Fix a race where a pending interrupt could be received and the handler
called before the handler's data has been setup, by converting to
irq_set_chained_handler_and_data().

Search and conversion was done with coccinelle:

@@
expression E1, E2, E3;
@@
(
-if (irq_set_chained_handler(E1, E3) != 0)
-   BUG();
|
-irq_set_chained_handler(E1, E3);
)
-irq_set_handler_data(E1, E2);
+irq_set_chained_handler_and_data(E1, E3, E2);

@@
expression E1, E2, E3;
@@
(
-if (irq_set_chained_handler(E1, E3) != 0)
-   BUG();
...
|
-irq_set_chained_handler(E1, E3);
...
)
-irq_set_handler_data(E1, E2);
+irq_set_chained_handler_and_data(E1, E3, E2);
Reported-by: NRussell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: NThomas Gleixner <tglx@linutronix.de>
Cc: Julia Lawall <Julia.Lawall@lip6.fr>
Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: Maxime Ripard <maxime.ripard@free-electrons.com>
Cc: Hans de Goede <hdegoede@redhat.com>
Cc: Chen-Yu Tsai <wens@csie.org>
Cc: Fan Wu <fwu@marvell.com>
Cc: abdoulaye berthe <berthe.ab@gmail.com>
Cc: Alexandre Courbot <acourbot@nvidia.com>
Cc: linux-gpio@vger.kernel.org
Cc: linux-arm-kernel@lists.infradead.org

On sun4i-a10, when GPIOs are configured as external interrupt the value for
them in the data register does not seem to get updated, so set their mux to
input (and restore afterwards) when reading the pin.

Missed edges seem to be buffered, so this does not introduce a race
condition.

I've also tested this on sun5i-a13 and sun7i-a20 and those do not seem to
be affected, the input value representation in the data register does seem
to correctly get updated to the actual pin value while in irq mode there.
Signed-off-by: NHans de Goede <hdegoede@redhat.com>
Acked-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

This function actually operates on a gpio_chip, so its prefix should
reflect that fact for consistency with other functions defined in
gpio/driver.h.
Signed-off-by: NAlexandre Courbot <acourbot@nvidia.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

commit 2243a87d
"pinctrl: avoid duplicated calling enable_pinmux_setting for a pin"
removed the .disable callback from the struct pinmux_ops,
making the .enable() callback the only remaining callback.

However .enable() is a bad name as it seems to imply that a
muxing can also be disabled. Rename the callback to .set_mux()
and also take this opportunity to clean out any remaining
mentions of .disable() from the documentation.
Acked-by: NStephen Warren <swarren@nvidia.com>
Acked-by: NBjorn Andersson <bjorn.andersson@sonymobile.com>
Acked-by: NFan Wu <fwu@marvell.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

The pinctrl-sunxi driver originally used the pin number as the gpio
range offset. This resulted in large, bogus gpio numbers for the
new sun6i-a31-r pinctrl devices.

This patch makes the driver number the gpios ranges starting from an
offset of 0, by subtracting the pin_base number from the pin number.
This also makes the system-wide gpio number match the pin number.

Tested on sun8i with sysfs exported gpios.

This patch also changes the GPIO bindings for R_PIO:

    gpios = <&r_pio B N flag>;

Where B originally was the pinbank label (L or M) counted from A,
with this patch it becomes (L or M) counted from its pinbank base (L).

Thus

    gpios = <&r_pio 10 11 0>; /* PL11 */

becomes

    gpios = <&r_pio 0 11 0>; /* PL11 */

IMO this is correct, as the binding shows the bank offset and pin offset
within the bank for the GPIO controller. But I'm worried it might be a
bit confusing.
Signed-off-by: NChen-Yu Tsai <wens@csie.org>
Acked-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

When an IRQ is started on a GPIO line, mark this GPIO as IRQ in
the gpiolib so we can keep track of the usage centrally.
Signed-off-by: NChen-Yu Tsai <wens@csie.org>
Acked-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Signed-off-by: Nabdoulaye berthe <berthe.ab@gmail.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

When mapping the interrupts, the gpio_to_irq function did not consider
the bank number of the gpio pin in question, only the offset or the
interrupt number in the bank. As a result, requests for interrupts in
the later banks get mapped to the first bank.

This issue was discovered while enabling mmc on the new sun8i platform.
The tablet I have uses a pin/interrupt from the second bank to do mmc
card detection. Tested on this very device with register inspection and
actual mmc card insertion/removal.
Signed-off-by: NChen-Yu Tsai <wens@csie.org>
Acked-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Some drivers use disable_irq / enable_irq and do the work
clearing the source in another thread instead of using a threaded
interrupt handler.

The irqchip used not having irq_disable and irq_enable
callbacks in this case, will lead to unnecessary spurious
interrupts:

On a disable_irq in a chip without a handler for this, the irq
core will remember the disable, but not actually call into the
irqchip. With a level triggered interrupt (where the source has
not been cleared) this will lead to an immediate retrigger, at
which point the irq-core will mask the irq. So having an
irq_disable callback in the irqchip will save us the interrupt
firing a 2nd time for nothing.

