pm_runtime_get_sync will increment pm usage counter even it failed.
Forgetting to putting operation will result in reference leak here.
Fix it by replacing it with pm_runtime_resume_and_get to keep usage
counter balanced.
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Reviewed-by: Robert Foss <robert.foss@linaro.org>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210531135622.3348252-1-yukuai3@huawei.com
The Tegra clock driver contains legacy code which deasserts hardware reset
when peripheral clocks are enabled. This behaviour comes from a pre-CCF
era of the Tegra drivers. This is unacceptable for modern kernel drivers
which use generic CCF and reset-control APIs because it breaks assumptions
of the drivers about clk/reset sequences and about reset-propagation
delays. Hence remove the awkward legacy behaviour from the clk driver.
In particular PMC driver assumes that hardware blocks remains in reset
while power domain is turning on, but the clk driver deasserts the reset
before power clamp is removed, hence breaking the driver's assumption.
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
The external clocks don't have reset bits as they don't belong to any
specific hardware unit. Mark them as not having reset control for
consistency.
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Check whether thermal DIV2 throttle is active in order to report
the CPU frequency properly. This very useful for userspace tools
like cpufreq-info which show actual frequency asserted from hardware.
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Zero clock rate doesn't make sense for PLLs and tegra-clk driver enters
into infinite loop on trying to calculate PLL parameters for zero rate.
Make code to error out if requested rate is zero.
Originally this trouble was found by Robert Yang while he was trying to
bring up upstream kernel on Samsung Galaxy Tab, which happened due to a
bug in Tegra DRM driver that erroneously sets PLL rate to zero. This
issues came over again recently during of kernel bring up on ASUS TF700T.
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Higher SCLK rates on Tegra20 require high core voltage. The higher
clock rate may have a positive performance effect only for AHB DMA
transfers and AVP CPU, but both aren't used by upstream kernel at all.
Halve SCLK rate on Tegra20 in order to remove the high core voltage
requirement.
Acked-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
The PLLU (USB) consists of the PLL configuration itself and configuration
of the PLLU outputs. The PLLU programming is inconsistent on T30 vs T114,
where T114 immediately bails out if PLLU is enabled and T30 re-enables
a potentially already enabled PLL (left after bootloader) and then fully
reprograms it, which could be unsafe to do. The correct way should be to
skip enabling of the PLL if it's already enabled and then apply
configuration to the outputs. This patch doesn't fix any known problems,
it's a minor improvement.
Acked-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
The refcounting of the gate clocks has a bug causing the enable_refcnt
to underflow when unused clocks are disabled. This happens because clk
provider erroneously bumps the refcount if clock is enabled at a boot
time, which it shouldn't be doing, and it does this only for the gate
clocks, while peripheral clocks are using the same gate ops and the
peripheral clocks are missing the initial bump. Hence the refcount of
the peripheral clocks is 0 when unused clocks are disabled and then the
counter is decremented further by the gate ops, causing the integer
underflow.
Fix this problem by removing the erroneous bump and by implementing the
disable_unused() callback, which disables the unused gates properly.
The visible effect of the bug is such that the unused clocks are never
gated if a loaded kernel module grabs the unused clocks and starts to use
them. In practice this shouldn't cause any real problems for the drivers
and boards supported by the kernel today.
Acked-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
The 600MHz is a too high clock rate for some SoC versions for the video
decoder hardware and this may cause stability issues. Use 300MHz for the
video decoder by default, which is supported by all hardware versions.
Fixes: ed1a2459e2 ("clk: tegra: Add Tegra20/30 EMC clock implementation")
Acked-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
The hci_sock_dev_event() function will cleanup the hdev object for
sockets even if this object may still be in used within the
hci_sock_bound_ioctl() function, result in UAF vulnerability.
This patch replace the BH context lock to serialize these affairs
and prevent the race condition.
Signed-off-by: Lin Ma <linma@zju.edu.cn>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Parameterize code in several places to allow scaling of windows. Note
that this currently still relies on static programming of the various
metering and memory pool allocation registers. This seems to work for
the common cases, but may eventually need to be updated to support
use-cases with multiple windows and higher bandwidth and latency
requirements.
