drm/amd/display: Fixes for mcache programming in DML21

[WHY & HOW]
- Fix indexing phantom planes for mcache programming in the wrapper
- Fix phantom mcache allocations to align with HW guidance
- Fix mcache assignment for chroma plane for multi-planar formats

Reviewed-by: Austin Zheng <Austin.Zheng@amd.com>
Signed-off-by: Dillon Varone <dillon.varone@amd.com>
Signed-off-by: Alex Hung <alex.hung@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

4 files changed, 53 insertions, 9 deletions

diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.c
index d68b4567e218..ec0beb139200 100644
--- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.c
+++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.c
@@ -254,7 +254,8 @@ static void expand_implict_subvp(const struct display_configuation_with_meta *di
 static void pack_mode_programming_params_with_implicit_subvp(struct dml2_core_instance *core, const struct display_configuation_with_meta *display_cfg,
 	const struct dml2_display_cfg *svp_expanded_display_cfg, struct dml2_display_cfg_programming *programming, struct dml2_core_scratch *scratch)
 {
-	unsigned int stream_index, plane_index, pipe_offset, stream_already_populated_mask, main_plane_index;
+	unsigned int stream_index, plane_index, pipe_offset, stream_already_populated_mask, main_plane_index, mcache_index;
+	unsigned int total_main_mcaches_required = 0;
 	int total_pipe_regs_copied = 0;
 	int dml_internal_pipe_index = 0;
 	const struct dml2_plane_parameters *main_plane;
@@ -325,6 +326,13 @@ static void pack_mode_programming_params_with_implicit_subvp(struct dml2_core_in
 
 		dml2_core_calcs_get_mall_allocation(&core->clean_me_up.mode_lib, &programming->plane_programming[plane_index].surface_size_mall_bytes, dml_internal_pipe_index);
 
+		memcpy(&programming->plane_programming[plane_index].mcache_allocation,
+				&display_cfg->stage2.mcache_allocations[plane_index],
+				sizeof(struct dml2_mcache_surface_allocation));
+		total_main_mcaches_required += programming->plane_programming[plane_index].mcache_allocation.num_mcaches_plane0 +
+				programming->plane_programming[plane_index].mcache_allocation.num_mcaches_plane1 -
+				(programming->plane_programming[plane_index].mcache_allocation.last_slice_sharing.plane0_plane1 ? 1 : 0);
+
 		for (pipe_offset = 0; pipe_offset < programming->plane_programming[plane_index].num_dpps_required; pipe_offset++) {
 			// Assign storage for this pipe's register values
 			programming->plane_programming[plane_index].pipe_regs[pipe_offset] = &programming->pipe_regs[total_pipe_regs_copied];
@@ -363,6 +371,22 @@ static void pack_mode_programming_params_with_implicit_subvp(struct dml2_core_in
 		memcpy(&programming->plane_programming[main_plane_index].phantom_plane.descriptor, phantom_plane, sizeof(struct dml2_plane_parameters));
 
 		dml2_core_calcs_get_mall_allocation(&core->clean_me_up.mode_lib, &programming->plane_programming[main_plane_index].svp_size_mall_bytes, dml_internal_pipe_index);
+
+		/* generate mcache allocation, phantoms use identical mcache configuration, but in the MALL set and unique mcache ID's beginning after all main ID's */
+		memcpy(&programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation,
+				&programming->plane_programming[main_plane_index].mcache_allocation,
+				sizeof(struct dml2_mcache_surface_allocation));
+		for (mcache_index = 0; mcache_index < programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.num_mcaches_plane0; mcache_index++) {
+			programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_plane0[mcache_index] += total_main_mcaches_required;
+			programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_mall_plane0[mcache_index] =
+					programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_plane0[mcache_index];
+		}
+		for (mcache_index = 0; mcache_index < programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.num_mcaches_plane1; mcache_index++) {
+			programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_plane1[mcache_index] += total_main_mcaches_required;
+			programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_mall_plane1[mcache_index] =
+					programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_plane1[mcache_index];
+		}
+
 		for (pipe_offset = 0; pipe_offset < programming->plane_programming[main_plane_index].num_dpps_required; pipe_offset++) {
 			// Assign storage for this pipe's register values
 			programming->plane_programming[main_plane_index].phantom_plane.pipe_regs[pipe_offset] = &programming->pipe_regs[total_pipe_regs_copied];
@@ -572,6 +596,10 @@ bool core_dcn4_mode_programming(struct dml2_core_mode_programming_in_out *in_out
 
 				dml2_core_calcs_get_mall_allocation(&core->clean_me_up.mode_lib, &in_out->programming->plane_programming[plane_index].surface_size_mall_bytes, dml_internal_pipe_index);
 
