Re: [PATCH v3 4/5] iio: adc: ltc2378: Enable high-speed data capture

[David Lechner <dlechner@baylibre.com>]

On 6/16/26 9:04 PM, Marcelo Schmitt wrote:
> Make use of SPI transfer offloading to speed up data capture, enabling data
> acquisition at faster sample rates (up to 2 MSPS).
> 
> Signed-off-by: Marcelo Schmitt <marcelo.schmitt@analog.com>
> ---
> Change log v2 -> v3:
> - Fixed the evaluation loop conditions for CNV PWM and SPI Engine trigger PWM,
>   avoiding potential infinite loop if and CPU stall.
> - Added comment to about initial PWM disable.
> - Adjusted SPI offload setup initialization to not print error on a valid condition.
> - Fully initialize IIO channel scan_type.
> - Reworked to make offload support not imply all dependencies to be built in.
> - Made sampling_frequency a buffer attribute.
> - Made offload support not require DMA and other features to be built in.
> - Now using same scan_type configuration for all use cases.
> 
>  drivers/iio/adc/Kconfig                  |  19 ++
>  drivers/iio/adc/Makefile                 |   6 +
>  drivers/iio/adc/ltc2378-lib-core.c       |  35 +++
>  drivers/iio/adc/ltc2378-offload-buffer.c | 305 +++++++++++++++++++++++
>  drivers/iio/adc/ltc2378.c                |  46 ++++
>  drivers/iio/adc/ltc2378.h                |  42 ++++
>  6 files changed, 453 insertions(+)
>  create mode 100644 drivers/iio/adc/ltc2378-lib-core.c
>  create mode 100644 drivers/iio/adc/ltc2378-offload-buffer.c
> 
> diff --git a/drivers/iio/adc/Kconfig b/drivers/iio/adc/Kconfig
> index 2b8203451367..f96d9262b891 100644
> --- a/drivers/iio/adc/Kconfig
> +++ b/drivers/iio/adc/Kconfig
> @@ -944,6 +944,9 @@ config LTC2378
>  	tristate "Analog Devices LTC2378 ADC driver"
>  	depends on SPI
>  	depends on GPIOLIB
> +	select LTC2378_LIB
> +	select LTC2378_LIB_OFFLOAD_BUFFER if SPI_OFFLOAD && PWM && SPI_OFFLOAD_TRIGGER_PWM && IIO_BUFFER && IIO_BUFFER_DMAENGINE
> +	select LTC2378_LIB_TRIGGERED_BUFFER if IIO_BUFFER
>  	help
>  	  Say yes here to build support for Analog Devices LTC2378-20 and
>  	  similar analog to digital converters.
> @@ -2027,3 +2030,19 @@ config XILINX_AMS
>  	  xilinx-ams.
>  
>  endmenu
> +
> +config LTC2378_LIB
> +	tristate
> +	help
> +	  Say yes here to build support for buffered data capture with LTC2378
> +
> +config LTC2378_LIB_OFFLOAD_BUFFER
> +	bool
> +	help
> +	  Say yes here to build support for high speed data capture with LTC2378
> +
> +config LTC2378_LIB_TRIGGERED_BUFFER
> +	bool
> +	select IIO_TRIGGERED_BUFFER
> +	help

Why do these need to be compile-time options to only select one or the other?
In all other SPI offload ADCs we've done so far, they always support both and
gets selected at runtime based on devicetree config.

> +	  Say yes here to build support for buffered data capture with LTC2378
> diff --git a/drivers/iio/adc/Makefile b/drivers/iio/adc/Makefile
> index 1814fb78dde3..109cd39237c9 100644
> --- a/drivers/iio/adc/Makefile
> +++ b/drivers/iio/adc/Makefile
> @@ -82,6 +82,12 @@ obj-$(CONFIG_LPC18XX_ADC) += lpc18xx_adc.o
>  obj-$(CONFIG_LPC32XX_ADC) += lpc32xx_adc.o
>  obj-$(CONFIG_LTC2309) += ltc2309.o
>  obj-$(CONFIG_LTC2378) += ltc2378.o
> +
> +ltc2378_lib-y += ltc2378-lib-core.o
> +ltc2378_lib-$(CONFIG_LTC2378_LIB_OFFLOAD_BUFFER) += ltc2378-offload-buffer.o
> +ltc2378_lib-$(CONFIG_LTC2378_LIB_TRIGGERED_BUFFER) += ltc2378-triggered-buffer.o
> +obj-$(CONFIG_LTC2378_LIB) += ltc2378_lib.o

Why do these need to be split into separate files? The driver isn't that
long, so seems better to just do it all in one file to make it easier to
read.

> +
>  obj-$(CONFIG_LTC2471) += ltc2471.o
>  obj-$(CONFIG_LTC2485) += ltc2485.o
>  obj-$(CONFIG_LTC2496) += ltc2496.o ltc2497-core.o
> diff --git a/drivers/iio/adc/ltc2378-lib-core.c b/drivers/iio/adc/ltc2378-lib-core.c
> new file mode 100644
> index 000000000000..1160f4324d01
> --- /dev/null
> +++ b/drivers/iio/adc/ltc2378-lib-core.c
> @@ -0,0 +1,35 @@
> +// SPDX-License-Identifier: GPL-2.0+
> +/*
> + * Analog Devices LTC2378 ADC series driver
> + *
> + * Copyright (C) 2026 Analog Devices Inc.
> + * Author: Marcelo Schmitt <marcelo.schmitt@analog.com>
> + */
> +
> +#include <linux/err.h>
> +#include <linux/iio/iio.h>
> +
> +#include "ltc2378.h"
> +
> +int ltc2378_lib_buffer_setup(struct iio_dev *indio_dev, struct ltc2378_state *st)
> +{
> +	struct device *dev = &st->spi->dev;
> +	int ret;
> +
> +	ret = __ltc2378_set_offload_ops(st);
> +	if (ret == -EOPNOTSUPP)
> +		return 0; /* Let device setup complete without buffer support */
> +
> +	if (!ret)
> +		ret = st->ops->buffer_setup(indio_dev, st);
> +
> +	if (ret)
> +		return dev_err_probe(dev, ret, "error on SPI offload setup\n");

Would be better to just return early instead of doing the !ret check.

> +
> +	return 0;
> +}
> +EXPORT_SYMBOL_NS_GPL(ltc2378_lib_buffer_setup, "IIO_LTC2378");
> +

...

> diff --git a/drivers/iio/adc/ltc2378.c b/drivers/iio/adc/ltc2378.c
> index 88582bdcd6a6..bf17b202230b 100644
> --- a/drivers/iio/adc/ltc2378.c
> +++ b/drivers/iio/adc/ltc2378.c
> @@ -17,6 +17,7 @@
>  #include <linux/regulator/consumer.h>
>  #include <linux/spi/spi.h>
>  #include <linux/types.h>
> +#include <linux/units.h>
>  
>  #include <linux/iio/iio.h>
>  #include <linux/iio/types.h>
> @@ -26,120 +27,160 @@
>  static const struct ltc2378_chip_info ltc2338_18_chip_info = {
>  	.name = "ltc2338-18",
>  	.resolution = 18,
> +	.max_sample_rate_hz = HZ_PER_MHZ,

1 * HZ_PER_MHZ would make a bit more sense

> +	.tconv_ns = 527,
>  	.bipolar = true,
>  };
>