For TI power management chip TPS40422, READ_TEMPERATURE_2 command is supported on
page 1 of the chip, but the original driver(pmbus.c) only tried to detect this command
on page 0, this will lead to a result that the temperature sensor in page 1 couldn't
be detected. This change is to isolate the tps40422 driver from pmbus.c into a solo
front-end driver.
Signed-off-by: NZhu Laiwen <richard.zhu@nsn.com>
[Guenter Roeck: Dropped unnecessary license text (fixes checkpatch warning)]
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

The chip's programming interface is quite similar to LTC3880
and supports the same set of sensors.
Reviewed-by: NRobert Coulson <rob.coulson@gmail.com>
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

LTC2974 datasheet revision C lists the chip ID for LTC2974 as 0x0213.
Reviewed-by: NRobert Coulson <rob.coulson@gmail.com>
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

Some chips use different exponents for sensors on different pages
or rails. Detect and store exponent per page to support this situation.

This fixes a problem with wrong voltages seen on UCD90120.
Reported-by: NSoren Brinkmann <soren.brinkmann@xilinx.com>
Tested-by: NSoren Brinkmann <soren.brinkmann@xilinx.com>
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

Detect LTC2978A chip ID. Treat it as LC2978.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

LTC2977 is a pin compatible replacement for LTC2978.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

Signed-off-by: NLABBE Corentin <clabbe.montjoie@gmail.com>
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

pmbus code currently crashes the kernel if it detects an internal
implementation error. While the detected condition suggests that there
is a bug in the code, it is hardly fatal. Therefore, it should not
trigger a crash. Replace BUG() with WARN().
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

Use the wrapper function for retrieving the platform data instead of
accessing dev->platform_data directly.
Signed-off-by: NJingoo Han <jg1.han@samsung.com>
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

Initialize device specific coefficients from table instead of hard-coding it
to simplify adding additional chips.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

Limits on all supported sensors and chips have to be within 0..0x0fff,
and limits are always positive.

Clamp written values in chip driver. Also clear value cache to ensure
that the actually written value is read back and reported correctly.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

So far the driver reported the voltage on VAUX as "vout2". This was not
entirely appropriate as it is not an output voltage, and complicates
the code. Use the new virtual "VMON" register set and report the voltage
as "vmon" instead.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

Use u16 instead of int to store cached limit attributes.
This reduces allocated data size per driver instance by 48 bytes.
Use defines for the number of pages supported by individual chips.
Use ARRAY_SIZE for loops to initialize array variables, and
initialize all variables in the same code block.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

On LTC2978, only READ_TEMPERATURE is supported. It reports
the internal junction temperature. This register is unpaged.

On LTC3880, READ_TEMPERATURE and READ_TEMPERATURE2 are supported.
READ_TEMPERATURE is paged and reports external temperatures.
READ_TEMPERATURE2 is unpaged and reports the internal junction
temperature.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>
Cc: stable@vger.kernel.org # 3.2+
Acked-by: NJean Delvare <khali@linux-fr.org>

If krealloc() returns NULL, it *doesn't* free the original. So any code
of the form 'foo = krealloc(foo, …);' is almost certainly a bug.
Signed-off-by: NDavid Woodhouse <David.Woodhouse@intel.com>
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

We read the chip ID from the chip, use it to determine if the chip ID provided
to the driver is correct, and report it if wrong. We should also use the
correct chip ID to select supported functionality.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>
Cc: stable@vger.kernel.org # 3.2+
Acked-by: NJean Delvare <khali@linux-fr.org>

Peak attributes were not initialized and cleared correctly.
Also, temp2_max is only supported on page 0 and thus does not need to be
an array.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>
Cc: stable@vger.kernel.org # 3.2+
Acked-by: NJean Delvare <khali@linux-fr.org>

Rearranged some data structures, and merged some common functions.
Overall code and data size reduction by more than 900 bytes.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

MAX34460 is a PMBus 12-Channel Voltage Monitor & Sequencer.
MAX34461 is a PMBus 16-Channel Voltage Monitor & Sequencer.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

Not all PMBus devices support the byte status register at 0x78.
Try to use the word status register at 0x79 instead if that is the case.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

Some of the ZL6100 compatible chips support monitoring a separate voltage pin,
VMON (ZL2004) or VDRV (ZL91xx). Report it as in2 / vmon.

