| Message ID | 6186e311cb6f64a787f87fd41e49a73f409b789c.1730136799.git.robin.murphy@arm.com |
|---|---|
| State | New |
| Headers | show |
| Series | Fix omap-iommu bitrot | expand |
On Mon, Oct 28, 2024 at 05:58:35PM +0000, Robin Murphy wrote: > It's no longer practical for the OMAP IOMMU driver to trick > arm_setup_iommu_dma_ops() into ignoring its presence, so let's use the > same tactic as other IOMMU API users on 32-bit ARM and explicitly kick > the arch code's dma_iommu_mapping out of the way to avoid problems. > > Fixes: 4720287c7bf7 ("iommu: Remove struct iommu_ops *iommu from arch_setup_dma_ops()") > Signed-off-by: Robin Murphy <robin.murphy@arm.com> > --- > drivers/remoteproc/omap_remoteproc.c | 17 +++++++++++++++++ > 1 file changed, 17 insertions(+) > > diff --git a/drivers/remoteproc/omap_remoteproc.c b/drivers/remoteproc/omap_remoteproc.c > index 9ae2e831456d..3260dd512491 100644 > --- a/drivers/remoteproc/omap_remoteproc.c > +++ b/drivers/remoteproc/omap_remoteproc.c > @@ -37,6 +37,10 @@ > > #include <linux/platform_data/dmtimer-omap.h> > > +#ifdef CONFIG_ARM_DMA_USE_IOMMU > +#include <asm/dma-iommu.h> > +#endif > + > #include "omap_remoteproc.h" > #include "remoteproc_internal.h" > > @@ -1323,6 +1327,19 @@ static int omap_rproc_probe(struct platform_device *pdev) > /* All existing OMAP IPU and DSP processors have an MMU */ > rproc->has_iommu = true; > > +#ifdef CONFIG_ARM_DMA_USE_IOMMU > + /* > + * Throw away the ARM DMA mapping that we'll never use, so it doesn't > + * interfere with the core rproc->domain and we get the right DMA ops. > + */ > + if (pdev->dev.archdata.mapping) { > + struct dma_iommu_mapping *mapping = to_dma_iommu_mapping(&pdev->dev); > + > + arm_iommu_detach_device(&pdev->dev); > + arm_iommu_release_mapping(mapping); > + } > +#endif > + Despite acknowledging this patch I never applied it, something I have corrected now. Thanks for Beleswar for bringing this to my attention. Regards, Mathieu > ret = omap_rproc_of_get_internal_memories(pdev, rproc); > if (ret) > return ret; > -- > 2.39.2.101.g768bb238c484.dirty >
diff --git a/drivers/remoteproc/omap_remoteproc.c b/drivers/remoteproc/omap_remoteproc.c index 9ae2e831456d..3260dd512491 100644 --- a/drivers/remoteproc/omap_remoteproc.c +++ b/drivers/remoteproc/omap_remoteproc.c @@ -37,6 +37,10 @@ #include <linux/platform_data/dmtimer-omap.h> +#ifdef CONFIG_ARM_DMA_USE_IOMMU +#include <asm/dma-iommu.h> +#endif + #include "omap_remoteproc.h" #include "remoteproc_internal.h" @@ -1323,6 +1327,19 @@ static int omap_rproc_probe(struct platform_device *pdev) /* All existing OMAP IPU and DSP processors have an MMU */ rproc->has_iommu = true; +#ifdef CONFIG_ARM_DMA_USE_IOMMU + /* + * Throw away the ARM DMA mapping that we'll never use, so it doesn't + * interfere with the core rproc->domain and we get the right DMA ops. + */ + if (pdev->dev.archdata.mapping) { + struct dma_iommu_mapping *mapping = to_dma_iommu_mapping(&pdev->dev); + + arm_iommu_detach_device(&pdev->dev); + arm_iommu_release_mapping(mapping); + } +#endif + ret = omap_rproc_of_get_internal_memories(pdev, rproc); if (ret) return ret;
It's no longer practical for the OMAP IOMMU driver to trick arm_setup_iommu_dma_ops() into ignoring its presence, so let's use the same tactic as other IOMMU API users on 32-bit ARM and explicitly kick the arch code's dma_iommu_mapping out of the way to avoid problems. Fixes: 4720287c7bf7 ("iommu: Remove struct iommu_ops *iommu from arch_setup_dma_ops()") Signed-off-by: Robin Murphy <robin.murphy@arm.com> --- drivers/remoteproc/omap_remoteproc.c | 17 +++++++++++++++++ 1 file changed, 17 insertions(+)