Commit 5abc100e authored by Russell King's avatar Russell King

[PATCH] ARM: Ensure DMA-bounced buffers are properly written to RAM

When DMA bounce buffers were unmapped and the data was memcpy'd to
the original buffer, we were not ensuring that the data was written
to RAM.  This means that there was the potential for page cache
pages to have different cache states depending whether they've been
bounced or not.
Signed-off-by: default avatarRussell King <rmk+kernel@arm.linux.org.uk>
parent b8a9b66f
...@@ -302,12 +302,24 @@ unmap_single(struct device *dev, dma_addr_t dma_addr, size_t size, ...@@ -302,12 +302,24 @@ unmap_single(struct device *dev, dma_addr_t dma_addr, size_t size,
DO_STATS ( device_info->bounce_count++ ); DO_STATS ( device_info->bounce_count++ );
if ((dir == DMA_FROM_DEVICE) || if (dir == DMA_FROM_DEVICE || dir == DMA_BIDIRECTIONAL) {
(dir == DMA_BIDIRECTIONAL)) { unsigned long ptr;
dev_dbg(dev, dev_dbg(dev,
"%s: copy back safe %p to unsafe %p size %d\n", "%s: copy back safe %p to unsafe %p size %d\n",
__func__, buf->safe, buf->ptr, size); __func__, buf->safe, buf->ptr, size);
memcpy(buf->ptr, buf->safe, size); memcpy(buf->ptr, buf->safe, size);
/*
* DMA buffers must have the same cache properties
* as if they were really used for DMA - which means
* data must be written back to RAM. Note that
* we don't use dmac_flush_range() here for the
* bidirectional case because we know the cache
* lines will be coherent with the data written.
*/
ptr = (unsigned long)buf->ptr;
dmac_clean_range(ptr, ptr + size);
} }
free_safe_buffer(device_info, buf); free_safe_buffer(device_info, buf);
} }
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment