uORBDeviceNode: use px4::atomic instead of volatile for _generation

_generation is read in a multi-threaded context w/o locking.

src/modules/uORB/uORBDeviceNode.hpp

@@ -39,6 +39,7 @@
 #include <lib/cdev/CDev.hpp>
 
 #include <containers/List.hpp>
+#include <px4_platform_common/atomic.h>
 
 namespace uORB
 {
@@ -185,7 +186,7 @@ public:
 
 	uint32_t lost_message_count() const { return _lost_messages; }
 
-	unsigned published_message_count() const { return _generation; }
+	unsigned published_message_count() const { return _generation.load(); }
 
 	const orb_metadata *get_meta() const { return _meta; }
 
@@ -267,7 +268,7 @@ private:
 	const uint8_t _instance; /**< orb multi instance identifier */
 	uint8_t     *_data{nullptr};   /**< allocated object buffer */
 	hrt_abstime   _last_update{0}; /**< time the object was last updated */
-	volatile unsigned   _generation{0};  /**< object generation count */
+	px4::atomic<unsigned>  _generation{0};  /**< object generation count */
 	List<uORB::SubscriptionCallback *>	_callbacks;
 	uint8_t   _priority;  /**< priority of the topic */
 	bool _published{false};  /**< has ever data been published */