We don't need to specify any ranges when allocating IDs so we can switch
to ida_alloc() and ida_free() instead of the ida_simple_ counterparts.

ida_simple_get(ida, 0, 0, gfp) is equivalent to
ida_alloc_range(ida, 0, UINT_MAX, gfp) which is equivalent to
ida_alloc(ida, gfp). Note: IDR will never actually allocate an ID
larger than INT_MAX.
Signed-off-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>
Reviewed-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>

Use DEFINE_SEQ_ATTRIBUTE macro to simplify the code.
Signed-off-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>

Replace usage of atomic_notifier_call_chain with
blocking_notifier_call_chain as the notifier function,
lineinfo_changed_notify, calls gpio_desc_to_lineinfo,
which calls pinctrl_gpio_can_use_line, which can sleep.

The chain isn't being called from an atomic context so the
the blocking notifier is a suitable substitute.
Signed-off-by: NKent Gibson <warthog618@gmail.com>
Signed-off-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>

Move gpiolib-sysfs function declarations into their own header.

These functions are in gpiolib-sysfs.c, and are only required by gpiolib.c,
and so should be in a module header, not gpiolib.h.

This brings gpiolib-sysfs into line with gpiolib-cdev, and is another step
towards removing the sysfs inferface.
Signed-off-by: NKent Gibson <warthog618@gmail.com>
Signed-off-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>

bitmap_full() is a shortcut to find_first_zero_bit().
Thus, no need to call it twice.
Signed-off-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>

Split the cdev specific functionality out of gpiolib.c and into
gpiolib-cdev.c. This improves the readability and maintainability of both
the cdev and core gpiolib code.
Suggested-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>
Signed-off-by: NKent Gibson <warthog618@gmail.com>
Link: https://lore.kernel.org/r/20200616093615.5167-1-warthog618@gmail.comSigned-off-by: NLinus Walleij <linus.walleij@linaro.org>

The function connects an IRQ domain to a gpiochip and reuses
gpiochip_to_irq() which is provided by gpiolib.

gpiochip_irqchip_* and regmap_irq partially provide the same
functionality. This function will help to connect just the
minimal functionality of the gpiochip_irqchip which is needed to
work together with regmap-irq.
Signed-off-by: NMichael Walle <michael@walle.cc>
Reviewed-by: NAndy Shevchenko <andy.shevchenko@gmail.com>
Link: https://lore.kernel.org/r/20200528145845.31436-2-michael@walle.ccSigned-off-by: NLinus Walleij <linus.walleij@linaro.org>

We provided the right semantics on open drain lines being
by definition output but incidentally the irq set up function
would only allow IRQs on lines that were "not output".

Fix the semantics to allow output open drain lines to be used
for IRQs.
Reported-by: NHans Verkuil <hverkuil@xs4all.nl>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>
Signed-off-by: NHans Verkuil <hverkuil@xs4all.nl>
Tested-by: NHans Verkuil <hverkuil@xs4all.nl>
Cc: Russell King <linux@armlinux.org.uk>
Cc: stable@vger.kernel.org # v5.3+
Link: https://lore.kernel.org/r/20200527140758.162280-1-linus.walleij@linaro.orgSigned-off-by: NLinus Walleij <linus.walleij@linaro.org>

With 'commit 461c1a7d ("gpiolib: override irq_enable/disable")' gpiolib
overrides irqchip's irq_enable and irq_disable callbacks. If irq_disable
callback is implemented then genirq takes unlazy path to disable irq.

Underlying irqchip may not want to implement irq_disable callback to lazy
disable irq when client drivers invokes disable_irq(). By overriding
irq_disable callback, gpiolib ends up always unlazy disabling IRQ.

Allow gpiolib to lazy disable IRQs by overriding irq_disable callback only
if irqchip implemented irq_disable. In cases where irq_disable is not
implemented irq_mask is overridden. Similarly override irq_enable callback
only if irqchip implemented irq_enable otherwise irq_unmask is overridden.

