linux/drivers/pinctrl/intel/pinctrl-cherryview.c

1787 lines
50 KiB

pinctrl: cherryview: Preserve CHV_PADCTRL1_INVRXTX_TXDATA flag on GPIOs One some devices the GPIO should output the inverted value from what device-drivers / ACPI code expects. The reason for this is unknown, perhaps these systems use an external buffer chip on the GPIO which inverts the signal. The BIOS makes this work by setting the CHV_PADCTRL1_INVRXTX_TXDATA flag. Before this commit we would unconditionally clear all INVRXTX flags, including the CHV_PADCTRL1_INVRXTX_TXDATA flag when a GPIO is requested by a driver (from chv_gpio_request_enable()). This breaks systems using this setup. Specifically it is causing problems for systems with a goodix touchscreen, where the BIOS sets the INVRXTX_TXDATA flag on the GPIO used for the touchscreen's reset pin. The goodix touchscreen driver by defaults configures this pin as input (relying on the pull-up to keep it high), but the clearing of the INVRXTX_TXDATA flag done by chv_gpio_request_enable() causes it to be driven low for a brief time before the GPIO gets set to input mode. This causes the touchscreen controller to get reset. On most CHT devs with this touchscreen this leads to: [ 31.596534] Goodix-TS i2c-GDIX1001:00: i2c test failed attempt 1: -121 The driver retries this though and then everything is fine. But during reset the touchscreen uses its interrupt pin as bootstrap to determine which i2c address to use and on the Acer One S1003 the spurious reset caused by the clearing of the INVRXTX_TXDATA flag causes the controller to come back up again on the wrong i2c address, breaking things. This commit fixes both the -121 errors, as well as the total breakage on the Acer One S1003, by making chv_gpio_clear_triggering() not clear the INVRXTX_TXDATA flag if the pin is already configured as a GPIO. Note that chv_pinmux_set_mux() does still unconditionally clear the flag, so this only affects GPIO usage. Fixes: a7d4b171660c ("Input: goodix - add support for getting IRQ + reset GPIOs on Cherry Trail devices") Signed-off-by: Hans de Goede <hdegoede@redhat.com> Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
2020-09-04 19:21:41 +02:00
pinctrl: cherryview: Preserve CHV_PADCTRL1_INVRXTX_TXDATA flag on GPIOs One some devices the GPIO should output the inverted value from what device-drivers / ACPI code expects. The reason for this is unknown, perhaps these systems use an external buffer chip on the GPIO which inverts the signal. The BIOS makes this work by setting the CHV_PADCTRL1_INVRXTX_TXDATA flag. Before this commit we would unconditionally clear all INVRXTX flags, including the CHV_PADCTRL1_INVRXTX_TXDATA flag when a GPIO is requested by a driver (from chv_gpio_request_enable()). This breaks systems using this setup. Specifically it is causing problems for systems with a goodix touchscreen, where the BIOS sets the INVRXTX_TXDATA flag on the GPIO used for the touchscreen's reset pin. The goodix touchscreen driver by defaults configures this pin as input (relying on the pull-up to keep it high), but the clearing of the INVRXTX_TXDATA flag done by chv_gpio_request_enable() causes it to be driven low for a brief time before the GPIO gets set to input mode. This causes the touchscreen controller to get reset. On most CHT devs with this touchscreen this leads to: [ 31.596534] Goodix-TS i2c-GDIX1001:00: i2c test failed attempt 1: -121 The driver retries this though and then everything is fine. But during reset the touchscreen uses its interrupt pin as bootstrap to determine which i2c address to use and on the Acer One S1003 the spurious reset caused by the clearing of the INVRXTX_TXDATA flag causes the controller to come back up again on the wrong i2c address, breaking things. This commit fixes both the -121 errors, as well as the total breakage on the Acer One S1003, by making chv_gpio_clear_triggering() not clear the INVRXTX_TXDATA flag if the pin is already configured as a GPIO. Note that chv_pinmux_set_mux() does still unconditionally clear the flag, so this only affects GPIO usage. Fixes: a7d4b171660c ("Input: goodix - add support for getting IRQ + reset GPIOs on Cherry Trail devices") Signed-off-by: Hans de Goede <hdegoede@redhat.com> Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
