This makes the driver use the data pointer added to the gpio_chip
to store a pointer to the state container instead of relying on
container_of().

Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Lee Jones <lee.jones@linaro.org>

tps65010 driver's initcall cannot succeed when the driver is built-in,
because it expects that the I2C probe is completed at initcall time;
this cannot happen as MFD is initialized before I2C. Also on systems
where the chip is not present there is unnecessary 30 ms delay during
the boot.

Instead of waiting for probe to finish, just register the I2C device.
If some boards need retry mechanism for startup glitches, that should be
done in the actual probe function. Also delete the driver banner message.

The patch allows to use tps65010 again with OMAP1 (where it's required
to be built-in) and enables e.g. USB and LED functionality on OMAP5912 OSK.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Lee Jones <lee.jones@linaro.org>

Pull GPIO updates from Linus Walleij:
 "Here is the bulk of GPIO changes for v4.5.

  Notably there are big refactorings mostly by myself, aimed at getting
  the gpio_chip into a shape that makes me believe I can proceed to
  preserve state for a proper userspace ABI (character device) that has
  already been proposed once, but resulted in the feedback that I need
  to go back and restructure stuff.  So I've been restructuring stuff.
  On the way I ran into brokenness (return code from the get_value()
  callback) and had to fix it.  Also, refactored generic GPIO to be
  simpler.

  Some of that is still waiting to trickle down from the subsystems all
  over the kernel that provide random gpio_chips, I've touched every
  single GPIO driver in the kernel now, oh man I didn't know I was
  responsible for so much...

  Apart from that we're churning along as usual.

  I took some effort to test and retest so it should merge nicely and we
  shook out a couple of bugs in -next.

  Infrastructural changes:

   - In struct gpio_chip, rename the .dev node to .parent to better
     reflect the fact that this is not the GPIO struct device
     abstraction.  We will add that soon so this would be totallt
     confusing.

   - It was noted that the driver .get_value() callbacks was sometimes
     reporting negative -ERR values to the gpiolib core, expecting them
     to be propagated to consumer gpiod_get_value() and gpio_get_value()
     calls.  This was not happening, so as there was a mess of drivers
     returning negative errors and some returning "anything else than
     zero" to indicate that a line was active.  As some would have bit
     31 set to indicate "line active" it clashed with negative error
     codes.  This is fixed by the largeish series clamping values in all
     drivers with !!value to [0,1] and then augmenting the code to
     propagate error codes to consumers.  (Includes some ACKed patches
     in other subsystems.)

   - Add a void *data pointer to struct gpio_chip.  The container_of()
     design pattern is indeed very nice, but we want to reform the
     struct gpio_chip to be a non-volative, stateless business, and keep
     states internal to the gpiolib to be able to hold on to the state
     when adding a proper userspace ABI (character device) further down
     the road.  To achieve this, drivers need a handle at the internal
     state that is not dependent on their struct gpio_chip() so we add
     gpiochip_add_data() and gpiochip_get_data() following the pattern
     of many other subsystems.  All the "use gpiochip data pointer"
     patches transforms drivers to this scheme.

   - The Generic GPIO chip header has been merged into the general
     <linux/gpio/driver.h> header, and the custom header for that
     removed.  Instead of having a separate mm_gpio_chip struct for
     these generic drivers, merge that into struct gpio_chip,
     simplifying the code and removing the need for separate and
     confusing includes.

  Misc improvements:

   - Stabilize the way GPIOs are looked up from the ACPI legacy
     specification.

   - Incremental driver features for PXA, PCA953X, Lantiq (patches from
     the OpenWRT community), RCAR, Zynq, PL061, 104-idi-48

  New drivers:

   - Add a GPIO chip to the ALSA SoC AC97 driver.

   - Add a new Broadcom NSP SoC driver (this lands in the pinctrl dir,
     but the branch is merged here too to account for infrastructural
     changes).