+				memcpy(&in_out->programming->plane_programming[plane_index].mcache_allocation,
+						&in_out->display_cfg->stage2.mcache_allocations[plane_index],
+						sizeof(struct dml2_mcache_surface_allocation));
+
 				for (pipe_offset = 0; pipe_offset < in_out->programming->plane_programming[plane_index].num_dpps_required; pipe_offset++) {
 					in_out->programming->plane_programming[plane_index].plane_descriptor = &in_out->programming->display_config.plane_descriptors[plane_index];
 
diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c
index e96a13dc43d4..47872c6f657e 100644
--- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c
+++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c
@@ -2637,6 +2637,9 @@ static void calculate_mcache_setting(
 	// Luma/Chroma combine in the last mcache
 	// In the case of Luma/Chroma combine-mCache (with lc_comb_mcache==1), all mCaches except the last segment are filled as much as possible, when stay aligned to mvmpg boundary
 	if (*p->lc_comb_mcache && l->is_dual_plane) {
+		/* if luma and chroma planes share an mcache, increase total chroma mcache count */
+		*p->num_mcaches_c = *p->num_mcaches_c + 1;
+
 		for (n = 0; n < *p->num_mcaches_l - 1; n++)
 			p->mcache_offsets_l[n] = (n + 1) * l->mvmpg_per_mcache_lb_l * l->mvmpg_access_width_l;
 		p->mcache_offsets_l[*p->num_mcaches_l - 1] = l->full_vp_access_width_l;
diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c
index a3324f7b9ba6..15c906c42ec4 100644
--- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c
+++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c
@@ -1082,12 +1082,21 @@ static bool all_timings_support_svp(const struct dml2_pmo_instance *pmo,
 	const struct dml2_fams2_meta *stream_fams2_meta;
 	unsigned int microschedule_vlines;
 	unsigned int i;
+	unsigned int mcaches_per_plane;
+	unsigned int total_mcaches_required = 0;
 
 	unsigned int num_planes_per_stream[DML2_MAX_PLANES] = { 0 };
 
 	/* confirm timing it is not a centered timing */
 	for (i = 0; i < display_config->display_config.num_planes; i++) {
 		plane_descriptor = &display_config->display_config.plane_descriptors[i];
+		mcaches_per_plane = 0;
+
+		if (plane_descriptor->surface.dcc.enable) {
+			mcaches_per_plane += display_config->stage2.mcache_allocations[i].num_mcaches_plane0 +
+					display_config->stage2.mcache_allocations[i].num_mcaches_plane1 -
+					(display_config->stage2.mcache_allocations[i].last_slice_sharing.plane0_plane1 ? 1 : 0);
+		}
 
 		if (is_bit_set_in_bitfield(mask, (unsigned char)plane_descriptor->stream_index)) {
 			num_planes_per_stream[plane_descriptor->stream_index]++;
@@ -1098,7 +1107,19 @@ static bool all_timings_support_svp(const struct dml2_pmo_instance *pmo,
 					plane_descriptor->composition.rotation_angle != dml2_rotation_0) {
 				return false;
 			}
+
+			/* phantom requires same number of mcaches as main */
+			if (plane_descriptor->surface.dcc.enable) {
+				mcaches_per_plane *= 2;
+			}
 		}
+
+		total_mcaches_required += mcaches_per_plane;
+	}
+
+	if (total_mcaches_required > pmo->soc_bb->num_dcc_mcaches) {
+		/* too many mcaches required */
+		return false;
 	}
 
 	for (i = 0; i < DML2_MAX_PLANES; i++) {
diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.c
index a8f58f8448e4..dc2ce5e77f57 100644
--- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.c
+++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.c
@@ -831,7 +831,6 @@ static bool dml2_top_soc15_build_mode_programming(struct dml2_build_mode_program
 	bool uclk_pstate_success = false;
 	bool vmin_success = false;
 	bool stutter_success = false;
-	unsigned int i;
 
 	memset(l, 0, sizeof(struct dml2_build_mode_programming_locals));
 	memset(in_out->programming, 0, sizeof(struct dml2_display_cfg_programming));
@@ -977,13 +976,6 @@ static bool dml2_top_soc15_build_mode_programming(struct dml2_build_mode_program
 	}
 
 	/*
-	* Populate mcache programming
-	*/
-	for (i = 0; i < in_out->display_config->num_planes; i++) {
-		in_out->programming->plane_programming[i].mcache_allocation = l->base_display_config_with_meta.stage2.mcache_allocations[i];
-	}
-
-	/*
 	* Call DPMM to map all requirements to minimum clock state
 	*/
 	if (result) {