These newer PM macros allow the compiler to see what code it can remove
if !CONFIG_PM_SLEEP. This allows the removal of __maybe_unused markings
whilst achieving the same result. In this case it also lets the
structure itself be removed.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Cc: Nuno Sá <nuno.sa@analog.com>
Link: https://lore.kernel.org/r/20220925172759.3573439-9-jic23@kernel.org
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
These newer PM macros allow the compiler to see what code it can remove
if !CONFIG_PM_SLEEP. This allows the removal of __maybe_unused markings
whilst achieving the same result.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Cc: Guenter Roeck <linux@roeck-us.net>
Link: https://lore.kernel.org/r/20220925172759.3573439-8-jic23@kernel.org
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
These newer PM macros allow the compiler to see what code it can remove
if !CONFIG_PM_SLEEP. This allows the removal of __maybe_unused markings
whilst achieving the same result.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/20220925172759.3573439-7-jic23@kernel.org
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
These newer PM macros allow the compiler to see what code it can remove
if !CONFIG_PM_SLEEP. This allows the removal of messy #ifdef barriers whilst
achieving the same result.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Cc: Linus Walleij <linus.walleij@linaro.org>
Link: https://lore.kernel.org/r/20220925172759.3573439-6-jic23@kernel.org
[groeck: Drop #ifdef from struct gpio_fan_data]
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
These newer PM macros allow the compiler to see what code it can remove
if !CONFIG_PM_SLEEP. This allows the removal of messy #ifdef barriers whilst
achieving the same result.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Cc: Lars-Peter Clausen <lars@metafoo.de>
Link: https://lore.kernel.org/r/20220925172759.3573439-5-jic23@kernel.org
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
These newer PM macros allow the compiler to see what code it can remove
if !CONFIG_PM_SLEEP. This allows the removal of messy #ifdef barriers whilst
achieving the same result.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Cc: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Link: https://lore.kernel.org/r/20220925172759.3573439-4-jic23@kernel.org
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
These newer PM macros allow the compiler to see what code it can remove
if !CONFIG_PM_SLEEP. This allows the removal of messy #ifdef barriers whilst
achieving the same result.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Cc: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20220925172759.3573439-3-jic23@kernel.org
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
These newer PM macros allow the compiler to see what code it can remove
if !CONFIG_PM_SLEEP. This allows the removal of messy #ifdef barriers.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Cc: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20220925172759.3573439-2-jic23@kernel.org
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Declarations for static symbols are useless code repetition (unless
there are cyclic dependencies).
Reorder some functions and variables which allows to get rid of 42
forward declarations.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Declarations for static symbols are useless repetition unless there are
cyclic dependencies.
Reorder the functions and variables to get rid of 4 forward declarations.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Link: https://lore.kernel.org/r/20220924135738.234051-2-u.kleine-koenig@pengutronix.de
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Make use of the cpp symbol DRIVER_NAME to set the driver's name and use
it instead of all explicit usages of the same string. Also make use of
it instead of sis5595_driver.driver.name which breaks a cyclic dependency
between sis5595_probe() and sis5595_driver that in the next commit allows
to drop some forward declarations. For an amd64 allyesconfig this even
reduces the size of the driver by 3 bytes.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Link: https://lore.kernel.org/r/20220924135738.234051-1-u.kleine-koenig@pengutronix.de
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
In
adm9240_read()
adm9240_fan_read()
adm9240_write_fan_div(),
it assumes that the caller of adm9240_write_fan_div() must hold
data->update_lock. Otherwise, it may cause data races when data is
updated by other threads.
Signed-off-by: Li Zhong <floridsleeves@gmail.com>
Link: https://lore.kernel.org/r/20220924001751.1726369-1-floridsleeves@gmail.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
On 32-bit platforms, long is 32 bits, so (long)UINT_MAX is less than
(long)SHT4X_MIN_POLL_INTERVAL, which means the clamping operation is
bogus. Fix this by clamping at INT_MAX, so that the upperbound is the
same on all platforms.
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Link: https://lore.kernel.org/r/20220924101151.4168414-1-Jason@zx2c4.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Declarations for static symbols are useless repetition unless there are
cyclic dependencies.
Reorder the functions and variables to get rid of 6 forward declarations.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Link: https://lore.kernel.org/r/20220922074900.2763331-2-u.kleine-koenig@pengutronix.de
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Make use of the cpp symbol DRIVER_NAME to set the driver's name and use
it instead of all explicit usages of the same string. Also make use of
it instead of sis5595_driver.driver.name which breaks a cyclic dependency
between sis5595_probe() and sis5595_driver that in the next commit allows
to drop some forward declarations.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Link: https://lore.kernel.org/r/20220922074900.2763331-1-u.kleine-koenig@pengutronix.de
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Several drivers manually register a devm handler to disable their clk.
