drm/amd/display: Defer LUT memory powerdown until LUT bypass latches

[WHY]
Blnd, 3dlut, and shaper LUT select registers are double buffered, however
their accompanying LUT memory shutdown registers are not. As a result,
shutting down LUT memory immediately after setting a block to bypass causes
corruption as bypass only happens at next Vupdate.

[HOW]
Re-enable mem low power for CM block
Force optimization on next flip and disable LUT memory during optimization
sequence if LUT select field is then set to bypass

v2: squash in CONFIG_DRM_AMD_DC_DCN fix (Alex)
Reviewed-by: Eric Yang <Eric.Yang2@amd.com>
Acked-by: Anson Jacob <Anson.Jacob@amd.com>
Signed-off-by: Michael Strauss <michael.strauss@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/core/dc.c

@@ -1793,6 +1793,25 @@ static bool is_flip_pending_in_pipes(struct dc *dc, struct dc_state *context)
 	return false;
 }
 
+/* Perform updates here which need to be deferred until next vupdate
+ *
+ * i.e. blnd lut, 3dlut, and shaper lut bypass regs are double buffered
+ * but forcing lut memory to shutdown state is immediate. This causes
+ * single frame corruption as lut gets disabled mid-frame unless shutdown
+ * is deferred until after entering bypass.
+ */
+static void process_deferred_updates(struct dc *dc)
+{
+#ifdef CONFIG_DRM_AMD_DC_DCN
+	int i;
+
+	if (dc->debug.enable_mem_low_power.bits.cm)
+		for (i = 0; i < dc->dcn_ip->max_num_dpp; i++)
+			if (dc->res_pool->dpps[i]->funcs->dpp_deferred_update)
+				dc->res_pool->dpps[i]->funcs->dpp_deferred_update(dc->res_pool->dpps[i]);
+#endif
+}
+
 void dc_post_update_surfaces_to_stream(struct dc *dc)
 {
 	int i;
@@ -1818,6 +1837,8 @@ void dc_post_update_surfaces_to_stream(struct dc *dc)
 			dc->hwss.disable_plane(dc, &context->res_ctx.pipe_ctx[i]);
 		}
 
+	process_deferred_updates(dc);
+
 	dc->hwss.optimize_bandwidth(dc, context);
 
 	dc->optimized_required = false;

drivers/gpu/drm/amd/display/dc/dcn30/dcn30_dpp.c

@@ -488,6 +488,40 @@ void dpp3_cnv_set_bias_scale(
 	REG_UPDATE(FCNV_FP_SCALE_B, FCNV_FP_SCALE_B, bias_and_scale->scale_blue);
 }
 
+void dpp3_deferred_update(
+	struct dpp *dpp_base)
+{
+	int bypass_state;
+	struct dcn3_dpp *dpp = TO_DCN30_DPP(dpp_base);
+
+	if (dpp_base->deferred_reg_writes.bits.disable_blnd_lut) {
+		REG_GET(CM_BLNDGAM_CONTROL, CM_BLNDGAM_MODE_CURRENT, &bypass_state);
+		if (bypass_state == 0) {	// only program if bypass was latched
+			REG_UPDATE(CM_MEM_PWR_CTRL, BLNDGAM_MEM_PWR_FORCE, 3);
+		} else
+			ASSERT(0); // LUT select was updated again before vupdate
+		dpp_base->deferred_reg_writes.bits.disable_blnd_lut = false;
+	}
+
+	if (dpp_base->deferred_reg_writes.bits.disable_3dlut) {
+		REG_GET(CM_3DLUT_MODE, CM_3DLUT_MODE_CURRENT, &bypass_state);
+		if (bypass_state == 0) {	// only program if bypass was latched
+			REG_UPDATE(CM_MEM_PWR_CTRL2, HDR3DLUT_MEM_PWR_FORCE, 3);
+		} else
+			ASSERT(0); // LUT select was updated again before vupdate
+		dpp_base->deferred_reg_writes.bits.disable_3dlut = false;
+	}
+
+	if (dpp_base->deferred_reg_writes.bits.disable_shaper) {
+		REG_GET(CM_SHAPER_CONTROL, CM_SHAPER_MODE_CURRENT, &bypass_state);
+		if (bypass_state == 0) {	// only program if bypass was latched
+			REG_UPDATE(CM_MEM_PWR_CTRL2, SHAPER_MEM_PWR_FORCE, 3);
+		} else
+			ASSERT(0); // LUT select was updated again before vupdate
+		dpp_base->deferred_reg_writes.bits.disable_shaper = false;
+	}
+}
+
 static void dpp3_power_on_blnd_lut(
 	struct dpp *dpp_base,
 	bool power_on)
@@ -495,9 +529,13 @@ static void dpp3_power_on_blnd_lut(
 	struct dcn3_dpp *dpp = TO_DCN30_DPP(dpp_base);
 
 	if (dpp_base->ctx->dc->debug.enable_mem_low_power.bits.cm) {
-		REG_UPDATE(CM_MEM_PWR_CTRL, BLNDGAM_MEM_PWR_FORCE, power_on ? 0 : 3);
-		if (power_on)
+		if (power_on) {
+			REG_UPDATE(CM_MEM_PWR_CTRL, BLNDGAM_MEM_PWR_FORCE, 0);
 			REG_WAIT(CM_MEM_PWR_STATUS, BLNDGAM_MEM_PWR_STATE, 0, 1, 5);
+		} else {
+			dpp_base->ctx->dc->optimized_required = true;
+			dpp_base->deferred_reg_writes.bits.disable_blnd_lut = true;
+		}
 	} else {
 		REG_SET(CM_MEM_PWR_CTRL, 0,
 				BLNDGAM_MEM_PWR_FORCE, power_on == true ? 0 : 1);
@@ -511,9 +549,13 @@ static void dpp3_power_on_hdr3dlut(
 	struct dcn3_dpp *dpp = TO_DCN30_DPP(dpp_base);
 
 	if (dpp_base->ctx->dc->debug.enable_mem_low_power.bits.cm) {
-		REG_UPDATE(CM_MEM_PWR_CTRL2, HDR3DLUT_MEM_PWR_FORCE, power_on ? 0 : 3);
-		if (power_on)
+		if (power_on) {
+			REG_UPDATE(CM_MEM_PWR_CTRL2, HDR3DLUT_MEM_PWR_FORCE, 0);
 			REG_WAIT(CM_MEM_PWR_STATUS2, HDR3DLUT_MEM_PWR_STATE, 0, 1, 5);
+		} else {
+			dpp_base->ctx->dc->optimized_required = true;
+			dpp_base->deferred_reg_writes.bits.disable_3dlut = true;
+		}
 	}
 }
 
@@ -524,9 +566,13 @@ static void dpp3_power_on_shaper(
 	struct dcn3_dpp *dpp = TO_DCN30_DPP(dpp_base);
 
 	if (dpp_base->ctx->dc->debug.enable_mem_low_power.bits.cm) {
-		REG_UPDATE(CM_MEM_PWR_CTRL2, SHAPER_MEM_PWR_FORCE, power_on ? 0 : 3);
-		if (power_on)
+		if (power_on) {
+			REG_UPDATE(CM_MEM_PWR_CTRL2, SHAPER_MEM_PWR_FORCE, 0);
 			REG_WAIT(CM_MEM_PWR_STATUS2, SHAPER_MEM_PWR_STATE, 0, 1, 5);
+		} else {
+			dpp_base->ctx->dc->optimized_required = true;
+			dpp_base->deferred_reg_writes.bits.disable_shaper = true;
+		}
 	}
 }
 
@@ -1400,6 +1446,7 @@ static struct dpp_funcs dcn30_dpp_funcs = {
 	.dpp_program_blnd_lut = dpp3_program_blnd_lut,
 	.dpp_program_shaper_lut = dpp3_program_shaper,
 	.dpp_program_3dlut = dpp3_program_3dlut,
+	.dpp_deferred_update = dpp3_deferred_update,
 	.dpp_program_bias_and_scale	= NULL,
 	.dpp_cnv_set_alpha_keyer	= dpp2_cnv_set_alpha_keyer,
 	.set_cursor_attributes		= dpp3_set_cursor_attributes,

drivers/gpu/drm/amd/display/dc/dcn31/dcn31_resource.c

@@ -1013,7 +1013,7 @@ static const struct dc_debug_options debug_defaults_drv = {
 			.i2c = true,
 			.dmcu = false, // This is previously known to cause hang on S3 cycles if enabled
 			.dscl = true,
-			.cm = false, // visible flicker on OLED eDPs
+			.cm = true,
 			.mpc = true,
 			.optc = true,
 			.vpg = true,

drivers/gpu/drm/amd/display/dc/inc/hw/dpp.h

@@ -29,6 +29,15 @@
 
 #include "transform.h"
 
+union defer_reg_writes {
+	struct {
+		bool disable_blnd_lut:1;
+		bool disable_3dlut:1;
+		bool disable_shaper:1;
+	} bits;
+	uint32_t raw;
+};
+
 struct dpp {
 	const struct dpp_funcs *funcs;
 	struct dc_context *ctx;
@@ -43,6 +52,7 @@ struct dpp {
 	struct pwl_params regamma_params;
 	struct pwl_params degamma_params;
 	struct dpp_cursor_attributes cur_attr;
+	union defer_reg_writes deferred_reg_writes;
 
 	struct pwl_params shaper_params;
 	bool cm_bypass_mode;
@@ -245,6 +255,8 @@ struct dpp_funcs {
 			bool dppclk_div,
 			bool enable);
 
+	void (*dpp_deferred_update)(
+			struct dpp *dpp);
 	bool (*dpp_program_blnd_lut)(
 			struct dpp *dpp,
 			const struct pwl_params *params);