Drivers using disable / enable_irq like this, will call
enable_irq when they finally have cleared the interrupt source,
without an enable_irq callback, this will turn into an unmask,
at which point the irq will trigger immediately because when it
was originally acked the level was still high, so the ack was
a nop.
Signed-off-by: NHans de Goede <hdegoede@redhat.com>
Acked-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

For level triggered gpio interrupts we need to use handle_fasteoi_irq,
like we do with the irq-sunxi-nmi driver. This is necessary to give threaded
interrupt handlers a chance to actuall clear the source of the interrupt
(which may involve sleeping waiting for i2c / spi / mmc transfers), before
acknowledging the interrupt.
Signed-off-by: NHans de Goede <hdegoede@redhat.com>
Acked-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

With level triggered interrupt mask / unmask will get called for each
interrupt, doing the somewhat expensive mux setting on each unmask thus is
not a good idea. Instead add a request_resources callback and do it there.
Signed-off-by: NHans de Goede <hdegoede@redhat.com>
Acked-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

The sunxi pinctrl irq chip driver does not support wakeup at the
moment. Adding IRQCHIP_SKIP_SET_WAKE lets the irqs work with drivers
using wakeup.
Signed-off-by: NChen-Yu Tsai <wens@csie.org>
Signed-off-by: NHans de Goede <hdegoede@redhat.com>
Acked-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

The A23 and A31 support multiple interrupt banks. Support it by adding a linear
domain covering all the banks. It's trickier than it should because there's an
interrupt per bank, so we have multiple interrupts using the same domain.
Signed-off-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

If irq_mask_ack is not defined, mask_ack_irq will call irq_mask and then
irq_ack. In order to avoid code duplication, between irq_mask_ack and irq_mask,
just declare irq_ack.
Signed-off-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

kzalloc can fail. Add a null check to avoid null pointer
dereference error while accessing the pointer later.
Signed-off-by: NSachin Kamat <sachin.kamat@linaro.org>
Acked-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

The irq/pin mapping is used to lookup the pin to mux to the irq
function when the irq is enabled. It is created when gpio_to_irq
is called. Creating the mapping during init allows us to map the
interrupts directly from the device tree.

Originally the IRQ to pin mapping was created when gpio_to_irq
was called with a GPIO handle. The mapping in turn is used to mux
the pin into EINT mode.

If the mapping is created during gpio_to_irq, we can't use the
interrupts directly, i.e. through the DT with "interrupts = <&pio A 4>".

Instead we'd have to use "gpios = <&pio A B>", then pass the gpio
through to gpio_to_irq.
Signed-off-by: NChen-Yu Tsai <wens@csie.org>
Signed-off-by: NHans de Goede <hdegoede@redhat.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

The pin numbers passed to sunxi_*_reg helpers to get the correct
registers should be the pin offset for the PIO block, not the
absolute number we use that is based on the alphanumeric labels
Allwinner uses.

This patch subtracts .pin_base from the pin number passed to these
functions, so the driver accesses the correct registers.
Signed-off-by: NChen-Yu Tsai <wens@csie.org>
Reviewed-by: NBoris Brezillon <boris.brezillon@free-electrons.com>
Acked-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

When the support for the PRCM muxer on the A31 has been added, the global
static pinctl_desc definition has been left as is. Unfortunately, this
structure is used to register the pinctrl device, and prior to this
registration, we set the name and pins field.

Since this structure is shared across instances, that means that the latest
registered pinctrl device wins in setting the name, pins and pins numbers,
which is not really a good thing.
Signed-off-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

The way that reset is handled right now is that it is made optional for every
pinctrl driver, while actually, it isn't used at all for the main pin
controllers so far, and while it's mandatory for the A31's secondary pin
controller.

Move the reset functions out of the core and in the driver, where they can be
made mandatory.
Signed-off-by: NMaxime Ripard <maxime.ripard@free-electrons.com>

Move the pin description to a driver specific to be.

This is the final step toward retiring pinctrl-sunxi-pins.h that used to define
all the pins for all the Allwinner SoCs in a single header, that would have in
turn result in having these structures in the final binary as many times as the
header was included.

We can finally remove that header, and remove all the driver part of the
pinctrl-sunxi core.
Signed-off-by: NMaxime Ripard <maxime.ripard@free-electrons.com>

Move the pin description to a driver specific to be. This is one more step
toward retiring pinctrl-sunxi-pins.h that used to define all the pins for all
the Allwinner SoCs in a single header, that would have in turn result in having
these structures in the final binary as many times as the header was included.
Signed-off-by: NMaxime Ripard <maxime.ripard@free-electrons.com>

Move the pin description to a driver specific to be. This is one more step
toward retiring pinctrl-sunxi-pins.h that used to define all the pins for all
the Allwinner SoCs in a single header, that would have in turn result in having
these structures in the final binary as many times as the header was included.
Signed-off-by: NMaxime Ripard <maxime.ripard@free-electrons.com>

