Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932793Ab2FVDEp (ORCPT ); Thu, 21 Jun 2012 23:04:45 -0400 Received: from mail-pz0-f46.google.com ([209.85.210.46]:60682 "EHLO mail-pz0-f46.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1761138Ab2FVDEm (ORCPT ); Thu, 21 Jun 2012 23:04:42 -0400 From: Huacai Chen To: Ralf Baechle Cc: linux-mips@linux-mips.org, linux-kernel@vger.kernel.org, Fuxin Zhang , Zhangjin Wu , Huacai Chen , Hongliang Tao , Hua Yan Subject: [PATCH V3 09/16] MIPS: Loongson: Add swiotlb to support big memory (>4GB). Date: Fri, 22 Jun 2012 11:01:06 +0800 Message-Id: <1340334073-17804-10-git-send-email-chenhc@lemote.com> X-Mailer: git-send-email 1.7.7.3 In-Reply-To: <1340334073-17804-1-git-send-email-chenhc@lemote.com> References: <1340334073-17804-1-git-send-email-chenhc@lemote.com> Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 10555 Lines: 333 This is probably a workaround because Loongson doesn't support DMA address above 4GB. If memory is more than 4GB, CONFIG_SWIOTLB and ZONE_DMA32 should be selected. In this way, DMA pages are allocated below 4GB preferably. However, CONFIG_SWIOTLB+ZONE_DMA32 is not enough, so, we provide a platform-specific dma_map_ops::set_dma_mask() to make sure each driver's dma_mask and coherent_dma_mask is below 32-bit. Signed-off-by: Huacai Chen Signed-off-by: Hongliang Tao Signed-off-by: Hua Yan --- arch/mips/include/asm/dma-mapping.h | 5 + .../mips/include/asm/mach-loongson/dma-coherence.h | 25 +++- arch/mips/loongson/common/Makefile | 5 + arch/mips/loongson/common/dma-swiotlb.c | 159 ++++++++++++++++++++ arch/mips/mm/dma-default.c | 13 ++- 5 files changed, 202 insertions(+), 5 deletions(-) create mode 100644 arch/mips/loongson/common/dma-swiotlb.c diff --git a/arch/mips/include/asm/dma-mapping.h b/arch/mips/include/asm/dma-mapping.h index be39a12..35f91bc 100644 --- a/arch/mips/include/asm/dma-mapping.h +++ b/arch/mips/include/asm/dma-mapping.h @@ -46,9 +46,14 @@ static inline int dma_mapping_error(struct device *dev, u64 mask) static inline int dma_set_mask(struct device *dev, u64 mask) { + struct dma_map_ops *ops = get_dma_ops(dev); + if(!dev->dma_mask || !dma_supported(dev, mask)) return -EIO; + if (ops->set_dma_mask) + return ops->set_dma_mask(dev, mask); + *dev->dma_mask = mask; return 0; diff --git a/arch/mips/include/asm/mach-loongson/dma-coherence.h b/arch/mips/include/asm/mach-loongson/dma-coherence.h index e143305..b1dc286 100644 --- a/arch/mips/include/asm/mach-loongson/dma-coherence.h +++ b/arch/mips/include/asm/mach-loongson/dma-coherence.h @@ -13,26 +13,43 @@ struct device; +extern dma_addr_t phys_to_dma(struct device *dev, phys_addr_t paddr); +extern phys_addr_t dma_to_phys(struct device *dev, dma_addr_t daddr); static inline dma_addr_t plat_map_dma_mem(struct device *dev, void *addr, size_t size) { +#ifdef CONFIG_CPU_LOONGSON3 + return virt_to_phys(addr) < 0x10000000 ? + (virt_to_phys(addr) | 0x0000000080000000) : virt_to_phys(addr); +#else return virt_to_phys(addr) | 0x80000000; +#endif } static inline dma_addr_t plat_map_dma_mem_page(struct device *dev, struct page *page) { +#ifdef CONFIG_CPU_LOONGSON3 + return page_to_phys(page) < 0x10000000 ? + (page_to_phys(page) | 0x0000000080000000) : page_to_phys(page); +#else return page_to_phys(page) | 0x80000000; +#endif } static inline unsigned long plat_dma_addr_to_phys(struct device *dev, dma_addr_t dma_addr) { -#if defined(CONFIG_CPU_LOONGSON2F) && defined(CONFIG_64BIT) +#if defined(CONFIG_64BIT) +#if defined(CONFIG_CPU_LOONGSON3) + return (dma_addr < 0x90000000 && dma_addr >= 