2020-09-04 19:21:41 +02:00
pinctrl: cherryview: Preserve CHV_PADCTRL1_INVRXTX_TXDATA flag on GPIOs One some devices the GPIO should output the inverted value from what device-drivers / ACPI code expects. The reason for this is unknown, perhaps these systems use an external buffer chip on the GPIO which inverts the signal. The BIOS makes this work by setting the CHV_PADCTRL1_INVRXTX_TXDATA flag. Before this commit we would unconditionally clear all INVRXTX flags, including the CHV_PADCTRL1_INVRXTX_TXDATA flag when a GPIO is requested by a driver (from chv_gpio_request_enable()). This breaks systems using this setup. Specifically it is causing problems for systems with a goodix touchscreen, where the BIOS sets the INVRXTX_TXDATA flag on the GPIO used for the touchscreen's reset pin. The goodix touchscreen driver by defaults configures this pin as input (relying on the pull-up to keep it high), but the clearing of the INVRXTX_TXDATA flag done by chv_gpio_request_enable() causes it to be driven low for a brief time before the GPIO gets set to input mode. This causes the touchscreen controller to get reset. On most CHT devs with this touchscreen this leads to: [ 31.596534] Goodix-TS i2c-GDIX1001:00: i2c test failed attempt 1: -121 The driver retries this though and then everything is fine. But during reset the touchscreen uses its interrupt pin as bootstrap to determine which i2c address to use and on the Acer One S1003 the spurious reset caused by the clearing of the INVRXTX_TXDATA flag causes the controller to come back up again on the wrong i2c address, breaking things. This commit fixes both the -121 errors, as well as the total breakage on the Acer One S1003, by making chv_gpio_clear_triggering() not clear the INVRXTX_TXDATA flag if the pin is already configured as a GPIO. Note that chv_pinmux_set_mux() does still unconditionally clear the flag, so this only affects GPIO usage. Fixes: a7d4b171660c ("Input: goodix - add support for getting IRQ + reset GPIOs on Cherry Trail devices") Signed-off-by: Hans de Goede <hdegoede@redhat.com> Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
2020-09-04 19:21:41 +02:00
pinctrl: cherryview: Preserve CHV_PADCTRL1_INVRXTX_TXDATA flag on GPIOs One some devices the GPIO should output the inverted value from what device-drivers / ACPI code expects. The reason for this is unknown, perhaps these systems use an external buffer chip on the GPIO which inverts the signal. The BIOS makes this work by setting the CHV_PADCTRL1_INVRXTX_TXDATA flag. Before this commit we would unconditionally clear all INVRXTX flags, including the CHV_PADCTRL1_INVRXTX_TXDATA flag when a GPIO is requested by a driver (from chv_gpio_request_enable()). This breaks systems using this setup. Specifically it is causing problems for systems with a goodix touchscreen, where the BIOS sets the INVRXTX_TXDATA flag on the GPIO used for the touchscreen's reset pin. The goodix touchscreen driver by defaults configures this pin as input (relying on the pull-up to keep it high), but the clearing of the INVRXTX_TXDATA flag done by chv_gpio_request_enable() causes it to be driven low for a brief time before the GPIO gets set to input mode. This causes the touchscreen controller to get reset. On most CHT devs with this touchscreen this leads to: [ 31.596534] Goodix-TS i2c-GDIX1001:00: i2c test failed attempt 1: -121 The driver retries this though and then everything is fine. But during reset the touchscreen uses its interrupt pin as bootstrap to determine which i2c address to use and on the Acer One