Completion callback should happen after dma_descriptor_unmap() has happened. This allow the cache invalidate to happen and ensure that the data accessed by the upper layer is in memory that was from DMA rather than stale data. On some architecture this is done by the hardware, however we should make the code consistent to not cause confusion. Signed-off-by: Dave Jiang <dave.jiang@xxxxxxxxx> --- drivers/dma/ppc4xx/adma.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/dma/ppc4xx/adma.c b/drivers/dma/ppc4xx/adma.c index d4c77da..77a19de 100644 --- a/drivers/dma/ppc4xx/adma.c +++ b/drivers/dma/ppc4xx/adma.c @@ -1482,11 +1482,11 @@ static dma_cookie_t ppc440spe_adma_run_tx_complete_actions( cookie = desc->async_tx.cookie; desc->async_tx.cookie = 0; + dma_descriptor_unmap(&desc->async_tx); /* call the callback (must not sleep or submit new * operations to this channel) */ dmaengine_desc_get_callback_invoke(&desc->async_tx, NULL); - dma_descriptor_unmap(&desc->async_tx); } /* run dependent operations */ -- To unsubscribe from this list: send the line "unsubscribe dmaengine" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
