…t64-for-4.15' into sunxi/for-next

…xi/for-next

…nxi/for-next

…i/for-next

…isc-next' into sunxi-next

…i/for-next

…i/for-next

…xi/for-next

We should abort probe if we cannot talk to the display. This
prevents creating a fb device which does not work and prevents
a lot of i2c communication error messages as a consequence.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

For autoprobing to work on devicetree based systems, the compatible
name is used when probing i2c devices. Because the device id is now not
the same, the ssd1307 does not get autoprobed.

By renaming the device id, autoprobing is now functional on devicetree
based systems.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

The ssd1306 driver is a little inconsistent in the debug messages.
Common is to not capitalize the first letter and to not add any periods
etc at the end. Also add a quick shortcut to access the device
structure shortening our error messages.

Finally take this cleanup opportunity to add the missing device.h
header.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

When we fail to send messages to our display, we print an error.
However this 'spam' can continue heavily and add a heavy load on the
system just to print these messages.

Lets use the ratelimited variant of dev_err to lessen the system load.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

Currently, the driver always successfully manages to load,
even if there is no actual chip present (or communication
is not possible over i2c).

This has the side-effect, that the node is always available in
sysfs, and writing data to it which results in i2c errors.

To improve behavior here, we try to read the first byte from
the device, and if this was okay, we at least know there is some
device on the bus, even though we do not know which one.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

Re-order headers so they are in alphabetical order.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

This patch adds some more defines so that the driver can receive
a little more future work. These new defines are then used throughout the
existing code the remove some magic values.

This patch does not produce any binary changes.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

Currently, we blast sleep mode or 0 into the mode1 register, without
taking notice of whatever was in it before.

Lets be a bit more proper here and read the register and modify only
the bit which we are concerned.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

Let's improve readability a little bit by reordering initial LED output
configuration.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

This patch does some whitespace fixing to make the entire driver more
consistent, especially with regards to alignment. Because of this it
also reduces quite some of the checkpatch warnings.

This did came at the cost of some minor 80 char warnings however to satisfy
the alignment and improve readability.

This patch does not introduce any binary changes.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

The PCA963x series of chips have several operating modes controlled via
the LEDOUT register. LDR controls whether the leds are completely off
(0x0), completely on (0x1), only controlled via the PWM registers (0x2)
or a special 'grouped' mode (0x3).

In the grouped mode, the leds can either blink, or PWM in
synchronization. Since the LED framework currently does not support
linked LEDs and thus we cannot PWM in synchronization, set the default group
to blink, to ensure that hardware based blinking at least is synchronized
when blinking.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

Add the bitops.h header as we need it, alphabetize header order.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

The AXP20X_OFF define is an actual specific bit, define it as such.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

The axp20x_usb_power driver uses BIT() operations but lacks the include
for it. Include the bitops.h header file.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

Make use of the recommended BIT() macro for bit defines.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

The AXP20X driver currently has several masks defined throughout the
code. Use nice defines to make them clean and more descriptive.

Additionally include bitops.h which was missing before.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

The current axp20x names the ramping register 'scal' which probably
means scaling. Since the register really has nothing to do with
scaling, but really is the voltage ramp we rename it appropriately.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

When a regulator is marked as always on, it is enabled early on, when
checking and setting up constraints. It makes the assumption that the
bootloader properly initialized the regulator, and just in case enables
the regulator anyway.

Some constraints however currently get missed, such as the soft-start
and ramp-delay. This causes the regulator to be enabled, without the
soft-start and ramp-delay being applied, which in turn can cause
high-currents or other start-up problems.

By moving the always-enabled constraints later in the constraints check,
we can at least ensure all constraints for the regulator are followed.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

The AXP209 supports ramping up voltages on several regulators such as
DCDC2 and LDO3.

This patch adds preliminary support for the regulator-ramp-delay
property these 2 regulators. Note that the voltage ramp only works when
the regulator is already enabled. E.g. when going from say 0.7 V to 3.6 V.

When turning on the regulator, no voltage ramp is performed by the hardware.

What this means, is that if the bootloader brings up the voltage at 0.7 V,
the ramp delay property is properly applied. If however, the bootloader
leaves the power off, no ramp delay is applied when the power is
enabled by the regulator framework.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

In the past, there have been words on various lists that if LDO3 is
disabled in u-boot, but enabled in the DTS, the axp driver would fail
to continue/hang. Several enable/disable patches have been issues to
devicetree's in both the kernel and u-boot to address this issue.

What really happened however, was that the AXP shuts down, without
notice, without setting an interrupt. This is caused when LDO3 gets
overloaded, for example with large capacitors on the LDO3 output.

Normally, we would expect that the AXP would source 200 mA as per
datasheet and set and trigger an interrupt when being overloaded. For
some reason however, this does not happen.

