GitLab

Philippe Gerum authored Sep 07, 2020 and Jan Kiszka committed Mar 07, 2021

This patch enables the Intel pinctrl/GPIO core driver to operate in a
pipelined interrupt system. However, it does not allow chained GPIO
IRQs to be handled from the head stage of such pipeline yet. In other
words, the chained GPIO interrupts can safely be handled from the
in-band stage when CONFIG_IPIPE is turned on, but cannot be routed to
a real-time application.

Enabling full support will require the I-pipe core to handle IRQs
chained from a shared parent interrupt natively, which it is not
implemented at the moment.
Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>

[ Upstream commit f3c75e7a9349d1d33eb53ddc1b31640994969f73 ]

When GPIO library asks pin control to set the bias, it doesn't pass
any value of it and argument is considered boolean (and this is true
for ACPI GpioIo() / GpioInt() resources, by the way). Thus, individual
drivers must behave well, when they got the resistance value of 1 Ohm,
i.e. transforming it to sane default.

In case of Intel pin control hardware the 5 kOhm sounds plausible
because on one hand it's a minimum of resistors present in all
hardware generations and at the same time it's high enough to minimize
leakage current (will be only 200 uA with the above choice).

Fixes: e57725ea

 ("pinctrl: intel: Add support for hardware debouncer")
Reported-by: Jamie McClymont <jamie@kwiius.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

When consumer requests a pin, in order to be on the safest side,
we switch it first to GPIO mode followed by immediate transition
to the input state. Due to posted writes it's luckily to be a single
I/O transaction.

However, if firmware or boot loader already configures the pin
to the GPIO mode, user expects no glitches for the requested pin.
We may check if the pin is pre-configured and leave it as is
till the actual consumer toggles its state to avoid glitches.

Fixes: 7981c001 ("pinctrl: intel: Add Intel Sunrisepoint pin controller and GPIO support")
Depends-on: f5a26acf

 ("pinctrl: intel: Initialize GPIO properly when used through irqchip")
Cc: stable@vger.kernel.org
Cc: fei.yang@intel.com
Reported-by: Oliver Barta <oliver.barta@aptiv.com>
Reported-by: Malin Jonsson <malin.jonsson@ericsson.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>

Keeping the IRQ chip definition static shares it with multiple instances of
the GPIO chip in the system. This is bad and now we get this warning from
GPIO library:

"detected irqchip that is shared with multiple gpiochips: please fix the driver."

Hence, move the IRQ chip definition from being driver static into the struct
intel_pinctrl. So a unique IRQ chip is used for each GPIO chip instance.

Fixes: ee1a6ca4 ("pinctrl: intel: Add Intel Broxton pin controller support")
Depends-on: 5ff56b01

 ("pinctrl: intel: Disable GPIO pin interrupts in suspend")
Reported-by: Federico Ricchiuto <fed.ricchiuto@gmail.com>
Suggested-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>

The intel_pin_to_gpio() function is only called by the
PM support functions and causes a warning when those are disabled:

drivers/pinctrl/intel/pinctrl-intel.c:841:12: error: unused function 'intel_pin_to_gpio' [-Werror,-Wunused-function]

Mark it __maybe_unused to suppress the warning.
Suggested-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Chris Chiu <chiu@endlessm.com>
Acked-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>

On Asus X571GT, GPIO 297 is configured as an interrupt and serves
for the touchpad. The touchpad will report input events much less
than expected after S3 suspend/resume, which results in extremely
slow cursor movement. However, the number of interrupts observed
from /proc/interrupts increases much more than expected even no
touching touchpad.

This is due to the value of PADCFG0 of PIN 225 for the interrupt
has been changed from 0x80800102 to 0x80100102. The GPIROUTIOXAPIC
is toggled on which results in the spurious interrupts. The PADCFG0
of PIN 225 is expected to be saved during suspend, but the 297 is
saved instead because the gpiochip_line_is_irq() expect the GPIO
offset but what's really passed to it is PIN number. In this case,
the /sys/kernel/debug/pinctrl/INT3450:00/gpio-ranges shows

288: INT3450:00 GPIOS [436 - 459] PINS [216 - 239]

So gpiochip_line_is_irq() returns true for GPIO offset 297, the
suspend routine spuriously saves the content for PIN 297 which
we expect to save for PIN 225.

