Re: [PATCH v3 32/32] iommu: remove iova_to_phys from domain_ops and io_pgtable_ops

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] The commit removes `.iova_to_phys` from `struct iommu_domain_ops` but fails to update the VeriSilicon IOMMU driver (`drivers/iommu/vsi-iommu.c`), causing a build failure and breaking IOVA translation.
--

commit 07242ed991899ab6da27ae54405dbc3801f90876
Author: Guanghui Feng <guanghuifeng@linux.alibaba.com>

iommu: remove iova_to_phys from domain_ops and io_pgtable_ops

This patch removes the deprecated .iova_to_phys interface from
struct iommu_domain_ops and struct io_pgtable_ops. It also cleans up the
fallback path in iommu_iova_to_phys_length() since all drivers are expected
to have migrated to .iova_to_phys_length.

> Now that all drivers implement iova_to_phys_length and all callers
> have migrated, remove the deprecated interfaces:

[Severity: High]
Is it possible the VeriSilicon IOMMU driver was missed during this migration?

Looking at drivers/iommu/vsi-iommu.c, it does not implement
.iova_to_phys_length and still assigns the removed field, which will cause
a build failure:

drivers/iommu/vsi-iommu.c:vsi_iommu_ops {
    ...
    .iova_to_phys       = vsi_iommu_iova_to_phys,
    ...
};

> diff --git a/drivers/iommu/iommu.c b/drivers/iommu/iommu.c
> index 320ea13488e7b..1ad4787925cd3 100644
> --- a/drivers/iommu/iommu.c
> +++ b/drivers/iommu/iommu.c
[ ... ]
> @@ -2572,20 +2570,10 @@ phys_addr_t iommu_iova_to_phys_length(struct iommu_domain *domain,
>  	if (mapped_length)
>  		*mapped_length = 0;
>  
> -	if (domain->ops->iova_to_phys_length)
> -		return domain->ops->iova_to_phys_length(domain, iova, mapped_length);
> -
> -	/* Fallback to legacy iova_to_phys without length info */
> -	if (!domain->ops->iova_to_phys)
> +	if (!domain->ops->iova_to_phys_length)
>  		return PHYS_ADDR_MAX;

[Severity: High]
Because vsi-iommu.c doesn't implement .iova_to_phys_length, won't this 
cause iommu_iova_to_phys_length() to unconditionally return PHYS_ADDR_MAX
for that driver?  This would break its IOVA-to-physical translation once the
build error is fixed.

> -	phys = domain->ops->iova_to_phys(domain, iova);
> -	if (!phys)
> -		return PHYS_ADDR_MAX;
> -
> -	if (mapped_length)
> -		*mapped_length = PAGE_SIZE;
> -	return phys;
> +	return domain->ops->iova_to_phys_length(domain, iova, mapped_length);
>  }
>  EXPORT_SYMBOL_GPL(iommu_iova_to_phys_length);

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260603151804.1963871-1-guanghuifeng@linux.alibaba.com?part=32