[PATCH AUTOSEL 6.12 09/31] drm/amd/display: Populate chroma prefetch parameters, DET buffer fix

[PATCH AUTOSEL 6.12 09/31] drm/amd/display: Populate chroma prefetch parameters, DET buffer fix

[Sasha Levin]

From: Ausef Yousof <Ausef.Yousof@amd.com>

[ Upstream commit 70fec46519fca859aa209f5f02e7e0a0123aca4a ]

[WHY]
Soft hang/lag observed during 10bit playback + moving cursor, corruption
observed in other tickets for same reason, also failing MPO.

1. Currently, we are always running
   calculate_lowest_supported_state_for_temp_read which is only
   necessary on dGPU
2. Fast validate path does not apply DET buffer allocation policy
3. Prefetch UrgBFactor chroma parameter not populated in prefetch
   calculation

[HOW]
1. Add a check to see if we are on APU, if so, skip the code
2. Add det buffer alloc policy checks to fast validate path
3. Populate UrgentBurstChroma param in call to calculate
   UrgBChroma prefetch values

-revision commits: small formatting/brackets/null check addition + remove test change + dGPU code

Reviewed-by: Charlene Liu <charlene.liu@amd.com>
Signed-off-by: Ausef Yousof <Ausef.Yousof@amd.com>
Signed-off-by: Fangzhi Zuo <jerry.zuo@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 .../amd/display/dc/dml2/display_mode_core.c   |  5 ++-
 .../drm/amd/display/dc/dml2/dml2_wrapper.c    | 35 +++++++++++++------
 2 files changed, 29 insertions(+), 11 deletions(-)

diff --git a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c b/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c
index 8dabb1ac0b684..be87dc0f07799 100644
--- a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c
+++ b/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c
@@ -6434,7 +6434,7 @@ static void dml_prefetch_check(struct display_mode_lib_st *mode_lib)
 							/* Output */
 							&mode_lib->ms.UrgentBurstFactorCursorPre[k],
 							&mode_lib->ms.UrgentBurstFactorLumaPre[k],
-							&mode_lib->ms.UrgentBurstFactorChroma[k],
+							&mode_lib->ms.UrgentBurstFactorChromaPre[k],
 							&mode_lib->ms.NotUrgentLatencyHidingPre[k]);
 
 					mode_lib->ms.cursor_bw_pre[k] = mode_lib->ms.cache_display_cfg.plane.NumberOfCursors[k] * mode_lib->ms.cache_display_cfg.plane.CursorWidth[k] *
@@ -9190,6 +9190,8 @@ void dml_core_mode_programming(struct display_mode_lib_st *mode_lib, const struc
 			&locals->FractionOfUrgentBandwidth,
 			&s->dummy_boolean[0]); // dml_bool_t *PrefetchBandwidthSupport
 
+
+
 		if (s->VRatioPrefetchMoreThanMax != false || s->DestinationLineTimesForPrefetchLessThan2 != false) {
 			dml_print("DML::%s: VRatioPrefetchMoreThanMax                   = %u\n", __func__, s->VRatioPrefetchMoreThanMax);
 			dml_print("DML::%s: DestinationLineTimesForPrefetchLessThan2    = %u\n", __func__, s->DestinationLineTimesForPrefetchLessThan2);
@@ -9204,6 +9206,7 @@ void dml_core_mode_programming(struct display_mode_lib_st *mode_lib, const struc
 			}
 		}
 
+
 		if (locals->PrefetchModeSupported == true && mode_lib->ms.support.ImmediateFlipSupport == true) {
 			locals->BandwidthAvailableForImmediateFlip = CalculateBandwidthAvailableForImmediateFlip(
 																	mode_lib->ms.num_active_planes,
diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c b/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c
index 866b0abcff1ba..4d64c45930da4 100644
--- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c
+++ b/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c
@@ -533,14 +533,21 @@ static bool optimize_pstate_with_svp_and_drr(struct dml2_context *dml2, struct d
 static bool call_dml_mode_support_and_programming(struct dc_state *context)
 {
 	unsigned int result = 0;
-	unsigned int min_state;
+	unsigned int min_state = 0;
 	int min_state_for_g6_temp_read = 0;
+
+
+	if (!context)
+		return false;
+
 	struct dml2_context *dml2 = context->bw_ctx.dml2;
 	struct dml2_wrapper_scratch *s = &dml2->v20.scratch;
 