S1003 the spurious reset caused by the clearing of the INVRXTX_TXDATA flag causes the controller to come back up again on the wrong i2c address, breaking things. This commit fixes both the -121 errors, as well as the total breakage on the Acer One S1003, by making chv_gpio_clear_triggering() not clear the INVRXTX_TXDATA flag if the pin is already configured as a GPIO. Note that chv_pinmux_set_mux() does still unconditionally clear the flag, so this only affects GPIO usage. Fixes: a7d4b171660c ("Input: goodix - add support for getting IRQ + reset GPIOs on Cherry Trail devices") Signed-off-by: Hans de Goede <hdegoede@redhat.com> Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
2020-09-04 19:21:41 +02:00
pinctrl: cherryview: Do not allow the same interrupt line to be used by 2 pins It is impossible to use the same interrupt line for 2 pins, this will result in the interrupts only being delivered to the IRQ handler for the pin for which chv_gpio_irq_type() was called last. The pinctrl-cherryview.c code relies on the BIOS to correctly setup the interrupt line, but there is a BIOS bug on at least the Medion Akoya E1239T and the GPD win models where both INT33FF:02 pin 8, used by the powerbutton and INT33FF:02 pin 21 used as IRQ input for the accelerometer are mapped to interrupt line 0. This causes 2 problems: 1. The accelerometer IRQ does not work, since the power button is probed later taking over the intr_lines[0] slot. 2. Since the accelerometer IRQ is not marked as wakeup, interrupt line 0 gets masked on suspend, causing the power button to not work to wake the system from suspend. Likewise on the Lenovo Yogabook, which has a touchscreen as keyboard and the keyboard half of the tablet also has a Wacom digitizer, the BIOS by default assigns the same interrupt line to the GPIOs used for their interrupts. Fix these problems by adding a check for this and assigning a new interrupt line to the 2nd pin for which chv_gpio_irq_type() gets called. With this fix in place the following 2 messages show up in dmesg on the Medion Akoya E1239T and the GPD win: cherryview-pinctrl INT33FF:02: interrupt line 0 is used by both pin 21 and pin 8 cherryview-pinctrl INT33FF:02: changing the interrupt line for pin 8 to 15 And the following gets logged on the Lenovo Yogabook: cherryview-pinctrl INT33FF:01: interrupt-line 0 is used by both pin 49 and pin 56 cherryview-pinctrl INT33FF:01: changing the interrupt line for pin 56 to 7 Note commit 9747070c11d6 ("Input: axp20x-pek - always register interrupt handlers") was added as a work around for the power button not being able to wakeup the system. This relies on using the PMIC's connection to the power button but that only works on systems with the AXP288 PMIC. Once this fix has been merged that workaround can be removed. Cc: Yauhen Kharuzhy <jekhor@gmail.com> Signed-off-by: Hans de Goede <hdegoede@redhat.com> Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
2021-11-18 11:56:49 +01:00
pinctrl: cherryview: Do not allow the same interrupt line to be used by 2 pins It is impossible to use the same interrupt line for 2 pins, this will result in the interrupts only being delivered to the IRQ handler for the pin for which chv_gpio_irq_type() was called last. The pinctrl-cherryview.c code relies on the BIOS to correctly setup the interrupt line, but there is a BIOS bug on at least the Medion Akoya E1239T and the GPD win models where both INT33FF:02 pin 8, used by the powerbutton and INT33FF:02 pin 21 used as IRQ input for the accelerometer are mapped to interrupt line 0. This causes 2 problems: 1. The accelerometer IRQ does not work, since the power button is probed later taking over the intr_lines[0] slot. 2. Since the accelerometer IRQ is not marked as wakeup, interrupt line 0 gets masked on suspend, causing the power button to not work to wake the system from suspend. Likewise on the Lenovo Yogabook, which has a touchscreen