Fixes: 461c1a7d ("gpiolib: override irq_enable/disable")
Signed-off-by: NMaulik Shah <mkshah@codeaurora.org>
Tested-by: NHans Verkuil <hverkuil-cisco@xs4all.nl>
Link: https://lore.kernel.org/r/1590253873-11556-2-git-send-email-mkshah@codeaurora.orgSigned-off-by: NLinus Walleij <linus.walleij@linaro.org>

We already have two conditionals inside the outer one to check
if the command is GPIO_GET_LINEINFO_WATCH_IOCTL. I think it's
time to actually do what I have proposed in the first place, i.e.
to separate GPIO_GET_LINEINFO_WATCH_IOCTL from GPIO_GET_LINEINFO_IOCTL.
It's +13 LOCs, and surprisingly only +13 bytes of binary on x86_32,
but for the price of much better readability.
Signed-off-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>
Link: https://lore.kernel.org/r/20200525194028.74236-1-andriy.shevchenko@linux.intel.comSigned-off-by: NLinus Walleij <linus.walleij@linaro.org>

Currently a GPIO lookup table can only refer to a specific GPIO by a
tuple, consisting of a GPIO controller label and a GPIO offset inside
the controller.

However, a GPIO may also carry a line name, defined by DT or ACPI.
If present, the line name is the most use-centric way to refer to a
GPIO.  Hence add support for looking up GPIOs by line name.
Note that there is no guarantee that GPIO line names are globally
unique, so this will use the first match found.

Implement this by reusing the existing gpiod_lookup infrastructure.
Rename gpiod_lookup.chip_label to gpiod_lookup.key, to make it clear
that this field can have two meanings, and update the kerneldoc and
GPIO_LOOKUP*() macros.
Signed-off-by: NGeert Uytterhoeven <geert+renesas@glider.be>
Tested-by: NEugeniu Rosca <erosca@de.adit-jv.com>
Reviewed-by: NUlrich Hecht <uli+renesas@fpond.eu>
Reviewed-by: NEugeniu Rosca <erosca@de.adit-jv.com>
Link: https://lore.kernel.org/r/20200511145257.22970-4-geert+renesas@glider.beSigned-off-by: NLinus Walleij <linus.walleij@linaro.org>

gpiochip_set_desc_names() no longer rejects GPIO line name collisions.
Hence GPIO line names are not guaranteed to be globally unique.
In case of multiple GPIO lines with the same name, gpio_name_to_desc()
will return the first match found.

Update the comments for gpio_name_to_desc() and
gpiochip_set_desc_names() to match reality.

Fixes: f881bab0 ("gpio: keep the GPIO line names internal")
Signed-off-by: NGeert Uytterhoeven <geert+renesas@glider.be>
Link: https://lore.kernel.org/r/20200511101828.30046-1-geert+renesas@glider.beSigned-off-by: NLinus Walleij <linus.walleij@linaro.org>

Currently we emit the REQUESTED line state event after the line is
requested but before the flags are configured. This is obviously wrong
as we want to pass the updated lineinfo to user-space together with the
event.

Since the flags can be configured in different ways depending on how the
line is being requested - we need to call the notifier chain in different
places separately.

Fixes: 51c1064e ("gpiolib: add new ioctl() for monitoring changes in line info")
Signed-off-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>

The symbol 'gpio_of_notifier' doesn't exist without both CONFIG_OF_GPIO
and CONFIG_OF_DYNAMIC enabled, but is referenced when only
CONFIG_OF_DYNAMIC is enabled.

This broke building with 'make ARCH=um allyesconfig':
---------------
/usr/bin/ld: drivers/gpio/gpiolib.o: in function `gpiolib_dev_init':
./drivers/gpio/gpiolib.c:5293: undefined reference to `gpio_of_notifier'
collect2: error: ld returned 1 exit status
---------------

Fixes: 63636d95 ("gpio: of: Add DT overlay support for GPIO hogs")
Signed-off-by: NDavid Gow <davidgow@google.com>
Reviewed-by: NGeert Uytterhoeven <geert+renesas@glider.be>
Link: https://lore.kernel.org/r/20200425044655.166257-1-davidgow@google.comSigned-off-by: NLinus Walleij <linus.walleij@linaro.org>

Simplify the printing of kernel messages and make the messages more
accurate by using the most appropriate {dev,chip,gpiod}_*() helpers.

