radeonsi: implement AMD_performance_monitor for CIK+

Expose most of the performance counter groups that are exposed by Catalyst.
Ideally, the driver will work with GPUPerfStudio at some point, but we are not
quite there yet. In any case, this is the reason for grouping multiple
instances of hardware blocks in the way it is implemented.

The counters can also be shown using the Gallium HUD. If one is interested to
see how work is distributed across multiple shader engines, one can set the
environment variable RADEON_PC_SEPARATE_SE=1 to obtain finer-grained performance
counter groups.

Part of the implementation is in radeon because an implementation for
older hardware would largely follow along the same lines, but exposing
a different set of blocks which are programmed slightly differently.

Reviewed-by: Marek Olšák <[email protected]>

1 files changed, 18 insertions, 3 deletions

diff --git a/src/gallium/drivers/radeon/r600_query.c b/src/gallium/drivers/radeon/r600_query.c
index 38bbbbf8a5e..09eabab0e7d 100644
--- a/src/gallium/drivers/radeon/r600_query.c
+++ b/src/gallium/drivers/radeon/r600_query.c
@@ -1141,11 +1141,15 @@ static int r600_get_driver_query_info(struct pipe_screen *screen,
 	struct r600_common_screen *rscreen = (struct r600_common_screen*)screen;
 	unsigned num_queries = r600_get_num_queries(rscreen);
 
-	if (!info)
-		return num_queries;
+	if (!info) {
+		unsigned num_perfcounters =
+			r600_get_perfcounter_info(rscreen, 0, NULL);
+
+		return num_queries + num_perfcounters;
+	}
 
 	if (index >= num_queries)
-		return 0;
+		return r600_get_perfcounter_info(rscreen, index - num_queries, info);
 
 	*info = r600_driver_query_list[index];
 
@@ -1166,9 +1170,19 @@ static int r600_get_driver_query_info(struct pipe_screen *screen,
 	return 1;
 }
 
+static int r600_get_driver_query_group_info(struct pipe_screen *screen,
+					    unsigned index,
+					    struct pipe_driver_query_group_info *info)
+{
+	struct r600_common_screen *rscreen = (struct r600_common_screen *)screen;
+
+	return r600_get_perfcounter_group_info(rscreen, index, info);
+}
+
 void r600_query_init(struct r600_common_context *rctx)
 {
 	rctx->b.create_query = r600_create_query;
+	rctx->b.create_batch_query = r600_create_batch_query;
 	rctx->b.destroy_query = r600_destroy_query;
 	rctx->b.begin_query = r600_begin_query;
 	rctx->b.end_query = r600_end_query;
@@ -1185,4 +1199,5 @@ void r600_query_init(struct r600_common_context *rctx)
 void r600_init_screen_query_functions(struct r600_common_screen *rscreen)
 {
 	rscreen->b.get_driver_query_info = r600_get_driver_query_info;
+	rscreen->b.get_driver_query_group_info = r600_get_driver_query_group_info;
 }