Some GPIO chips (e.g. the DLN2 USB adapter) have blocking get/set
operation but do not need a threaded irq handler.
Signed-off-by: NOctavian Purdila <octavian.purdila@intel.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Signed-off-by: NAlexander Shiyan <shc_work@mail.ru>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

This avoids handling gpiochip remove error in device
remove handler.
Signed-off-by: NAbdoulaye Berthe <berthe.ab@gmail.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

The current prototype of gpiochip_request_own_desc() requires to obtain
a pointer to a descriptor. This is in contradiction to all other GPIO
request schemes, and imposes an extra step of obtaining a descriptor to
drivers. Most drivers actually cannot even perform that step since the
function that does it (gpichip_get_desc()) is gpiolib-private.

Change gpiochip_request_own_desc() to return a descriptor from a
(chip, hwnum) tuple and update users of this function (currently
gpiolib-acpi only).
Signed-off-by: NAlexandre Courbot <acourbot@nvidia.com>
Tested-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

The huge majority of GPIOs have their direction and initial value set
right after being obtained by one of the gpiod_get() functions. The
integer GPIO API had gpio_request_one() that took a convenience flags
parameter allowing to specify an direction and value applied to the
returned GPIO. This feature greatly simplifies client code and ensures
errors are always handled properly.

A similar feature has been requested for the gpiod API. Since setting
the direction of a GPIO is so often the very next action done after
obtaining its descriptor, we prefer to extend the existing functions
instead of introducing new functions that would raise the
number of gpiod getters to 16 (!).

The drawback of this approach is that all gpiod clients need to be
updated. To limit the pain, temporary macros are introduced that allow
gpiod_get*() to be called with or without the extra flags argument. They
will be removed once all consumer code has been updated.
Signed-off-by: NAlexandre Courbot <acourbot@nvidia.com>
Reviewed-by: NMark Brown <broonie@linaro.org>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Since now we have irqchip helpers that the GPIO chip drivers are supposed
to use if possible, we can move the registration of ACPI events to happen
in these helpers. This seems to be more natural place and might encourage
GPIO chip driver writers to take advantage of the irqchip helpers.

We make the functions available to GPIO chip drivers via private gpiolib.h,
just in case generic irqchip helpers are not suitable.
Signed-off-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

As per example from the regulator subsystem: put all defines and
functions related to registering board info for GPIO descriptors
into a separate <linux/gpio/machine.h> header.

Cc: Andrew Victor <linux@maxim.org.za>
Cc: Nicolas Ferre <nicolas.ferre@atmel.com>
Cc: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Thierry Reding <thierry.reding@gmail.com>
Acked-by: NStephen Warren <swarren@wwwdotorg.org>
Reviewed-by: NAlexandre Courbot <gnurou@gmail.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Both functions were introduced to let gpio drivers request their own
gpio pins. Without exporting the functions, this can however only be
used by gpio drivers built into the kernel.

Secondary impact is that the functions can not currently be used by
platform initialization code associated with the gpio-pca953x driver.
This code permits auto-export of gpio pins through platform data, but
if this functionality is used, the module can no longer be unloaded due
to the problem solved with the introduction of gpiochip_request_own_desc
and gpiochip_free_own_desc.

Export both function so they can be used from modules and from
platform initialization code.
Reviewed-by: NAlexandre Courbot <acourbot@nvidia.com>
Reviewed-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

gpio_ensure_requested() only makes sense when using the integer-based
GPIO API, so make sure it is called from there instead of the gpiod
API which we know cannot be called with a non-requested GPIO anyway.

The uses of gpio_ensure_requested() in the gpiod API were kind of
out-of-place anyway, so putting them in gpio-legacy.c helps clearing the
code.

Actually, considering the time this ensure_requested mechanism has been
around, maybe we should just turn this patch into "remove
gpio_ensure_requested()" if we know for sure that no user depend on it
anymore?
Signed-off-by: NAlexandre Courbot <acourbot@nvidia.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

gpio_lock/unlock_as_irq() are working with (chip, offset) arguments and
are thus not using the old integer namespace. Therefore, there is no
reason to have gpiod variants of these functions working with
descriptors, especially since the (chip, offset) tuple is more suitable
to the users of these functions (GPIO drivers, whereas GPIO descriptors
are targeted at GPIO consumers).
Signed-off-by: NAlexandre Courbot <acourbot@nvidia.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

As GPIO descriptors are not going to remain unique anymore, having this
function public is not safe. Restrain its use to gpiolib since we have
no user outside of it.
Signed-off-by: NAlexandre Courbot <acourbot@nvidia.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

The old integer GPIO interface is, in effect, a privileged user of the
gpiod interface. Reflect this fact further by moving legacy GPIO support
into its own source file. This makes the code clearer and will allow us
to disable legacy GPIO support in the (far) future.
Signed-off-by: NAlexandre Courbot <acourbot@nvidia.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

sysfs support is currently entangled within the core GPIO support, while
it should relly just be a (privileged) user of the integer GPIO API.
This patch is a first step towards making the gpiolib code more readable
by splitting it into logical parts.