Sample impact:

    -gpiochip_setup_dev: registered GPIOs 496 to 511 on device: gpiochip0 (e6050000.gpio)
    +gpio gpiochip0: registered GPIOs 496 to 511 on e6050000.gpio

    -no flags found for gpios
    +gpio-953 (?): no flags found for gpios

    -GPIO line 355 (PCIE/SATA switch) hogged as output/low
    +gpio-355 (PCIE/SATA switch): hogged as output/low
Signed-off-by: NGeert Uytterhoeven <geert+renesas@glider.be>
Link: https://lore.kernel.org/r/20200424141432.11400-1-geert+renesas@glider.beSigned-off-by: NLinus Walleij <linus.walleij@linaro.org>

We must not call pinctrl_gpio_can_use_line() with the gpio_lock taken
as it takes a mutex internally. Let's move the call before taking the
spinlock and store the return value.

This isn't perfect - there's a moment between calling
pinctrl_gpio_can_use_line() and taking the spinlock where the situation
can change but it isn't a regression either: previously this part wasn't
protected at all and it only affects the information user-space is
seeing.
Reported-by: NGeert Uytterhoeven <geert@linux-m68k.org>
Fixes: d2ac2579 ("gpiolib: provide a dedicated function for setting lineinfo")
Signed-off-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>
Acked-by: NLinus Walleij <linus.walleij@linaro.org>

This makes the new ioctl() a bit more robust - we now check if a line
is already being watched and return -EBUSY if the user-space tries to
start watching it again. Same for unwatch - return -EBUSY if user-space
tries to unwatch a line that's not being watched.

Fixes: 51c1064e ("gpiolib: add new ioctl() for monitoring changes in line info")
Signed-off-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>
Reviewed-by: NLinus Walleij <linus.walleij@linaro.org>

There is struct gpio *gc, *chip and *gpiochip, and yes
I am responsible for some of the inconsistencies. I want
this to be just gc everywhere for minimizing cognitive
resistance when reading the code: more compact function
signatures and less clutter.

Purely syntactic changes intended. No semantic effects.
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>
Link: https://lore.kernel.org/r/20200329140405.52276-1-linus.walleij@linaro.orgSigned-off-by: NLinus Walleij <linus.walleij@linaro.org>

Do not use the struct gpio_device's .pin_ranges field if the PINCTRL
Kconfig symbol is not selected to avoid build failures.

Fixes: 2ab73c6d ("gpio: Support GPIO controllers without pin-ranges")
Reported-by: NStephen Rothwell <sfr@canb.auug.org.au>
Reported-by: Nkbuild test robot <lkp@intel.com>
Signed-off-by: NThierry Reding <treding@nvidia.com>
Link: https://lore.kernel.org/r/20200330090257.2332864-1-thierry.reding@gmail.comReviewed-by: NGeert Uytterhoeven <geert@linux-m68k.org>
Tested-by: NGeert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

gpio_set_bias() no longer uses the passed gpio_chip pointer parameter.
Remove it.
Signed-off-by: NGeert Uytterhoeven <geert+renesas@glider.be>
Link: https://lore.kernel.org/r/20200325100439.14000-3-geert+renesas@glider.beSigned-off-by: NLinus Walleij <linus.walleij@linaro.org>

All callers of gpio_set_config() have to convert a gpio_desc to a
gpio_chip and offset.  Avoid these duplicated conversion steps by
letting gpio_set_config() take a gpio_desc pointer directly.
Signed-off-by: NGeert Uytterhoeven <geert+renesas@glider.be>
Link: https://lore.kernel.org/r/20200325100439.14000-2-geert+renesas@glider.beSigned-off-by: NLinus Walleij <linus.walleij@linaro.org>

The GPIO Aggregator will need a method to forward a .set_config() call
to its parent gpiochip.  This requires obtaining the gpio_chip and
offset for a given gpio_desc.  While gpiod_to_chip() is public,
gpio_chip_hwgpio() is not, so there is currently no method to obtain the
needed GPIO offset parameter.

Hence introduce a public gpiod_set_config() helper, which invokes the
.set_config() callback through a gpio_desc pointer, like is done for
most other gpio_chip callbacks.

