diff options
Diffstat (limited to 'arch/arc/mm/dma.c')
| -rw-r--r-- | arch/arc/mm/dma.c | 131 |
1 files changed, 88 insertions, 43 deletions
diff --git a/arch/arc/mm/dma.c b/arch/arc/mm/dma.c index 8c1071840979..c75d5c3470e3 100644 --- a/arch/arc/mm/dma.c +++ b/arch/arc/mm/dma.c @@ -6,20 +6,17 @@ * published by the Free Software Foundation. */ -/* - * DMA Coherent API Notes - * - * I/O is inherently non-coherent on ARC. So a coherent DMA buffer is - * implemented by accessing it using a kernel virtual address, with - * Cache bit off in the TLB entry. - * - * The default DMA address == Phy address which is 0x8000_0000 based. - */ - #include <linux/dma-noncoherent.h> #include <asm/cache.h> #include <asm/cacheflush.h> +/* + * ARCH specific callbacks for generic noncoherent DMA ops (dma/noncoherent.c) + * - hardware IOC not available (or "dma-coherent" not set for device in DT) + * - But still handle both coherent and non-coherent requests from caller + * + * For DMA coherent hardware (IOC) generic code suffices + */ void *arch_dma_alloc(struct device *dev, size_t size, dma_addr_t *dma_handle, gfp_t gfp, unsigned long attrs) { @@ -27,42 +24,29 @@ void *arch_dma_alloc(struct device *dev, size_t size, dma_addr_t *dma_handle, 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; + bool need_coh = !(attrs & DMA_ATTR_NON_CONSISTENT); /* - * 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 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 - * (vs. always going to memory - thus are faster) + * __GFP_HIGHMEM flag is cleared by upper layer functions + * (in include/linux/dma-mapping.h) so we should never get a + * __GFP_HIGHMEM here. */ - if ((is_isa_arcv2() && ioc_enable) || - (attrs & DMA_ATTR_NON_CONSISTENT)) - need_coh = 0; + BUG_ON(gfp & __GFP_HIGHMEM); - /* - * - 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; + page = alloc_pages(gfp, order); + if (!page) + return NULL; /* This is linear addr (0x8000_0000 based) */ paddr = page_to_phys(page); *dma_handle = paddr; - /* This is kernel Virtual address (0x7000_0000 based) */ - if (need_kvaddr) { + /* + * A coherent buffer needs MMU mapping to enforce non-cachability. + * kvaddr is kernel Virtual address (0x7000_0000 based). + */ + if (need_coh) { kvaddr = ioremap_nocache(paddr, size); if (kvaddr == NULL) { __free_pages(page, order); @@ -93,12 +77,8 @@ void arch_dma_free(struct device *dev, size_t size, void *vaddr, { phys_addr_t paddr = dma_handle; struct page *page = virt_to_page(paddr); - int is_non_coh = 1; - - is_non_coh = (attrs & DMA_ATTR_NON_CONSISTENT) || - (is_isa_arcv2() && ioc_enable); - if (PageHighMem(page) || !is_non_coh) + if (!(attrs & DMA_ATTR_NON_CONSISTENT)) iounmap((void __force __iomem *)vaddr); __free_pages(page, get_order(size)); @@ -129,14 +109,79 @@ int arch_dma_mmap(struct device *dev, struct vm_area_struct *vma, return ret; } +/* + * Cache operations depending on function and direction argument, inspired by + * https://lkml.org/lkml/2018/5/18/979 + * "dma_sync_*_for_cpu and direction=TO_DEVICE (was Re: [PATCH 02/20] + * dma-mapping: provide a generic dma-noncoherent implementation)" + * + * | map == for_device | unmap == for_cpu + * |---------------------------------------------------------------- + * TO_DEV | writeback writeback | none none + * FROM_DEV | invalidate invalidate | invalidate* invalidate* + * BIDIR | writeback+inv writeback+inv | invalidate invalidate + * + * [*] needed for CPU speculative prefetches + * + * NOTE: we don't check the validity of direction argument as it is done in + * upper layer functions (in include/linux/dma-mapping.h) + */ + void arch_sync_dma_for_device(struct device *dev, phys_addr_t paddr, size_t size, enum dma_data_direction dir) { - dma_cache_wback(paddr, size); + switch (dir) { + case DMA_TO_DEVICE: + dma_cache_wback(paddr, size); + break; + + case DMA_FROM_DEVICE: + dma_cache_inv(paddr, size); + break; + + case DMA_BIDIRECTIONAL: + dma_cache_wback_inv(paddr, size); + break; + + default: + break; + } } void arch_sync_dma_for_cpu(struct device *dev, phys_addr_t paddr, size_t size, enum dma_data_direction dir) { - dma_cache_inv(paddr, size); + switch (dir) { + case DMA_TO_DEVICE: + break; + + /* FROM_DEVICE invalidate needed if speculative CPU prefetch only */ + case DMA_FROM_DEVICE: + case DMA_BIDIRECTIONAL: + dma_cache_inv(paddr, size); + break; + + default: + break; + } +} + +/* + * Plug in coherent or noncoherent dma ops + */ +void arch_setup_dma_ops(struct device *dev, u64 dma_base, u64 size, + const struct iommu_ops *iommu, bool coherent) +{ + /* + * IOC hardware snoops all DMA traffic keeping the caches consistent + * with memory - eliding need for any explicit cache maintenance of + * DMA buffers - so we can use dma_direct cache ops. + */ + if (is_isa_arcv2() && ioc_enable && coherent) { + set_dma_ops(dev, &dma_direct_ops); + dev_info(dev, "use dma_direct_ops cache ops\n"); + } else { + set_dma_ops(dev, &dma_noncoherent_ops); + dev_info(dev, "use dma_noncoherent_ops cache ops\n"); + } } |