queues

commit 746285cf81dc19502ab238249d75f5990bd2d231 upstream.

Using not existing queues can panic the kernel with rtl8180/rtl8185 cards.
Ignore the skb priority for those cards, they only have one tx queue. Pierre
Asselin (pa@panix.com) reported the kernel crash in the Gentoo forum:

https://forums.gentoo.org/viewtopic-t-1147832-postdays-0-postorder-asc-start-25.html

He also confirmed that this patch fixes the issue. In summary this happened:

After updating wpa_supplicant from 2.9 to 2.10 the kernel crashed with a
"divide error: 0000" when connecting to an AP. Control port tx now tries to
use IEEE80211_AC_VO for the priority, which wpa_supplicants starts to use in
2.10.

Since only the rtl8187se part of the driver supports QoS, the priority
of the skb is set to IEEE80211_AC_BE (2) by mac80211 for rtl8180/rtl8185
cards.

rtl8180 is then unconditionally reading out the priority and finally crashes on
drivers/net/wireless/realtek/rtl818x/rtl8180/dev.c line 544 without this
patch:
idx = (ring->idx + skb_queue_len(&ring->queue)) % ring->entries

"ring->entries" is zero for rtl8180/rtl8185 cards, tx_ring[2] never got
initialized.

Cc: stable@vger.kernel.org
Reported-by: pa@panix.com
Tested-by: pa@panix.com
Signed-off-by: Alexander Wetzel <alexander@wetzel-home.de>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220422145228.7567-1-alexander@wetzel-home.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

xtensa/simdisk: fix proc_read_simdisk()

commit b011946d039d66bbc7102137e98cc67e1356aa87 upstream.

The commit a69755b18774 ("xtensa simdisk: switch to proc_create_data()")
split read operation into two parts, first retrieving the path when it's
non-null and second retrieving the trailing '\n'. However when the path
is non-null the first simple_read_from_buffer updates ppos, and the
second simple_read_from_buffer returns 0 if ppos is greater than 1 (i.e.
almost always). As a result reading from that proc file is almost always
empty.

Fix it by making a temporary copy of the path with the trailing '\n' and
using simple_read_from_buffer on that copy.

Cc: stable@vger.kernel.org
Fixes: a69755b18774 ("xtensa simdisk: switch to proc_create_data()")
Signed-off-by: Yi Yang <yiyang13@huawei.com>
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

mm/memremap: fix missing call to untrack_pfn() in pagemap_range()

commit a04e1928e2ead144dc2f369768bc0a0f3110af89 upstream.

We forget to call untrack_pfn() to pair with track_pfn_remap() when range
is not allowed to hotplug.  Fix it by jump err_kasan.

Link: https://lkml.kernel.org/r/20220531122643.25249-1-linmiaohe@huawei.com
Fixes: bca3feaa0764 ("mm/memory_hotplug: prevalidate the address range being added with platform")
Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Acked-by: Muchun Song <songmuchun@bytedance.com>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

hugetlb: fix huge_pmd_unshare address update

commit 48381273f8734d28ef56a5bdf1966dd8530111bc upstream.

The routine huge_pmd_unshare() is passed a pointer to an address
associated with an area which may be unshared.  If unshare is successful
this address is updated to 'optimize' callers iterating over huge page
addresses.  For the optimization to work correctly, address should be
updated to the last huge page in the unmapped/unshared area.  However, in
the common case where the passed address is PUD_SIZE aligned, the address
is incorrectly updated to the address of the preceding huge page.  That
wastes CPU cycles as the unmapped/unshared range is scanned twice.

Link: https://lkml.kernel.org/r/20220524205003.126184-1-mike.kravetz@oracle.com
Fixes: 39dde65c9940 ("shared page table for hugetlb page")
Signed-off-by: Mike Kravetz <mike.kravetz@oracle.com>
Acked-by: Muchun Song <songmuchun@bytedance.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

nodemask.h: fix compilation error with GCC12

commit 37462a920392cb86541650a6f4121155f11f1199 upstream.

With gcc version 12.0.1 20220401 (Red Hat 12.0.1-0), building with
defconfig results in the following compilation error:

|   CC      mm/swapfile.o
| mm/swapfile.c: In function `setup_swap_info':
| mm/swapfile.c:2291:47: error: array subscript -1 is below array bounds
|  of `struct plist_node[]' [-Werror=array-bounds]
|  2291 |                                 p->avail_lists[i].prio = 1;
|       |                                 ~~~~~~~~~~~~~~^~~
| In file included from mm/swapfile.c:16:
| ./include/linux/swap.h:292:27: note: while referencing `avail_lists'
|   292 |         struct plist_node avail_lists[]; /*
|       |                           ^~~~~~~~~~~

This is due to the compiler detecting that the mask in
node_states[__state] could theoretically be zero, which would lead to
first_node() returning -1 through find_first_bit.