Move all sysfs support to their own source file, and share static
members of gpiolib that need to be in the private gpiolib.h file. In
the future we will want to put some of them back into gpiolib.c, but this
first patch let us at least identify them.
Signed-off-by: NAlexandre Courbot <acourbot@nvidia.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Compiling out GPIO labels results in a space gain so small that it can
hardly be justified. Labels can also be useful for printing debug
messages, so always keep them around.
Signed-off-by: NAlexandre Courbot <acourbot@nvidia.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Commit dd34c37a (gpio: of: Allow -gpio suffix for property names)
added parsing for both -gpio and -gpios suffix but also changed
the handling for deferred probe unintentionally. Because of the
looping the second name will now return -ENOENT instead of
-EPROBE_DEFER. Fix the issue by breaking out of the loop if
-EPROBE_DEFER is encountered.
Signed-off-by: NTony Lindgren <tony@atomide.com>
Reviewed-by: NThierry Reding <treding@nvidia.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Introduce gpiod_get_optional() and gpiod_get_index_optional() helpers
that make it easier for drivers to handle optional GPIOs.

Currently in order to handle optional GPIOs, a driver needs to special
case error handling for -ENOENT, such as this:

	gpio = gpiod_get(dev, "foo");
	if (IS_ERR(gpio)) {
		if (PTR_ERR(gpio) != -ENOENT)
			return PTR_ERR(gpio);

		gpio = NULL;
	}

	if (gpio) {
		/* set up GPIO */
	}

With these new helpers the above is reduced to:

	gpio = gpiod_get_optional(dev, "foo");
	if (IS_ERR(gpio))
		return PTR_ERR(gpio);

	if (gpio) {
		/* set up GPIO */
	}

While at it, device-managed variants of these functions are also
provided.
Signed-off-by: NThierry Reding <treding@nvidia.com>
Reviewed-by: NAlexandre Courbot <acourbot@nvidia.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

The function is called gpiod_get(), not gpio_get(). Fix the kernel-doc
comment to match the name.
Signed-off-by: NThierry Reding <treding@nvidia.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Many bindings use the -gpio suffix in property names. Support this in
addition to the -gpios suffix when requesting GPIOs using the new
descriptor-based API.
Signed-off-by: NThierry Reding <treding@nvidia.com>
Acked-by: NAlexandre Courbot <acourbot@nvidia.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

of_find_gpio() is always called under an IS_ENABLED(CONFIG_OF), so the
dummy implementation provided for !OF configurations is not needed.
Signed-off-by: NThierry Reding <treding@nvidia.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Some GPIO irqchip drivers exploit the irqdomain mapping
function to set up the IRQ default type in the hardware,
make sure that if we pass IRQ_TYPE_NONE, no hardware setup
whatsoever takes place (this should be the norm) until
later when the IRQ gets utilized.

Cc: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Cc: Santosh Shilimkar <santosh.shilimkar@ti.com>
Cc: linux-omap <linux-omap@vger.kernel.org>
Tested-by: NPeter Ujfalusi <peter.ujfalusi@ti.com>
Tested-by: NJavier Martinez Canillas <javier@dowhile0.org>
Tested-by: NNishanth Menon <nm@ti.com>
Acked-by: NNishanth Menon <nm@ti.com>
Tested-by: NTony Lindgren <tony@atomide.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Another feature that is duplicated in a number of GPIO irqchips
is that these cascades IRQs are assigned their own lock class
so as to avoid warnings about lockdep recursions. Do this also
in the generic GPIO irqchip helpers for smooth transition to
this core infrastructure.

Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Some off-chip GPIO expanders need to be communicated by I2C or
SPI traffic, but may still support IRQs. By the sleeping nature
of such buses, such IRQ handlers need to be threaded. Support
such handlers in the gpiochip irqchip helpers by flagging IRQs
as threaded if the .can_sleep property of the gpiochip is
true.

Helpfully deny registration of chained IRQ handlers if the
.can_sleep property is set, as such chips will invariably need
a nested handler rather than a chained handler.

Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

During irq mapping, in irq_set_chip_and_handler() the process
of setting this up may incur calls to lock the irqchip, which
in turn may need to dereference and use the chip data. So set
the data first, then set the chip and handler.
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

When using the irqchip helper inside the gpiolib, make sure
the IRQs are unmapped/disposed before the irqdomain is removed
as part of removing the gpiochip.
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

This provides a function gpiochip_irqchip_add() to set
up an irqchip for a GPIO controller, and a function
gpiochip_set_chained_irqchip() to chain it to a parent
irqchip.