as keyboard and the keyboard half of the tablet also has a Wacom digitizer, the BIOS by default assigns the same interrupt line to the GPIOs used for their interrupts. Fix these problems by adding a check for this and assigning a new interrupt line to the 2nd pin for which chv_gpio_irq_type() gets called. With this fix in place the following 2 messages show up in dmesg on the Medion Akoya E1239T and the GPD win: cherryview-pinctrl INT33FF:02: interrupt line 0 is used by both pin 21 and pin 8 cherryview-pinctrl INT33FF:02: changing the interrupt line for pin 8 to 15 And the following gets logged on the Lenovo Yogabook: cherryview-pinctrl INT33FF:01: interrupt-line 0 is used by both pin 49 and pin 56 cherryview-pinctrl INT33FF:01: changing the interrupt line for pin 56 to 7 Note commit 9747070c11d6 ("Input: axp20x-pek - always register interrupt handlers") was added as a work around for the power button not being able to wakeup the system. This relies on using the PMIC's connection to the power button but that only works on systems with the AXP288 PMIC. Once this fix has been merged that workaround can be removed. Cc: Yauhen Kharuzhy <jekhor@gmail.com> Signed-off-by: Hans de Goede <hdegoede@redhat.com> Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
2021-11-18 11:56:49 +01:00
pinctrl: cherryview: Do not allow the same interrupt line to be used by 2 pins It is impossible to use the same interrupt line for 2 pins, this will result in the interrupts only being delivered to the IRQ handler for the pin for which chv_gpio_irq_type() was called last. The pinctrl-cherryview.c code relies on the BIOS to correctly setup the interrupt line, but there is a BIOS bug on at least the Medion Akoya E1239T and the GPD win models where both INT33FF:02 pin 8, used by the powerbutton and INT33FF:02 pin 21 used as IRQ input for the accelerometer are mapped to interrupt line 0. This causes 2 problems: 1. The accelerometer IRQ does not work, since the power button is probed later taking over the intr_lines[0] slot. 2. Since the accelerometer IRQ is not marked as wakeup, interrupt line 0 gets masked on suspend, causing the power button to not work to wake the system from suspend. Likewise on the Lenovo Yogabook, which has a touchscreen as keyboard and the keyboard half of the tablet also has a Wacom digitizer, the BIOS by default assigns the same interrupt line to the GPIOs used for their interrupts. Fix these problems by adding a check for this and assigning a new interrupt line to the 2nd pin for which chv_gpio_irq_type() gets called. With this fix in place the following 2 messages show up in dmesg on the Medion Akoya E1239T and the GPD win: cherryview-pinctrl INT33FF:02: interrupt line 0 is used by both pin 21 and pin 8 cherryview-pinctrl INT33FF:02: changing the interrupt line for pin 8 to 15 And the following gets logged on the Lenovo Yogabook: cherryview-pinctrl INT33FF:01: interrupt-line 0 is used by both pin 49 and pin 56 cherryview-pinctrl INT33FF:01: changing the interrupt line for pin 56 to 7 Note commit 9747070c11d6 ("Input: axp20x-pek - always register interrupt handlers") was added as a work around for the power button not being able to wakeup the system. This relies on using the PMIC's connection to the power button but that only works on systems with the AXP288 PMIC. Once this fix has been merged that workaround can be removed. Cc: Yauhen Kharuzhy <jekhor@gmail.com> Signed-off-by: Hans de Goede <hdegoede@redhat.com> Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
2021-11-18 11:56:49 +01:00