Convert them to devm_clk_get_enabled().
Acked-by: Guenter Roeck <linux@roeck-us.net>
Reviewed-by: Nuno Sá <nuno.sa@analog.com>
Acked-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Declarations for static symbols are useless repetition unless there are
cyclic dependencies.
Reorder the functions and variables to get rid of 5 forward declarations.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Link: https://lore.kernel.org/r/20220920135617.1046361-2-u.kleine-koenig@pengutronix.de
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Make use of the cpp symbol DRIVER_NAME to set the driver's name and use
it instead of all explicit usages of the same string. Also make use of
it instead of vt8231_driver.driver.name which breaks a cyclic dependency
between vt8231_probe() and vt8231_driver that in the next commit allows
to drop some forward declarations.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Link: https://lore.kernel.org/r/20220920135617.1046361-1-u.kleine-koenig@pengutronix.de
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Declarations for static symbols are useless repetition unless there are
cyclic dependencies.
Reorder the functions and variables to get rid of 6 forward declarations.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Link: https://lore.kernel.org/r/20220919103155.795151-3-u.kleine-koenig@pengutronix.de
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Make use of the cpp symbol DRIVER_NAME to set the driver's name and also
as name for devm_request_region(). While at it add a module alias using
the new define.
This is a preparation for the next cleanup commit that removes a cyclic
dependency between pc87360_driver (which references pc87360_probe in
.probe) and pc87360_probe() (which used pc87360_driver.driver.name).
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Link: https://lore.kernel.org/r/20220919103155.795151-2-u.kleine-koenig@pengutronix.de
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
This adds the enable attribute which is used to select if zero PWM duty
means to switch off regulator and PWM or to keep them enabled but
at inactive PWM output level.
Depending on the select enable mode, turn off the regulator and PWM if
the PWM duty is zero, or keep them enabled.
This is especially important for fan using inverted PWM signal polarity.
Having regulator supplied and PWM disabled, some PWM controllers provide
the active, rather than inactive signal.
With this change the shutdown as well as suspend/resume paths require
modifcations as well.
Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Link: https://lore.kernel.org/r/20220914153137.613982-6-alexander.stein@ew.tq-group.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Regular calls to set_pwm don't hold the mutex, but the upcoming
update_enable support needs to call set_pwm with the mutex being held.
So provide the previous behavior in set_pwm (handling the lock), while
adding __set_pwm which assumes the lock being held.
Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Reviewed-by: Guenter Roeck <linux@roeck-us.net>
Link: https://lore.kernel.org/r/20220914153137.613982-5-alexander.stein@ew.tq-group.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
This handles enabling/disabling the regulator in a single function, while
keeping the enables/disabled balanced. This is a preparation when
regulator is switched from different code paths.
Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Link: https://lore.kernel.org/r/20220914153137.613982-4-alexander.stein@ew.tq-group.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Instead of comparing the current to the new pwm duty to decide whether to
enable the PWM, use a dedicated flag. Also apply the new PWM duty in any
case. This is a preparation to enable/disable the regulator dynamically.
Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Link: https://lore.kernel.org/r/20220914153137.613982-3-alexander.stein@ew.tq-group.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
In preparation for dynamically switching regulator, split the power on
and power off sequence into separate functions.
Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Reviewed-by: Guenter Roeck <linux@roeck-us.net>
Link: https://lore.kernel.org/r/20220914153137.613982-2-alexander.stein@ew.tq-group.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Add the pmbus driver for TEXAS tps546d24 Buck Converter.
The vout mode of tps546d24 supported relative data format,
which is not supported by the PMBus core.
Signed-off-by: Duke Du <dukedu83@gmail.com>
Link: https://lore.kernel.org/r/1662951668-9849-1-git-send-email-Duke.Du@quantatw.com
[groeck: Add __maybe_unused to tps546d24_of_match declaration]
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
MAX31760 is a precision fan speed controller with nonvolatile lookup table.
Device has one internal and one external temperature sensor support.
Controls two fans and measures their speeds. Generates hardware alerts when
programmable max and critical temperatures are exceeded.
Signed-off-by: Ibrahim Tilki <Ibrahim.Tilki@analog.com>
Reviewed-by: Nurettin Bolucu <Nurettin.Bolucu@analog.com>
Link: https://lore.kernel.org/r/20220910171945.48088-2-Ibrahim.Tilki@analog.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Fix: "ERROR: space required before the open parenthesis '('"
All of the errors were introduced before this series of patches.