This commit maps the PIN number to GPIO offset first in the
intel_pinctrl_should_save() to make sure the values for the
specific PINs can be correctly saved and then restored.

Fixes: c538b943

 ("pinctrl: intel: Only restore pins that are used by the driver")
Signed-off-by: Chris Chiu <chiu@endlessm.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>

Some firmwares would like to protect pads from being modified by OS
and at the same time provide them to OS as a resource. So, the driver
in such circumstances may request pad and may not change its state.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>

We don't need dev_err() messages when platform_get_irq() fails now that
platform_get_irq() prints an error message itself when something goes
wrong. Let's remove these prints with a simple semantic patch.

// <smpl>
@@
expression ret;
struct platform_device *E;
@@

ret =
(
platform_get_irq(E, ...)
|
platform_get_irq_byname(E, ...)
);

if ( \( ret < 0 \| ret <= 0 \) )
{
(
-if (ret != -EPROBE_DEFER)
-{ ...
-dev_err(...);
-... }
|
...
-dev_err(...);
)
...
}
// </smpl>

While we're here, remove braces on if statements that only have one
statement (manually).

Cc: Linus Walleij <linus.walleij@linaro.org>
Cc: linux-gpio@vger.kernel.org
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Mika Westerberg <mika.westerberg@linux.intel.com>
Cc: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Stephen Boyd <swboyd@chromium.org>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>

Replace hard coded constants with self-explanatory names, i.e.
use NSEC_PER_USEC for debounce calculus.

While here, add a unit suffix to debounce period constant.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>

Since some of the GPIO controllers use different Interrupt Status offset,
it make sense to provide it explicitly in the drivers.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>

There is more generic and simpler validation just against the nregs.
Using it allows to drop customization from the intel_get_padcfg().
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>

There is no need to duplicate the check which is done in the common
intel_pinctrl_probe().
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>

Use the new helper that wraps the calls to platform_get_resource()
and devm_ioremap_resource() together.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>

Commit a939bb57

 ("pinctrl: intel: implement gpio_irq_enable") was
added because clearing interrupt status bit is required to avoid
unexpected behavior.

Turns out the unmask callback also needs the fix, which can solve weird
IRQ triggering issues on I2C touchpad ELAN1200.
Signed-off-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>

Use GENMASK() macro for all definitions.
No functional change intended.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>

We better to use usual pattern for read-modify-update,
than doing some operations in definition block.

No functional change.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>

The touchpad of the ASUS laptops E403NA, X540NA, X541NA are not
responsive after suspend/resume. The following error message
shows after resume.
 i2c_hid i2c-ELAN1200:00: failed to reset device.

On these laptops, the touchpad interrupt is connected via a GPIO
pin which is controlled by Intel pinctrl. After system resumes,
the GPIO is in ACPI mode and no longer works as an IRQ.

This commit saves the HOSTSW_OWN value during suspend, make sure
the HOSTSW_OWN mode remains the same after resume.
Signed-off-by: Chris Chiu <chiu@endlessm.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>

In current driver, SET_LATE_SYSTEM_SLEEP_PM_OPS is used to install the
callbacks for suspend/resume.
GPIO pin may be used as the interrupt pin by some device. However, using
SET_LATE_SYSTEM_SLEEP_PM_OPS() to install the callbacks, the resume
callback is called after resume_device_irqs(). Unintended interrupts may
arrive due to resuming device irqs first, but the GPIO controller is not
properly restored.

Normally, for a SMP system, there are multiple cores, so even when there are
unintended interrupts, BSP gets the chance to initialize the GPIO chip soon.
But when there is only 1 core is active (other cores are offlined or
single core) during resume, it is more easily to observe the unintended
interrupts.

This patch renames the suspend/resume function by adding suffix "_noirq",
and installs the callbacks using SET_NOIRQ_SYSTEM_SLEEP_PM_OPS().
Signed-off-by: Binbin Wu <binbin.wu@intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>

Since there are no more users, unexport it and make static.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>

We should get 'driver_data' from 'struct device' directly. Going via
platform_device is an unneeded step back and forth.
Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>

The parameter 'community' had been spelled incorrectly.
Fix it here.