pinctrl: cherryview: Do not allow the same interrupt line to be used by 2 pins It is impossible to use the same interrupt line for 2 pins, this will result in the interrupts only being delivered to the IRQ handler for the pin for which chv_gpio_irq_type() was called last. The pinctrl-cherryview.c code relies on the BIOS to correctly setup the interrupt line, but there is a BIOS bug on at least the Medion Akoya E1239T and the GPD win models where both INT33FF:02 pin 8, used by the powerbutton and INT33FF:02 pin 21 used as IRQ input for the accelerometer are mapped to interrupt line 0. This causes 2 problems: 1. The accelerometer IRQ does not work, since the power button is probed later taking over the intr_lines[0] slot. 2. Since the accelerometer IRQ is not marked as wakeup, interrupt line 0 gets masked on suspend, causing the power button to not work to wake the system from suspend. Likewise on the Lenovo Yogabook, which has a touchscreen as keyboard and the keyboard half of the tablet also has a Wacom digitizer, the BIOS by default assigns the same interrupt line to the GPIOs used for their interrupts. Fix these problems by adding a check for this and assigning a new interrupt line to the 2nd pin for which chv_gpio_irq_type() gets called. With this fix in place the following 2 messages show up in dmesg on the Medion Akoya E1239T and the GPD win: cherryview-pinctrl INT33FF:02: interrupt line 0 is used by both pin 21 and pin 8 cherryview-pinctrl INT33FF:02: changing the interrupt line for pin 8 to 15 And the following gets logged on the Lenovo Yogabook: cherryview-pinctrl INT33FF:01: interrupt-line 0 is used by both pin 49 and pin 56 cherryview-pinctrl INT33FF:01: changing the interrupt line for pin 56 to 7 Note commit 9747070c11d6 ("Input: axp20x-pek - always register interrupt handlers") was added as a work around for the power button not being able to wakeup the system. This relies on using the PMIC's connection to the power button but that only works on systems with the AXP288 PMIC. Once this fix has been merged that workaround can be removed. Cc: Yauhen Kharuzhy <jekhor@gmail.com> Signed-off-by: Hans de Goede <hdegoede@redhat.com> Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
2021-11-18 11:56:49 +01:00
pinctrl: cherryview: Do not allow the same interrupt line to be used by 2 pins It is impossible to use the same interrupt line for 2 pins, this will result in the interrupts only being delivered to the IRQ handler for the pin for which chv_gpio_irq_type() was called last. The pinctrl-cherryview.c code relies on the BIOS to correctly setup the interrupt line, but there is a BIOS bug on at least the Medion Akoya E1239T and the GPD win models where both INT33FF:02 pin 8, used by the powerbutton and INT33FF:02 pin 21 used as IRQ input for the accelerometer are mapped to interrupt line 0. This causes 2 problems: 1. The accelerometer IRQ does not work, since the power button is probed later taking over the intr_lines[0] slot. 2. Since the accelerometer IRQ is not marked as wakeup, interrupt line 0 gets masked on suspend, causing the power button to not work to wake the system from suspend. Likewise on the Lenovo Yogabook, which has a touchscreen as keyboard and the keyboard half of the tablet also has a Wacom digitizer, the BIOS by default assigns the same interrupt line to the GPIOs used for their interrupts. Fix these problems by adding a check for this and assigning a new interrupt line to the 2nd pin for which chv_gpio_irq_type() gets called. With this fix in place the following 2 messages show up in dmesg on the Medion Akoya E1239T and the GPD win: cherryview-pinctrl INT33FF:02: interrupt line 0 is used by both pin 21 and pin 8 cherryview-pinctrl INT33FF:02: changing the interrupt line for pin 8 to 15 And the following gets logged on the Lenovo Yogabook: cherryview-pinctrl INT33FF:01: interrupt-line 0 is used by both pin 49 and pin 56 cherryview-pinctrl INT33FF:01: changing the interrupt line for pin 56 to 7 Note commit 9747070c11d6 ("Input: axp20x-pek - always register interrupt handlers") was added as a work around for the power button not being able to wakeup the system. This relies on using the PMIC's connection to the power button but that only works on systems with the AXP288 PMIC. Once this fix has been merged that workaround can be removed. Cc: Yauhen Kharuzhy <jekhor@gmail.com> Signed-off-by: Hans de Goede <hdegoede@redhat.com> Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