-	min_state_for_g6_temp_read = calculate_lowest_supported_state_for_temp_read(dml2, context);
+	if (!context->streams[0]->sink->link->dc->caps.is_apu) {
+		min_state_for_g6_temp_read = calculate_lowest_supported_state_for_temp_read(dml2, context);
 
-	ASSERT(min_state_for_g6_temp_read >= 0);
+		ASSERT(min_state_for_g6_temp_read >= 0);
+	}
 
 	if (!dml2->config.use_native_pstate_optimization) {
 		result = optimize_pstate_with_svp_and_drr(dml2, context);
@@ -551,14 +558,20 @@ static bool call_dml_mode_support_and_programming(struct dc_state *context)
 	/* Upon trying to sett certain frequencies in FRL, min_state_for_g6_temp_read is reported as -1. This leads to an invalid value of min_state causing crashes later on.
 	 * Use the default logic for min_state only when min_state_for_g6_temp_read is a valid value. In other cases, use the value calculated by the DML directly.
 	 */
-	if (min_state_for_g6_temp_read >= 0)
-		min_state = min_state_for_g6_temp_read > s->mode_support_params.out_lowest_state_idx ? min_state_for_g6_temp_read : s->mode_support_params.out_lowest_state_idx;
-	else
-		min_state = s->mode_support_params.out_lowest_state_idx;
-
-	if (result)
-		result = dml_mode_programming(&dml2->v20.dml_core_ctx, min_state, &s->cur_display_config, true);
+	if (!context->streams[0]->sink->link->dc->caps.is_apu) {
+		if (min_state_for_g6_temp_read >= 0)
+			min_state = min_state_for_g6_temp_read > s->mode_support_params.out_lowest_state_idx ? min_state_for_g6_temp_read : s->mode_support_params.out_lowest_state_idx;
+		else
+			min_state = s->mode_support_params.out_lowest_state_idx;
+	}
 
+	if (result) {
+		if (!context->streams[0]->sink->link->dc->caps.is_apu) {
+			result = dml_mode_programming(&dml2->v20.dml_core_ctx, min_state, &s->cur_display_config, true);
+		} else {
+			result = dml_mode_programming(&dml2->v20.dml_core_ctx, s->mode_support_params.out_lowest_state_idx, &s->cur_display_config, true);
+		}
+	}
 	return result;
 }
 
@@ -687,6 +700,8 @@ static bool dml2_validate_only(struct dc_state *context)
 	build_unoptimized_policy_settings(dml2->v20.dml_core_ctx.project, &dml2->v20.dml_core_ctx.policy);
 
 	map_dc_state_into_dml_display_cfg(dml2, context, &dml2->v20.scratch.cur_display_config);
+	 if (!dml2->config.skip_hw_state_mapping)
+		 dml2_apply_det_buffer_allocation_policy(dml2, &dml2->v20.scratch.cur_display_config);
 
 	result = pack_and_call_dml_mode_support_ex(dml2,
 		&dml2->v20.scratch.cur_display_config,
-- 
2.39.5

[PATCH AUTOSEL 6.12 10/31] drm/amd/display: Overwriting dualDPP UBF values before usage

[Sasha Levin]

From: Ausef Yousof <Ausef.Yousof@amd.com>

[ Upstream commit 24909d9ec7c3afa8da2f3c9afa312e7a4a61f250 ]

[WHY]
Right now in dml2 mode validation we are calculating UBF parameters for
prefetch calculation for single and dual DPP scenarios. Data structure
to store such values are just 1D arrays, the single DPP values are
overwritten by the dualDPP values, and we end up using dualDPP for
prefetch calculations twice (once in place of singleDPP support check
and again for dual).