As a side effect it satisfies static checkers that issue
the following warnings:

drivers/pinctrl/intel/pinctrl-intel.c:845: warning: Function parameter or member 'community' not described in 'intel_gpio_to_pin'
drivers/pinctrl/intel/pinctrl-intel.c:845: warning: Excess function parameter 'commmunity' description in 'intel_gpio_to_pin'
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

Simple type conversion with no functional change implied.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

This reverts commit 55aedef5.

Commit 55aedef5 ("pinctrl: intel: Do pin translation when lock IRQ")
added special translation from GPIO number to hardware pin number to
irq_reqres/relres hooks to avoid failure when IRQs are requested. The
actual failure happened inside gpiochip_lock_as_irq() because it calls
gpiod_get_direction() and pinctrl-intel.c::intel_gpio_get_direction()
implementation originally missed the translation so the two hooks made
it work by skipping the ->get_direction() call entirely (it overwrote
the default GPIOLIB provided functions).

The proper fix that adds translation to GPIO callbacks was merged with
commit 96147db1 ("pinctrl: intel: Do pin translation in other GPIO
operations as well"). This allows us to use the default GPIOLIB provided
functions again.

In addition as find out by Benjamin Tissoires the two functions
(intel_gpio_irq_reqres()/intel_gpio_irq_relres()) now cause problems of
their own because they operate on pin numbers and pass that pin number
to gpiochip_lock_as_irq() which actually expects a GPIO number.

Link: https://bugzilla.kernel.org/show_bug.cgi?id=199911
Fixes: 55aedef5

 ("pinctrl: intel: Do pin translation when lock IRQ")
Reported-and-tested-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Acked-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

For some reason I thought GPIOLIB handles translation from GPIO ranges
to pinctrl pins but it turns out not to be the case. This means that
when GPIOs operations are performed for a pin controller having a custom
GPIO base such as Cannon Lake and Ice Lake incorrect pin number gets
used internally.

Fix this in the same way we did for lock/unlock IRQ operations and
translate the GPIO number to pin before using it.

Fixes: a60eac32

 ("pinctrl: intel: Allow custom GPIO base for pad groups")
Reported-by: Rajat Jain <rajatja@google.com>
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Tested-by: Rajat Jain <rajatja@google.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

Introduce intel_pinctrl_probe_by_hid() internal API to simplify drivers,
which are using ACPI _HID to distinguish which SoC data needs to be used
when being probed.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

Introduce intel_pinctrl_probe_by_uid() internal API to simplify drivers,
which are using ACPI _UID to distinguish which SoC data needs to be used
when being probed.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

The parameter 'community' had been spelled incorrectly.
Fix it here.

As a side effect it satisfies static checkers that issue
the following warnings:

drivers/pinctrl/intel/pinctrl-intel.c:845: warning: Function parameter or member 'community' not described in 'intel_gpio_to_pin'
drivers/pinctrl/intel/pinctrl-intel.c:845: warning: Excess function parameter 'commmunity' description in 'intel_gpio_to_pin'
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

gpiochip_lock_as_irq() may return a few error codes,
do not shadow them by -EINVAL and let caller to decide.

No functional change intended.

Cc: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

Default GPIOLIB callbacks for request and release IRQ do not do a GPIO
to pin translation which is necessary for Intel hardware, such as Intel
Cannonlake. Absence of the translation prevents some pins to be locked
as IRQ due to direction check. Introduce own callbacks to make
translation possible to avoid above issue.

Fixes: a60eac32

 ("pinctrl: intel: Allow custom GPIO base for pad groups")
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

Reduce size of duplicated comments by switching to use SPDX identifier.

No functional change.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

Allows querying GPIO direction from the pad config register.
If the pad is not in GPIO mode, return an error.
Signed-off-by: Javier Arteaga <javier@emutex.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

In the (unlikely) event that community->ngpps is zero, or if every
gpp->gpio_base is less than zero, then an ininitialized value in
ret is returned by function intel_gpio_add_pin_ranges. Fix this by
ensuring ret is initialized to zero.  It's a moot point, but I think
it is worthwhile ensuring this corner case is fixed.