Most GPIOs are of the type where a number of lines form
a cascaded interrupt controller chained onto
the primary system interrupt controller (or further down the
chain) so let's add this helper and factor the code to
request the lines to be used as IRQs, the .to_irq() function
and the irqdomain into the core as well.
Acked-by: NThomas Gleixner <tglx@linutronix.de>
Acked-by: NAlexandre Courbot <acourbot@nvidia.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Nothing prevents GPIO drivers from returning values outside the
boolean range, and as it turns out a few drivers are actually doing so.
These values were passed as-is to unsuspecting consumers and created
confusion.

This patch makes the internal _gpiod_get_raw_value() function return a
bool, effectively clamping the GPIO value to the boolean range no
matter what the driver does.

While we are at it, we also change the value parameter of
_gpiod_set_raw_value() to bool type before drivers start doing funny
things with it as well.

Another way to fix this would be to change the prototypes of the driver
interface to use bool directly, but this would require a huge
cross-systems patch so this simpler solution is preferred.

Changes since v1:
- Change local variable type to bool as well, use boolean values in
  code
- Also change prototype of open drain/open source setting functions
  since they are only called from _gpiod_set_raw_value()

This probably calls for a larger booleanization of gpiolib, but let's
keep that for a latter change - right now we need to address the issue
of non-boolean values returned by drivers.
Signed-off-by: NAlexandre Courbot <acourbot@nvidia.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

The current ACPI GPIO event handling code was never tested against real
hardware with functioning GPIO triggered events (at the time such hardware
wasn't available). Thus it misses certain things like requesting the GPIOs
properly, passing correct flags to the interrupt handler and so on.

This patch reworks ACPI GPIO event handling so that we:

 1) Use struct acpi_gpio_event for all GPIO signaled events.
 2) Switch to use GPIO descriptor API and request GPIOs by calling
    gpiochip_request_own_desc() that we added in a previous patch.
 3) Pass proper flags from ACPI GPIO resource to request_threaded_irq().

Also instead of open-coding the _AEI iteration loop we can use
acpi_walk_resources(). This simplifies the code a bit and fixes memory leak
that was caused by missing kfree() for buffer returned by
acpi_get_event_resources().

Since the remove path now calls gpiochip_free_own_desc() which takes GPIO
spinlock we need to call acpi_gpiochip_remove() outside of that lock
(analogous to acpi_gpiochip_add() path where the lock is released before
those funtions are called).
Signed-off-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Reviewed-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Sometimes it is useful to allow GPIO chips themselves to request GPIOs they
own through gpiolib API. One use case is ACPI ASL code that should be able
to toggle GPIOs through GPIO operation regions.

We can't use gpio_request() because it will pin the module to the kernel
forever (it calls try_module_get()). To solve this we move module refcount
manipulation to gpiod_request() and let __gpiod_request() handle the actual
request. This changes the sequence a bit as now try_module_get() is called
outside of gpio_lock (I think this is safe, try_module_get() handles
serialization it needs already).

Then we provide gpiolib internal functions gpiochip_request/free_own_desc()
that do the same as gpio_request() but don't manipulate module refrence
count. This allows the GPIO chip driver to request and free descriptors it
owns without being pinned to the kernel forever.
Signed-off-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Reviewed-by: NAlexandre Courbot <acourbot@nvidia.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

It is currently debated where the functions to lock a certain
GPIO line as used for IRQs should be called. Delete all
misleading documentation.
Reported-by: NThomas Gleixner <tglx@linutronix.de>
Cc: Jean-Jacques Hiblot <jjhiblot@traphandler.com>
Acked-by: NAlexandre Courbot <acourbot@nvidia.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Some drivers dealing with a gpio_chip might need to act on its
descriptors directly; one example is pinctrl drivers that need to lock a
GPIO for being used as IRQ using gpiod_lock_as_irq().

This patch exports a gpiochip_get_desc() function that returns the
GPIO descriptor at the requested index. It also sweeps the
gpio_to_chip() function out of the consumer interface since any holder
of a gpio_chip reference can manipulate its GPIOs way beyond what a
consumer should be allowed to do.

As a result, gpio_chip is not visible anymore to simple GPIO consumers.
Signed-off-by: NAlexandre Courbot <acourbot@nvidia.com>
Reviewed-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Reviewed-by: NJean-Jacques Hiblot <jjhiblot@traphandler.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Signed-off-by: NRahul Bedarkar <rahulbedarkar89@gmail.com>
Reviewed-by: NAlexandre Courbot <acourbot@nvidia.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

The documentation was not clear about whether
gpio_direction_output should take a logical value or the physical
level on the output line, i.e. whether the ACTIVE_LOW status
would be taken into account.

