Sometimes it is not worth for the iommu allocating big chunks. Here we enable DMA_ATTR_ALLOC_SINGLE_PAGES which could help avoid to allocate big chunks while iommu allocating buffer.
More information about this attribute, please check Doug's commit[1]. [1]: https://lkml.org/lkml/2016/1/11/720 Cc: Robin Murphy <robin.mur...@arm.com> Suggested-by: Douglas Anderson <diand...@chromium.org> Signed-off-by: Yong Wu <yong...@mediatek.com> --- Our video drivers may soon use this. arch/arm64/mm/dma-mapping.c | 4 ++-- drivers/iommu/dma-iommu.c | 14 ++++++++++---- include/linux/dma-iommu.h | 4 ++-- 3 files changed, 14 insertions(+), 8 deletions(-) diff --git a/arch/arm64/mm/dma-mapping.c b/arch/arm64/mm/dma-mapping.c index 331c4ca..3225e3ca 100644 --- a/arch/arm64/mm/dma-mapping.c +++ b/arch/arm64/mm/dma-mapping.c @@ -562,8 +562,8 @@ static void *__iommu_alloc_attrs(struct device *dev, size_t size, struct page **pages; pgprot_t prot = __get_dma_pgprot(attrs, PAGE_KERNEL, coherent); - pages = iommu_dma_alloc(dev, iosize, gfp, ioprot, handle, - flush_page); + pages = iommu_dma_alloc(dev, iosize, gfp, ioprot, attrs, + handle, flush_page); if (!pages) return NULL; diff --git a/drivers/iommu/dma-iommu.c b/drivers/iommu/dma-iommu.c index 72d6182..3569cb6 100644 --- a/drivers/iommu/dma-iommu.c +++ b/drivers/iommu/dma-iommu.c @@ -190,7 +190,8 @@ static void __iommu_dma_free_pages(struct page **pages, int count) kvfree(pages); } -static struct page **__iommu_dma_alloc_pages(unsigned int count, gfp_t gfp) +static struct page **__iommu_dma_alloc_pages(unsigned int count, gfp_t gfp, + struct dma_attrs *attrs) { struct page **pages; unsigned int i = 0, array_size = count * sizeof(*pages); @@ -203,6 +204,10 @@ static struct page **__iommu_dma_alloc_pages(unsigned int count, gfp_t gfp) if (!pages) return NULL; + /* Go straight to 4K chunks if caller says it's OK. */ + if (dma_get_attr(DMA_ATTR_ALLOC_SINGLE_PAGES, attrs)) + order = 0; + /* IOMMU can map any pages, so himem can also be used here */ gfp |= __GFP_NOWARN | __GFP_HIGHMEM; @@ -268,6 +273,7 @@ void iommu_dma_free(struct device *dev, struct page **pages, size_t size, * @size: Size of buffer in bytes * @gfp: Allocation flags * @prot: IOMMU mapping flags + * @attrs: DMA attributes flags * @handle: Out argument for allocated DMA handle * @flush_page: Arch callback which must ensure PAGE_SIZE bytes from the * given VA/PA are visible to the given non-coherent device. @@ -278,8 +284,8 @@ void iommu_dma_free(struct device *dev, struct page **pages, size_t size, * Return: Array of struct page pointers describing the buffer, * or NULL on failure. */ -struct page **iommu_dma_alloc(struct device *dev, size_t size, - gfp_t gfp, int prot, dma_addr_t *handle, +struct page **iommu_dma_alloc(struct device *dev, size_t size, gfp_t gfp, + int prot, struct dma_attrs *attrs, dma_addr_t *handle, void (*flush_page)(struct device *, const void *, phys_addr_t)) { struct iommu_domain *domain = iommu_get_domain_for_dev(dev); @@ -292,7 +298,7 @@ struct page **iommu_dma_alloc(struct device *dev, size_t size, *handle = DMA_ERROR_CODE; - pages = __iommu_dma_alloc_pages(count, gfp); + pages = __iommu_dma_alloc_pages(count, gfp, attrs); if (!pages) return NULL; diff --git a/include/linux/dma-iommu.h b/include/linux/dma-iommu.h index fc48103..08d9603 100644 --- a/include/linux/dma-iommu.h +++ b/include/linux/dma-iommu.h @@ -38,8 +38,8 @@ int dma_direction_to_prot(enum dma_data_direction dir, bool coherent); * These implement the bulk of the relevant DMA mapping callbacks, but require * the arch code to take care of attributes and cache maintenance */ -struct page **iommu_dma_alloc(struct device *dev, size_t size, - gfp_t gfp, int prot, dma_addr_t *handle, +struct page **iommu_dma_alloc(struct device *dev, size_t size, gfp_t gfp, + int prot, struct dma_attrs *attrs, dma_addr_t *handle, void (*flush_page)(struct device *, const void *, phys_addr_t)); void iommu_dma_free(struct device *dev, struct page **pages, size_t size, dma_addr_t *handle); -- 1.8.1.1.dirty