drm/amd/display: use uclk pstate latency for fw assisted mclk validation dcn32

[WHY?]
DCN32 uses fclk pstate watermarks for dummy pstate, and must always be
supported.

[HOW?]
Validation needs to be run with fclk pstate latency set
as the dummy pstate latency to get correct prefetch and bandwidth outputs.

Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Dillon Varone <Dillon.Varone@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.c b/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.c
index 659323ebd79d846de075e0f00960c6a9316cac9f..12e17bcfca3f05cf07bff004190afd3cf9ed258a 100644 (file)
--- a/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.c
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.c
@@ -1803,6 +1803,12 @@ void dcn32_calculate_wm_and_dlg_fpu(struct dc *dc, struct dc_state *context,
                         */
                        context->bw_ctx.dml.soc.dram_clock_change_latency_us =
                                        dc->clk_mgr->bw_params->wm_table.nv_entries[WM_A].dml_input.pstate_latency_us;
+                       /* For DCN32/321 need to validate with fclk pstate change latency equal to dummy so
+                        * prefetch is scheduled correctly to account for dummy pstate.
+                        */
+                       if (dummy_latency_index == 0)
+                               context->bw_ctx.dml.soc.fclk_change_latency_us =
+                                               dc->clk_mgr->bw_params->dummy_pstate_table[dummy_latency_index].dummy_pstate_latency_us;
                        dcn32_internal_validate_bw(dc, context, pipes, &pipe_cnt, &vlevel, false);
                        maxMpcComb = context->bw_ctx.dml.vba.maxMpcComb;
                        dcfclk_from_fw_based_mclk_switching = context->bw_ctx.dml.vba.DCFCLKState[vlevel][context->bw_ctx.dml.vba.maxMpcComb];
@@ -1990,6 +1996,10 @@ void dcn32_calculate_wm_and_dlg_fpu(struct dc *dc, struct dc_state *context,
 
        context->perf_params.stutter_period_us = context->bw_ctx.dml.vba.StutterPeriod;
 
+       if (context->bw_ctx.bw.dcn.clk.fw_based_mclk_switching && dummy_latency_index == 0)
+               context->bw_ctx.dml.soc.fclk_change_latency_us =
+                               dc->clk_mgr->bw_params->dummy_pstate_table[dummy_latency_index].dummy_pstate_latency_us;
+
        dcn32_calculate_dlg_params(dc, context, pipes, pipe_cnt, vlevel);
 
        if (!pstate_en)
@@ -1997,8 +2007,12 @@ void dcn32_calculate_wm_and_dlg_fpu(struct dc *dc, struct dc_state *context,
                context->bw_ctx.dml.soc.dram_clock_change_latency_us =
                                dc->clk_mgr->bw_params->wm_table.nv_entries[WM_A].dml_input.pstate_latency_us;
 
-       if (context->bw_ctx.bw.dcn.clk.fw_based_mclk_switching)
+       if (context->bw_ctx.bw.dcn.clk.fw_based_mclk_switching) {
                dcn30_setup_mclk_switch_using_fw_based_vblank_stretch(dc, context);
+               if (dummy_latency_index == 0)
+                       context->bw_ctx.dml.soc.fclk_change_latency_us =
+                                       dc->clk_mgr->bw_params->wm_table.nv_entries[WM_A].dml_input.fclk_change_latency_us;
+       }
 }
 
 static void dcn32_get_optimal_dcfclk_fclk_for_uclk(unsigned int uclk_mts,