| Message ID | 20190111181731.11782-2-hch@lst.de (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [1/3] dma-mapping: remove the default map_resource implementation | expand |
On 11/01/2019 18:17, Christoph Hellwig wrote: > Just returning the physical address when not map_resource method is > present is highly dangerous as it doesn't take any offset in the > direct mapping into account and does the completely wrong thing for > IOMMUs. Instead provide a proper implementation in the direct mapping > code, and also wire it up for arm and powerpc. Ignoring the offset was kind of intentional there, because at the time I was primarily thinking about it in terms of the Keystone 2 platform where the peripherals are all in the same place (0-2GB) in both the bus and CPU physical address maps, and only the view of RAM differs between the two (2-4GB vs. 32-34GB). However, on something like BCM283x, the peripherals region is also offset from its bus address in the CPU view, but at a *different* offset relative to that of RAM. Fortunately, I'm not aware of any platform which has a DMA engine behind an IOMMU (and thus *needs* to use dma_map_resource() to avoid said IOMMU blocking the slave device register reads/writes) and also has any nonzero offsets, and AFAIK the IOMMU-less platforms above aren't using dma_map_resource() at all, so this change shouldn't actually break anything, but I guess we have a bit of a problem making it truly generic and robust :( Is this perhaps another shove in the direction of overhauling dma_pfn_offset into an arbitrary "DMA ranges" lookup table? > Signed-off-by: Christoph Hellwig <hch@lst.de> > --- > arch/arm/mm/dma-mapping.c | 2 ++ > arch/powerpc/kernel/dma-swiotlb.c | 1 + > arch/powerpc/kernel/dma.c | 1 + > include/linux/dma-mapping.h | 12 +++++++----- > kernel/dma/direct.c | 14 ++++++++++++++ > 5 files changed, 25 insertions(+), 5 deletions(-) > > diff --git a/arch/arm/mm/dma-mapping.c b/arch/arm/mm/dma-mapping.c > index f1e2922e447c..3c8534904209 100644 > --- a/arch/arm/mm/dma-mapping.c > +++ b/arch/arm/mm/dma-mapping.c > @@ -188,6 +188,7 @@ const struct dma_map_ops arm_dma_ops = { > .unmap_page = arm_dma_unmap_page, > .map_sg = arm_dma_map_sg, > .unmap_sg = arm_dma_unmap_sg, > + .map_resource = dma_direct_map_resource, > .sync_single_for_cpu = arm_dma_sync_single_for_cpu, > .sync_single_for_device = arm_dma_sync_single_for_device, > .sync_sg_for_cpu = arm_dma_sync_sg_for_cpu, > @@ -211,6 +212,7 @@ const struct dma_map_ops arm_coherent_dma_ops = { > .get_sgtable = arm_dma_get_sgtable, > .map_page = arm_coherent_dma_map_page, > .map_sg = arm_dma_map_sg, > + .map_resource = dma_direct_map_resource, > .dma_supported = arm_dma_supported, > }; > EXPORT_SYMBOL(arm_coherent_dma_ops); > diff --git a/arch/powerpc/kernel/dma-swiotlb.c b/arch/powerpc/kernel/dma-swiotlb.c > index 7d5fc9751622..fbb2506a414e 100644 > --- a/arch/powerpc/kernel/dma-swiotlb.c > +++ b/arch/powerpc/kernel/dma-swiotlb.c > @@ -55,6 +55,7 @@ const struct dma_map_ops powerpc_swiotlb_dma_ops = { > .dma_supported = swiotlb_dma_supported, > .map_page = dma_direct_map_page, > .unmap_page = dma_direct_unmap_page, > + .map_resource = dma_direct_map_resource, > .sync_single_for_cpu = dma_direct_sync_single_for_cpu, > .sync_single_for_device = dma_direct_sync_single_for_device, > .sync_sg_for_cpu = dma_direct_sync_sg_for_cpu, > diff --git a/arch/powerpc/kernel/dma.c b/arch/powerpc/kernel/dma.c > index b1903ebb2e9c..258b9e8ebb99 100644 > --- a/arch/powerpc/kernel/dma.c > +++ b/arch/powerpc/kernel/dma.c > @@ -273,6 +273,7 @@ const struct dma_map_ops dma_nommu_ops = { > .dma_supported = dma_nommu_dma_supported, > .map_page = dma_nommu_map_page, > .unmap_page = dma_nommu_unmap_page, > + .map_resource = dma_direct_map_resource, > .get_required_mask = dma_nommu_get_required_mask, > #ifdef CONFIG_NOT_COHERENT_CACHE > .sync_single_for_cpu = dma_nommu_sync_single, > diff --git a/include/linux/dma-mapping.h b/include/linux/dma-mapping.h > index cef2127e1d70..d3087829a6df 100644 > --- a/include/linux/dma-mapping.h > +++ b/include/linux/dma-mapping.h > @@ -208,6 +208,8 @@ dma_addr_t dma_direct_map_page(struct device *dev, struct page *page, > unsigned long attrs); > int dma_direct_map_sg(struct device *dev, struct scatterlist *sgl, int nents, > enum dma_data_direction dir, unsigned long attrs); > +dma_addr_t dma_direct_map_resource(struct device *dev, phys_addr_t paddr, > + size_t size, enum dma_data_direction dir, unsigned long attrs); > > #if defined(CONFIG_ARCH_HAS_SYNC_DMA_FOR_DEVICE) || \ > defined(CONFIG_SWIOTLB) > @@ -346,19 +348,19 @@ static inline dma_addr_t dma_map_resource(struct device *dev, > unsigned long attrs) > { > const struct dma_map_ops *ops = get_dma_ops(dev); > - dma_addr_t addr; > + dma_addr_t addr = DMA_MAPPING_ERROR; > > BUG_ON(!valid_dma_direction(dir)); > > /* Don't allow RAM to be mapped */ > BUG_ON(pfn_valid(PHYS_PFN(phys_addr))); > > - addr = phys_addr; > - if (ops && ops->map_resource) > + if (dma_is_direct(ops)) > + addr = dma_direct_map_resource(dev, phys_addr, size, dir, attrs); > + else if (ops->map_resource) > addr = ops->map_resource(dev, phys_addr, size, dir, attrs); Might it be reasonable to do: if (!dma_is_direct(ops) && ops->map_resource) addr = ops->map_resource(...); else addr = dma_direct_map_resource(...); and avoid having to explicitly wire up the dma_direct callback elsewhere? Robin. > > debug_dma_map_resource(dev, phys_addr, size, dir, addr); > - > return addr; > } > > @@ -369,7 +371,7 @@ static inline void dma_unmap_resource(struct device *dev, dma_addr_t addr, > const struct dma_map_ops *ops = get_dma_ops(dev); > > BUG_ON(!valid_dma_direction(dir)); > - if (ops && ops->unmap_resource) > + if (!dma_is_direct(ops) && ops->unmap_resource) > ops->unmap_resource(dev, addr, size, dir, attrs); > debug_dma_unmap_resource(dev, addr, size, dir); > } > diff --git a/kernel/dma/direct.c b/kernel/dma/direct.c > index 355d16acee6d..8e0359b04957 100644 > --- a/kernel/dma/direct.c > +++ b/kernel/dma/direct.c > @@ -356,6 +356,20 @@ int dma_direct_map_sg(struct device *dev, struct scatterlist *sgl, int nents, > } > EXPORT_SYMBOL(dma_direct_map_sg); > > +dma_addr_t dma_direct_map_resource(struct device *dev, phys_addr_t paddr, > + size_t size, enum dma_data_direction dir, unsigned long attrs) > +{ > + dma_addr_t dma_addr = phys_to_dma(dev, paddr); > + > + if (unlikely(!dma_direct_possible(dev, dma_addr, size))) { > + report_addr(dev, dma_addr, size); > + return DMA_MAPPING_ERROR; > + } > + > + return dma_addr; > +} > +EXPORT_SYMBOL(dma_direct_map_resource); > + > /* > * Because 32-bit DMA masks are so common we expect every architecture to be > * able to satisfy them - either by not supporting more physical memory, or by >
