From 7f307a6c2c9fd25670d48560695f0f358efc3b96 Mon Sep 17 00:00:00 2001 From: James Hogan Date: Fri, 5 Oct 2012 16:27:03 +0100 Subject: [PATCH] metag: DMA Add DMA mapping code. Signed-off-by: James Hogan --- arch/metag/include/asm/dma-mapping.h | 183 ++++++++++ arch/metag/kernel/dma.c | 507 +++++++++++++++++++++++++++ 2 files changed, 690 insertions(+) create mode 100644 arch/metag/include/asm/dma-mapping.h create mode 100644 arch/metag/kernel/dma.c diff --git a/arch/metag/include/asm/dma-mapping.h b/arch/metag/include/asm/dma-mapping.h new file mode 100644 index 0000000000000..b5f80a62fe8b0 --- /dev/null +++ b/arch/metag/include/asm/dma-mapping.h @@ -0,0 +1,183 @@ +#ifndef _ASM_METAG_DMA_MAPPING_H +#define _ASM_METAG_DMA_MAPPING_H + +#include + +#include +#include +#include +#include + +#define dma_alloc_noncoherent(d, s, h, f) dma_alloc_coherent(d, s, h, f) +#define dma_free_noncoherent(d, s, v, h) dma_free_coherent(d, s, v, h) + +void *dma_alloc_coherent(struct device *dev, size_t size, + dma_addr_t *dma_handle, gfp_t flag); + +void dma_free_coherent(struct device *dev, size_t size, + void *vaddr, dma_addr_t dma_handle); + +void dma_sync_for_device(void *vaddr, size_t size, int dma_direction); +void dma_sync_for_cpu(void *vaddr, size_t size, int dma_direction); + +int dma_mmap_coherent(struct device *dev, struct vm_area_struct *vma, + void *cpu_addr, dma_addr_t dma_addr, size_t size); + +int dma_mmap_writecombine(struct device *dev, struct vm_area_struct *vma, + void *cpu_addr, dma_addr_t dma_addr, size_t size); + +static inline dma_addr_t +dma_map_single(struct device *dev, void *ptr, size_t size, + enum dma_data_direction direction) +{ + BUG_ON(!valid_dma_direction(direction)); + WARN_ON(size == 0); + dma_sync_for_device(ptr, size, direction); + return virt_to_phys(ptr); +} + +static inline void +dma_unmap_single(struct device *dev, dma_addr_t dma_addr, size_t size, + enum dma_data_direction direction) +{ + BUG_ON(!valid_dma_direction(direction)); + dma_sync_for_cpu(phys_to_virt(dma_addr), size, direction); +} + +static inline int +dma_map_sg(struct device *dev, struct scatterlist *sglist, int nents, + enum dma_data_direction direction) +{ + struct scatterlist *sg; + int i; + + BUG_ON(!valid_dma_direction(direction)); + WARN_ON(nents == 0 || sglist[0].length == 0); + + for_each_sg(sglist, sg, nents, i) { + BUG_ON(!sg_page(sg)); + + sg->dma_address = sg_phys(sg); + dma_sync_for_device(sg_virt(sg), sg->length, direction); + } + + return nents; +} + +static inline dma_addr_t +dma_map_page(struct device *dev, struct page *page, unsigned long offset, + size_t size, enum dma_data_direction direction) +{ + BUG_ON(!valid_dma_direction(direction)); + dma_sync_for_device((void *)(page_to_phys(page) + offset), size, + direction); + return page_to_phys(page) + offset; +} + +static inline void +dma_unmap_page(struct device *dev, dma_addr_t dma_address, size_t size, + enum dma_data_direction direction) +{ + BUG_ON(!valid_dma_direction(direction)); + dma_sync_for_cpu(phys_to_virt(dma_address), size, direction); +} + + +static inline void +dma_unmap_sg(struct device *dev, struct scatterlist *sglist, int nhwentries, + enum dma_data_direction direction) +{ + struct scatterlist *sg; + int i; + + BUG_ON(!valid_dma_direction(direction)); + WARN_ON(nhwentries == 0 || sglist[0].length == 0); + + for_each_sg(sglist, sg, nhwentries, i) { + BUG_ON(!sg_page(sg)); + + sg->dma_address = sg_phys(sg); + dma_sync_for_cpu(sg_virt(sg), sg->length, direction); + } +} + +static