The chips support implicit warning limits for VMON/VDRV, as percentage of the
respective critical voltage. Support by reading/writing the critical voltages
and calculating the associated warning voltages.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

For PMBus chips, modifying one limit register may affect other limits.
Since limits are all cached in the PMBus core driver, related changes
are not reflected in reported limits.

Introduce function to clear the attribute cache. After calling this function,
the core pmbus driver re-reads all cached values.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

Some PMBus chips support monitoring an additional non-standard voltage. While
this voltage can in many cases be supported by simulating an additional sensor
page, this does not work in all cases. Specifically, it is problematic if the
data format is linear and the voltage is reported in LINEAR11 format. Since
output voltages use LINEAR16, and the exponent for LINEAR16 data is chip-wide
and fixed, this can result in overflows.

To solve this problem, add support for an additional virtual input voltage,
call it 'vmon', and treat this voltage as input voltage (which, when the chip
supports linear data format, uses LINEAR11).
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

So far, attribute memory was allocated by pre-calculating the maximum possible
amount of attributes. Not only does this waste memory, it is also risky because
the calculation might be wrong. It also requires a lot of defines to specify
the maximum number of attributes per class.

Allocate attribute memory using krealloc() instead. That means we have to use
kfree(), since devm_krealloc() does not exist, but that is still less costly
and less risky than trying to predict the number of attributes at the beginning.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

Since memory is now allocated with dev_ functions, we no longer need to keep
track of allocated memory. Sensor memory allocation can therefore be
simplified significantly.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

Boolean handling depends on storing the sensor data index in sensor_device_attr
as part of the index variable. This limits the number of sensor attributes to
256, and means the sensor sequence number actually has to be maintained to be
able to access sensor data from boolean functions.

Rework the code to store sensor pointers in the pmbus_boolean data structure
directly. With this approach, the number of supportable sensors is now
unlimited.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

Since memory is now allocated with dev_ functions, we no longer need to keep
track of allocated memory. Memory allocation for booleans and labels can
therefore be simplified substantially by allocating it only as needed.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

This simplifies the code and makes it a bit smaller.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

Fix:
ERROR: Macros with multiple statements should be enclosed in a do - while loop

by unwinding the problematic macros.

As a side effect, this patch reduces code size on x86_64 by 160 bytes and bss
size by 64 bytes.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

Drop error messages due to implementation errors and due to memory allocation
errors.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

SENSORS_LIMIT and the generic clamp_val have the same functionality,
and clamp_val is more efficient.

This patch reduces text size by 9052 bytes and bss size by 11624 bytes
for x86_64 builds.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>
Acked-by: NGeorge Joseph <george.joseph@fairview5.com>
Acked-by: NJean Delvare <khali@linux-fr.org>

This config item has not carried much meaning for a while now and is
almost always enabled by default. As agreed during the Linux kernel
summit, remove it.

CC: Guenter Roeck <linux@roeck-us.net>
CC: Jean Delvare <khali@linux-fr.org>
Signed-off-by: NKees Cook <keescook@chromium.org>
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>

These drivers use no sleep or delay functions so they don't need to
include <linux/delay.h>.
Signed-off-by: NJean Delvare <khali@linux-fr.org>
Acked-by: NGuenter Roeck <linux@roeck-us.net>
Cc: Rudolf Marek <r.marek@assembler.cz>

Many hwmon drivers use jiffies but omit the inclusion of the header
file. Fix that, and also fix one driver which was including the header
file but didn't need it.
Signed-off-by: NJean Delvare <khali@linux-fr.org>
Acked-by: NGuenter Roeck <linux@roeck-us.net>
Acked-by: NLuca Tettamanti <kronos.it@gmail.com>
Cc: Marc Hulsman <m.hulsman@tudelft.nl>
Cc: Rudolf Marek <r.marek@assembler.cz>

Some configurations produce the following compiler warning:

drivers/hwmon/pmbus/pmbus_core.c: In function 'pmbus_show_boolean':
drivers/hwmon/pmbus/pmbus_core.c:752: warning: 'val' may be used uninitialized in this function

While this is a false positive, it can easily be fixed by overloading the return
value from pmbus_get_boolean with both val and error return code (val is a
boolean and thus never negative).
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>
Reviewed-by: NRobert Coulson <robert.coulson@ericsson.com>

ZL9101M and ZL9117M are compatible to ZL6100. Add support to the zl6100 driver.
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>
Acked-by: NJean Delvare <khali@linux-fr.org>