[PATCH 3/5] drm/tegra: Setup shared IOMMU domain after initialization

[Dmitry Osipenko]

23.01.2019 12:39, Thierry Reding пишет:
> From: Thierry Reding <treding at nvidia.com>
> 
> Move initialization of the shared IOMMU domain after the host1x device
> has been initialized. At this point all the Tegra DRM clients have been
> attached to the shared IOMMU domain.
> 
> This is important because Tegra186 and later use an ARM SMMU, for which
> the driver defers setting up the geometry for a domain until a device is
> attached to it. This is to ensure that the domain is properly set up for
> a specific ARM SMMU instance, which is unknown at allocation time.
> 
> Signed-off-by: Thierry Reding <treding at nvidia.com>
> ---
>  drivers/gpu/drm/tegra/drm.c | 54 ++++++++++++++++++++-----------------
>  1 file changed, 29 insertions(+), 25 deletions(-)
> 
> diff --git a/drivers/gpu/drm/tegra/drm.c b/drivers/gpu/drm/tegra/drm.c
> index 61dcbd218ffc..271c7a5fc954 100644
> --- a/drivers/gpu/drm/tegra/drm.c
> +++ b/drivers/gpu/drm/tegra/drm.c
> @@ -92,10 +92,6 @@ static int tegra_drm_load(struct drm_device *drm, unsigned long flags)
>  		return -ENOMEM;
>  
>  	if (iommu_present(&platform_bus_type)) {
> -		u64 carveout_start, carveout_end, gem_start, gem_end;
> -		struct iommu_domain_geometry *geometry;
> -		unsigned long order;
> -
>  		tegra->domain = iommu_domain_alloc(&platform_bus_type);
>  		if (!tegra->domain) {
>  			err = -ENOMEM;
> @@ -105,27 +101,6 @@ static int tegra_drm_load(struct drm_device *drm, unsigned long flags)
>  		err = iova_cache_get();
>  		if (err < 0)
>  			goto domain;
> -
> -		geometry = &tegra->domain->geometry;
> -		gem_start = geometry->aperture_start;
> -		gem_end = geometry->aperture_end - CARVEOUT_SZ;
> -		carveout_start = gem_end + 1;
> -		carveout_end = geometry->aperture_end;
> -
> -		order = __ffs(tegra->domain->pgsize_bitmap);
> -		init_iova_domain(&tegra->carveout.domain, 1UL << order,
> -				 carveout_start >> order);
> -
> -		tegra->carveout.shift = iova_shift(&tegra->carveout.domain);
> -		tegra->carveout.limit = carveout_end >> tegra->carveout.shift;
> -
> -		drm_mm_init(&tegra->mm, gem_start, gem_end - gem_start + 1);
> -		mutex_init(&tegra->mm_lock);
> -
> -		DRM_DEBUG("IOMMU apertures:\n");
> -		DRM_DEBUG("  GEM: %#llx-%#llx\n", gem_start, gem_end);
> -		DRM_DEBUG("  Carveout: %#llx-%#llx\n", carveout_start,
> -			  carveout_end);
>  	}
>  
>  	mutex_init(&tegra->clients_lock);
> @@ -159,6 +134,35 @@ static int tegra_drm_load(struct drm_device *drm, unsigned long flags)
>  	if (err < 0)
>  		goto fbdev;
>  
> +	if (tegra->domain) {
> +		u64 carveout_start, carveout_end, gem_start, gem_end;
> +		dma_addr_t start, end;
> +		unsigned long order;
> +
> +		start = tegra->domain->geometry.aperture_start;
> +		end = tegra->domain->geometry.aperture_end;
> +
> +		gem_start = start;
> +		gem_end = end - CARVEOUT_SZ;
> +		carveout_start = gem_end + 1;
> +		carveout_end = end;

Given that seems falcon can't address >32bit at least right now, will make sense to reserve carveout from the bottom.

[snip]