drm/amd/display: Use correct plane for CAB cursor size allocation
authorAurabindo Pillai <aurabindo.pillai@amd.com>
Fri, 19 Aug 2022 19:11:19 +0000 (15:11 -0400)
committerAlex Deucher <alexander.deucher@amd.com>
Tue, 30 Aug 2022 21:08:07 +0000 (17:08 -0400)
[Why&How]
plane and stream variables used for cursor size allocation calculation
were stale from previous iteration. Redo the iteration to find the
correct cursor plane for the calculation.

Reviewed-by: Alvin Lee <Alvin.Lee2@amd.com>
Acked-by: Brian Chang <Brian.Chang@amd.com>
Signed-off-by: Aurabindo Pillai <aurabindo.pillai@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hwseq.c

index 1828774..8d9d96c 100644 (file)
@@ -295,8 +295,8 @@ static uint32_t dcn32_calculate_cab_allocation(struct dc *dc, struct dc_state *c
                }
 
                // Include cursor size for CAB allocation
-               for (i = 0; i < dc->res_pool->pipe_count; i++) {
-                       struct pipe_ctx *pipe = &ctx->res_ctx.pipe_ctx[i];
+               for (j = 0; j < dc->res_pool->pipe_count; j++) {
+                       struct pipe_ctx *pipe = &ctx->res_ctx.pipe_ctx[j];
                        struct hubp *hubp = pipe->plane_res.hubp;
 
                        if (pipe->stream && pipe->plane_state && hubp)
@@ -339,15 +339,25 @@ static uint32_t dcn32_calculate_cab_allocation(struct dc *dc, struct dc_state *c
        if (cache_lines_used % lines_per_way > 0)
                num_ways++;
 
-       if (stream->cursor_position.enable &&
-           !plane->address.grph.cursor_cache_addr.quad_part &&
-           cursor_size > 16384)
-               /* Cursor caching is not supported since it won't be on the same line.
-                * So we need an extra line to accommodate it. With large cursors and a single 4k monitor
-                * this case triggers corruption. If we're at the edge, then dont trigger display refresh
-                * from MALL. We only need to cache cursor if its greater that 64x64 at 4 bpp.
-                */
-               num_ways++;
+       for (i = 0; i < ctx->stream_count; i++) {
+               stream = ctx->streams[i];
+               for (j = 0; j < ctx->stream_status[i].plane_count; j++) {
+                       plane = ctx->stream_status[i].plane_states[j];
+
+                       if (stream->cursor_position.enable && plane &&
+                               !plane->address.grph.cursor_cache_addr.quad_part &&
+                               cursor_size > 16384) {
+                               /* Cursor caching is not supported since it won't be on the same line.
+                                * So we need an extra line to accommodate it. With large cursors and a single 4k monitor
+                                * this case triggers corruption. If we're at the edge, then dont trigger display refresh
+                                * from MALL. We only need to cache cursor if its greater that 64x64 at 4 bpp.
+                                */
+                               num_ways++;
+                               /* We only expect one cursor plane */
+                               break;
+                       }
+               }
+       }
 
        return num_ways;
 }