0x80000000) ? + (dma_addr & 0x0fffffff) : dma_addr; +#elif defined(CONFIG_CPU_LOONGSON2F) return (dma_addr > 0x8fffffff) ? dma_addr : (dma_addr & 0x0fffffff); +#endif /* CONFIG_CPU_LOONGSON3 */ #else return dma_addr & 0x7fffffff; -#endif +#endif /* CONFIG_64BIT */ } static inline void plat_unmap_dma_mem(struct device *dev, dma_addr_t dma_addr, @@ -65,7 +82,11 @@ static inline int plat_dma_mapping_error(struct device *dev, static inline int plat_device_is_coherent(struct device *dev) { +#ifdef CONFIG_DMA_NONCOHERENT return 0; +#else + return 1; +#endif /* CONFIG_DMA_NONCOHERENT */ } #endif /* __ASM_MACH_LOONGSON_DMA_COHERENCE_H */ diff --git a/arch/mips/loongson/common/Makefile b/arch/mips/loongson/common/Makefile index e526488..3a26109 100644 --- a/arch/mips/loongson/common/Makefile +++ b/arch/mips/loongson/common/Makefile @@ -25,3 +25,8 @@ obj-$(CONFIG_CS5536) += cs5536/ # obj-$(CONFIG_LOONGSON_SUSPEND) += pm.o + +# +# Big Memory Support +# +obj-$(CONFIG_LOONGSON_BIGMEM) += dma-swiotlb.o diff --git a/arch/mips/loongson/common/dma-swiotlb.c b/arch/mips/loongson/common/dma-swiotlb.c new file mode 100644 index 0000000..b87a21e --- /dev/null +++ b/arch/mips/loongson/common/dma-swiotlb.c @@ -0,0 +1,159 @@ +#include +#include +#include +#include +#include +#include + +#include +#include + +static void *loongson_dma_alloc_coherent(struct device *dev, size_t size, + dma_addr_t *dma_handle, gfp_t gfp, struct dma_attrs *attrs) +{ + void *ret; + + if (dma_alloc_from_coherent(dev, size, dma_handle, &ret)) + return ret; + + /* ignore region specifiers */ + gfp &= ~(__GFP_DMA | __GFP_DMA32 | __GFP_HIGHMEM); + +#ifdef CONFIG_ZONE_DMA + if (dev == NULL) + gfp |= __GFP_DMA; + else if (dev->coherent_dma_mask <= DMA_BIT_MASK(24)) + gfp |= __GFP_DMA; + else +#endif +#ifdef CONFIG_ZONE_DMA32 + if (dev->coherent_dma_mask <= DMA_BIT_MASK(32)) + gfp |= __GFP_DMA32; + else +#endif + ; + gfp |= __GFP_NORETRY; + + ret = swiotlb_alloc_coherent(dev, size, dma_handle, gfp); + mb(); + return ret; +} + +static void loongson_dma_free_coherent(struct device *dev, size_t size, + void *vaddr, dma_addr_t dma_handle, struct dma_attrs *attrs) +{ + int order = get_order(size); + + if (dma_release_from_coherent(dev, order, vaddr)) + return; + + swiotlb_free_coherent(dev, size, vaddr, dma_handle); +} + +static dma_addr_t loongson_dma_map_page(struct device *dev, struct page *page, + unsigned long offset, size_t size, + enum dma_data_direction dir, + struct dma_attrs *attrs) +{ + dma_addr_t daddr = swiotlb_map_page(dev, page, offset, size, + dir, attrs); + mb(); + return daddr; +} + +static int loongson_dma_map_sg(struct device *dev, struct scatterlist *sg, + int nents, enum dma_data_direction dir, + struct dma_attrs *attrs) +{ + int r = swiotlb_map_sg_attrs(dev, sg, nents, dir, NULL); + mb(); + + return r; +} + +static void loongson_dma_sync_single_for_device(struct device *dev, + dma_addr_t dma_handle, size_t size, + enum dma_data_direction dir) +{ + swiotlb_sync_single_for_device(dev, dma_handle, size, dir); + mb(); +} + +static void loongson_dma_sync_sg_for_device(struct device *dev, + struct scatterlist *sg, int nents, + enum dma_data_direction dir) +{ + swiotlb_sync_sg_for_device(dev, sg, nents, dir); + mb(); +} + +static dma_addr_t loongson_unity_phys_to_dma(struct device *dev, phys_addr_t paddr) +{ + return (paddr < 0x10000000) ? + (paddr | 0x0000000080000000) : paddr; +} + +static phys_addr_t loongson_unity_dma_to_phys(struct device *dev, dma_addr_t daddr) +{ + return (daddr < 0x90000000 && daddr >= 0x80000000) ? + (daddr & 0x0fffffff) : daddr; +} + +struct