Signed-off-by: Thierry Reding <treding@nvidia.com>
The driver currently exposes several YUV formats but fails to properly
program all the registers needed to display such formats. Add the right
programming sequences so that overlay windows can be used to accelerate
color format conversions in multimedia playback use-cases.
Signed-off-by: Thierry Reding <treding@nvidia.com>
The format modifier is 64bit, while DRM_FORMAT_MOD_NVIDIA_SECTOR_LAYOUT
uses BIT() macro that is 32bit on ARM32.
The (modifier &= ~DRM_FORMAT_MOD_NVIDIA_SECTOR_LAYOUT) doesn't work as
expected on ARM32 and tegra_fb_get_tiling() fails for the tiled formats
on 32bit Tegra because modifier mask isn't applied properly. Use the
BIT_ULL() macro to fix this trouble.
Fixes: 7b6f846785 ("drm/tegra: Support sector layout on Tegra194")
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
These functions haven't been used, so just remove them. The patch
just uses grep to verify.
Signed-off-by: Guo Ren <guoren@linux.alibaba.com>
Reviewed-by: Anup Patel <anup@brainfault.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Palmer Dabbelt <palmerdabbelt@google.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Michal Simek <monstr@monstr.eu>
Link: https://lore.kernel.org/r/1622350408-44875-3-git-send-email-guoren@kernel.org
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
pm_runtime_get_sync(), contradictory to intuition, does not drop the
runtime PM usage counter on errors which lead to several wrong usages in
drivers (missing the put). pm_runtime_resume_and_get() was added as a
better implementation so document the preference of using it, hoping it
will stop bad patterns.
Suggested-by: Marek Szyprowski <m.szyprowski@samsung.com>
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
[ rjw: Documentation change edits ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Make turning off unused power resources (after the enumeration of
devices and during system-wide resume from S3) more straightforward
by using the observation that the power resource state stored in
struct acpi_power_resource can be used to determine whether or not
the give power resource has any users.
Namely, when the state of the power resource is unknown, its _STA
method has never been evaluated (or the evaluation of it has failed)
and its _ON and _OFF methods have never been executed (or they have
failed to execute), so for all practical purposes it can be assumed
to have no users (or to be unusable). Therefore, instead of checking
the number of power resource users, it is sufficient to check if its
state is known.
Moreover, if the last known state of a given power resource is "off",
it is not necessary to turn it off, because it has been used to
initialize the power state or the wakeup power resources list of at
least one device and either its _STA method has returned 0 ("off"),
or its _OFF method has been successfully executed already.
Accordingly, modify acpi_turn_off_unused_power_resources() to do the
above checks (which are suitable for both uses of it) instead of
using the number of power resource users or evaluating its _STA
method, drop its argument (which is not useful any more) and update
its callers.
Also drop the users field from struct acpi_power_resource as it is
not useful any more.
Tested-by: Dave Olsthoorn <dave@bewaar.me>
Tested-by: Shujun Wang <wsj20369@163.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Currently, there are two ways to check the state of an ACPI power
resource and they may not be consistent with each other. The first
one is to evaluate the power resource's _STA object and the other one
is to check its reference counter value. However, on some systems
the value returned by _STA may not be consistent with the value of
the power resource's reference counter (for example, on some systems
it returns the same value every time for certain power resources).
Moreover, evaluating _STA is unnecessary overhead for a power
resource for which it has been evaluated already or whose state is
otherwise known, because either the _ON or the _OFF method has been
executed for it.
For this reason, save the state of each power resource in its
struct acpi_power_resource object and use the saved value whenever
its state needs to be checked, except when its stats is unknown, in
which case the _STA method is evaluated for it and the value
returned by that method is saved as the last known state of
the power resource.
Moreover, drop the power resource _STA method evaluation from
acpi_add_power_resource(), so as to avoid doing that unnecessarily
for power resources that will never be used.
Tested-by: Dave Olsthoorn <dave@bewaar.me>
Tested-by: Shujun Wang <wsj20369@163.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Use u8 as the data type for representing the state of an ACPI
power resource.