On Mon, Jan 14, 2019 at 01:12:33PM +0000, Robin Murphy wrote: > Ignoring the offset was kind of intentional there, because at the time I > was primarily thinking about it in terms of the Keystone 2 platform where > the peripherals are all in the same place (0-2GB) in both the bus and CPU > physical address maps, and only the view of RAM differs between the two > (2-4GB vs. 32-34GB). However, on something like BCM283x, the peripherals > region is also offset from its bus address in the CPU view, but at a > *different* offset relative to that of RAM. I was more thinking of the PCIe P2P case, where we need to apply a consistent offset to translate between the CPU and the bus view. But this isn't really used for PCIe P2P, so I guess keeping the original sematics might be a better idea. That being said the videobuf code seems to rely on these offsets, so we might be between a rock and a hard place. > Fortunately, I'm not aware of any platform which has a DMA engine behind an > IOMMU (and thus *needs* to use dma_map_resource() to avoid said IOMMU > blocking the slave device register reads/writes) and also has any nonzero > offsets, and AFAIK the IOMMU-less platforms above aren't using > dma_map_resource() at all, so this change shouldn't actually break > anything, but I guess we have a bit of a problem making it truly generic > and robust :( Note that we don't actually use the code in this patch for ARM/ARM64 platforms with IOMMUs, as both the ARM and the ARM64 iommu code have their own implementations of ->map_resource that actually program the iommu (which at least for the PCIe P2P case would be wrong). > Is this perhaps another shove in the direction of overhauling > dma_pfn_offset into an arbitrary "DMA ranges" lookup table? It is long overdue anyway. >> addr = ops->map_resource(dev, phys_addr, size, dir, attrs); > > Might it be reasonable to do: > > if (!dma_is_direct(ops) && ops->map_resource) > addr = ops->map_resource(...); > else > addr = dma_direct_map_resource(...); > > and avoid having to explicitly wire up the dma_direct callback elsewhere? No, I absolutely _want_ the callback explicitly wired up. That is the only way to ensure we actually intentionally support it and don't just get a default that often won't work. Same issue for ->mmap and ->get_sgtable.
diff --git a/arch/arm/mm/dma-mapping.c b/arch/arm/mm/dma-mapping.c index f1e2922e447c..3c8534904209 100644 --- a/arch/arm/mm/dma-mapping.c +++ b/arch/arm/mm/dma-mapping.c @@ -188,6 +188,7 @@ const struct dma_map_ops arm_dma_ops = { .unmap_page = arm_dma_unmap_page, .map_sg = arm_dma_map_sg, .unmap_sg = arm_dma_unmap_sg, + .map_resource = dma_direct_map_resource, .sync_single_for_cpu = arm_dma_sync_single_for_cpu, .sync_single_for_device = arm_dma_sync_single_for_device, .sync_sg_for_cpu = arm_dma_sync_sg_for_cpu, @@ -211,6 +212,7 @@ const struct dma_map_ops arm_coherent_dma_ops = { .get_sgtable = arm_dma_get_sgtable, .map_page = arm_coherent_dma_map_page, .map_sg = arm_dma_map_sg, + .map_resource = dma_direct_map_resource, .dma_supported = arm_dma_supported, }; EXPORT_SYMBOL(arm_coherent_dma_ops); diff --git a/arch/powerpc/kernel/dma-swiotlb.c b/arch/powerpc/kernel/dma-swiotlb.c index 7d5fc9751622..fbb2506a414e 100644 --- a/arch/powerpc/kernel/dma-swiotlb.c +++ b/arch/powerpc/kernel/dma-swiotlb.c @@ -55,6 +55,7 @@ const struct dma_map_ops powerpc_swiotlb_dma_ops = { .dma_supported = swiotlb_dma_supported, .map_page = dma_direct_map_page, .unmap_page = dma_direct_unmap_page, + .map_resource = dma_direct_map_resource, .sync_single_for_cpu = dma_direct_sync_single_for_cpu, .sync_single_for_device = dma_direct_sync_single_for_device, .sync_sg_for_cpu = dma_direct_sync_sg_for_cpu, diff --git a/arch/powerpc/kernel/dma.c b/arch/powerpc/kernel/dma.c index