Re: [PATCH 1/2] mfd: Constify mfd_cell arrays

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] mfd: wm97xx-core: rodata write and const discard in wm97xx_ac97_probe()
--

commit d1e87419971e113856988eb45a5e61c55d61ecd6
Author: Krzysztof Kozlowski <krzysztof.kozlowski@oss.qualcomm.com>

mfd: Constify mfd_cell arrays

This commit marks several struct mfd_cell arrays across drivers/mfd/ as
const so they can reside in read-only memory when passed to
mfd_add_devices().

> diff --git a/drivers/mfd/wm97xx-core.c b/drivers/mfd/wm97xx-core.c
> index 1566a9b04b6a0..c0ddcd5219b57 100644
> --- a/drivers/mfd/wm97xx-core.c
> +++ b/drivers/mfd/wm97xx-core.c
> @@ -104,7 +104,7 @@ static const struct regmap_config wm9705_regmap_config = {
>  	.writeable_reg = wm97xx_writeable_reg,
>  };
>  
> -static struct mfd_cell wm9705_cells[] = {
> +static const struct mfd_cell wm9705_cells[] = {
>  	{ .name = "wm9705-codec", },
>  	{ .name = "wm97xx-ts", },
>  };

[Severity: High]
Can wm9705_cells be made const here?

In drivers/mfd/wm97xx-core.c, wm97xx_ac97_probe() declares cells as a
non-const pointer, assigns wm9705_cells to it when probing a
WM9705_VENDOR_ID device, and then mutates the array elements in place:

drivers/mfd/wm97xx-core.c:wm97xx_ac97_probe() {
	...
	struct mfd_cell *cells;
	...
	case WM9705_VENDOR_ID:
		config = &wm9705_regmap_config;
		cells = wm9705_cells;
		nb_cells = ARRAY_SIZE(wm9705_cells);
		break;
	...
	for (i = 0; i < nb_cells; i++) {
		cells[i].platform_data = codec_pdata;
		cells[i].pdata_size = sizeof(*codec_pdata);
	}
}

Won't this discard the const qualifier at build time and fault when writing
to read-only memory during WM9705 device probe?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261030071137.100954-3-krzysztof.kozlowski@oss.qualcomm.com?part=1