kernel/linux.git/drivers/iio/adc, branch v6.18.22

iio: adc: aspeed: clear reference voltage bits before configuring vref

2026-04-11T12:26:40+00:00

commit 7cf2f6ed8e7a3bf481ef70b6b4a2edb8abfa5c57 upstream. Ensures the reference voltage bits are cleared in the ADC engine control register before configuring the voltage reference. This avoids potential misconfigurations caused by residual bits. Fixes: 1b5ceb55fec2 ("iio: adc: aspeed: Support ast2600 adc.") Signed-off-by: Billy Tsai Cc: Signed-off-by: Jonathan Cameron Signed-off-by: Greg Kroah-Hartman

iio: adc: ade9000: move mutex init before IRQ registration

2026-04-11T12:26:40+00:00

commit 0206dd36418c104c0b3dea4ed7047e21eccb30b0 upstream. Move devm_mutex_init() before ade9000_request_irq() calls so that st->lock is initialized before any handler that depends on it can run. Fixes: 81de7b4619fc ("iio: adc: add ade9000 support") Signed-off-by: Antoniu Miclaus Reviewed-by: Andy Shevchenko Cc: Signed-off-by: Jonathan Cameron Signed-off-by: Greg Kroah-Hartman

iio: adc: ade9000: fix wrong register in CALIBBIAS case for active power

2026-04-11T12:26:39+00:00

commit 86133fb1ec36b2f5cec29d71fbae84877c3a1358 upstream. The switch statement in ade9000_write_raw() attempts to match chan->address against ADE9000_REG_AWATTOS (0x00F) to dispatch the calibration offset write for active power channels. However, chan->address is set via ADE9000_ADDR_ADJUST(ADE9000_REG_AWATT, num), so after masking the phase bits, tmp holds ADE9000_REG_AWATT (0x210), which never matches 0x00F. As a result, writing IIO_CHAN_INFO_CALIBBIAS for IIO_POWER always falls through to the default case and returns -EINVAL, making active power offset calibration silently broken. Fix this by matching against ADE9000_REG_AWATT instead, which is the actual base address stored in chan->address for watt channels. Reference:ADE9000 datasheet (Rev. B), AWATTOS is the offset correction register at 0x00F (p. 44), while AWATT is the total active power register at 0x210 (p. 48). Fixes: 81de7b4619fc ("iio: adc: add ade9000 support") Signed-off-by: Giorgi Tchankvetadze Reviewed-by: Antoniu Miclaus Cc: Signed-off-by: Jonathan Cameron Signed-off-by: Greg Kroah-Hartman

iio: adc: ade9000: fix wrong return type in streaming push

2026-04-11T12:26:39+00:00

commit 57b207e38d414a27fda9fff638a0d3e7ef16b917 upstream. The else branch of ade9000_iio_push_streaming() incorrectly returns IRQ_HANDLED on regmap_write failure. This function returns int (0 on success, negative errno on failure), so IRQ_HANDLED (1) would be misinterpreted as a non-error by callers. Return ret instead, consistent with every other error path in the function. Fixes: 81de7b4619fc ("iio: adc: add ade9000 support") Signed-off-by: Giorgi Tchankvetadze Reviewed-by: Antoniu Miclaus Cc: Signed-off-by: Jonathan Cameron Signed-off-by: Greg Kroah-Hartman

iio: adc: ti-ads7950: do not clobber gpio state in ti_ads7950_get()

