Mathieu Poirier [off-list ref] writes:

Implementing perf related APIs to activate and terminate
a trace session.  More specifically dealing with the sink
buffer's internal mechanic along with perf's API to start
and stop interactions with the ring buffers.

A matter of preference, but I'd say that it would be easier to review
this part if you merged all the buffer related patches together.

+static void etb_reset_buffer(struct coresight_device *csdev,
+			     struct perf_output_handle *handle,
+			     void *sink_config)
+{
+	struct cs_buffers *buf = sink_config;
+	struct etb_drvdata *drvdata = dev_get_drvdata(csdev->dev.parent);
+
+	if (buf) {
+		/*
+		 * In snapshot mode ->data_size holds the new address of the
+		 * ring buffer's head.  The size itself is the whole address
+		 * range since we want the latest information.
+		 */
+		if (buf->snapshot)
+			handle->head = local_xchg(&buf->data_size,
+						  buf->nr_pages << PAGE_SHIFT);

Does it make sense to do this in etb_update_buffer() instead?

+		perf_aux_output_end(handle, local_xchg(&buf->data_size, 0),
+				    local_xchg(&buf->lost, 0));

The corresponding perf_aux_output_begin() is done in etm_event_add(),
I'd suggest that you do this in etm_event_del(),
unconditionally. Otherwise you're risking ending up with a refcount leak
and all sorts of horror.

Regards,
--
Alex