It is s not necessary to use int for that and because subsequent
changes are going to use ACPI_POWER_RESOURCE_STATE_UNKNOWN, it is
better to adjust the data type so that the "unknown" state is
represented by the "all ones" value.
While at it, clean up acpi_power_get_state() somewhat.
No intentional functional impact.
Tested-by: Dave Olsthoorn <dave@bewaar.me>
Tested-by: Shujun Wang <wsj20369@163.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
The USB4 Configuration Manager guide suggests that the USB4 port wakes
are configured in a certain way, like that when the port is configured
the wake-on-connect should not be set and so forth, so align the driver
with this.
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Latest USB4 spec added a new wake bit for DisplayPort so add this to the
driver when runtime suspending. This way wake up the domain when a new
monitor is plugged in to any of the device routers.
Also do the same for pre-USB4 devices through the link controller
registers as documented in chapter 13 of the USB4 spec.
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Align the "What" entries along with the rest and move
nvm_authenticate_on_disconnect to correct place.
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Now that we have generic functionality available in nvm.c make the DMA
port code call it instead of duplicating the functionality.
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
We do this for Thunderbolt 2/3 devices through DMA port, USB4 devices
and retimers pretty much the same way. Only the actual block read/write
is different. For this reason split out the NVM read/write functions
from usb4.c to nvm.c and make USB4 device code call these when needed.
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Add new device known to support self-authenticate on disconnect.
Signed-off-by: Crag Wang <crag.wang@dell.com>
Reviewed-by: Mario Limonciello <mario.limonciello@outlook.com>
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
The Lontium bridge is including legacy header files for GPIO
but not using them. Delete the includes.
Cc: Adrien Grassein <adrien.grassein@gmail.com>
Cc: Robert Foss <robert.foss@linaro.org>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Adren Grassein <adrien.grassein@gmail.com>
Signed-off-by: Robert Foss <robert.foss@linaro.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20210529002759.468964-1-linus.walleij@linaro.org
The GLF_LRU flag is checked under lru_lock in gfs2_glock_remove_from_lru() to
remove the glock from the lru list in __gfs2_glock_put().
On the shrink scan path, the same flag is cleared under lru_lock but because
of cond_resched_lock(&lru_lock) in gfs2_dispose_glock_lru(), progress on the
put side can be made without deleting the glock from the lru list.
Keep GLF_LRU across the race window opened by cond_resched_lock(&lru_lock) to
ensure correct behavior on both sides - clear GLF_LRU after list_del under
lru_lock.
Reported-by: syzbot <syzbot+34ba7ddbf3021981a228@syzkaller.appspotmail.com>
Signed-off-by: Hillf Danton <hdanton@sina.com>
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Current dts files with 'grf' nodes are manually verified.
In order to automate this process grf.txt has to be
converted to YAML.
Most compatibility strings are in use with "simple-mfd" added.
Changed compatibles:
"rockchip,rk3066-grf", "syscon", "simple-mfd"
"rockchip,rk3188-grf", "syscon", "simple-mfd"
Add description already in use:
"rockchip,rv1108-pmugrf", "syscon"
Add new descriptions for:
"rockchip,rk3568-grf", "syscon", "simple-mfd"
"rockchip,rk3568-pmugrf", "syscon", "simple-mfd"
Signed-off-by: Johan Jonker <jbx6244@gmail.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20210512122346.9463-3-jbx6244@gmail.com
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Memory-related Tegra devfreq devices now could be used as a cooling
devices. Enable CONFIG_DEVFREQ_THERMAL by default since this option
enables cooling functionality of the devfreq drivers.
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
dwmac-sun8i supports v3s and
Licheepi-zero Dock provides an ethernet port
furthermore, align nodes in alphabetical order
Signed-off-by: Andreas Rehn <rehn.andreas86@gmail.com>
Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20210525173159.183415-1-rehn.andreas86@gmail.com
The ACTMON module monitors activity of memory clients and then devfreq
driver makes decisions about a required memory frequency based on info
from ACTMON. Add ACTMON device to the thermal zone of Ouya in order to
use it as a cooling device which throttles memory freq on overheat.