As a work-around, we use the soft-start constraint of the regulator node
to first bring up the LDO3 to the lowest possible voltage and then
enable the LDO. After that, we can set the requested voltage as per
normal.

Combining this setting with the regulator-ramp-delay allows LDO3 to
come up slowly and staggered, potentially reducing overall inrush current.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

…ime2

The OLinuXino Lime2 has a big capacitor on its LDO3 output. It is
actually too large, causing the PMIC to shutdown when toggling the LDO3.

By enabling soft-start and ramp delay we increase the time for the
capacitor to charge lowering the current drain on the power regulator.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

With commit b43776d ("ARM: dts: sunxi: Use axp209.dtsi for
Olinuxino Lime2") we force them an arbitrary 2.8 volts. Granted, for
LDO3 this may be less arbitrary, but for LDO4 this is just wrong.

In the defense of LDO3, LDO3 is the regulator that feeds port bank E,
which has no other purpose then a CSI/TS interface, however the case
may still be, that the connected IO may be just as well be 3.3 volts.
The big misnomer is however, that the schematic names GPIO-2 pin4
LDO3_2.8V, rather then VDD-CSI0 or similar.

This is much worse for LDO4 however, which is not referenced on any
pin, is now set to 2.8 volts, but port bank G can also support various
other peripherals such as UARTS etc.

By having 2.8 volts however for LDO4, we thus now have peripherals that
no longer function properly all of the time.

Ideally, we want to set a supply voltage for each port bank, but the
monolithic nature of the sunxi pinctroller currently prevents this and
as such, the board should at least configure the LDO4 with the proper
ranges.

Until we can set the consumer at the port bank level, a child
device-tree has to do something like:

&reg_ldo4 {
    regulator-min-microvolt = <3300000>;
    regulator-max-microvolt = <3300000>;
};

While doing this the same way results in the same solution currently,
we force the hack into the final devicetree rather then having it wrong
at the board level.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

The lradc's analog reference voltage is set to 3.0 volt in the
hardware. This is more or less set in stone for atleast lradc0. Set the
property in the dts to ensure the lradc is referenced properly.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

The old pwm_config call, did some parameter sanitation such as
preventing to change the pwm config if the period is 0.

With the new atomic framework, we now not only apply configuration
parameters, but also the polarity setting.

The polarity setting however can be applied both with an enabled or
disabled PWM, in other words, setting the polarity is perfectly
acceptable when the period is 0.

This patch removes the check if the period is valid and instead, sets
the state disabled if the period is 0. It now thus becomes possible to
set the polarity, before changing any other PWM parameters.

The reason this is needed is that if someone calls pwm_set_polarity
with the period set to 0, pwm_set_polarity would fail. Having period of 0
is however perfectly acceptable here.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

Fix alignment whitespace, no code changes where performed.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

To clean up some whitespace issues and make room for future additions
some variable names where shortened.

This patch does not invoke any code changes.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

The parameter touchscreen-size-x sits oddly next to
touchscreen-max-pressure for example. Further more when considering
that the actual parameter modified in absinfo is the 'maximum'.

Let us thus rename the touchscreen-max-size-[xy] property in the
devicetree. The new name takes precedence over the earlier
touchscreen-size-[xy] in case both are found.
The former is now deprecated.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

In certain configurations, touchscreens are bigger then the underlying
display. This can happen for various reasons, but an interesting design
choice is to improve the edge detection of touch events. To make sure
we don't have strange offsets we want to pass this information along to
the input framework via the of_touchscreen layer. To this, we need to
also parse the 'minimum' property.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

The current ft5x06 driver happily supports the ft5426 controller chip.
Lets add a compatible for it.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

Alphabetize and add missing headers to the edt-ft5x06 driver. No code
changes where done in this commit.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

The input framework has callbacks for when an input device gets
opened/closed to do certain actions. Use these to enable/disable
interrupts when they are not in use. This can then be improved in the
future by doing power savings when the device is not in use.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

This patch just cleans up some ident/whitespacing issues. No code
changes where performed.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

The struct device is often referenced in the probe function via
&client->dev. We can shorten this by introducing a compile time removed
pointer to be able to just type 'dev'. This reduces some of our longer
function line lengths improving readability just a little bit.

While touching said lines, (mostly dev_*() print strings) make the
strings more consistent (with itself and the kernel output) by dropping
caps at the start and periods at the end having at least a concistent
style within the same function.

No logical changes where performed.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

Drop the long _REGISTER_ string in the M09_ macro's. It makes our
register defines really long and doesn't add anything. By shortening
the defines we can introduce more macro's that are more descriptive and
potentially longer.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

We can simply add some more defines to remove some magic values and
thus create more readable code.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

Currently when we get a deferral request on an optional GPIO, we print
out a failure message making the user think something bad has happened.