2026-04-11T12:26:39+00:00

commit d20bbae6e5d408a8a7c2a4344d76dd1ac557a149 upstream. GPIO state was inadvertently overwritten by the result of spi_sync(), resulting in ti_ads7950_get() only returning 0 as GPIO state (or error). Fix this by introducing a separate variable to hold the state. Fixes: c97dce792dc8 ("iio: adc: ti-ads7950: add GPIO support") Reported-by: David Lechner Signed-off-by: Dmitry Torokhov Cc: Signed-off-by: Jonathan Cameron Signed-off-by: Greg Kroah-Hartman

iio: adc: ti-ads7950: normalize return value of gpio_get

2026-04-11T12:26:39+00:00

commit e2fa075d5ce1963e7cb7b0ac708ba567e5af66db upstream. The GPIO get callback is expected to return 0 or 1 (or a negative error code). Ensure that the value returned by ti_ads7950_get() for output pins is normalized to the [0, 1] range. Fixes: 86ef402d805d ("gpiolib: sanitize the return value of gpio_chip::get()") Reviewed-by: Andy Shevchenko Reviewed-by: Bartosz Golaszewski Reviewed-by: Linus Walleij Signed-off-by: Dmitry Torokhov Cc: Signed-off-by: Jonathan Cameron Signed-off-by: Greg Kroah-Hartman

iio: adc: ti-ads1119: Replace IRQF_ONESHOT with IRQF_NO_THREAD

2026-04-11T12:26:37+00:00

commit 36f6d4db3c5cb0f58fb02b1f54f9e86522d2f918 upstream. As there is no threaded handler, replace devm_request_threaded_irq() with devm_request_irq(), and as the handler calls iio_trigger_poll() which may not be called from a threaded handler replace IRQF_ONESHOT with IRQF_NO_THREAD. Since commit aef30c8d569c ("genirq: Warn about using IRQF_ONESHOT without a threaded handler"), the IRQ core checks IRQF_ONESHOT flag in IRQ request and gives a warning if there is no threaded handler. Fixes: a9306887eba4 ("iio: adc: ti-ads1119: Add driver") Signed-off-by: Felix Gu Reviewed-by: David Lechner Cc: Signed-off-by: Jonathan Cameron Signed-off-by: Greg Kroah-Hartman

iio: adc: ti-ads1119: Reinit completion before wait_for_completion_timeout()

2026-04-11T12:26:37+00:00

commit 2f168094177f8553a36046afce139001801ca917 upstream. The completion is not reinit before wait_for_completion_timeout(), so wait_for_completion_timeout() will return immediately after the first successful completion. Fixes: a9306887eba4 ("iio: adc: ti-ads1119: Add driver") Signed-off-by: Felix Gu Reviewed-by: Francesco Dolcini Cc: Signed-off-by: Jonathan Cameron Signed-off-by: Greg Kroah-Hartman

iio: adc: ti-ads1119: Fix unbalanced pm reference count in ds1119_single_conversion()

2026-04-11T12:26:37+00:00

commit 48a5c36577ebe0144f8ede70e59b59ea18b75089 upstream. In ads1119_single_conversion(), if pm_runtime_resume_and_get() fails, the code jumps to the pdown label, which calls pm_runtime_put_autosuspend(). Since pm_runtime_resume_and_get() automatically decrements the usage counter on failure, the subsequent call to pm_runtime_put_autosuspend() causes an unbalanced reference counter. Fixes: a9306887eba4 ("iio: adc: ti-ads1119: Add driver") Signed-off-by: Felix Gu Reviewed-by: João Paulo Gonçalves Cc: Signed-off-by: Jonathan Cameron Signed-off-by: Greg Kroah-Hartman

iio: adc: ti-adc161s626: use DMA-safe memory for spi_read()

2026-04-11T12:26:36+00:00

commit 768461517a28d80fe81ea4d5d03a90cd184ea6ad upstream. Add a DMA-safe buffer and use it for spi_read() instead of a stack memory. All SPI buffers must be DMA-safe. Since we only need up to 3 bytes, we just use a u8[] instead of __be16 and __be32 and change the conversion functions appropriately. Fixes: 4d671b71beef ("iio: adc: ti-adc161s626: add support for TI 1-channel differential ADCs") Signed-off-by: David Lechner Reviewed-by: Andy Shevchenko Cc: Signed-off-by: Jonathan Cameron Signed-off-by: Greg Kroah-Hartman