Re: [PATCH v6 6/8] dma-mapping: detect and configure IOMMU in of_dma_configure

[Will Deacon <will.deacon-5wv7dgnIgG8@public.gmane.org>]

On Wed, Dec 17, 2014 at 12:19:45AM +0000, Laurent Pinchart wrote:
> On Monday 15 December 2014 11:32:52 Will Deacon wrote:
> > On Sun, Dec 14, 2014 at 03:51:13PM +0000, Laurent Pinchart wrote:
> > > On Monday 01 December 2014 16:57:12 Will Deacon wrote:
> > > > +	of_dma_configure(&dev->dev);
> > > >  	if (of_device_add(dev) != 0) {
> > > > +		of_dma_deconfigure(&dev->dev);
> > > 
> > > Don't you also need to call of_dma_deconfigure() when the device is
> > > destroyed ? Otherwise the default domain created by arch_setup_dma_ops()
> > > will be leaked.
> >
> > Something like below?
> 
> Yes, something like that. This will however cause a dev_warn("Not attached") 
> message to be printed for devices that have no IOMMU, that should be fixed in 
> the arch code.

Good catch, additional patch below to fix that up.

Will

--->8

diff --git a/arch/arm/mm/dma-mapping.c b/arch/arm/mm/dma-mapping.c
index 09645f00bd17..c757f3d2d01e 100644
--- a/arch/arm/mm/dma-mapping.c
+++ b/arch/arm/mm/dma-mapping.c
@@ -2023,7 +2023,10 @@ static bool arm_setup_iommu_dma_ops(struct device *dev, u64 dma_base, u64 size,
 
 static void arm_teardown_iommu_dma_ops(struct device *dev)
 {
-       struct dma_iommu_mapping *mapping = dev->archdata.mapping;
+       struct dma_iommu_mapping *mapping = to_dma_iommu_mapping(dev);
+
+       if (!mapping)
+               return;
 
        arm_iommu_detach_device(dev);
        arm_iommu_release_mapping(mapping);