Re: [PATCH 3/4] iio: adc: ade9000: add support for ADE9078

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Pre-existing issues:
- [High] Duplicate and overlapping `scan_index` values assigned to phase B and C channels break the IIO buffer subsystem.
--

--- Patch [3]: [PATCH 3/4] iio: adc: ade9000: add support for ADE9078 ---
commit b9467a190f2a74b5ff3993276ef11d90860fe13b
Author: Antoniu Miclaus <antoniu.miclaus@analog.com>

iio: adc: ade9000: add support for ADE9078

This commit adds support for the ADE9078 polyphase energy metering device,
which is register-compatible with the ADE9000 but lacks dip/swell detection.
It introduces a dedicated channel table to omit those events and adds matching
chip_info for the full-scale codes.

> diff --git a/drivers/iio/adc/ade9000.c b/drivers/iio/adc/ade9000.c
> index 52ad87ef05d0c..0d5aa26f6667b 100644
> --- a/drivers/iio/adc/ade9000.c
> +++ b/drivers/iio/adc/ade9000.c
[ ... ]
> +#define ADE9000_DECLARE_CHANNELS(_name, altvoltage_rms)			      \
> +	static const struct iio_chan_spec _name[] = {			      \
> +		/* Phase A channels */					      \
> +		ADE9000_CURRENT_CHANNEL(ADE9000_PHASE_A_NR),		      \
> +		ADE9000_VOLTAGE_CHANNEL(ADE9000_PHASE_A_NR),		      \
[ ... ]
> +		/* Phase B channels */					      \
> +		ADE9000_CURRENT_CHANNEL(ADE9000_PHASE_B_NR),		      \
> +		ADE9000_VOLTAGE_CHANNEL(ADE9000_PHASE_B_NR),		      \

[Severity: High]
This is a pre-existing issue, but does this channel declaration cause the
driver probe to fail during IIO device registration?

Looking at the channel macros in drivers/iio/adc/ade9000.c:

ADE9000_CURRENT_CHANNEL(num) sets .scan_index = num
ADE9000_VOLTAGE_CHANNEL(num) sets .scan_index = num + 1

With ADE9000_PHASE_A_NR (0) and ADE9000_PHASE_B_NR (1), Phase A's voltage
channel gets scan_index 1, which perfectly collides with Phase B's current
channel (also scan_index 1).

Since ade9000_probe() calls devm_iio_kfifo_buffer_setup(), the
INDIO_BUFFER_SOFTWARE mode flag is set. When devm_iio_device_register() is
called, the IIO core invokes iio_check_unique_scan_index(), detects the
duplicate scan_index, and unconditionally returns -EINVAL.

It looks like ade9000_waveform_buffer_config() expects Phase B Current to
map to BIT(2), so shouldn't these macros use num * 2 and num * 2 + 1
instead?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260720110754.7674-1-antoniu.miclaus@analog.com?part=3