   - The sx150x driver now supports the sx1502"

* tag 'gpio-v4.5-1' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-gpio: (220 commits)
  gpio: generic: make bgpio_pdata always visible
  gpiolib: fix chip order in gpio list
  gpio: mpc8xxx: Do not use gpiochip_get_data() in mpc8xxx_gpio_save_regs()
  gpio: mm-lantiq: Do not use gpiochip_get_data() in ltq_mm_save_regs()
  gpio: brcmstb: Allow building driver for BMIPS_GENERIC
  gpio: brcmstb: Set endian flags for big-endian MIPS
  gpio: moxart: fix build regression
  gpio: xilinx: Do not use gpiochip_get_data() in xgpio_save_regs()
  leds: pca9532: use gpiochip data pointer
  leds: tca6507: use gpiochip data pointer
  hid: cp2112: use gpiochip data pointer
  bcma: gpio: use gpiochip data pointer
  avr32: gpio: use gpiochip data pointer
  video: fbdev: via: use gpiochip data pointer
  gpio: pch: Optimize pch_gpio_get()
  Revert "pinctrl: lantiq: Implement gpio_chip.to_irq"
  pinctrl: nsp-gpio: use gpiochip data pointer
  pinctrl: vt8500-wmt: use gpiochip data pointer
  pinctrl: exynos5440: use gpiochip data pointer
  pinctrl: at91-pio4: use gpiochip data pointer
  ...

As we want gpio_chip .get() calls to be able to return negative
error codes and propagate to drivers, we need to go over all
drivers and make sure their return values are clamped to [0,1].
We do this by using the ret = !!(val) design pattern.

This also start to propagate the negative error code from the
smbus call if there is one, as the last commit of this series
will make the gpiolib core deal with that properly.

Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Lee Jones <lee.jones@linaro.org>

The name .dev in a struct is normally reserved for a struct device
that is let us say a superclass to the thing described by the struct.
struct gpio_chip stands out by confusingly using a struct device *dev
to point to the parent device (such as a platform_device) that
represents the hardware. As we want to give gpio_chip:s real devices,
this is not working. We need to rename this member to parent.

This was done by two coccinelle scripts, I guess it is possible to
combine them into one, but I don't know such stuff. They look like
this:

@@
struct gpio_chip *var;
@@
-var->dev
+var->parent

and:

@@
struct gpio_chip var;
@@
-var.dev
+var.parent

and:

@@
struct bgpio_chip *var;
@@
-var->gc.dev
+var->gc.parent

Plus a few instances of bgpio that I couldn't figure out how
to teach Coccinelle to rewrite.

This patch hits all over the place, but I *strongly* prefer this
solution to any piecemal approaches that just exercise patch
mechanics all over the place. It mainly hits drivers/gpio and
drivers/pinctrl which is my own backyard anyway.

Cc: Haavard Skinnemoen <hskinnemoen@gmail.com>
Cc: Rafał Miłecki <zajec5@gmail.com>
Cc: Richard Purdie <rpurdie@rpsys.net>
Cc: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Cc: Alek Du <alek.du@intel.com>
Cc: Jaroslav Kysela <perex@perex.cz>
Cc: Takashi Iwai <tiwai@suse.com>
Acked-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Acked-by: Lee Jones <lee.jones@linaro.org>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Acked-by: Hans-Christian Egtvedt <egtvedt@samfundet.no>
Acked-by: Jacek Anaszewski <j.anaszewski@samsung.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

Even if bus is not hot-pluggable, the devices can be unbound from the
driver via sysfs, so we should not be using __exit annotations on
remove() methods. The only exception is drivers registered with
platform_driver_probe() which specifically disables sysfs bind/unbind
attributes.

Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Lee Jones <lee.jones@linaro.org>

Use devm_*() functions to make cleanup paths simpler.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Lee Jones <lee.jones@linaro.org>

There is no need to use a per CPU workqueue to poll, especially with the
5s delay used, so allow the scheduler to use any CPU.

Signed-off-by: Mark Brown <broonie@linaro.org>
Signed-off-by: Lee Jones <lee.jones@linaro.org>

Use the wrapper function for retrieving the platform data instead of
accessing dev->platform_data directly.

Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Lee Jones <lee.jones@linaro.org>

With the changes in the random tree, IRQF_SAMPLE_RANDOM is now a
no-op; interrupt randomness is now collected unconditionally in a very
low-overhead fashion; see commit 775f4b297b.  The IRQF_SAMPLE_RANDOM
flag was scheduled to be removed in 2009 on the
feature-removal-schedule, so this patch is preparation for the final
removal of this flag.

Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Cc: Samuel Ortiz <sameo@linux.intel.com>