Signed-off-by: Eliav Farber <farbere@amazon.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Link: https://lore.kernel.org/r/20220908152449.35457-22-farbere@amazon.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
This change adds debugfs to read and write temperature sensor coefficients
- g, h, j and cal5.
The coefficients can vary between product and product, so it can be very
useful to be able to modify them on the fly during the calibration
process.
e.g.:
cat /sys/kernel/debug/940f23d0000.pvt/ts_coeff_cal5
4096
echo 83000 > sys/kernel/debug/940f23d0000.pvt/ts_coeff_g
Signed-off-by: Eliav Farber <farbere@amazon.com>
Link: https://lore.kernel.org/r/20220908152449.35457-21-farbere@amazon.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Use thermal coefficients from the device tree if they exist.
Otherwise, use default values according to the series (5 or 6).
All coefficients can be used or only part of them.
The coefficients shall be used for fine tuning the default values since
coefficients can vary between product and product.
Signed-off-by: Eliav Farber <farbere@amazon.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Link: https://lore.kernel.org/r/20220908152449.35457-20-farbere@amazon.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
The current equation used in code is aligned to series 5:
T = G + H * (n / cal5 - 0.5) + J * F
Where:
G = 60, H = 200, cal5 = 4094, J = -0.1, F = frequency clock in MHz
Series 6 has a slightly different equation:
T = G + H * (n / cal5 - 0.5)
and a different set of coefficients:
G = 57.4, H = 249.4, cal5 = 4096
This change supports equation and coefficients for both series.
(for series 6, J is set to 0).
The series is determined according to “moortec,ts-series” property in
the device tree.
If absent, series 5 is assumed to be the default.
Signed-off-by: Eliav Farber <farbere@amazon.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Link: https://lore.kernel.org/r/20220908152449.35457-18-farbere@amazon.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
The SENSORS_ASPEED is part of the Aspeed silicon so it makes
sense to depend on ARCH_ASPEED and for compile testing.
Signed-off-by: Peter Robinson <pbrobinson@gmail.com>
Link: https://lore.kernel.org/r/20220916120936.372591-1-pbrobinson@gmail.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Modify the equation and coefficients used to convert the digital output
to temperature according to series 5 of the Moortec Embedded Temperature
Sensor (METS) datasheet:
T = G + H * (n / cal5 - 0.5) + J * F
Where:
*) G = 60, H = 200, cal5 = 4094, J = -0.1.
*) F = frequency clock in MHz.
*) n is the digital output.
In code, the G, H and J coefficients are multiplied by a factor of 1000
to get the temperature in milli-Celsius.
Final result is clamped in case it exceeds min/max thresholds.
Change is done since it is unclear where the current equation and
coefficients came from.
Signed-off-by: Eliav Farber <farbere@amazon.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Link: https://lore.kernel.org/r/20220908152449.35457-16-farbere@amazon.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Add support for mr76006 pre-scaler which provides divide-by-2 scaling
of the input voltage, so that it can be presented to the VM for
measurement within its range (the VM input range is limited from -0.1V
to 1V).
The driver reads from the device-tree all the channels that use the
mr76006 pre-scaler and multiplies the voltage result by a factor of 2,
to represent to the user with the actual voltage input source.
Channels that are not in the device-tree are multiplied by a factor
of 1.
Signed-off-by: Eliav Farber <farbere@amazon.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Link: https://lore.kernel.org/r/20220908152449.35457-15-farbere@amazon.com
[groeck: Addressed conflicts against commit d59eacaac9]
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Add active channel support per voltage monitor.
The number of active channels is read from the device-tree.
When absent in device-tree, all channels are assumed to be used.
This shall be useful to expose sysfs only for inputs that are connected
to a voltage source.
Setting number of active channels to 0, means that entire VM sensor is
not used.
Signed-off-by: Eliav Farber <farbere@amazon.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Link: https://lore.kernel.org/r/20220908152449.35457-13-farbere@amazon.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Static analyzers report:
drivers/hwmon/emc2305.c:194 emc2305_set_cur_state()
warn: impossible condition '(val > 255) => (0-255 > 255)'
'val' is u8 and thus can never be larger than 255. In theory
the operation calculating 'val' could result in a value larger
than 255, but this won't happen because its parameter has already
been range checked and it is guaranteed that the result never exceeds
255. Remove the unnecessary value check.
Cc: Michael Shych <michaelsh@nvidia.com>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Don't fail the probe function and don't deassert the reset controller if
a "reset" property doesn't exist in the device tree.
Change is done for SOCs that don't support a reset controller.
Signed-off-by: Eliav Farber <farbere@amazon.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Link: https://lore.kernel.org/r/20220908152449.35457-10-farbere@amazon.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Extend aquacomputer_d5next driver to expose various hardware
sensors of the Aquacomputer High Flow Next flow sensor, which
communicates through a proprietary USB HID protocol.
