diff --git a/drivers/dma-buf/heaps/system_heap.c b/drivers/dma-buf/heaps/system_heap.c index 3a5fbbda630c..9525087a6879 100644 --- a/drivers/dma-buf/heaps/system_heap.c +++ b/drivers/dma-buf/heaps/system_heap.c @@ -22,6 +22,7 @@ #include static struct dma_heap *sys_heap; +static struct dma_heap *sys_uncached_heap; struct system_heap_buffer { struct dma_heap *heap; @@ -31,6 +32,8 @@ struct system_heap_buffer { struct sg_table sg_table; int vmap_cnt; void *vaddr; + + bool uncached; }; struct dma_heap_attachment { @@ -38,6 +41,8 @@ struct dma_heap_attachment { struct sg_table *table; struct list_head list; bool mapped; + + bool uncached; }; #define LOW_ORDER_GFP (GFP_HIGHUSER | __GFP_ZERO) @@ -100,7 +105,7 @@ static int system_heap_attach(struct dma_buf *dmabuf, a->dev = attachment->dev; INIT_LIST_HEAD(&a->list); a->mapped = false; - + a->uncached = buffer->uncached; attachment->priv = a; mutex_lock(&buffer->lock); @@ -130,9 +135,13 @@ static struct sg_table *system_heap_map_dma_buf(struct dma_buf_attachment *attac { struct dma_heap_attachment *a = attachment->priv; struct sg_table *table = a->table; + int attr = attachment->dma_map_attrs; int ret; - ret = dma_map_sgtable(attachment->dev, table, direction, 0); + if (a->uncached) + attr |= DMA_ATTR_SKIP_CPU_SYNC; + + ret = dma_map_sgtable(attachment->dev, table, direction, attr); if (ret) return ERR_PTR(ret); @@ -145,9 +154,12 @@ static void system_heap_unmap_dma_buf(struct dma_buf_attachment *attachment, enum dma_data_direction direction) { struct dma_heap_attachment *a = attachment->priv; + int attr = attachment->dma_map_attrs; + if (a->uncached) + attr |= DMA_ATTR_SKIP_CPU_SYNC; a->mapped = false; - dma_unmap_sgtable(attachment->dev, table, direction, 0); + dma_unmap_sgtable(attachment->dev, table, direction, attr); } static int system_heap_dma_buf_begin_cpu_access(struct dma_buf *dmabuf, @@ -161,10 +173,12 @@ static int system_heap_dma_buf_begin_cpu_access(struct dma_buf *dmabuf, if (buffer->vmap_cnt) invalidate_kernel_vmap_range(buffer->vaddr, buffer->len); - list_for_each_entry(a, &buffer->attachments, list) { - if (!a->mapped) - continue; - dma_sync_sgtable_for_cpu(a->dev, a->table, direction); + if (!buffer->uncached) { + list_for_each_entry(a, &buffer->attachments, list) { + if (!a->mapped) + continue; + dma_sync_sgtable_for_cpu(a->dev, a->table, direction); + } } mutex_unlock(&buffer->lock); @@ -182,10 +196,12 @@ static int system_heap_dma_buf_end_cpu_access(struct dma_buf *dmabuf, if (buffer->vmap_cnt) flush_kernel_vmap_range(buffer->vaddr, buffer->len); - list_for_each_entry(a, &buffer->attachments, list) { - if (!a->mapped) - continue; - dma_sync_sgtable_for_device(a->dev, a->table, direction); + if (!buffer->uncached) { + list_for_each_entry(a, &buffer->attachments, list) { + if (!a->mapped) + continue; + dma_sync_sgtable_for_device(a->dev, a->table, direction); + } } mutex_unlock(&buffer->lock); @@ -252,6 +268,11 @@ system_heap_dma_buf_begin_cpu_access_partial(struct dma_buf *dmabuf, if (buffer->vmap_cnt) invalidate_kernel_vmap_range(buffer->vaddr, buffer->len); + if (buffer->uncached) { + mutex_unlock(&buffer->lock); + return 0; + } + ret = system_heap_sgl_sync_range(dma_heap_get_dev(heap), table, offset, len, direction, true); mutex_unlock(&buffer->lock); @@ -274,6 +295,11 @@ system_heap_dma_buf_end_cpu_access_partial(struct dma_buf *dmabuf, if (buffer->vmap_cnt) flush_kernel_vmap_range(buffer->vaddr, buffer->len); + if (buffer->uncached) { + mutex_unlock(&buffer->lock); + return 0; + } + ret = system_heap_sgl_sync_range(dma_heap_get_dev(heap), table, offset, len, direction, false); mutex_unlock(&buffer->lock); @@ -289,6 +315,9 @@ static int system_heap_mmap(struct dma_buf *dmabuf, struct vm_area_struct *vma) struct sg_page_iter piter; int ret; + if (buffer->uncached) + vma->vm_page_prot = pgprot_writecombine(vma->vm_page_prot); + for_each_sgtable_page(table, &piter, vma->vm_pgoff) { struct page *page = sg_page_iter_page(&piter); @@ -310,17 +339,21 @@ static void *system_heap_do_vmap(struct system_heap_buffer *buffer) struct page **pages = vmalloc(sizeof(struct page *) * npages); struct page **tmp = pages; struct sg_page_iter piter; + pgprot_t pgprot = PAGE_KERNEL; void *vaddr; if (!pages) return ERR_PTR(-ENOMEM); + if (buffer->uncached) + pgprot = pgprot_writecombine(PAGE_KERNEL); + for_each_sgtable_page(table, &piter, 0) { WARN_ON(tmp - pages >= npages); *tmp++ = sg_page_iter_page(&piter); } - vaddr = vmap(pages, npages, VM_MAP, PAGE_KERNEL); + vaddr = vmap(pages, npages, VM_MAP, pgprot); vfree(pages); if (!vaddr) @@ -424,10 +457,11 @@ static struct page *alloc_largest_available(unsigned long size, return NULL; } -static struct dma_buf *system_heap_allocate(struct dma_heap *heap, - unsigned long len, - unsigned long fd_flags, - unsigned long heap_flags) +static struct dma_buf *system_heap_do_allocate(struct dma_heap *heap, + unsigned long len, + unsigned long fd_flags, + unsigned long heap_flags, + bool uncached) { struct system_heap_buffer *buffer; DEFINE_DMA_BUF_EXPORT_INFO(exp_info); @@ -448,6 +482,7 @@ static struct dma_buf *system_heap_allocate(struct dma_heap *heap, mutex_init(&buffer->lock); buffer->heap = heap; buffer->len = len; + buffer->uncached = uncached; INIT_LIST_HEAD(&pages); i = 0; @@ -493,6 +528,18 @@ static struct dma_buf *system_heap_allocate(struct dma_heap *heap, ret = PTR_ERR(dmabuf); goto free_pages; } + + /* + * For uncached buffers, we need to initially flush cpu cache, since + * the __GFP_ZERO on the allocation means the zeroing was done by the + * cpu and thus it is likely cached. Map (and implicitly flush) and + * unmap it now so we don't get corruption later on. + */ + if (buffer->uncached) { + dma_map_sgtable(dma_heap_get_dev(heap), table, DMA_BIDIRECTIONAL, 0); + dma_unmap_sgtable(dma_heap_get_dev(heap), table, DMA_BIDIRECTIONAL, 0); + } + return dmabuf; free_pages: @@ -510,10 +557,39 @@ free_buffer: return ERR_PTR(ret); } +static struct dma_buf *system_heap_allocate(struct dma_heap *heap, + unsigned long len, + unsigned long fd_flags, + unsigned long heap_flags) +{ + return system_heap_do_allocate(heap, len, fd_flags, heap_flags, false); +} static const struct dma_heap_ops system_heap_ops = { .allocate = system_heap_allocate, }; +static struct dma_buf *system_uncached_heap_allocate(struct dma_heap *heap, + unsigned long len, + unsigned long fd_flags, + unsigned long heap_flags) +{ + return system_heap_do_allocate(heap, len, fd_flags, heap_flags, true); +} + +/* Dummy function to be used until we can call coerce_mask_and_coherent */ +static struct dma_buf *system_uncached_heap_not_initialized(struct dma_heap *heap, + unsigned long len, + unsigned long fd_flags, + unsigned long heap_flags) +{ + return ERR_PTR(-EBUSY); +} + +static struct dma_heap_ops system_uncached_heap_ops = { + /* After system_heap_create is complete, we will swap this */ + .allocate = system_uncached_heap_not_initialized, +}; + static int system_heap_create(void) { struct dma_heap_export_info exp_info; @@ -526,6 +602,18 @@ static int system_heap_create(void) if (IS_ERR(sys_heap)) return PTR_ERR(sys_heap); + exp_info.name = "system-uncached"; + exp_info.ops = &system_uncached_heap_ops; + exp_info.priv = NULL; + + sys_uncached_heap = dma_heap_add(&exp_info); + if (IS_ERR(sys_uncached_heap)) + return PTR_ERR(sys_uncached_heap); + + dma_coerce_mask_and_coherent(dma_heap_get_dev(sys_uncached_heap), DMA_BIT_MASK(64)); + mb(); /* make sure we only set allocate after dma_mask is set */ + system_uncached_heap_ops.allocate = system_uncached_heap_allocate; + return 0; } module_init(system_heap_create); diff --git a/include/linux/dma-buf.h b/include/linux/dma-buf.h index f7d05558c9ae..1e80020d5742 100644 --- a/include/linux/dma-buf.h +++ b/include/linux/dma-buf.h @@ -580,6 +580,8 @@ struct dma_buf_attach_ops { * @importer_ops: importer operations for this attachment, if provided * dma_buf_map/unmap_attachment() must be called with the dma_resv lock held. * @importer_priv: importer specific attachment data. + * @dma_map_attrs: DMA attributes to be used when the exporter maps the buffer + * through dma_buf_map_attachment. * * This structure holds the attachment information between the dma_buf buffer * and its user device(s). The list contains one attachment struct per device @@ -600,6 +602,7 @@ struct dma_buf_attachment { const struct dma_buf_attach_ops *importer_ops; void *importer_priv; void *priv; + unsigned long dma_map_attrs; }; /**