Coverity found that we were checking an unsigned variable for
>= zero. Type it correctly so that the check works as intended.
Signed-off-by: NVirupax Sadashivpetimath <virupax.sadashivpetimath@stericsson.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Suspend/resume didn't take care of pull-up and pull-down
settings and writing back the DAT register at resume can
change pull up/down settings, depending on pin input value.
Output values are now also restored.
Signed-off-by: NJonas Aaberg <jonas.aberg@stericsson.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

When GPIOs wake up the system from sleep mode, the normal GPIO interrupt
handler does not hit and the normal interrupt status register does not
contain the status. Instead the secondary GPIO handler does, and the
interrupt status needs to be retrieved from the wakeup status saved by
the suspend/resume code.
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
[Added constant 32-pin assignment in platform data]
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Signed-off-by: NLennert Buytenhek <buytenh@secretlab.ca>
Acked-by: NRabin Vincent <rabin.vincent@stericsson.com>

Platforms may choose not to provide an additional name for the GPIO
block.
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NLinus Walleij <linus.walleij@stericsson.com>

In the nomadik GPIO pin configuration, allow the sleep mode direction
and pull configurations to differ from the ones for the normal state.
PIN_SLPM_PULL_*, PIN_SLPM_INPUT, PIN_SLPM_OUTPUT* macros are provided
for this.

Since the hardware does not allow seperate configurations for sleep mode
and normal mode, this is implemented by having software remux the
configurations as necessary.
Reviewed-by: NSrinidhi KASAGAR <srinidhi.kasagar@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NMian Yousaf Kaukab <mian.yousaf.kaukab@stericsson.com>
Signed-off-by: NLinus Walleij <linus.walleij@stericsson.com>

This fixes a section mismatch reported by modpost:

	WARNING: arch/arm/plat-nomadik/built-in.o(.data+0x100): Section mismatch in reference from the variable nmk_gpio_driver to the function .init.text:nmk_gpio_probe()
	The variable nmk_gpio_driver references
	the function __init nmk_gpio_probe()
	If the reference is valid then annotate the
	variable with __init* or __refdata (see linux/init.h) or name the variable:
	*_template, *_timer, *_sht, *_ops, *_probe, *_probe_one, *_console,

According to Rabin VINCENT the alternative to use platform_driver_probe
won't work because the "platform devices are not yet registered when
nmk_gpio_init() is called."
Acked-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NUwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

DB8500v2 allows control of direction and pull up/down configuration in
sleep mode, instead of switching the pin to input with pull up/down
enabled.
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

The SLPM register enables/disables wakeup detection on DB8500v2.
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

arch/arm/plat-nomadik/gpio.c: In function 'nmk_gpio_remove':
arch/arm/plat-nomadik/gpio.c:630: warning: ignoring return value of 'gpiochip_remove', declared with attribute warn_unused_result

Because this function is marked __exit and this code can't be built as a
module, this function is never used.  So just remove it, fixing the
warning in the process.
Acked-by: NAlessandro Rubini <rubini@unipv.it>
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

So that gpio_to_irq() works.
Acked-by: NAlessandro Rubini <rubini@unipv.it>
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

So that set_irq_wake() works.

Cc: Alessandro Rubini <rubini@unipv.it>
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Move the platform driver registration to a core_initcall, instead of an
arch_initcall.  This will allow us to use gpio related calls in
init_machine() (which is an arch_initcall) after adding the gpio
platform devices.

Cc: Alessandro Rubini <rubini@unipv.it>
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Add a pin configuration API to all pin-related configuration to be
specified with a single macro and groups of pins to be configured
at one go.  Based on the PXA MFP implementation.

Cc: Alessandro Rubini <rubini@unipv.it>
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Extract out a portion of nmk_gpio_set_mode() that can be called with the
lock already held.  This will be used from the pincfg API.

Cc: Alessandro Rubini <rubini@unipv.it>
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Cc: Alessandro Rubini <rubini@unipv.it>
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Cc: Alessandro Rubini <rubini@unipv.it>
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

The ->direction_output() callback needs to set the value it was passed,
in addition to setting the direction as output.  On this peripheral, we
can't set the value before setting the direction, so set it after.
Acked-by: NAlessandro Rubini <rubini@unipv.it>
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Add clocks with appropriate names in platforms that use it, and use the
clk API in nomadik-gpio.
Acked-by: NAlessandro Rubini <rubini@unipv.it>
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Acked-by: NSrinidhi Kasagar <srinidhi.kasagar@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Set a default trigger type for interrupts, otherwise if request_irq is
called without specifiying a trigger type, the interrupt will not
actually be enabled.
Acked-by: NAlessandro Rubini <rubini@unipv.it>
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

On this peripheral, setting the trigger type enables the interrupt, and
the current set_type() implementation unconditionally enables the
interrupt, even if it is called when the interrupt is disabled.  Fix
set_type() to:

 - if the interrupt is disabled, defer the actual trigger setting to
   when it is unmasked

 - if the interrupt is enabled, change the type immediately by clearing
   the old type and then re-enabling with the new type.
Acked-by: NAlessandro Rubini <rubini@unipv.it>
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Remove some nearly-duplicated code to make the following patch simpler.
Acked-by: NAlessandro Rubini <rubini@unipv.it>
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Mask/unmask should modify the interrupt mask register (RIMSC
and FIMSC) not the wakeup mask registers (RWIMSC and FWIMSC).
Acked-by: NAlessandro Rubini <rubini@unipv.it>
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

On the U8500 platform there are four GPIO blocks, each with a 4K address
space, including the peripheral identification.  However, each of these
blocks have a varying number of banks, each of which have 32 GPIOs and
an interrupt line.

The current nomadik-gpio driver implementation can handle each of these
sub-banks easily with one instance each, but cannot as-is be hooked up
to them because it is an AMBA driver and it expects to see a peripheral
with the appropriate peripheral ids but having only one bank and only
one interrupt.

Solve this by converting the driver to a platform driver.
Acked-by: NAlessandro Rubini <rubini@unipv.it>
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Since we register as a chained handler, we need to take care of acking
the parent irq.  Current code had the acking disabled because just
acking the interrupt disables it on some interrupt controllers (like the
VIC used on Nomadik).  However, not acking at all is incorrect because
interrupt controllers like the GIC need an EOI indication, which is done
in the ack routine.

Solve this by maskacking and unmasking it appropriately, similar to how
handle_level_irq operates.
Acked-by: NAlessandro Rubini <rubini@unipv.it>
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Fix the incorrect spinlock usage which leads to these compiler warnings:

arch/arm/plat-nomadik/gpio.c:60: warning: passing argument 1 of `spinlock_check' from incompatible pointer type
arch/arm/plat-nomadik/gpio.c:69: warning: passing argument 1 of `spin_unlock_irqrestore' from incompatible pointer type
Acked-by: NAlessandro Rubini <rubini@unipv.it>
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Move the Nomadik GPIO driver to plat-nomadik so that it can also be used
on the Ux500 platform.

The mach-nomadik include/mach/gpio.h is replaced by one that includes
the one now in plat-nomadik, so that code doesn't need to include the
one in plat specifically, and can instead use <linux/gpio.h> as usual.
Acked-by: NAlessandro Rubini <rubini@unipv.it>
Acked-by: NLinus Walleij <linus.walleij@stericsson.com>
Signed-off-by: NRabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Signed-off-by: NAlessandro Rubini <rubini@unipv.it>
Acked-by: NAndrea Gallo <andrea.gallo@stericsson.com>
Signed-off-by: NRussell King <rmk+kernel@arm.linux.org.uk>