Rewrite the existing gpiod_set_debounce() helper as a wrapper around
gpiod_set_config(), to avoid duplication.
Signed-off-by: NGeert Uytterhoeven <geert+renesas@glider.be>
Link: https://lore.kernel.org/r/20200324135653.6676-5-geert+renesas@glider.beSigned-off-by: NLinus Walleij <linus.walleij@linaro.org>

Use a different markup for the ERR_PTR, as %FOO doesn't work
if there are parenthesis. So, use, instead:

	``ERR_PTR(-EINVAL)``

This fixes the following warning:

	./drivers/gpio/gpiolib.c:139: WARNING: Inline literal start-string without end-string.
Signed-off-by: NMauro Carvalho Chehab <mchehab+huawei@kernel.org>
Link: https://lore.kernel.org/r/51197e3568f073e22c280f0584bfa20b44436708.1584456635.git.mchehab+huawei@kernel.orgReviewed-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Wake gpiochip_generic_request() call into the pinctrl helpers only if a
GPIO controller had any pin-ranges assigned to it. This allows a driver
to unconditionally use this helper if it supports multiple devices of
which only a subset have pin-ranges assigned to them.
Signed-off-by: NThierry Reding <treding@nvidia.com>
Link: https://lore.kernel.org/r/20200319122737.3063291-2-thierry.reding@gmail.comTested-by: NVidya Sagar <vidyas@nvidia.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Since name == NULL can't ever match, move the check out of
IRQ-disabled region.
Signed-off-by: NMichał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>

These two devres functions devm_gpiochip_[add|remove]()
were in the wrong file. They should be in gpiolib-devres.c
not gpiolib.c.

Link: https://lore.kernel.org/r/20200313081522.35143-1-linus.walleij@linaro.orgReviewed-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

The implementation if .irq_disable() which kicks in between
the gpiolib and the driver is not properly mimicking the
expected semantics of the irqchip core: the irqchip will
call .irq_disable() if that exists, else it will call
mask_irq() which first checks if .irq_mask() is defined
before calling it.

Since we are calling it unconditionally, we get this bug
from drivers/pinctrl/qcom/pinctrl-ssbi-gpio.c, as it only
defines .irq_mask_ack and not .irq_mask:

  Unable to handle kernel NULL pointer dereference at virtual address 00000000
  pgd = (ptrval)
  (...)
  PC is at 0x0
  LR is at gpiochip_irq_disable+0x20/0x30

Fix this by only calling .irq_mask() if it exists.

Cc: Brian Masney <masneyb@onstation.org>
Cc: Hans Verkuil <hans.verkuil@cisco.com>
Cc: stable@vger.kernel.org
Reviewed-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>
Fixes: 461c1a7d ("gpiolib: override irq_enable/disable")
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>
Link: https://lore.kernel.org/r/20200306132326.1329640-1-linus.walleij@linaro.org

The optimization to check for requested lines actually optimized for the
uncomon error case, where one of the GPIO lines is still in use.
Hence the error message must be printed when the loop is terminated
early, not when it went through all available GPIO lines.

Fixes: 869233f8 ("gpiolib: Optimize gpiochip_remove() when check for requested line")
Signed-off-by: NGeert Uytterhoeven <geert+renesas@glider.be>
Link: https://lore.kernel.org/r/20200302082448.11795-1-geert+renesas@glider.beAcked-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

When operating on the bits of watched_lines bitmap, we're using
desc_to_gpio() which returns the GPIO number from the global numberspace.
This leads to all sorts of memory corruptions and invalid behavior. We
should switch to using gpio_chip_hwgpio() instead.

Fixes: 51c1064e ("gpiolib: add new ioctl() for monitoring changes in line info")
Reported-by: NKent Gibson <warthog618@gmail.com>
Signed-off-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>
Tested-by: NKent Gibson <warthog618@gmail.com>

Fix the field having a bit cleared by the unwatch ioctl().

Fixes: 51c1064e ("gpiolib: add new ioctl() for monitoring changes in line info")
Signed-off-by: NKent Gibson <warthog618@gmail.com>
Signed-off-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>

Rename 'event' to 'ge' to be consistent with other use.
Signed-off-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>

Here are the following optimizations have been done:
 - break the loop after first found requested line
  - due to above, drop redundant boolean variable
 - replace open coded variant of gpiochip_is_requested()
  - due to above, drop redundant pointer to struct gpio_desc
 - use 'unsigned int' instead of 'unsigned' for loop counter