b1903ebb2e9c..258b9e8ebb99 100644 --- a/arch/powerpc/kernel/dma.c +++ b/arch/powerpc/kernel/dma.c @@ -273,6 +273,7 @@ const struct dma_map_ops dma_nommu_ops = { .dma_supported = dma_nommu_dma_supported, .map_page = dma_nommu_map_page, .unmap_page = dma_nommu_unmap_page, + .map_resource = dma_direct_map_resource, .get_required_mask = dma_nommu_get_required_mask, #ifdef CONFIG_NOT_COHERENT_CACHE .sync_single_for_cpu = dma_nommu_sync_single, diff --git a/include/linux/dma-mapping.h b/include/linux/dma-mapping.h index cef2127e1d70..d3087829a6df 100644 --- a/include/linux/dma-mapping.h +++ b/include/linux/dma-mapping.h @@ -208,6 +208,8 @@ dma_addr_t dma_direct_map_page(struct device *dev, struct page *page, unsigned long attrs); int dma_direct_map_sg(struct device *dev, struct scatterlist *sgl, int nents, enum dma_data_direction dir, unsigned long attrs); +dma_addr_t dma_direct_map_resource(struct device *dev, phys_addr_t paddr, + size_t size, enum dma_data_direction dir, unsigned long attrs); #if defined(CONFIG_ARCH_HAS_SYNC_DMA_FOR_DEVICE) || \ defined(CONFIG_SWIOTLB) @@ -346,19 +348,19 @@ static inline dma_addr_t dma_map_resource(struct device *dev, unsigned long attrs) { const struct dma_map_ops *ops = get_dma_ops(dev); - dma_addr_t addr; + dma_addr_t addr = DMA_MAPPING_ERROR; BUG_ON(!valid_dma_direction(dir)); /* Don't allow RAM to be mapped */ BUG_ON(pfn_valid(PHYS_PFN(phys_addr))); - addr = phys_addr; - if (ops && ops->map_resource) + if (dma_is_direct(ops)) + addr = dma_direct_map_resource(dev, phys_addr, size, dir, attrs); + else if (ops->map_resource) addr = ops->map_resource(dev, phys_addr, size, dir, attrs); debug_dma_map_resource(dev, phys_addr, size, dir, addr); - return addr; } @@ -369,7 +371,7 @@ static inline void dma_unmap_resource(struct device *dev, dma_addr_t addr, const struct dma_map_ops *ops = get_dma_ops(dev); BUG_ON(!valid_dma_direction(dir)); - if (ops && ops->unmap_resource) + if (!dma_is_direct(ops) && ops->unmap_resource) ops->unmap_resource(dev, addr, size, dir, attrs); debug_dma_unmap_resource(dev, addr, size, dir); } diff --git a/kernel/dma/direct.c b/kernel/dma/direct.c index 355d16acee6d..8e0359b04957 100644 --- a/kernel/dma/direct.c +++ b/kernel/dma/direct.c @@ -356,6 +356,20 @@ int dma_direct_map_sg(struct device *dev, struct scatterlist *sgl, int nents, } EXPORT_SYMBOL(dma_direct_map_sg); +dma_addr_t dma_direct_map_resource(struct device *dev, phys_addr_t paddr, + size_t size, enum dma_data_direction dir, unsigned long attrs) +{ + dma_addr_t dma_addr = phys_to_dma(dev, paddr); + + if (unlikely(!dma_direct_possible(dev, dma_addr, size))) { + report_addr(dev, dma_addr, size); + return DMA_MAPPING_ERROR; + } + + return dma_addr; +} +EXPORT_SYMBOL(dma_direct_map_resource); + /* * Because 32-bit DMA masks are so common we expect every architecture to be * able to satisfy them - either by not supporting more physical memory, or by
Just returning the physical address when not map_resource method is present is highly dangerous as it doesn't take any offset in the direct mapping into account and does the completely wrong thing for IOMMUs. Instead provide a proper implementation in the direct mapping code, and also wire it up for arm and powerpc. Signed-off-by: Christoph Hellwig <hch@lst.de> --- arch/arm/mm/dma-mapping.c | 2 ++ arch/powerpc/kernel/dma-swiotlb.c | 1 + arch/powerpc/kernel/dma.c | 1 + include/linux/dma-mapping.h | 12 +++++++----- kernel/dma/direct.c | 14 ++++++++++++++ 5 files changed, 25 insertions(+), 5 deletions(-)