Tested-by: Peter Geis <pgwipeout@gmail.com>
Tested-by: Matt Merhar <mattmerhar@protonmail.com>
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
The ACTMON module monitors activity of memory clients and then devfreq
driver makes decisions about a required memory frequency based on info
from ACTMON. Add ACTMON device to the thermal zone of Nexus 7 in order
to use it as a cooling device which throttles memory freq on overheat.
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
The ACTMON module monitors activity of memory clients and decisions
about a minimum required memory frequency are made based on info from
ACTMON. Add cooling cells to ACTMON device-tree node in order to turn
it into a cooling device that will throttle memory freq on overheat.
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
The 3v3 regulator GPIO is GP6 and not GP7, which is the DDR regulator.
Both regulators are always-on, nevertheless the DT model needs to be
corrected, fix it.
Reported-by: Svyatoslav Ryhel <clamor95@gmail.com>
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
The bq27541 Linux kernel driver will try to reprogram controller based
on the values from monitored-battery node, but it fails to do so because
controller was locked by manufacturer. Still this is a very undesirable
behaviour, hence let's remove the optional battery node.
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Remove unused thermal zone just to clean up device-tree and set critical
temperature further apart from the passive cooling trip point since
during or thermal testing of Asus Transformer devices we found that CPU
could reach the critical temperature in a certain kernel configurations
for a brief moment if critical trip point is set close to the passive
trip point and then device will be immediately shut off without getting
a chance to cool down using passive cooling.
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Add i2c-thermtrip node which enables emergency shutdown by PMC on
SoC die overheat detected by TSENSOR.
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Add thermal zone with a passive cooling trip for CPU. Attach it to the
LM90 sensor which monitors CPU temperature. Now CPU frequencies will be
throttled once trip point is reached, preventing critical overheat.
Tested-by: Agneli <poczt@protonmail.ch>
Tested-by: Paul Fertser <fercerpav@gmail.com>
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
All Tegra boards which use WM8903 audio codec are specifying a wrong
polarity for the headphones detection GPIO. The kernel driver hardcodes
the polarity to active-low, which is the correct polarity, so we can fix
the device-trees safely.
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
The reg property is now specified for the emc-tables nodes in the Tegra20
device-tree binding. Add reg property to the EMC table device-tree nodes
of Tegra20 board device-trees in order to silence dt_binding_check warning
about the missing property.
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
It's possible to hit the temperature of the thermal zone in a very warm
environment under a constant load, like watching a video using software
decoding. It's even easier to hit the limit with a slightly overclocked
CPU. Bump the temperature limit by 10C in order to improve user
experience. Acer A500 has a large board and 10" display panel which are
used for the heat dissipation, the SoC is placed far away from battery,
hence we can safely bump the temperature limit.
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Tegra20 has v2.00 SDMMC controller which doesn't support voltage
switching and the WiFi SDIO bus voltage is fixed to 1.8v in accordance
to the board's schematics, while MMC core confusingly saying that it's
3.3v because of the v2.00. Let's correct the voltage in the device-tree
just for consistency. This is a minor improvement which doesn't fix any
problems.
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Use edge-triggered interrupt and set delay to 100ms for microphone hook
detection. This doesn't fix any known problems, but there is a smaller
chance to miss insertion of the microphone now, which previously happened
rarely.
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Implement support for pull-up (most pins, excl. DU_DOTCLKIN and EXTALR)
and pull-down (most pins, excl. JTAG) handling for the R-Car V3M SoC,
using the common R-Car bias handling.
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Link: https://lore.kernel.org/r/bcfad447624d874258a45a92554574b8fe9f712f.1619785375.git.geert+renesas@glider.be
Implement support for pull-up (most pins) and pull-down (ASEBRK#/ACK)
handling for R-Car E2 and RZ/G1E SoCs, using the common R-Car bias
handling.
Note that on RZ/G1E, the "ASEBRK#/ACK" pin is called "ACK", but the code
doesn't handle that naming difference. Hence users should use the R-Car
naming in DTS files.
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Tested-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Reviewed-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Link: https://lore.kernel.org/r/f78da2ba937ce98ae9196f4ee54149a5214fd545.1619785375.git.geert+renesas@glider.be