2021-11-18 11:56:49 +01:00
pinctrl: cherryview: Do not allow the same interrupt line to be used by 2 pins It is impossible to use the same interrupt line for 2 pins, this will result in the interrupts only being delivered to the IRQ handler for the pin for which chv_gpio_irq_type() was called last. The pinctrl-cherryview.c code relies on the BIOS to correctly setup the interrupt line, but there is a BIOS bug on at least the Medion Akoya E1239T and the GPD win models where both INT33FF:02 pin 8, used by the powerbutton and INT33FF:02 pin 21 used as IRQ input for the accelerometer are mapped to interrupt line 0. This causes 2 problems: 1. The accelerometer IRQ does not work, since the power button is probed later taking over the intr_lines[0] slot. 2. Since the accelerometer IRQ is not marked as wakeup, interrupt line 0 gets masked on suspend, causing the power button to not work to wake the system from suspend. Likewise on the Lenovo Yogabook, which has a touchscreen as keyboard and the keyboard half of the tablet also has a Wacom digitizer, the BIOS by default assigns the same interrupt line to the GPIOs used for their interrupts. Fix these problems by adding a check for this and assigning a new interrupt line to the 2nd pin for which chv_gpio_irq_type() gets called. With this fix in place the following 2 messages show up in dmesg on the Medion Akoya E1239T and the GPD win: cherryview-pinctrl INT33FF:02: interrupt line 0 is used by both pin 21 and pin 8 cherryview-pinctrl INT33FF:02: changing the interrupt line for pin 8 to 15 And the following gets logged on the Lenovo Yogabook: cherryview-pinctrl INT33FF:01: interrupt-line 0 is used by both pin 49 and pin 56 cherryview-pinctrl INT33FF:01: changing the interrupt line for pin 56 to 7 Note commit 9747070c11d6 ("Input: axp20x-pek - always register interrupt handlers") was added as a work around for the power button not being able to wakeup the system. This relies on using the PMIC's connection to the power button but that only works on systems with the AXP288 PMIC. Once this fix has been merged that workaround can be removed. Cc: Yauhen Kharuzhy <jekhor@gmail.com> Signed-off-by: Hans de Goede <hdegoede@redhat.com> Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
2021-11-18 11:56:49 +01:00
pinctrl: cherryview: Do not allow the same interrupt line to be used by 2 pins It is impossible to use the same interrupt line for 2 pins, this will result in the interrupts only being delivered to the IRQ handler for the pin for which chv_gpio_irq_type() was called last. The pinctrl-cherryview.c code relies on the BIOS to correctly setup the interrupt line, but there is a BIOS bug on at least the Medion Akoya E1239T and the GPD win models where both INT33FF:02 pin 8, used by the powerbutton and INT33FF:02 pin 21 used as IRQ input for the accelerometer are mapped to interrupt line 0. This causes 2 problems: 1. The accelerometer IRQ does not work, since the power button is probed later taking over the intr_lines[0] slot. 2. Since the accelerometer IRQ is not marked as wakeup, interrupt line 0 gets masked on suspend, causing the power button to not work to wake the system from suspend. Likewise on the Lenovo Yogabook, which has a touchscreen as keyboard and the keyboard half of the tablet also has a Wacom digitizer, the BIOS by default assigns the same interrupt line to the GPIOs used for their interrupts. Fix these problems by adding a check for this and assigning a new interrupt line to the 2nd pin for which chv_gpio_irq_type() gets called. With this fix in place the following 2 messages show up in dmesg on the Medion Akoya E1239T and the GPD win: cherryview-pinctrl INT33FF:02: interrupt line 0 is used by both pin 21 