The High Flow Next exposes +5V voltages, water quality, conductivity
and flow readings. A temperature sensor can be connected to it, in
which case it provides its reading and an estimation of the
dissipated/absorbed power in the liquid cooling loop.
Additionally, serial number and firmware version are exposed through
debugfs.
Registry offsets were discovered and tested by users on Github [1] [2].
[1] https://github.com/aleksamagicka/aquacomputer_d5next-hwmon/issues/8
[2] https://github.com/aleksamagicka/aquacomputer_d5next-hwmon/pull/34
Signed-off-by: Aleksa Savic <savicaleksa83@gmail.com>
Link: https://lore.kernel.org/r/20220907100739.806571-1-savicaleksa83@gmail.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
On systems with more than one tps23861, creating the debugfs directory
for additional devices fails with
debugfs: Directory 'tps23861' with parent '/' already present!
To resolve this, include the hwmon device name in the directory name.
Since the name is unique, this guarantees that the debugfs directory
is unique.
Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-by: Guenter Roeck <linux@roeck-us.net>
Link: https://lore.kernel.org/r/20220907015405.16547-2-mr.nuke.me@gmail.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
When reading the 'port_status' debugfs entry, some I2C registers were
read more than once. This looks inefficient in an I2C trace.
To reduce I2C traffic, update tps23861_port_status_show() to only read
each register once. Indexing the port number from 0 instead of 1 also
allows simplifying things a bit.
Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Link: https://lore.kernel.org/r/20220907015405.16547-1-mr.nuke.me@gmail.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Add driver for Microchip EMC2301/2/3/5 RPM-based PWM Fan Speed Controller.
Modify Makefile and Kconfig to support Microchip EMC2305 RPM-based
PWM Fan Speed Controller.
Signed-off-by: Michael Shych <michaelsh@nvidia.com>
Reviewed-by: Vadim Pasternak <vadimp@nvidia.com>
Link: https://lore.kernel.org/r/20220810171552.56417-3-michaelsh@nvidia.com
[groeck: Drop unnecessary () around DIV_ROUND_CLOSEST()]
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
The MAX31790 has a tach input enable bit in each fan's configuration
register. This is only enabled by the driver if RPM mode is selected,
but the driver doesn't provide a way to independently enable tachometer
input regardless of the regulator mode.
By adding the fanN_enable sysfs files, we can decouple the tach input
from the regulator mode. Also update the documentation.
Signed-off-by: Justin Ledford <justinledford@google.com>
Link: https://lore.kernel.org/r/20220829195930.2521755-1-justinledford@google.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
There is no need to call OF specific devm_of_pwm_get() since
the device node parameter duplicates in the device parameter.
Hence we may safely replace it by plain devm_pwm_get() call.
This allows to drop devm_of_pwm_get() as no more users will be.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Link: https://lore.kernel.org/r/20220826172642.16404-1-andriy.shevchenko@linux.intel.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
The devm_clk_get_enabled() helper:
- calls devm_clk_get()
- calls clk_prepare_enable() and registers what is needed in order to
call clk_disable_unprepare() when needed, as a managed resource.
This simplifies the code, the error handling paths and avoid the need of
a dedicated function used with devm_add_action_or_reset().
Based on my test with allyesconfig, this reduces the .o size from:
text data bss dec hex filename
2419 1472 0 3891 f33 drivers/hwmon/sparx5-temp.o
down to:
2155 1472 0 3627 e2b drivers/hwmon/sparx5-temp.o
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Link: https://lore.kernel.org/r/cfe4c965074b5ecbe03830b05e038b4594c7b970.1661336689.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
When dell-smm-hwmon is loaded on a machine with a buggy BIOS
with the option "force" being enabled, it wrongly prints
that the buggy features where disabled. This may cause
users to wrongly assume that the driver still protects them
from these BIOS bugs even with "force" being enabled.
Replace the messages with two messages each which are depending
on the value of the "force" parameter. The messages which are
being printed when "force" is not set use dev_notice() instead
of dev_warn() since they only serve as a notice.
Tested on a Dell Inspiron 3505.
Signed-off-by: Armin Wolf <W_Armin@gmx.de>
Link: https://lore.kernel.org/r/20220822174053.8750-3-W_Armin@gmx.de
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Convert the module to be property provider agnostic and allow
it to be used on non-OF platforms.
Include mod_devicetable.h explicitly to replace the dropped of.h
which included mod_devicetable.h indirectly.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Link: https://lore.kernel.org/r/20220826173700.17395-1-andriy.shevchenko@linux.intel.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
