Re: [PATCH 02/19] iommu/tegra: Do not use struct page as the handle for pts

[Jason Gunthorpe <jgg@nvidia.com>]

On Wed, Feb 05, 2025 at 07:28:37PM +0000, Robin Murphy wrote:
> On 2025-02-04 6:34 pm, Jason Gunthorpe wrote:
> > Instead use the virtual address and dma_map_single() like as->pd
> > uses. Introduce a small struct tegra_pt instead of void * to have some
> > clarity what is using this API and add compile safety during the
> > conversion.
> 
> Hmm, I'm not sure it's all that clear to have a major discrepancy between
> how different levels of pagetable are referenced

Well, with a trivial conversion the pd would be u32* and the pt would
be u32**. There is 1 usage of a tegra_pd and 7 of a tegra_pt.

Spraying u32** all over the place was pretty ugly. Removing
one of the *'s was more than I wanted to do on this old driver.

So, lets a type for pd.

> > @@ -596,21 +600,21 @@ static u32 *as_get_pte(struct tegra_smmu_as *as, dma_addr_t iova,
> >   	if (!as->pts[pde]) {
> >   		dma_addr_t dma;
> > -		dma = dma_map_page(smmu->dev, page, 0, SMMU_SIZE_PT,
> > -				   DMA_TO_DEVICE);
> > +		dma = dma_map_single(smmu->dev, pt, SMMU_SIZE_PD,
> 
> SMMU_SIZE_PT (yeah they're numerically the same, but still...)

Oops, copy pasteo

> > +				     DMA_TO_DEVICE);
> >   		if (dma_mapping_error(smmu->dev, dma)) {
> > -			__iommu_free_pages(page, 0);
> > +			iommu_free_page(pt);
> >   			return NULL;
> >   		}
> >   		if (!smmu_dma_addr_valid(smmu, dma)) {
> >   			dma_unmap_page(smmu->dev, dma, SMMU_SIZE_PT,
> >   				       DMA_TO_DEVICE);
> 
> All the unmaps should be converted to _single as well (and in patch #1) -
> the APIs are not officially interchangeable.

Missed, that got it on both patches

Thanks,
Jason