inline void +dma_sync_single_for_cpu(struct device *dev, dma_addr_t dma_handle, size_t size, + enum dma_data_direction direction) +{ + dma_sync_for_cpu(phys_to_virt(dma_handle), size, direction); +} + +static inline void +dma_sync_single_for_device(struct device *dev, dma_addr_t dma_handle, + size_t size, enum dma_data_direction direction) +{ + dma_sync_for_device(phys_to_virt(dma_handle), size, direction); +} + +static inline void +dma_sync_single_range_for_cpu(struct device *dev, dma_addr_t dma_handle, + unsigned long offset, size_t size, + enum dma_data_direction direction) +{ + dma_sync_for_cpu(phys_to_virt(dma_handle)+offset, size, + direction); +} + +static inline void +dma_sync_single_range_for_device(struct device *dev, dma_addr_t dma_handle, + unsigned long offset, size_t size, + enum dma_data_direction direction) +{ + dma_sync_for_device(phys_to_virt(dma_handle)+offset, size, + direction); +} + +static inline void +dma_sync_sg_for_cpu(struct device *dev, struct scatterlist *sg, int nelems, + enum dma_data_direction direction) +{ + int i; + for (i = 0; i < nelems; i++, sg++) + dma_sync_for_cpu(sg_virt(sg), sg->length, direction); +} + +static inline void +dma_sync_sg_for_device(struct device *dev, struct scatterlist *sg, int nelems, + enum dma_data_direction direction) +{ + int i; + for (i = 0; i < nelems; i++, sg++) + dma_sync_for_device(sg_virt(sg), sg->length, direction); +} + +static inline int +dma_mapping_error(struct device *dev, dma_addr_t dma_addr) +{ + return 0; +} + +#define dma_supported(dev, mask) (1) + +static inline int +dma_set_mask(struct device *dev, u64 mask) +{ + if (!dev->dma_mask || !dma_supported(dev, mask)) + return -EIO; + + *dev->dma_mask = mask; + + return 0; +} + +/* + * dma_alloc_noncoherent() returns non-cacheable memory, so there's no need to + * do any flushing here. + */ +static inline void +dma_cache_sync(struct device *dev, void *vaddr, size_t size, + enum dma_data_direction direction) +{ +} + +#endif diff --git a/arch/metag/kernel/dma.c b/arch/metag/kernel/dma.c new file mode 100644 index 0000000000000..8c00dedadc54a --- /dev/null +++ b/arch/metag/kernel/dma.c @@ -0,0 +1,507 @@ +/* + * Meta version derived from arch/powerpc/lib/dma-noncoherent.c + * Copyright (C) 2008 Imagination Technologies Ltd. + * + * PowerPC version derived from arch/arm/mm/consistent.c + * Copyright (C) 2001 Dan Malek (dmalek@jlc.net) + * + * Copyright (C) 2000 Russell King + * + * Consistent memory allocators. Used for DMA devices that want to + * share uncached memory with the processor core. The function return + * is the virtual address and 'dma_handle' is the physical address. + * Mostly stolen from the ARM port, with some changes for PowerPC. + * -- Dan + * + * Reorganized to get rid of the arch-specific consistent_* functions + * and provide non-coherent implementations for the DMA API. -Matt + * + * Added in_interrupt() safe dma_alloc_coherent()/dma_free_coherent() + * implementation. This is pulled straight from ARM and barely + * modified. -Matt + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include + +#define CONSISTENT_OFFSET(x) (((unsigned long)(x) - CONSISTENT_START) \ + >> PAGE_SHIFT) + +static u64 get_coherent_dma_mask(struct device *dev) +{ + u64 mask = ~0ULL; + + if (dev) { + mask = dev->coherent_dma_mask; + + /* + * Sanity check the DMA mask - it must be non-zero, and + * must be able to be satisfied by a DMA allocation. + */ + if (mask == 0) { + dev_warn(dev, "coherent DMA mask is unset\n"); + return 