Move the pin description to a driver specific to be. This is one more step
toward retiring pinctrl-sunxi-pins.h that used to define all the pins for all
the Allwinner SoCs in a single header, that would have in turn result in having
these structures in the final binary as many times as the header was included.
Signed-off-by: NMaxime Ripard <maxime.ripard@free-electrons.com>

Move the pin description to a driver specific to be. This is one more step
toward retiring pinctrl-sunxi-pins.h that used to define all the pins for all
the Allwinner SoCs in a single header, that would have in turn result in having
these structures in the final binary as many times as the header was included.
Signed-off-by: NMaxime Ripard <maxime.ripard@free-electrons.com>

Move the pin description to a driver specific to be. This is one more step
toward retiring pinctrl-sunxi-pins.h that used to define all the pins for all
the Allwinner SoCs in a single header, that would have in turn result in having
these structures in the final binary as many times as the header was included.
Signed-off-by: NMaxime Ripard <maxime.ripard@free-electrons.com>

This will allow to have multiple drivers using the same core code, and
eventually, retire pinctrl-sunxi-pins.h
Signed-off-by: NMaxime Ripard <maxime.ripard@free-electrons.com>

The previous code was calling of_iomap, which doesn't do any resource
management, and doesn't call request_mem_region either. Use
devm_ioremap_resource that do both.
Signed-off-by: NMaxime Ripard <maxime.ripard@free-electrons.com>

This will allow to create numerous files without crippling the main pinctrl
directory.
Signed-off-by: NMaxime Ripard <maxime.ripard@free-electrons.com>

The pins description structure were declared as const, but the of_device_id
data magic was losing it silently.

Make sure we have it on both sides.

And now that we're using const, we can also remove the useless cast in probe.
Signed-off-by: NMaxime Ripard <maxime.ripard@free-electrons.com>

According to the MODULE_AUTHOR() comments, the author name should be
"Name <email>" or just "Name". Add the missing '>'.
Signed-off-by: NAntoine Ténart <antoine.tenart@free-electrons.com>
Acked-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

The A31 SoC define a reset line for the R_PIO block which needs to be
deasserted.

Try to retrieve a reset control and deassert if one was found.
Signed-off-by: NBoris BREZILLON <boris.brezillon@free-electrons.com>
Acked-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

The A31 SoC provides both PL and PM pio bank through the R_PIO block.

These pins all support gpio function and can bbe assigned to system
peripherals (like TWI, P2WI, JTAG, ...)
Signed-off-by: NBoris BREZILLON <boris.brezillon@free-electrons.com>
Acked-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Add support for multiple pin controller instances.

First remove the static definition of the sunxi gpio chip struct and fill
the dynamically struct instead.
Then define a new pin_base field in the sunxi_pinctrl_desc which will be
used to specify the gpiochip base pin.
Signed-off-by: NBoris BREZILLON <boris.brezillon@free-electrons.com>
Acked-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Disable the clk when failing to probe the pin controller device.
Signed-off-by: NBoris BREZILLON <boris.brezillon@free-electrons.com>
Acked-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Check the clk_prepare_enable return value to avoid false positive probe.
Signed-off-by: NBoris BREZILLON <boris.brezillon@free-electrons.com>
Acked-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Signed-off-by: NHans de Goede <hdegoede@redhat.com>
Acked-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

On tha Allwinner A20 SoC, the external interrupts on the pin controller
device are connected to the GIC. Without chained_irq_{enter, exit},
external GPIO interrupts, such as used by mmc core card detect, cause
the system to hang.

This issue was first encountered during my attempt to get out-of-band
interrupts for WiFi on the Cubietruck working. With David's new series
of sunci-mci using mmc slot-gpio for (GPIO interrupt based) card
detection, removing the SD card also causes my Cubietruck to hang. This
problem should extend to all Allwinner A20 based boards.

With this fix, the system no longer hangs when I remove or insert the
SD card. /proc/interrupts show that the interrupt has correctly fired.
However the system still does not detect card removal/insertion. I
believe this is another unrelated issue.

Cc: stable@vger.kernel.org
Signed-off-by: NChen-Yu Tsai <wens@csie.org>
Acked-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

Some GPIO users, such as fixed-regulator, request GPIO output with
initial value of 1. This was ignored by sunxi driver.

Cc: stable@vger.kernel.org
Signed-off-by: NChen-Yu Tsai <wens@csie.org>
Acked-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

This switches the two members of struct gpio_chip that were
defined as unsigned foo:1 to bool, because that is indeed what
they are. Switch all users in the gpio and pinctrl subsystems
to assign these values with true/false instead of 0/1. The
users outside these subsystems will survive since true/false
is 1/0, atleast we set some kind of more strict typing example.
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>

The valid offset range should be 0 ... chip->ngpio - 1.
Signed-off-by: NAxel Lin <axel.lin@ingics.com>
Signed-off-by: NLinus Walleij <linus.walleij@linaro.org>