I believe that the warning/error is legitimate.  I first tried adding a
test to check that the node mask is not emtpy, since a similar test exists
in the case where MAX_NUMNODES == 1.

However, adding the if statement causes other warnings to appear in
for_each_cpu_node_but, because it introduces a dangling else ambiguity.
And unfortunately, GCC is not smart enough to detect that the added test
makes the case where (node) == -1 impossible, so it still complains with
the same message.

This is why I settled on replacing that with a harmless, but relatively
useless (node) >= 0 test.  Based on the warning for the dangling else, I
also decided to fix the case where MAX_NUMNODES == 1 by moving the
condition inside the for loop.  It will still only be tested once.  This
ensures that the meaning of an else following for_each_node_mask or
derivatives would not silently have a different meaning depending on the
configuration.

Link: https://lkml.kernel.org/r/20220414150855.2407137-3-dinechin@redhat.com
Signed-off-by: Christophe de Dinechin <christophe@dinechin.org>
Signed-off-by: Christophe de Dinechin <dinechin@redhat.com>
Reviewed-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Ben Segall <bsegall@google.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Vincent Guittot <vincent.guittot@linaro.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Zhen Lei <thunder.leizhen@huawei.com>
Cc: Juri Lelli <juri.lelli@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

mm/page_alloc: always attempt to allocate at least one page during bulk allocation

commit c572e4888ad1be123c1516ec577ad30a700bbec4 upstream.

Peter Pavlisko reported the following problem on kernel bugzilla 216007.

When I try to extract an uncompressed tar archive (2.6 milion
files, 760.3 GiB in size) on newly created (empty) XFS file system,
after first low tens of gigabytes extracted the process hangs in
iowait indefinitely. One CPU core is 100% occupied with iowait,
the other CPU core is idle (on 2-core Intel Celeron G1610T).

It was bisected to c9fa563072e1 ("xfs: use alloc_pages_bulk_array() for
buffers") but XFS is only the messenger.  The problem is that nothing is
waking kswapd to reclaim some pages at a time the PCP lists cannot be
refilled until some reclaim happens.  The bulk allocator checks that there
are some pages in the array and the original intent was that a bulk
allocator did not necessarily need all the requested pages and it was best
to return as quickly as possible.

This was fine for the first user of the API but both NFS and XFS require
the requested number of pages be available before making progress.  Both
could be adjusted to call the page allocator directly if a bulk allocation
fails but it puts a burden on users of the API.  Adjust the semantics to
attempt at least one allocation via __alloc_pages() before returning so
kswapd is woken if necessary.

It was reported via bugzilla that the patch addressed the problem and that
the tar extraction completed successfully.  This may also address bug
215975 but has yet to be confirmed.

BugLink: https://bugzilla.kernel.org/show_bug.cgi?id=216007
BugLink: https://bugzilla.kernel.org/show_bug.cgi?id=215975
Link: https://lkml.kernel.org/r/20220526091210.GC3441@techsingularity.net
Fixes: 387ba26fb1cb ("mm/page_alloc: add a bulk page allocator")
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Cc: "Darrick J. Wong" <djwong@kernel.org>
Cc: Dave Chinner <dchinner@redhat.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Jesper Dangaard Brouer <brouer@redhat.com>
Cc: Chuck Lever <chuck.lever@oracle.com>
Cc: <stable@vger.kernel.org> [5.13+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Revert "mm/cma.c: remove redundant cma_mutex lock"

commit 60a60e32cf91169840abcb4a80f0b0df31708ba7 upstream.

This reverts commit a4efc174b382fcdb which introduced a regression issue
that when there're multiple processes allocating dma memory in parallel by
calling dma_alloc_coherent(), it may fail sometimes as follows:

Error log:
cma: cma_alloc: linux,cma: alloc failed, req-size: 148 pages, ret: -16
cma: number of available pages:
3@125+20@172+12@236+4@380+32@736+17@2287+23@2473+20@36076+99@40477+108@40852+44@41108+20@41196+108@41364+108@41620+
108@42900+108@43156+483@44061+1763@45341+1440@47712+20@49324+20@49388+5076@49452+2304@55040+35@58141+20@58220+20@58284+
7188@58348+84@66220+7276@66452+227@74525+6371@75549=> 33161 free of 81920 total pages

When issue happened, we saw there were still 33161 pages (129M) free CMA
memory and a lot available free slots for 148 pages in CMA bitmap that we
want to allocate.

When dumping memory info, we found that there was also ~342M normal
memory, but only 1352K CMA memory left in buddy system while a lot of
pageblocks were isolated.