0; + } + } + + return mask; +} +/* + * This is the page table (2MB) covering uncached, DMA consistent allocations + */ +static pte_t *consistent_pte; +static DEFINE_SPINLOCK(consistent_lock); + +/* + * VM region handling support. + * + * This should become something generic, handling VM region allocations for + * vmalloc and similar (ioremap, module space, etc). + * + * I envisage vmalloc()'s supporting vm_struct becoming: + * + * struct vm_struct { + * struct metag_vm_region region; + * unsigned long flags; + * struct page **pages; + * unsigned int nr_pages; + * unsigned long phys_addr; + * }; + * + * get_vm_area() would then call metag_vm_region_alloc with an appropriate + * struct metag_vm_region head (eg): + * + * struct metag_vm_region vmalloc_head = { + * .vm_list = LIST_HEAD_INIT(vmalloc_head.vm_list), + * .vm_start = VMALLOC_START, + * .vm_end = VMALLOC_END, + * }; + * + * However, vmalloc_head.vm_start is variable (typically, it is dependent on + * the amount of RAM found at boot time.) I would imagine that get_vm_area() + * would have to initialise this each time prior to calling + * metag_vm_region_alloc(). + */ +struct metag_vm_region { + struct list_head vm_list; + unsigned long vm_start; + unsigned long vm_end; + struct page *vm_pages; + int vm_active; +}; + +static struct metag_vm_region consistent_head = { + .vm_list = LIST_HEAD_INIT(consistent_head.vm_list), + .vm_start = CONSISTENT_START, + .vm_end = CONSISTENT_END, +}; + +static struct metag_vm_region *metag_vm_region_alloc(struct metag_vm_region + *head, size_t size, + gfp_t gfp) +{ + unsigned long addr = head->vm_start, end = head->vm_end - size; + unsigned long flags; + struct metag_vm_region *c, *new; + + new = kmalloc(sizeof(struct metag_vm_region), gfp); + if (!new) + goto out; + + spin_lock_irqsave(&consistent_lock, flags); + + list_for_each_entry(c, &head->vm_list, vm_list) { + if ((addr + size) < addr) + goto nospc; + if ((addr + size) <= c->vm_start) + goto found; + addr = c->vm_end; + if (addr > end) + goto nospc; + } + +found: + /* + * Insert this entry _before_ the one we found. + */ + list_add_tail(&new->vm_list, &c->vm_list); + new->vm_start = addr; + new->vm_end = addr + size; + new->vm_active = 1; + + spin_unlock_irqrestore(&consistent_lock, flags); + return new; + +nospc: + spin_unlock_irqrestore(&consistent_lock, flags); + kfree(new); +out: + return NULL; +} + +static struct metag_vm_region *metag_vm_region_find(struct metag_vm_region + *head, unsigned long addr) +{ + struct metag_vm_region *c; + + list_for_each_entry(c, &head->vm_list, vm_list) { + if (c->vm_active && c->vm_start == addr) + goto out; + } + c = NULL; +out: + return c; +} + +/* + * Allocate DMA-coherent memory space and return both the kernel remapped + * virtual and bus address for that space. + */ +void *dma_alloc_coherent(struct device *dev, size_t size, + dma_addr_t *handle, gfp_t gfp) +{ + struct page *page; + struct metag_vm_region *c; + unsigned long order; + u64 mask = get_coherent_dma_mask(dev); + u64 limit; + + if (!consistent_pte) { + pr_err("%s: not initialised\n", __func__); + dump_stack(); + return NULL; + } + + if (!mask) + goto no_page; + size = PAGE_ALIGN(size); + limit = (mask + 1) & ~mask; + if ((limit && size >= limit) + || size >= (CONSISTENT_END - CONSISTENT_START)) { + pr_warn("coherent allocation too big (requested %#x mask %#Lx)\n", + size, mask); + return NULL; + } + + order = get_order(size); + + if (mask != 0xffffffff) + gfp |= GFP_DMA; + + page = alloc_pages(gfp, order); + if (!page) + goto no_page; + + /* + * Invalidate any data that might be lurking in the + * kernel direct-mapped region for device DMA. + */ + { + void *kaddr = page_address(page); + memset(kaddr, 0, size); + flush_dcache_region(kaddr, size); + } + + /* + * Allocate a virtual address in the consistent mapping region. + */ + c = metag_vm_region_alloc(&consistent_head, size, + gfp & ~(__GFP_DMA | __GFP_HIGHMEM)); + if (c) { + unsigned long vaddr = c->vm_start; + pte_t *pte = consistent_pte + CONSISTENT_OFFSET(vaddr); + struct page *end = page + (1 << order); + + c->vm_pages = page; + split_page(page, order); + + /* + * Set the "dma handle" + */ + *handle = page_to_bus(page); + + do { + BUG_ON(!pte_none(*pte)); + + SetPageReserved(page); + set_pte_at(&init_mm, vaddr, + pte, mk_pte(page, + pgprot_writecombine + (PAGE_KERNEL))); + page++; + pte++; + vaddr += PAGE_SIZE; + } while (size -= PAGE_SIZE); + + /* + * Free the otherwise unused pages. + */ + while (page < end) { + __free_page(page); + page++; + } + + return (void *)c->vm_start; + } + + if (page) + __free_pages(page, order); +no_page: + return NULL; +} +EXPORT_SYMBOL(dma_alloc_coherent); + +/* + * free a page as defined by the above mapping. + */ +void dma_free_coherent(struct device *dev, size_t size, + void *vaddr, dma_addr_t dma_handle) +{ + struct metag_vm_region *c; + unsigned long flags, addr; + pte_t *ptep; + + size = PAGE_ALIGN(size); + + spin_lock_irqsave(&consistent_lock, flags); + + c = metag_vm_region_find(&consistent_head, (unsigned long)vaddr); + if (!c) + goto no_area; + + c->vm_active = 0; + if ((c->vm_end - c->vm_start) != size) { + pr_err("%s: freeing wrong coherent size (%ld != %d)\n", + __func__, c->vm_end - c->vm_start, size); + dump_stack(); + size = c->vm_end - c->vm_start; + } + + ptep = consistent_pte + CONSISTENT_OFFSET(c->vm_start); + addr = c->vm_start; + do { + pte_t pte = ptep_get_and_clear(&init_mm, addr, ptep); + unsigned long pfn; + + ptep++; + addr += PAGE_SIZE; + + if (!pte_none(pte) && pte_present(pte)) { + pfn = pte_pfn(pte); + + if (pfn_valid(pfn)) { + struct page *page = pfn_to_page(pfn); + ClearPageReserved(page); + + __free_page(page); + continue; + } + } + + pr_crit("%s: bad page in kernel page table\n", + __func__); + } while (size -= PAGE_SIZE); + + flush_tlb_kernel_range(c->vm_start, c->vm_end); + + list_del(&c->vm_list); + + spin_unlock_irqrestore(&consistent_lock, flags); + + kfree(c); + return; + +no_area: + spin_unlock_irqrestore(&consistent_lock, flags); + pr_err("%s: trying to free invalid coherent area: %p\n", + __func__, vaddr); + dump_stack(); +} +EXPORT_SYMBOL(dma_free_coherent); + + +static int dma_mmap(struct device *dev, struct vm_area_struct *vma, + void *cpu_addr, dma_addr_t dma_addr, size_t size) +{ + int ret = -ENXIO; + + unsigned long flags, user_size, kern_size; + struct metag_vm_region *c; + + user_size = (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; + + spin_lock_irqsave(&consistent_lock, flags); + c = metag_vm_region_find(&consistent_head, (unsigned long)cpu_addr); + spin_unlock_irqrestore(&consistent_lock, flags); + + if (c) { + unsigned long off = vma->vm_pgoff; + + kern_size = (c->vm_end - c->vm_start) >> PAGE_SHIFT; + + if (off < kern_size && + user_size <= (kern_size - off)) { + ret = remap_pfn_range(vma, vma->vm_start, + page_to_pfn(c->vm_pages) + off, + user_size << PAGE_SHIFT, + vma->vm_page_prot); + } + } + + + return ret; +} + +int dma_mmap_coherent(struct device *dev, struct vm_area_struct *vma, + void *cpu_addr, dma_addr_t dma_addr, size_t size) +{ + vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot); + return dma_mmap(dev, vma, cpu_addr, dma_addr, size); +} +EXPORT_SYMBOL(dma_mmap_coherent); + +int dma_mmap_writecombine(struct device *dev, struct vm_area_struct *vma, + void *cpu_addr, dma_addr_t dma_addr, size_t size) +{ + vma->vm_page_prot = pgprot_writecombine(vma->vm_page_prot); + return dma_mmap(dev, vma, cpu_addr, dma_addr, size); +} +EXPORT_SYMBOL(dma_mmap_writecombine); + + + + +/* + * Initialise the consistent memory allocation. + */ +static int __init dma_alloc_init(void) +{ + pgd_t *pgd, *pgd_k; + pud_t *pud, *pud_k; + pmd_t *pmd, *pmd_k; + pte_t *pte; + int ret = 0; + + do { + int offset = pgd_index(CONSISTENT_START); + pgd = pgd_offset(&init_mm, CONSISTENT_START); + pud = pud_alloc(&init_mm, pgd, CONSISTENT_START); + pmd = pmd_alloc(&init_mm, pud, CONSISTENT_START); + if (!pmd) { + pr_err("%s: no pmd tables\n", __func__); + ret = -ENOMEM; + break; + } + WARN_ON(!pmd_none(*pmd)); + + pte = pte_alloc_kernel(pmd, CONSISTENT_START); + if (!pte) { + pr_err("%s: no pte tables\n", __func__); + ret = -ENOMEM; + break; + } + + pgd_k = ((pgd_t *) mmu_get_base()) + offset; + pud_k = pud_offset(pgd_k, CONSISTENT_START); + pmd_k = pmd_offset(pud_k, CONSISTENT_START); + set_pmd(pmd_k, *pmd); + + consistent_pte = pte; + } while (0); + + return ret; +} +early_initcall(dma_alloc_init); + +/* + * make an area consistent to devices. + */ +void dma_sync_for_device(void *vaddr, size_t size, int dma_direction) +{ + /* + * Ensure any writes get through the write combiner. This is necessary + * even with DMA_FROM_DEVICE, or the write may dirty the cache after + * we've invalidated it and get written back during the DMA. + */ + + barrier(); + + switch (dma_direction) { + case DMA_BIDIRECTIONAL: + /* + * Writeback to ensure the device can see our latest changes and + * so that we have no dirty lines, and invalidate the cache + * lines too in preparation for receiving the buffer back + * (dma_sync_for_cpu) later. + */ + flush_dcache_region(vaddr, size); + break; + case DMA_TO_DEVICE: + /* + * Writeback to ensure the device can see our latest changes. + * There's no need to invalidate as the device shouldn't write + * to the buffer. + */ + writeback_dcache_region(vaddr, size); + break; + case DMA_FROM_DEVICE: + /* + * Invalidate to ensure we have no dirty lines that could get + * written back during the DMA. It's also safe to flush + * (writeback) here if necessary. + */ + invalidate_dcache_region(vaddr, size); + break; + case DMA_NONE: + BUG(); + } + + wmb(); +} +EXPORT_SYMBOL(dma_sync_for_device); + +/* + * make an area consistent to the core. + */ +void dma_sync_for_cpu(void *vaddr, size_t size, int dma_direction) +{ + /* + * Hardware L2 cache prefetch doesn't occur across 4K physical + * boundaries, however according to Documentation/DMA-API-HOWTO.txt + * kmalloc'd memory is DMA'able, so accesses in nearby memory could + * trigger a cache fill in the DMA buffer. + * + * This should never cause dirty lines, so a flush or invalidate should + * be safe to allow us to see data from the device. + */ + if (_meta_l2c_pf_is_enabled()) { + switch (dma_direction) { + case DMA_BIDIRECTIONAL: + case DMA_FROM_DEVICE: + invalidate_dcache_region(vaddr, size); + break; + case DMA_TO_DEVICE: + /* The device shouldn't have written to the buffer */ + break; + case DMA_NONE: + BUG(); + } + } + + rmb(); +} +EXPORT_SYMBOL(dma_sync_for_cpu); -- 2.39.5