[PATCH 431/459] drm/amd/display: always use 4 dp lanes for dml

Alex Deucher alexdeucher at gmail.com
Mon Jun 17 19:49:20 UTC 2019


From: Jun Lei <Jun.Lei at amd.com>

[why]
current DML logic uses currently trained setting for number
of dp lanes in DML calculations.  this is obviously flawed since
just because 1 lane is in use doesn't mean only 1 lane can be used

this causes mode validation to fail depending on current state,
which is incorrect

[how]
DML should always assume 4 lanes are available.  validation of
bandwidth is not supposed to be handled by DML, since we do
link validation without DML already

also, DML is expecting there to be a copy of the max state, this
state is removed when update_bounding_box is called to update
actual SKU clocks.  fix this as well by duping last state.

Signed-off-by: Jun Lei <Jun.Lei at amd.com>
Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin at amd.com>
Acked-by: Bhawanpreet Lakha <Bhawanpreet.Lakha at amd.com>
Acked-by: Eric Yang <eric.yang2 at amd.com>
Signed-off-by: Alex Deucher <alexander.deucher at amd.com>
---
 .../gpu/drm/amd/display/dc/dcn20/dcn20_resource.c    | 12 +-----------
 1 file changed, 1 insertion(+), 11 deletions(-)

diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c
index caebf4746475..266d2ea50882 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_resource.c
@@ -1619,7 +1619,6 @@ int dcn20_populate_dml_pipes_from_context(
 
 	for (i = 0, pipe_cnt = 0; i < dc->res_pool->pipe_count; i++) {
 		struct dc_crtc_timing *timing = &res_ctx->pipe_ctx[i].stream->timing;
-		struct dc_link *link;
 
 		if (!res_ctx->pipe_ctx[i].stream)
 			continue;
@@ -1665,16 +1664,7 @@ int dcn20_populate_dml_pipes_from_context(
 		if (timing->timing_3d_format == TIMING_3D_FORMAT_HW_FRAME_PACKING)
 			pipes[pipe_cnt].pipe.dest.pixel_rate_mhz *= 2;
 		pipes[pipe_cnt].pipe.dest.otg_inst = res_ctx->pipe_ctx[i].stream_res.tg->inst;
-
-		link = res_ctx->pipe_ctx[i].stream->link;
-		if (link->cur_link_settings.lane_count != LANE_COUNT_UNKNOWN) {
-			pipes[pipe_cnt].dout.dp_lanes = link->cur_link_settings.lane_count;
-		} else if (link->verified_link_cap.lane_count != LANE_COUNT_UNKNOWN) {
-			pipes[pipe_cnt].dout.dp_lanes = link->verified_link_cap.lane_count;
-		} else {
-			/* Unknown link capabilities, so assume max */
-			pipes[pipe_cnt].dout.dp_lanes = 4;
-		}
+		pipes[pipe_cnt].dout.dp_lanes = 4;
 		pipes[pipe_cnt].pipe.dest.vtotal_min = res_ctx->pipe_ctx[i].stream->adjust.v_total_min;
 		pipes[pipe_cnt].pipe.dest.vtotal_max = res_ctx->pipe_ctx[i].stream->adjust.v_total_max;
 
-- 
2.20.1



More information about the amd-gfx mailing list