Note, pointer to struct gpio_chip followed by pointer to struct gpio_device
is still valid, back link is not.
Signed-off-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Link: https://lore.kernel.org/r/20200225114725.839-1-andriy.shevchenko@linux.intel.comSigned-off-by: NLinus Walleij <linus.walleij@linaro.org>

The function was currently used internal by the gpiolib. Since commit
56cc3af4 ("pinctrl: da9062: add driver support") it is also used by
drivers so we need to export the symbol.
Signed-off-by: NMarco Felsch <m.felsch@pengutronix.de>
Link: https://lore.kernel.org/r/20200225093102.10964-1-m.felsch@pengutronix.deSigned-off-by: NLinus Walleij <linus.walleij@linaro.org>

As GPIO hogs are configured at GPIO controller initialization time,
adding/removing GPIO hogs in DT overlays does not work.

Add support for GPIO hogs described in DT overlays by registering an OF
reconfiguration notifier, to handle the addition and removal of GPIO hog
subnodes to/from a GPIO controller device node.

Note that when a GPIO hog device node is being removed, its "gpios"
properties is no longer available, so we have to keep track of which
node a hog belongs to, which is done by adding a pointer to the hog's
device node to struct gpio_desc.
Signed-off-by: NGeert Uytterhoeven <geert+renesas@glider.be>
Link: https://lore.kernel.org/r/20200220130149.26283-3-geert+renesas@glider.beReviewed-by: NFrank Rowand <frank.rowand@sony.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

The existing use of ktime_get_real_ns() in the timestamps from
the GPIO events is dubious.

We have had several discussions about this timestamp, and it is
unclear whether userspace has ever taken into account that a
timestamp from ktime_get_real_ns() can actually move backwards
in time relative the previous timetamp, and userspace is more
likely to expect a monotonic counter.

Background:
https://lore.kernel.org/linux-gpio/CAK8P3a1Skvm48sje8FNDPLYqyz9Lf8q0qX1QETWtyZTxuX4k1g@mail.gmail.com/
https://marc.info/?l=linux-gpio&m=151661955709074&w=2

The change is ABI incompatible, but incompatible in a way that
is IMO more likely to fix future bugs rather than break current
userspace. To the best of my knowledge all userspace expects
a monotonic timestamp and users are just lucky that they very
seldom move backwards in time.

Cc: Arnd Bergmann <arnd@arndb.de>
Acked-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Whenever retrieving a descriptor from a gpiochip: use the provided
helper which checks for errors.
Signed-off-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>
Link: https://lore.kernel.org/r/20200219094702.6463-1-brgl@bgdev.plSigned-off-by: NLinus Walleij <linus.walleij@linaro.org>

Currently there is no way for user-space to be informed about changes
in status of GPIO lines e.g. when someone else requests the line or its
config changes. We can only periodically re-read the line-info. This
is fine for simple one-off user-space tools, but any daemon that provides
a centralized access to GPIO chips would benefit hugely from an event
driven line info synchronization.

This patch adds a new ioctl() that allows user-space processes to reuse
the file descriptor associated with the character device for watching
any changes in line properties. Every such event contains the updated
line information.

Currently the events are generated on three types of status changes: when
a line is requested, when it's released and when its config is changed.
The first two are self-explanatory. For the third one: this will only
happen when another user-space process calls the new SET_CONFIG ioctl()
as any changes that can happen from within the kernel (i.e.
set_transitory() or set_debounce()) are of no interest to user-space.
Signed-off-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>
Reviewed-by: NLinus Walleij <linus.walleij@linaro.org>

We'll soon be filling out the gpioline_info structure in multiple
places. Add a separate function that given a gpio_desc sets all relevant
fields.
Signed-off-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>
Reviewed-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>

Currently if the line-event kfifo is full, we just silently drop any new
events. Add a ratelimited debug message so that we at least have some
trace in the kernel log of event overflow.
Signed-off-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>

The read_lock mutex is supposed to prevent collisions between reading
and writing to the line event kfifo but it's actually only taken when
the events are being read from it.

Drop the mutex entirely and reuse the spinlock made available to us in
the waitqueue struct. Take the lock whenever the fifo is modified or
inspected. Drop the call to kfifo_to_user() and instead first extract
the new element from kfifo when the lock is taken and only then pass
it on to the user after the spinlock is released.
Signed-off-by: NBartosz Golaszewski <bgolaszewski@baylibre.com>
Reviewed-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>