Memory info log:
Normal free:351096kB min:30000kB low:37500kB high:45000kB reserved_highatomic:0KB
    active_anon:98060kB inactive_anon:98948kB active_file:60864kB inactive_file:31776kB
    unevictable:0kB writepending:0kB present:1048576kB managed:1018328kB mlocked:0kB
    bounce:0kB free_pcp:220kB local_pcp:192kB free_cma:1352kB lowmem_reserve[]: 0 0 0
Normal: 78*4kB (UECI) 1772*8kB (UMECI) 1335*16kB (UMECI) 360*32kB (UMECI) 65*64kB (UMCI)
36*128kB (UMECI) 16*256kB (UMCI) 6*512kB (EI) 8*1024kB (UEI) 4*2048kB (MI) 8*4096kB (EI)
8*8192kB (UI) 3*16384kB (EI) 8*32768kB (M) = 489288kB

The root cause of this issue is that since commit a4efc174b382 ("mm/cma.c:
remove redundant cma_mutex lock"), CMA supports concurrent memory
allocation.  It's possible that the memory range process A trying to alloc
has already been isolated by the allocation of process B during memory
migration.

The problem here is that the memory range isolated during one allocation
by start_isolate_page_range() could be much bigger than the real size we
want to alloc due to the range is aligned to MAX_ORDER_NR_PAGES.

Taking an ARMv7 platform with 1G memory as an example, when
MAX_ORDER_NR_PAGES is big (e.g.  32M with max_order 14) and CMA memory is
relatively small (e.g.  128M), there're only 4 MAX_ORDER slot, then it's
very easy that all CMA memory may have already been isolated by other
processes when one trying to allocate memory using dma_alloc_coherent().
Since current CMA code will only scan one time of whole available CMA
memory, then dma_alloc_coherent() may easy fail due to contention with
other processes.

This patch simply falls back to the original method that using cma_mutex
to make alloc_contig_range() run sequentially to avoid the issue.

Link: https://lkml.kernel.org/r/20220509094551.3596244-1-aisheng.dong@nxp.com
Link: https://lore.kernel.org/all/20220315144521.3810298-2-aisheng.dong@nxp.com/
Fixes: a4efc174b382 ("mm/cma.c: remove redundant cma_mutex lock")
Signed-off-by: Dong Aisheng <aisheng.dong@nxp.com>
Acked-by: Minchan Kim <minchan@kernel.org>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Marek Szyprowski <m.szyprowski@samsung.com>
Cc: Lecopzer Chen <lecopzer.chen@mediatek.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: <stable@vger.kernel.org> [5.11+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

iommu/dma: Fix iova map result check bug

commit a3884774d731f03d3a3dd4fb70ec2d9341ceb39d upstream.

The data type of the return value of the iommu_map_sg_atomic
is ssize_t, but the data type of iova size is size_t,
e.g. one is int while the other is unsigned int.

When iommu_map_sg_atomic return value is compared with iova size,
it will force the signed int to be converted to unsigned int, if
iova map fails and iommu_map_sg_atomic return error code is less
than 0, then (ret < iova_len) is false, which will to cause not
do free iova, and the master can still successfully get the iova
of map fail, which is not expected.

Therefore, we need to check the return value of iommu_map_sg_atomic
in two cases according to whether it is less than 0.

Fixes: ad8f36e4b6b1 ("iommu: return full error code from iommu_map_sg[_atomic]()")
Signed-off-by: Yunfei Wang <yf.wang@mediatek.com>
Cc: <stable@vger.kernel.org> # 5.15.*
Reviewed-by: Robin Murphy <robin.murphy@arm.com>
Reviewed-by: Miles Chen <miles.chen@mediatek.com>
Link: https://lore.kernel.org/r/20220507085204.16914-1-yf.wang@mediatek.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

iommu/msm: Fix an incorrect NULL check on list iterator

commit 8b9ad480bd1dd25f4ff4854af5685fa334a2f57a upstream.

The bug is here:
if (!iommu || iommu->dev->of_node != spec->np) {

The list iterator value 'iommu' will *always* be set and non-NULL by
list_for_each_entry(), so it is incorrect to assume that the iterator
value will be NULL if the list is empty or no element is found (in fact,
it will point to a invalid structure object containing HEAD).

To fix the bug, use a new value 'iter' as the list iterator, while use
the old value 'iommu' as a dedicated variable to point to the found one,
and remove the unneeded check for 'iommu->dev->of_node != spec->np'
outside the loop.

Cc: stable@vger.kernel.org
Fixes: f78ebca8ff3d6 ("iommu/msm: Add support for generic master bindings")
Signed-off-by: Xiaomeng Tong <xiam0nd.tong@gmail.com>
Link: https://lore.kernel.org/r/20220501132823.12714-1-xiam0nd.tong@gmail.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>