This naturally leads to many problems, one of which validating a mode in
"singleDPP" (when we used dual DPP parameters) and sending the singleDPP
parameters to mode programming, if we cannot support then we observe the
corruption as described in the ticket.

[HOW]
UBF values need to have 2d arrays to store values specific to single and
dual DPP states to avoid single DPP values being overwritten. Other
parameters are recorded on a per state basis such as prefetch UBF values
but they are in the same loop used for calculation and at that point its
fine to overwrite them, its not the case for plain UBF values.

Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Ausef Yousof <Ausef.Yousof@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>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 .../amd/display/dc/dml2/display_mode_core.c   | 30 +++++++++----------
 .../dc/dml2/display_mode_core_structs.h       |  6 ++--
 2 files changed, 18 insertions(+), 18 deletions(-)

diff --git a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c b/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c
index be87dc0f07799..6822b07951204 100644
--- a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c
+++ b/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c
@@ -6301,9 +6301,9 @@ static void dml_prefetch_check(struct display_mode_lib_st *mode_lib)
 			mode_lib->ms.meta_row_bandwidth_this_state,
 			mode_lib->ms.dpte_row_bandwidth_this_state,
 			mode_lib->ms.NoOfDPPThisState,
-			mode_lib->ms.UrgentBurstFactorLuma,
-			mode_lib->ms.UrgentBurstFactorChroma,
-			mode_lib->ms.UrgentBurstFactorCursor);
+			mode_lib->ms.UrgentBurstFactorLuma[j],
+			mode_lib->ms.UrgentBurstFactorChroma[j],
+			mode_lib->ms.UrgentBurstFactorCursor[j]);
 
 		s->VMDataOnlyReturnBWPerState = dml_get_return_bw_mbps_vm_only(
 																	&mode_lib->ms.soc,
@@ -6458,9 +6458,9 @@ static void dml_prefetch_check(struct display_mode_lib_st *mode_lib)
 				mode_lib->ms.cursor_bw_pre,
 				mode_lib->ms.prefetch_vmrow_bw,
 				mode_lib->ms.NoOfDPPThisState,
-				mode_lib->ms.UrgentBurstFactorLuma,
-				mode_lib->ms.UrgentBurstFactorChroma,
-				mode_lib->ms.UrgentBurstFactorCursor,
+				mode_lib->ms.UrgentBurstFactorLuma[j],
+				mode_lib->ms.UrgentBurstFactorChroma[j],
+				mode_lib->ms.UrgentBurstFactorCursor[j],
 				mode_lib->ms.UrgentBurstFactorLumaPre,
 				mode_lib->ms.UrgentBurstFactorChromaPre,
 				mode_lib->ms.UrgentBurstFactorCursorPre,
@@ -6517,9 +6517,9 @@ static void dml_prefetch_check(struct display_mode_lib_st *mode_lib)
 						mode_lib->ms.cursor_bw,
 						mode_lib->ms.cursor_bw_pre,
 						mode_lib->ms.NoOfDPPThisState,
-						mode_lib->ms.UrgentBurstFactorLuma,
-						mode_lib->ms.UrgentBurstFactorChroma,
-						mode_lib->ms.UrgentBurstFactorCursor,
+						mode_lib->ms.UrgentBurstFactorLuma[j],
+						mode_lib->ms.UrgentBurstFactorChroma[j],
+						mode_lib->ms.UrgentBurstFactorCursor[j],
 						mode_lib->ms.UrgentBurstFactorLumaPre,
 						mode_lib->ms.UrgentBurstFactorChromaPre,
 						mode_lib->ms.UrgentBurstFactorCursorPre);
@@ -6586,9 +6586,9 @@ static void dml_prefetch_check(struct display_mode_lib_st *mode_lib)
 													mode_lib->ms.cursor_bw_pre,
 													mode_lib->ms.prefetch_vmrow_bw,
 													mode_lib->ms.NoOfDPP[j], // VBA_ERROR DPPPerSurface is not assigned at this point, should use NoOfDpp here
-													mode_lib->ms.UrgentBurstFactorLuma,
-													mode_lib->ms.UrgentBurstFactorChroma,
-													mode_lib->ms.UrgentBurstFactorCursor,
+													mode_lib->ms.UrgentBurstFactorLuma[j],
+													mode_lib->ms.UrgentBurstFactorChroma[j],
+													mode_lib->ms.UrgentBurstFactorCursor[j],
 													mode_lib->ms.UrgentBurstFactorLumaPre,
 													mode_lib->ms.UrgentBurstFactorChromaPre,
 													mode_lib->ms.UrgentBurstFactorCursorPre,
@@ -7809,9 +7809,9 @@ dml_bool_t dml_core_mode_support(struct display_mode_lib_st *mode_lib)
 				mode_lib->ms.DETBufferSizeYThisState[k],
 				mode_lib->ms.DETBufferSizeCThisState[k],
 				/* Output */
-				&mode_lib->ms.UrgentBurstFactorCursor[k],
-				&mode_lib->ms.UrgentBurstFactorLuma[k],
-				&mode_lib->ms.UrgentBurstFactorChroma[k],
+				&mode_lib->ms.UrgentBurstFactorCursor[j][k],
+				&mode_lib->ms.UrgentBurstFactorLuma[j][k],
+				&mode_lib->ms.UrgentBurstFactorChroma[j][k],
 				&mode_lib->ms.NotUrgentLatencyHiding[k]);
 		}
 
