[PATCH 362/459] drm/amd/display: Clean up locking in dcn*_apply_ctx_for_surface()

Alex Deucher alexdeucher at gmail.com
Mon Jun 17 19:45:03 UTC 2019


From: Leo Li <sunpeng.li at amd.com>

[Why]

dcn*_disable_plane() doesn't unlock the pipe anymore, making the extra
lock unnecessary.

In addition - during full plane updates - all necessary pipes should be
locked/unlocked together when modifying hubp to avoid tearing in
pipesplit setups.

[How]

Remove redundant locks, and add function to lock all pipes. If an
interdependent pipe update is required, lock down all pipes. Otherwise,
lock only the top pipe for the updated pipe tree.

Signed-off-by: Leo Li <sunpeng.li at amd.com>
Acked-by: Hawking Zhang <Hawking.Zhang at amd.com>
Signed-off-by: Alex Deucher <alexander.deucher at amd.com>
---
 .../amd/display/dc/dcn10/dcn10_hw_sequencer.c | 11 ++---
 .../drm/amd/display/dc/dcn20/dcn20_hwseq.c    | 46 +++++++------------
 2 files changed, 22 insertions(+), 35 deletions(-)

diff --git a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c
index 704de7369e86..a2d3c4fdc805 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_hw_sequencer.c
@@ -2456,6 +2456,11 @@ static void dcn10_apply_ctx_for_surface(
 	if (num_planes > 0)
 		program_all_pipe_in_tree(dc, top_pipe_to_program, context);
 
+#if defined(CONFIG_DRM_AMD_DC_DCN2_0)
+	/* Program secondary blending tree and writeback pipes */
+	if ((stream->num_wb_info > 0) && (dc->hwss.program_all_writeback_pipes_in_tree))
+		dc->hwss.program_all_writeback_pipes_in_tree(dc, stream, context);
+#endif
 	if (interdependent_update)
 		for (i = 0; i < dc->res_pool->pipe_count; i++) {
 			struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i];
@@ -2470,12 +2475,6 @@ static void dcn10_apply_ctx_for_surface(
 				&pipe_ctx->ttu_regs);
 		}
 
-#if defined(CONFIG_DRM_AMD_DC_DCN2_0)
-	/* Program secondary blending tree and writeback pipes */
-	if ((stream->num_wb_info > 0) && (dc->hwss.program_all_writeback_pipes_in_tree))
-		dc->hwss.program_all_writeback_pipes_in_tree(dc, stream, context);
-#endif
-
 	if (interdependent_update)
 		lock_all_pipes(dc, context, false);
 	else
diff --git a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c
index f2e5e4928119..ccb9f277911a 100644
--- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c
@@ -1270,8 +1270,6 @@ static void dcn20_pipe_control_lock(
 	}
 }
 
-
-
 static void dcn20_apply_ctx_for_surface(
 		struct dc *dc,
 		const struct dc_stream_state *stream,
@@ -1282,6 +1280,7 @@ static void dcn20_apply_ctx_for_surface(
 	int i;
 	struct timing_generator *tg;
 	bool removed_pipe[6] = { false };
+	bool interdependent_update = false;
 	struct pipe_ctx *top_pipe_to_program =
 			find_top_pipe_for_stream(dc, context, stream);
 	DC_LOGGER_INIT(dc->ctx->logger);
@@ -1291,7 +1290,13 @@ static void dcn20_apply_ctx_for_surface(
 
 	tg = top_pipe_to_program->stream_res.tg;
 
-	dcn20_pipe_control_lock(dc, top_pipe_to_program, true);
+	interdependent_update = top_pipe_to_program->plane_state &&
+		top_pipe_to_program->plane_state->update_flags.bits.full_update;
+
+	if (interdependent_update)
+		lock_all_pipes(dc, context, true);
+	else
+		dcn20_pipe_control_lock(dc, top_pipe_to_program, true);
 
 	if (num_planes == 0) {
 		/* OTG blank before remove all front end */
@@ -1311,16 +1316,9 @@ static void dcn20_apply_ctx_for_surface(
 		 */
 		if (pipe_ctx->plane_state && !old_pipe_ctx->plane_state) {
 			if (old_pipe_ctx->stream_res.tg == tg &&
-				old_pipe_ctx->plane_res.hubp &&
-				old_pipe_ctx->plane_res.hubp->opp_id != 0xf) {
+			    old_pipe_ctx->plane_res.hubp &&
+			    old_pipe_ctx->plane_res.hubp->opp_id != 0xf)
 				dcn20_disable_plane(dc, old_pipe_ctx);
-
-				/*
-				 * power down fe will unlock when calling reset, need
-				 * to lock it back here. Messy, need rework.
-				 */
-				pipe_ctx->stream_res.tg->funcs->lock(pipe_ctx->stream_res.tg);
-			}
 		}
 
 		if ((!pipe_ctx->plane_state ||
@@ -1343,35 +1341,25 @@ static void dcn20_apply_ctx_for_surface(
 	if ((stream->num_wb_info > 0) && (dc->hwss.program_all_writeback_pipes_in_tree))
 		dc->hwss.program_all_writeback_pipes_in_tree(dc, stream, context);
 
-	dcn20_pipe_control_lock(dc, top_pipe_to_program, false);
-
-	if (top_pipe_to_program->plane_state &&
-			top_pipe_to_program->plane_state->update_flags.bits.full_update)
+	if (interdependent_update)
 		for (i = 0; i < dc->res_pool->pipe_count; i++) {
 			struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i];
 
 			/* Skip inactive pipes and ones already updated */
-			if (!pipe_ctx->stream || pipe_ctx->stream == stream
-					|| !pipe_ctx->plane_state)
+			if (!pipe_ctx->stream || pipe_ctx->stream == stream ||
+			    !pipe_ctx->plane_state || !tg->funcs->is_tg_enabled(tg))
 				continue;
 
-			pipe_ctx->stream_res.tg->funcs->lock(pipe_ctx->stream_res.tg);
-
 			pipe_ctx->plane_res.hubp->funcs->hubp_setup_interdependent(
 				pipe_ctx->plane_res.hubp,
 				&pipe_ctx->dlg_regs,
 				&pipe_ctx->ttu_regs);
 		}
 
-	for (i = 0; i < dc->res_pool->pipe_count; i++) {
-		struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i];
-
-		if (!pipe_ctx->stream || pipe_ctx->stream == stream
-				|| !pipe_ctx->plane_state)
-			continue;
-
-		dcn20_pipe_control_lock(dc, pipe_ctx, false);
-	}
+	if (interdependent_update)
+		lock_all_pipes(dc, context, false);
+	else
+		dcn20_pipe_control_lock(dc, top_pipe_to_program, false);
 
 	for (i = 0; i < dc->res_pool->pipe_count; i++)
 		if (removed_pipe[i])
-- 
2.20.1



More information about the amd-gfx mailing list