loongson_dma_map_ops { + struct dma_map_ops dma_map_ops; + dma_addr_t (*phys_to_dma)(struct device *dev, phys_addr_t paddr); + phys_addr_t (*dma_to_phys)(struct device *dev, dma_addr_t daddr); +}; + +dma_addr_t phys_to_dma(struct device *dev, phys_addr_t paddr) +{ + struct loongson_dma_map_ops *ops = container_of(get_dma_ops(dev), + struct loongson_dma_map_ops, dma_map_ops); + + return ops->phys_to_dma(dev, paddr); +} + +phys_addr_t dma_to_phys(struct device *dev, dma_addr_t daddr) +{ + struct loongson_dma_map_ops *ops = container_of(get_dma_ops(dev), + struct loongson_dma_map_ops, dma_map_ops); + + return ops->dma_to_phys(dev, daddr); +} + +static int loongson_dma_set_mask(struct device *dev, u64 mask) +{ + /* Loongson doesn't support DMA above 32-bit */ + if (mask > DMA_BIT_MASK(32)) + return -EIO; + + *dev->dma_mask = mask; + + return 0; +} + +static struct loongson_dma_map_ops loongson_linear_dma_map_ops = { + .dma_map_ops = { + .alloc = loongson_dma_alloc_coherent, + .free = loongson_dma_free_coherent, + .map_page = loongson_dma_map_page, + .unmap_page = swiotlb_unmap_page, + .map_sg = loongson_dma_map_sg, + .unmap_sg = swiotlb_unmap_sg_attrs, + .sync_single_for_cpu = swiotlb_sync_single_for_cpu, + .sync_single_for_device = loongson_dma_sync_single_for_device, + .sync_sg_for_cpu = swiotlb_sync_sg_for_cpu, + .sync_sg_for_device = loongson_dma_sync_sg_for_device, + .mapping_error = swiotlb_dma_mapping_error, + .dma_supported = swiotlb_dma_supported, + .set_dma_mask = loongson_dma_set_mask + }, + .phys_to_dma = loongson_unity_phys_to_dma, + .dma_to_phys = loongson_unity_dma_to_phys +}; + +void __init plat_swiotlb_setup(void) +{ + swiotlb_init(1); + mips_dma_map_ops = &loongson_linear_dma_map_ops.dma_map_ops; +} diff --git a/arch/mips/mm/dma-default.c b/arch/mips/mm/dma-default.c index 3fab204..122f4f8 100644 --- a/arch/mips/mm/dma-default.c +++ b/arch/mips/mm/dma-default.c @@ -42,6 +42,13 @@ static inline int cpu_is_noncoherent_r10000(struct device *dev) current_cpu_type() == CPU_R12000); } +static inline int cpu_is_noncoherent_loongson(struct device *dev) +{ + return !plat_device_is_coherent(dev) && + (current_cpu_type() == CPU_LOONGSON2 || + current_cpu_type() == CPU_LOONGSON3); +} + static gfp_t massage_gfp_flags(const struct device *dev, gfp_t gfp) { gfp_t dma_flag; @@ -209,7 +216,7 @@ static inline void __dma_sync(struct page *page, static void mips_dma_unmap_page(struct device *dev, dma_addr_t dma_addr, size_t size, enum dma_data_direction direction, struct dma_attrs *attrs) { - if (cpu_is_noncoherent_r10000(dev)) + if (cpu_is_noncoherent_r10000(dev) || cpu_is_noncoherent_loongson(dev)) __dma_sync(dma_addr_to_page(dev, dma_addr), dma_addr & ~PAGE_MASK, size, direction); @@ -260,7 +267,7 @@ static void mips_dma_unmap_sg(struct device *dev, struct scatterlist *sg, static void mips_dma_sync_single_for_cpu(struct device *dev, dma_addr_t dma_handle, size_t size, enum dma_data_direction direction) { - if (cpu_is_noncoherent_r10000(dev)) + if (cpu_is_noncoherent_r10000(dev) || cpu_is_noncoherent_loongson(dev)) __dma_sync(dma_addr_to_page(dev, dma_handle), dma_handle & ~PAGE_MASK, size, direction); } @@ -281,7 +288,7 @@ static void mips_dma_sync_sg_for_cpu(struct device *dev, /* Make sure that gcc doesn't leave the empty loop body. */ for (i = 0; i < nelems; i++, sg++) { - if (cpu_is_noncoherent_r10000(dev)) + if (cpu_is_noncoherent_r10000(dev) || cpu_is_noncoherent_loongson(dev)) __dma_sync(sg_page(sg), sg->offset, sg->length, direction); } -- 1.7.7.3 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/