diff --git a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core_structs.h b/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core_structs.h
index f951936bb579e..504c427b3b319 100644
--- a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core_structs.h
+++ b/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core_structs.h
@@ -884,11 +884,11 @@ struct mode_support_st {
 	dml_uint_t meta_row_height[__DML_NUM_PLANES__];
 	dml_uint_t meta_row_height_chroma[__DML_NUM_PLANES__];
 	dml_float_t UrgLatency;
-	dml_float_t UrgentBurstFactorCursor[__DML_NUM_PLANES__];
+	dml_float_t UrgentBurstFactorCursor[2][__DML_NUM_PLANES__];
 	dml_float_t UrgentBurstFactorCursorPre[__DML_NUM_PLANES__];
-	dml_float_t UrgentBurstFactorLuma[__DML_NUM_PLANES__];
+	dml_float_t UrgentBurstFactorLuma[2][__DML_NUM_PLANES__];
 	dml_float_t UrgentBurstFactorLumaPre[__DML_NUM_PLANES__];
-	dml_float_t UrgentBurstFactorChroma[__DML_NUM_PLANES__];
+	dml_float_t UrgentBurstFactorChroma[2][__DML_NUM_PLANES__];
 	dml_float_t UrgentBurstFactorChromaPre[__DML_NUM_PLANES__];
 	dml_float_t MaximumSwathWidthInLineBufferLuma;
 	dml_float_t MaximumSwathWidthInLineBufferChroma;
-- 
2.39.5

[PATCH AUTOSEL 6.12 15/31] drm/amd/display: use eld_mutex to protect access to connector->eld

[Sasha Levin]

From: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>

[ Upstream commit 819bee01eea06282d7bda17d46caf29cae4f6d84 ]

Reading access to connector->eld can happen at the same time the
drm_edid_to_eld() updates the data. Take the newly added eld_mutex in
order to protect connector->eld from concurrent access.

