diff options
Diffstat (limited to 'arch/arc/mm/dma.c')
| -rw-r--r-- | arch/arc/mm/dma.c | 75 |
1 files changed, 50 insertions, 25 deletions
diff --git a/arch/arc/mm/dma.c b/arch/arc/mm/dma.c index 01eaf88bf..8c8e36fa5 100644 --- a/arch/arc/mm/dma.c +++ b/arch/arc/mm/dma.c @@ -24,22 +24,22 @@ static void *arc_dma_alloc(struct device *dev, size_t size, dma_addr_t *dma_handle, gfp_t gfp, struct dma_attrs *attrs) { - void *paddr, *kvaddr; - - /* This is linear addr (0x8000_0000 based) */ - paddr = alloc_pages_exact(size, gfp); - if (!paddr) + unsigned long order = get_order(size); + struct page *page; + phys_addr_t paddr; + void *kvaddr; + int need_coh = 1, need_kvaddr = 0; + + page = alloc_pages(gfp, order); + if (!page) return NULL; - /* This is bus address, platform dependent */ - *dma_handle = (dma_addr_t)paddr; - /* * IOC relies on all data (even coherent DMA data) being in cache * Thus allocate normal cached memory * * The gains with IOC are two pronged: - * -For streaming data, elides needs for cache maintenance, saving + * -For streaming data, elides need for cache maintenance, saving * cycles in flush code, and bus bandwidth as all the lines of a * buffer need to be flushed out to memory * -For coherent data, Read/Write to buffers terminate early in cache @@ -47,12 +47,31 @@ static void *arc_dma_alloc(struct device *dev, size_t size, */ if ((is_isa_arcv2() && ioc_exists) || dma_get_attr(DMA_ATTR_NON_CONSISTENT, attrs)) - return paddr; + need_coh = 0; + + /* + * - A coherent buffer needs MMU mapping to enforce non-cachability + * - A highmem page needs a virtual handle (hence MMU mapping) + * independent of cachability + */ + if (PageHighMem(page) || need_coh) + need_kvaddr = 1; + + /* This is linear addr (0x8000_0000 based) */ + paddr = page_to_phys(page); + + *dma_handle = plat_phys_to_dma(dev, paddr); /* This is kernel Virtual address (0x7000_0000 based) */ - kvaddr = ioremap_nocache((unsigned long)paddr, size); - if (kvaddr == NULL) - return NULL; + if (need_kvaddr) { + kvaddr = ioremap_nocache(paddr, size); + if (kvaddr == NULL) { + __free_pages(page, order); + return NULL; + } + } else { + kvaddr = (void *)(u32)paddr; + } /* * Evict any existing L1 and/or L2 lines for the backing page @@ -64,7 +83,8 @@ static void *arc_dma_alloc(struct device *dev, size_t size, * Currently flush_cache_vmap nukes the L1 cache completely which * will be optimized as a separate commit */ - dma_cache_wback_inv((unsigned long)paddr, size); + if (need_coh) + dma_cache_wback_inv(paddr, size); return kvaddr; } @@ -72,11 +92,16 @@ static void *arc_dma_alloc(struct device *dev, size_t size, static void arc_dma_free(struct device *dev, size_t size, void *vaddr, dma_addr_t dma_handle, struct dma_attrs *attrs) { - if (!dma_get_attr(DMA_ATTR_NON_CONSISTENT, attrs) && - !(is_isa_arcv2() && ioc_exists)) + struct page *page = virt_to_page(dma_handle); + int is_non_coh = 1; + + is_non_coh = dma_get_attr(DMA_ATTR_NON_CONSISTENT, attrs) || + (is_isa_arcv2() && ioc_exists); + + if (PageHighMem(page) || !is_non_coh) iounmap((void __force __iomem *)vaddr); - free_pages_exact((void *)dma_handle, size); + __free_pages(page, get_order(size)); } /* @@ -84,7 +109,7 @@ static void arc_dma_free(struct device *dev, size_t size, void *vaddr, * CPU accesses page via normal paddr, thus needs to explicitly made * consistent before each use */ -static void _dma_cache_sync(unsigned long paddr, size_t size, +static void _dma_cache_sync(phys_addr_t paddr, size_t size, enum dma_data_direction dir) { switch (dir) { @@ -98,7 +123,7 @@ static void _dma_cache_sync(unsigned long paddr, size_t size, dma_cache_wback_inv(paddr, size); break; default: - pr_err("Invalid DMA dir [%d] for OP @ %lx\n", dir, paddr); + pr_err("Invalid DMA dir [%d] for OP @ %pa[p]\n", dir, &paddr); } } @@ -106,9 +131,9 @@ static dma_addr_t arc_dma_map_page(struct device *dev, struct page *page, unsigned long offset, size_t size, enum dma_data_direction dir, struct dma_attrs *attrs) { - unsigned long paddr = page_to_phys(page) + offset; + phys_addr_t paddr = page_to_phys(page) + offset; _dma_cache_sync(paddr, size, dir); - return (dma_addr_t)paddr; + return plat_phys_to_dma(dev, paddr); } static int arc_dma_map_sg(struct device *dev, struct scatterlist *sg, @@ -127,13 +152,13 @@ static int arc_dma_map_sg(struct device *dev, struct scatterlist *sg, static void arc_dma_sync_single_for_cpu(struct device *dev, dma_addr_t dma_handle, size_t size, enum dma_data_direction dir) { - _dma_cache_sync(dma_handle, size, DMA_FROM_DEVICE); + _dma_cache_sync(plat_dma_to_phys(dev, dma_handle), size, DMA_FROM_DEVICE); } static void arc_dma_sync_single_for_device(struct device *dev, dma_addr_t dma_handle, size_t size, enum dma_data_direction dir) { - _dma_cache_sync(dma_handle, size, DMA_TO_DEVICE); + _dma_cache_sync(plat_dma_to_phys(dev, dma_handle), size, DMA_TO_DEVICE); } static void arc_dma_sync_sg_for_cpu(struct device *dev, @@ -144,7 +169,7 @@ static void arc_dma_sync_sg_for_cpu(struct device *dev, struct scatterlist *sg; for_each_sg(sglist, sg, nelems, i) - _dma_cache_sync((unsigned int)sg_virt(sg), sg->length, dir); + _dma_cache_sync(sg_phys(sg), sg->length, dir); } static void arc_dma_sync_sg_for_device(struct device *dev, @@ -155,7 +180,7 @@ static void arc_dma_sync_sg_for_device(struct device *dev, struct scatterlist *sg; for_each_sg(sglist, sg, nelems, i) - _dma_cache_sync((unsigned int)sg_virt(sg), sg->length, dir); + _dma_cache_sync(sg_phys(sg), sg->length, dir); } static int arc_dma_supported(struct device *dev, u64 dma_mask) |