drivers: dma: intel_adsp_gpdma: fix issue with stop and PM refcounts

The DMA interface allows start and stop to be called multiple
times and driver should ensure nothing bad happens if the calls
are not balanced.

Fix an issue where after a start-stop sequence the DMA would be
powered down, and then a subsequent stop would result in a crash
as driver accesses registers of a powered down hardware block.

Fix the issue by handling stop without actually reading the hw
registers to check channel status.

Link: thesofproject/sof#8503
Signed-off-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>

drivers/dma/dma_dw_common.c

@@ -12,6 +12,7 @@
 #include <zephyr/device.h>
 #include <zephyr/init.h>
 #include <zephyr/drivers/dma.h>
+#include <zephyr/pm/device.h>
 #include <zephyr/pm/device_runtime.h>
 #include <soc.h>
 #include "dma_dw_common.h"
@@ -546,14 +547,25 @@ int dw_dma_stop(const struct device *dev, uint32_t channel)
 	const struct dw_dma_dev_cfg *const dev_cfg = dev->config;
 	struct dw_dma_dev_data *dev_data = dev->data;
 	struct dw_dma_chan_data *chan_data = &dev_data->chan[channel];
+	enum pm_device_state pm_state;
 	int ret = 0;
 
 	if (channel >= DW_CHAN_COUNT) {
 		ret = -EINVAL;
 		goto out;
 	}
 
+	/*
+	 * skip if device is not active. if we get an error for state_get,
+	 * do not skip but check actual hardware state and stop if
+	 * needed
+	 */
+	ret = pm_device_state_get(dev, &pm_state);
+	if (!ret && pm_state != PM_DEVICE_STATE_ACTIVE)
+		goto out;
+
 	if (!dw_dma_is_enabled(dev, channel) && chan_data->state != DW_DMA_SUSPENDED) {
+		ret = 0;
 		goto out;
 	}