Detected by CoverityScan, CID#1462415 ("Uninitialized scalar variable")

Fixes: a60eac32

 ("pinctrl: intel: Allow custom GPIO base for pad groups")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

When a GPIO is requested using gpiod_get_* APIs the intel pinctrl driver
switches the pin to GPIO mode and makes sure interrupts are routed to
the GPIO hardware instead of IOAPIC. However, if the GPIO is used
directly through irqchip, as is the case with many I2C-HID devices where
I2C core automatically configures interrupt for the device, the pin is
not initialized as GPIO. Instead we rely that the BIOS configures the
pin accordingly which seems not to be the case at least in Asus X540NA
SKU3 with Focaltech touchpad.

When the pin is not properly configured it might result weird behaviour
like interrupts suddenly stop firing completely and the touchpad stops
responding to user input.

Fix this by properly initializing the pin to GPIO mode also when it is
used directly through irqchip.

Fixes: 7981c001

 ("pinctrl: intel: Add Intel Sunrisepoint pin controller and GPIO support")
Reported-by: Daniel Drake <drake@endlessm.com>
Reported-and-tested-by: Chris Chiu <chiu@endlessm.com>
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

Currently we always have direct mapping between GPIO numbers and the
hardware pin numbers. However, there are cases where that's not the case
anymore (more about this in the next patch). Instead we need to be able
to specify custom GPIO base for certain pad groups.

To support this, add a new field (gpio_base) to the pad group structure
and update the core Intel pinctrl driver to handle this accordingly.
Passing 0 as gpio_base will use direct mapping so the existing drivers
do not need to be modified. Passing -1 excludes the whole pad group from
having GPIO mapping.
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

In order to consolidate the multiple ways to associate an IRQ chip with
a GPIO chip, move more fields into the new struct gpio_irq_chip.
Signed-off-by: Thierry Reding <treding@nvidia.com>
Acked-by: Grygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

Some GPIO blocks have the interrupt status (GPI_IS) offset different
than it normally is, so make it configurable. If no offset is specified
we use the default.

While there remove two unused constants from the core driver.
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

In the same way as it's done in pinctrl-cherryview.c we would provide
a readback TX buffer state.

Fixes: 17fab473

 ("pinctrl: intel: Set pin direction properly")
Reported-by: "Bourque, Francis" <francis.bourque@intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Tested-by: "Bourque, Francis" <francis.bourque@intel.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

Decrease indentation in intel_gpio_set() to make it looking slightly better
and be in align with intel_gpio_get().
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

The fix prevents unintended wakes from second level GPIO pin interrupts.

On some Intel Kabylake platforms, it is observed that GPIO pin interrupts
can wake the platform from suspend-to-idle, even though the IRQ is not
configured as IRQF_NO_SUSPEND or enable_irq_wake().

This can cause undesired wakes on Mobile devices such as Laptops and
Chromebook devices. For example a headset jack insertion is not a desired
wake source on Chromebook devices.

The pinctrl-intel (GPIO controller) driver implements a "Shared IRQ" model.
All GPIO pin interrupts are OR'ed and mapped to a first level IRQ14 (or
IRQ15). The driver registers an irq_chip struct and maps an irq_domain for
the GPIO pin interrupts. The IRQ14 handler demuxes and calls the second
level IRQ for the respective pin.

In the suspend entry flow, at suspend_noirq stage, the kernel disables IRQs
that are not marked for wake. The pinctrl-intel driver does not implement a
irq_disable()  callback (to take advantage of lazy disabling). The
pinctrl-intel GPIO interrupts are not disabled in hardware during suspend
entry, and thus are able to wake the SoC out of suspend-to-idle.

This patch sets the IRQCHIP_MASK_ON_SUSPEND flag for the GPIO irq_chip, to
disable the second level interrupts at suspend_noirq stage via the irq_mask
callbacks. The irq_mask callback disables the IRQs in hardware by
programming the corresponding GPIO pad registers. Only IRQs that are not
marked for wake are disabled.
Signed-off-by: Rushikesh S Kadam <rushikesh.s.kadam@intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Reviewed-and-tested-by: Rajneesh Bhardwaj <rajneesh.bhardwaj@intel.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>

On some SoCs not all pins in a group use the same mode when a certain
function is muxed out of them. This makes it possible to specify mode per
pin as an array instead in addition to single integer.
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>