dmaengine: Introduce a device_config callback

The fact that the channel configuration is done in device_control is rather
misleading, since it's not really advertised as such, plus, the fact that the
framework exposes a function of its own makes it not really intuitive, while
we're losing the type checking whenever we pass that unsigned long argument.

Add a device_config callback to dma_device, with a fallback on the old
behaviour for now for existing drivers to opt in.

Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
diff --git a/include/linux/dmaengine.h b/include/linux/dmaengine.h
index 03a1feb..cf7c85d 100644
--- a/include/linux/dmaengine.h
+++ b/include/linux/dmaengine.h
@@ -608,6 +608,8 @@
  *	The function takes a buffer of size buf_len. The callback function will
  *	be called after period_len bytes have been transferred.
  * @device_prep_interleaved_dma: Transfer expression in a generic way.
+ * @device_config: Pushes a new configuration to a channel, return 0 or an error
+ *	code
  * @device_control: manipulate all pending operations on a channel, returns
  *	zero or error code
  * @device_tx_status: poll for transaction completion, the optional
@@ -674,6 +676,9 @@
 	struct dma_async_tx_descriptor *(*device_prep_interleaved_dma)(
 		struct dma_chan *chan, struct dma_interleaved_template *xt,
 		unsigned long flags);
+
+	int (*device_config)(struct dma_chan *chan,
+			     struct dma_slave_config *config);
 	int (*device_control)(struct dma_chan *chan, enum dma_ctrl_cmd cmd,
 		unsigned long arg);
 
@@ -697,6 +702,9 @@
 static inline int dmaengine_slave_config(struct dma_chan *chan,
 					  struct dma_slave_config *config)
 {
+	if (chan->device->device_config)
+		return chan->device->device_config(chan, config);
+
 	return dmaengine_device_control(chan, DMA_SLAVE_CONFIG,
 			(unsigned long)config);
 }