linux-kernel - Re: [PATCH] iio: adc: ad7606_spi: add offload scan mask check

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <aBSWmeRaPg5U8f7v@smile.fi.intel.com>
Date: Fri, 2 May 2025 12:55:37 +0300
From: Andy Shevchenko <andy@...nel.org>
To: David Lechner <dlechner@...libre.com>
Cc: Michael Hennerich <Michael.Hennerich@...log.com>,
	Jonathan Cameron <jic23@...nel.org>,
	Nuno Sá <nuno.sa@...log.com>,
	Angelo Dureghello <adureghello@...libre.com>,
	Jonathan Cameron <Jonathan.Cameron@...wei.com>,
	linux-iio@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] iio: adc: ad7606_spi: add offload scan mask check

On Wed, Apr 30, 2025 at 04:41:58PM -0500, David Lechner wrote:
> Validate the scan mask when SPI offloading is being used.
> 
> Since this family of ADCs is simultaneous sampling, there isn't a way
> to selectively disable channels when reading sample data. (Technically,
> AD7616 has a sequencer so could have some control, but that is for
> another day).
> 
> For "regular" IIO triggered buffer reads, this isn't a problem and the
> IIO core will demux the data and ignore data from disabled channels.
> However, since SPI offloading is done completely in hardware, we don't
> have a way to do the same. So before this patch, if less than all
> channels were enabled, the data would be misplaced in the buffer.
> 
> By adding a check in update_scan_mode, we can fail to enable the buffer
> instead of having bad data returned to userspace.

...

> +static int ad7606_spi_update_scan_mode(struct iio_dev *indio_dev,
> +				       const unsigned long *scan_mask)
> +{
> +	struct ad7606_state *st = iio_priv(indio_dev);
> +
> +	if (st->offload_en) {
> +		DECLARE_BITMAP(bitmap, AD760X_MAX_CHANNELS);
> +
> +		bitmap_fill(bitmap, AD760X_MAX_CHANNELS);
> +
> +		/*
> +		 * SPI offload requires that all channels are enabled since
> +		 * there isn't a way to selectively disable channels that get
> +		 * read (this is simultaneous sampling ADC) and the DMA buffer
> +		 * has no way of demuxing the data to filter out unwanted
> +		 * channels.
> +		 */
> +		if (!bitmap_equal(bitmap, scan_mask,
> +				  st->chip_info->num_adc_channels))
> +			return -EINVAL;


No need to have the local variable, just make sure that bitmap_weigth(FOO) == FOO.

	if (... && bitmap_weight(FOO) != FOO)
		return -EINVAL;

> +	}
> +
> +	return 0;
> +}

-- 
With Best Regards,
Andy Shevchenko