From: "George G. Davis" <george_da...@mentor.com>

We fix a bug in dma_mmap_from_coherent() that appears when we map non page
aligned DMA memory. It cuts off the non aligned part (this is different to
dma_alloc_coherent() that always rounds up to full pages). So for mappings
of less than a page we get -ENXIO as dma_mmap_from_coherent() assumes we
want to map zero pages.

Signed-off-by: George G. Davis <george_da...@mentor.com>
Signed-off-by: Jiada Wang <jiada_w...@mentor.com>
Signed-off-by: Mark Craske <mark_cra...@mentor.com>
---
 drivers/base/dma-coherent.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/base/dma-coherent.c b/drivers/base/dma-coherent.c
index bdf28f7..abd83b7 100644
--- a/drivers/base/dma-coherent.c
+++ b/drivers/base/dma-coherent.c
@@ -262,7 +262,7 @@ int dma_mmap_from_coherent(struct device *dev, struct 
vm_area_struct *vma,
                unsigned long off = vma->vm_pgoff;
                int start = (vaddr - mem->virt_base) >> PAGE_SHIFT;
                int user_count = (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
-               int count = size >> PAGE_SHIFT;
+               int count = PAGE_ALIGN(size) >> PAGE_SHIFT;
 
                *ret = -ENXIO;
                if (off < count && user_count <= count - off) {
-- 
1.7.9.5


Reply via email to