Re: [PATCH 6/7] remoteproc: sunxi: add KUnit tests for da_to_va address translation

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] remoteproc: sunxi: test logic diverges from driver and misses PubSRAM C DA view
--

commit 2e1cbebc465781878f472c97dd7ec7e6f704ea63
Author: Tim Michals <tcmichals@gmail.com>

remoteproc: sunxi: add KUnit tests for da_to_va address translation

This commit introduces a KUnit test suite for the sunxi_rproc da_to_va()
function. It verifies address translation logic across various memory
window types and boundary conditions.

> diff --git a/drivers/remoteproc/sunxi_rproc_test.c b/drivers/remoteproc/sunxi_rproc_test.c
> new file mode 100644
> index 0000000000000..ea1a3331bc264
> --- /dev/null
> +++ b/drivers/remoteproc/sunxi_rproc_test.c

[ ... ]

> +/*
> + * Mirror of sunxi_rproc_da_to_va() logic from drivers/remoteproc/sunxi_rproc.c.
> + * Must be kept in sync with the driver. Divergence is a bug.
> + */
> +static void *test_da_to_va(struct test_sunxi_rproc *priv,
> +			   u64 da, size_t len, bool *is_iomem)
> +{

[Severity: Medium]
By duplicating the static driver logic into test_da_to_va() instead of
testing the actual sunxi_rproc_da_to_va() code, does this circumvent the
System Under Test and fundamentally defeat the purpose of regression testing?

Any future changes to the actual driver will not be tested by this suite,
and it relies entirely on manual synchronization.

> +	if (len == 0)
> +		return NULL;
> +
> +	/* 1. Dedicated MCU Local SRAM Space 0 */

[ ... ]

> +		/* High SRAM Space 0 fallback views */
> +		if (da >= 0x3ff80000 && (da + len) <= (0x3ff80000 + priv->r_sram_size)) {
> +			if (is_iomem)
> +				*is_iomem = true;
> +			return priv->r_sram_va + (da - 0x3ff80000);
> +		}
> +		if (da >= 0x3ffc0000 && (da + len) <= (0x3ffc0000 + priv->r_sram_size)) {
> +			if (is_iomem)
> +				*is_iomem = true;
> +			return priv->r_sram_va + (da - 0x3ffc0000);
> +		}
> +	}
> +
> +	/* 2. Switchable MCU Local SRAM Space 1 */

[Severity: Medium]
Does this duplicated logic introduce a silent coverage gap by completely
omitting the PubSRAM C DA view (0x00020000) block that is present in the
actual driver's translation logic?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260922034711.190253-1-tcmichals@gmail.com?part=6