and pin 8 cherryview-pinctrl INT33FF:02: changing the interrupt line for pin 8 to 15 And the following gets logged on the Lenovo Yogabook: cherryview-pinctrl INT33FF:01: interrupt-line 0 is used by both pin 49 and pin 56 cherryview-pinctrl INT33FF:01: changing the interrupt line for pin 56 to 7 Note commit 9747070c11d6 ("Input: axp20x-pek - always register interrupt handlers") was added as a work around for the power button not being able to wakeup the system. This relies on using the PMIC's connection to the power button but that only works on systems with the AXP288 PMIC. Once this fix has been merged that workaround can be removed. Cc: Yauhen Kharuzhy <jekhor@gmail.com> Signed-off-by: Hans de Goede <hdegoede@redhat.com> Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
2021-11-18 11:56:49 +01:00
pinctrl: cherryview: Fix irq_valid_mask calculation Commit 03c4749dd6c7 ("gpio / ACPI: Drop unnecessary ACPI GPIO to Linux GPIO translation") has made the cherryview gpio numbers sparse, to get a 1:1 mapping between ACPI pin numbers and gpio numbers in Linux. This has greatly simplified things, but the code setting the irq_valid_mask was not updated for this, so the valid mask is still in the old "compressed" numbering with the gaps in the pin numbers skipped, which is wrong as irq_valid_mask needs to be expressed in gpio numbers. This results in the following error on devices using pin 24 (0x0018) on the north GPIO controller as an ACPI event source: [ 0.422452] cherryview-pinctrl INT33FF:01: Failed to translate GPIO to IRQ This has been reported (by email) to be happening on a Caterpillar CAT T20 tablet and I've reproduced this myself on a Medion Akoya e2215t 2-in-1. This commit uses the pin number instead of the compressed index into community->pins to clear the correct bits in irq_valid_mask for GPIOs using GPEs for interrupts, fixing these errors and in case of the Medion Akoya e2215t also fixing the LID switch not working. Cc: stable@vger.kernel.org Fixes: 03c4749dd6c7 ("gpio / ACPI: Drop unnecessary ACPI GPIO to Linux GPIO translation") Signed-off-by: Hans de Goede <hdegoede@redhat.com> Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
2019-10-18 11:08:42 +02:00
gpio / ACPI: Drop unnecessary ACPI GPIO to Linux GPIO translation We added acpi_gpiochip_pin_to_gpio_offset() because there was a need to translate from ACPI GpioIo/GpioInt number to Linux GPIO number in the Cherryview pinctrl driver. This translation is necessary because Cherryview has gaps in the pin list and the driver used continuous GPIO number space in Linux side as follows: created GPIO range 0->7 ==> INT33FF:03 PIN 0->7 created GPIO range 8->19 ==> INT33FF:03 PIN 15->26 created GPIO range 20->25 ==> INT33FF:03 PIN 30->35 created GPIO range 26->33 ==> INT33FF:03 PIN 45->52 created GPIO range 34->43 ==> INT33FF:03 PIN 60->69 created GPIO range 44->54 ==> INT33FF:03 PIN 75->85 For example when ACPI GpioInt resource refers to GPIO 81 (SDMMC3_CD_B) we translate from pin 81 to the corresponding Linux GPIO number, which is 50. This number is then used when the GPIO is accessed through gpiolib. It turns out, this is not necessary at all. We can just pass 1:1 mapping between Linux GPIO numbers and pin numbers (including gaps) and the pinctrl core handles all the details automatically: created GPIO range 0->7 ==> INT33FF:03 PIN 0->7 created GPIO range 15->26 ==> INT33FF:03 PIN 15->26 created GPIO range 30->35 ==> INT33FF:03 PIN 30->35 created GPIO range 45->52 ==> INT33FF:03 PIN 45->52 created GPIO range 60->69 ==> INT33FF:03 PIN 60->69 created GPIO range 75->85 ==> INT33FF:03 PIN 75->85 Here GPIO 81 is exactly same than the hardware pin 81 (SDMMC3_CD_B). As an added bonus this simplifies both the ACPI GPIO core code and the Cherryview pinctrl driver. Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com> Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
2017-11-27 16:54:42 +03:00