Instead, as commonly done, silence the -EPROBE_DEFER message as it is an
acceptable 'error'.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

It may happen, that tsdata or tsdata->input are NULL. One observed
example was that when requesting the IRQ, but not having all structures
setup properly. Add a simple sanity check for NULL, just in case.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

Multi-touch handling is currently incorrectly handled by the edt-ft5x06
driver.

According to Documentation/input/multi-touch-protocol.rst, we are
supposed to:
   ABS_MT_SLOT 0
   ABS_MT_TRACKING_ID 45
   ABS_MT_POSITION_X x[0]
   ABS_MT_POSITION_Y y[0]
   ABS_MT_SLOT 1
   ABS_MT_TRACKING_ID 46
   ABS_MT_POSITION_X x[1]
   ABS_MT_POSITION_Y y[1]
   SYN_REPORT

and:

Here is the sequence after lifting the contact in slot 0::

   ABS_MT_TRACKING_ID -1
   SYN_REPORT

What happens in our driver however, is:
   ABS_MT_SLOT 45
   ABS_MT_POSITION_X x[0]
   ABS_MT_POSITION_Y y[0]
   ABS_MT_SLOT 46
   ABS_MT_POSITION_X x[1]
   ABS_MT_POSITION_Y y[1]
   SYN_REPORT

and for lifting nothing is done.

This is obviously wrong, as we are allocating a slot based on the
tracking id, not based on the input (finger) and worse, never
reset tracking id to -1 to 'free' them.

Refactor the ISR to follow the spec by allocating a slot based on finger
and set/clear the tracking ID once the finger is no longer down.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

It may happen that we miss a touch !down event, for example the
interrupt fires, but we are to late to read the correct data. When this
happens, we may wrongfully set/clear ABS_MT_TRACKING_ID to -1.

The M09 firmware actually tells us how many fingers are being touched.
By using that, we also thus know how many events to focus on and to
clear unused tracking id's. This also has as a boon that we save some
processing time by not evaluating the registers for unused touches.

Note: this should not be used to reduce the number of bytes read, e.g.
read the number of touches register, followed by a read for the actual
registers, as that would result in a non-atomic read, and our data
could be out of sync. E.g. we are told to 4 inputs have been pressed,
but by the time we actually read these registers, 5 inputs are actually
being touched.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

The edt-ft5x06 is not required to be used with an interrupt. We can
just as easily poll the edt-ft5x06 using the polldev framework. This is
needed as some electrical designs may omit the IRQ line.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

Move the read and write functions higher and next to the read-write
function. This makes adding features easier without introducing
prototypes and keeps the relevant functions together.

This patch introduces no code changes.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

With the M09, the touchscreen enters sleep mode after a few seconds.
During this time the controller responds to all i2c commands normally,
but no input data is generated.

This patch forces active mode during each poll interval, to prevent
sleep mode (monitor mode) to be activated during normal operation.

This can be further improved by setting/clearing the bit when going
into suspend/resume mode. Some extra logic would be needed there
however to only go into suspend mode if a reset/wake line is available.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

Not sure what TOCUH is, but i'm fairly certain 'TOUCH' was meant.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

The PCA9570 is the most basic of the pca957x series.

It is a 4 bit i2c gpio expander without an interrupt line.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

Commit 2ad6f30 ("ASoC: sun4i-i2s: Add quirks to handle a31 compatible")
added support for reset control but support for sun4i was overlooked. For
these older SoC's it is quite common not have this reset control.

The reset control is first requested, and later checked if it IS_ERR.
If however there is no reset control configured, we can get NULL and
thus the checks pass with a NULL pointer passed the functions
afterwards.

Lets use IS_ERR_OR_NULL to ensure the reset control is only
(de)asserted when we actually have a valid reset control.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

Some very minor code cleanups, such as including the bitops header for
DW_UART_MCR_SIRE, use the BIT() macro as suggested by checkpatch and
removed some white spacing.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

It seems that at some point, someone made the assumption that the UART
Interrupt ID Register was a bitfield and started to check if certain
bits where set.

Actually however the register contains interrupt ID's where only the MSB
seems to be used singular and the rest share at least one bit. Thus
doing bitfield operations is wrong.

This patch cleans up the serial_reg include file by ordering it and
replacing the UART_IIR_ID 'mask' with a proper mask for the register.
The OMAP uart appears to have used the two commonly 'reserved' bits 4
and 5 and thus get an UART_IIR_EXT_MASK for these two bits.

This patch then goes over all UART_IIR_* users and changes the code from
bitfield checking, to ID checking instead.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

The tegra serial IP seems to be following the common layout and the
interrupt ID's match up nicely. Replace the magic values to match the
common serial_reg defines, with the addition of the Tegra unique End of
Data interrupt.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

The ddc items can be consolidated more into one place.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

No codechanges where done with this change, just some whitespace
cleanup.

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>

Signed-off-by: Olliver Schinagl <oliver@schinagl.nl>