This converts gpiod_direction_output to use the logical level
and adds a new gpiod_direction_output_raw for the raw value.
Signed-off-by: NPhilipp Zabel <p.zabel@pengutronix.de>
Reviewed-by: NAlexandre Courbot <acourbot@nvidia.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Now that all users of acpi_gpio.h have been moved to use either the GPIO
descriptor interface or to the internal gpiolib.h we can get rid of
acpi_gpio.h entirely.

Once this is done the only interface to get GPIOs to drivers enumerated
from ACPI namespace is the descriptor based interface.
Signed-off-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Acked-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: NAlexandre Courbot <acourbot@nvidia.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Instead of asking each driver to register to ACPI events we can just call
acpi_gpiochip_register_interrupts() for each chip that has an ACPI handle.
The function checks chip->to_irq and if it is set to NULL (a GPIO driver
that doesn't do interrupts) the function does nothing.

We also add the a new header drivers/gpio/gpiolib.h that is used for
functions internal to gpiolib and add ACPI GPIO chip registering functions
to that header.

Once that is done we can remove call to acpi_gpiochip_register_interrupts()
from its only user, pinctrl-baytrail.c
Signed-off-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Acked-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Some devices drivers make use of optional GPIO parameters. For such
drivers, it is important to discriminate between the case where no
GPIO mapping has been defined for the function they are requesting, and
the case where a mapping exists but an error occured while resolving it
or when acquiring the GPIO.

This patch changes the family of gpiod_get() functions such that they
will return -ENOENT if and only if no GPIO mapping is defined for the
requested function. Other error codes are used when an actual error
occured during the GPIO resolution.
Signed-off-by: NAlexandre Courbot <acourbot@nvidia.com>
Reviewed-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

kernfs has just been separated out from sysfs and we're already in
full conflict mode.  Nothing can make the situation any worse.  Let's
take the chance to name things properly.

This patch performs the following renames.

* s/sysfs_elem_dir/kernfs_elem_dir/
* s/sysfs_elem_symlink/kernfs_elem_symlink/
* s/sysfs_elem_attr/kernfs_elem_file/
* s/sysfs_dirent/kernfs_node/
* s/sd/kn/ in kernfs proper
* s/parent_sd/parent/
* s/target_sd/target/
* s/dir_sd/parent/
* s/to_sysfs_dirent()/rb_to_kn()/
* misc renames of local vars when they conflict with the above

Because md, mic and gpio dig into sysfs details, this patch ends up
modifying them.  All are sysfs_dirent renames and trivial.  While we
can avoid these by introducing a dummy wrapping struct sysfs_dirent
around kernfs_node, given the limited usage outside kernfs and sysfs
proper, I don't think such workaround is called for.

This patch is strictly rename only and doesn't introduce any
functional difference.

- mic / gpio renames were missing.  Spotted by kbuild test robot.
Signed-off-by: NTejun Heo <tj@kernel.org>
Cc: Neil Brown <neilb@suse.de>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: Ashutosh Dixit <ashutosh.dixit@intel.com>
Cc: kbuild test robot <fengguang.wu@intel.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

The patch just accents that @dev could be NULL.

There is no functional change.
Signed-off-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

In several places we are printing messages with prefix based on chip->label.
Introduced macros help us to do this easier and in uniform way.
Signed-off-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: NAlexandre Courbot <acourbot@nvidia.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

This patch includes the following amendments:
 1) use "?" as a label when the last one is not defined in gpiod_*;
 2) whenever it's possible gpiod_* are used;
 3) print a function name, if it's already used in other messages.

Additionally it fixes an indentation in few places.
Signed-off-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: NAlexandre Courbot <acourbot@nvidia.com>
Reviewed-by: NAlexandre Courbot <acourbot@nvidia.com>
Reviewed-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Change the format of the platform GPIO lookup tables to make them less
confusing and improve lookup efficiency.

The previous format was a single linked-list that required to compare
the device name and function ID of every single GPIO defined for each
lookup. Switch that to a list of per-device tables, so that the lookup
can be done in two steps, omitting the GPIOs that are not relevant for a
particular device.

The matching rules are now defined as follows:
- The device name must match *exactly*, and can be NULL for GPIOs not
  assigned to a particular device,
- If the function ID in the lookup table is NULL, the con_id argument of
  gpiod_get() will not be used for lookup. However, if it is defined, it
  must match exactly.
- The index must always match.
Signed-off-by: NAlexandre Courbot <acourbot@nvidia.com>
Acked-by: NMika Westerberg <mika.westerberg@linux.intel.com>
Reviewed-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>