Reviewed-by: Maxime Ripard <mripard@kernel.org>
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241206-drm-connector-eld-mutex-v2-4-c9bce1ee8bea@linaro.org
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
index ea403fece8392..84b801171fe5a 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
@@ -1036,8 +1036,10 @@ static int amdgpu_dm_audio_component_get_eld(struct device *kdev, int port,
 			continue;
 
 		*enabled = true;
+		mutex_lock(&connector->eld_mutex);
 		ret = drm_eld_size(connector->eld);
 		memcpy(buf, connector->eld, min(max_bytes, ret));
+		mutex_unlock(&connector->eld_mutex);
 
 		break;
 	}
-- 
2.39.5

[PATCH AUTOSEL 6.12 17/31] drm/radeon: use eld_mutex to protect access to connector->eld

[Sasha Levin]

From: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>

[ Upstream commit b54c14f82428c8a602392d4cae1958a71a578132 ]

Reading access to connector->eld can happen at the same time the
drm_edid_to_eld() updates the data. Take the newly added eld_mutex in
order to protect connector->eld from concurrent access.

Reviewed-by: Maxime Ripard <mripard@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20241206-drm-connector-eld-mutex-v2-8-c9bce1ee8bea@linaro.org
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 drivers/gpu/drm/radeon/radeon_audio.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/gpu/drm/radeon/radeon_audio.c b/drivers/gpu/drm/radeon/radeon_audio.c
index 5b69cc8011b42..8d64ba18572ec 100644
--- a/drivers/gpu/drm/radeon/radeon_audio.c
+++ b/drivers/gpu/drm/radeon/radeon_audio.c
@@ -775,8 +775,10 @@ static int radeon_audio_component_get_eld(struct device *kdev, int port,
 		if (!dig->pin || dig->pin->id != port)
 			continue;
 		*enabled = true;
+		mutex_lock(&connector->eld_mutex);
 		ret = drm_eld_size(connector->eld);
 		memcpy(buf, connector->eld, min(max_bytes, ret));
+		mutex_unlock(&connector->eld_mutex);
 		break;
 	}
 
-- 
2.39.5

[PATCH AUTOSEL 6.12 20/31] drm/amd/display: Fix Mode Cutoff in DSC Passthrough to DP2.1 Monitor

[Sasha Levin]

From: Fangzhi Zuo <Jerry.Zuo@amd.com>

[ Upstream commit e56ad45e991128bf4db160b75a1d9f647a341d8f ]

Source --> DP2.1 MST hub --> DP1.4/2.1 monitor

When change from DP1.4 to DP2.1 from monitor manual, modes higher than
4k120 are all cutoff by mode validation. Switch back to DP1.4 gets all
the modes up to 4k240 available to be enabled by dsc passthrough.

[why]
Compared to DP1.4 link from hub to monitor, DP2.1 link has larger
full_pbn value that causes overflow in the process of doing conversion
from pbn to kbps.

[how]
Change the data type accordingly to fit into the data limit during
conversion calculation.

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Reviewed-by: Wayne Lin <wayne.lin@amd.com>
Signed-off-by: Fangzhi Zuo <Jerry.Zuo@amd.com>
Signed-off-by: Rodrigo Siqueira <rodrigo.siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c
index 32b025c92c63c..6c6a908615052 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c
@@ -1684,16 +1684,16 @@ int pre_validate_dsc(struct drm_atomic_state *state,
 	return ret;
 }
 
-static unsigned int kbps_from_pbn(unsigned int pbn)
+static uint32_t kbps_from_pbn(unsigned int pbn)
 {
-	unsigned int kbps = pbn;
+	uint64_t kbps = (uint64_t)pbn;
 
 	kbps *= (1000000 / PEAK_FACTOR_X1000);
 	kbps *= 8;
 	kbps *= 54;
 	kbps /= 64;
 
-	return kbps;
+	return (uint32_t)kbps;
 }
 
 static bool is_dsc_common_config_possible(struct dc_stream_state *stream,
-- 
2.39.5

[PATCH AUTOSEL 6.12 21/31] drm/amdgpu: Don't enable sdma 4.4.5 CTXEMPTY interrupt

[Sasha Levin]

From: Philip Yang <Philip.Yang@amd.com>

[ Upstream commit b4b7271e5ca95b581f2fcc4ae852c4079215e92d ]

The sdma context empty interrupt is dropped in amdgpu_irq_dispatch
as unregistered interrupt src_id 243, this interrupt accounts to 1/3 of
total interrupts and causes IH primary ring overflow when running
stressful benchmark application. Disable this interrupt has no side
effect found.

Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c
index c77889040760a..4dd86c682ee6a 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c
@@ -953,10 +953,12 @@ static int sdma_v4_4_2_inst_start(struct amdgpu_device *adev,
 		/* set utc l1 enable flag always to 1 */
 		temp = RREG32_SDMA(i, regSDMA_CNTL);
 		temp = REG_SET_FIELD(temp, SDMA_CNTL, UTC_L1_ENABLE, 1);
-		/* enable context empty interrupt during initialization */
-		temp = REG_SET_FIELD(temp, SDMA_CNTL, CTXEMPTY_INT_ENABLE, 1);
-		WREG32_SDMA(i, regSDMA_CNTL, temp);
 
+		if (amdgpu_ip_version(adev, SDMA0_HWIP, 0) < IP_VERSION(4, 4, 5)) {
+			/* enable context empty interrupt during initialization */
+			temp = REG_SET_FIELD(temp, SDMA_CNTL, CTXEMPTY_INT_ENABLE, 1);
+			WREG32_SDMA(i, regSDMA_CNTL, temp);
+		}
 		if (!amdgpu_sriov_vf(adev)) {
 			if (adev->firmware.load_type != AMDGPU_FW_LOAD_PSP) {
 				/* unhalt engine */
-- 
2.39.5

[PATCH AUTOSEL 6.12 22/31] drm/amdkfd: Queue interrupt work to different CPU

[Sasha Levin]

From: Philip Yang <Philip.Yang@amd.com>

[ Upstream commit 34db5a32617d102e8042151bb87590e43c97132e ]

For CPX mode, each KFD node has interrupt worker to process ih_fifo to
send events to user space. Currently all interrupt workers of same adev
queue to same CPU, all workers execution are actually serialized and
this cause KFD ih_fifo overflow when CPU usage is high.

Use per-GPU unbounded highpri queue with number of workers equals to
number of partitions, let queue_work select the next CPU round robin
among the local CPUs of same NUMA.

Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 drivers/gpu/drm/amd/amdkfd/kfd_device.c    | 25 ++++++++--------------
 drivers/gpu/drm/amd/amdkfd/kfd_interrupt.c | 25 ++++++++--------------
 drivers/gpu/drm/amd/amdkfd/kfd_priv.h      |  3 ++-
 3 files changed, 20 insertions(+), 33 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device.c b/drivers/gpu/drm/amd/amdkfd/kfd_device.c
index b05be24531e18..d350c7ce35b3d 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device.c
@@ -637,6 +637,14 @@ static void kfd_cleanup_nodes(struct kfd_dev *kfd, unsigned int num_nodes)
 	struct kfd_node *knode;
 	unsigned int i;
 
+	/*
+	 * flush_work ensures that there are no outstanding
+	 * work-queue items that will access interrupt_ring. New work items
+	 * can't be created because we stopped interrupt handling above.
+	 */
+	flush_workqueue(kfd->ih_wq);
+	destroy_workqueue(kfd->ih_wq);
+
 	for (i = 0; i < num_nodes; i++) {
 		knode = kfd->nodes[i];
 		device_queue_manager_uninit(knode->dqm);
@@ -1058,21 +1066,6 @@ static int kfd_resume(struct kfd_node *node)
 	return err;
 }
 
-static inline void kfd_queue_work(struct workqueue_struct *wq,
-				  struct work_struct *work)
-{
-	int cpu, new_cpu;
-
-	cpu = new_cpu = smp_processor_id();
-	do {
-		new_cpu = cpumask_next(new_cpu, cpu_online_mask) % nr_cpu_ids;
-		if (cpu_to_node(new_cpu) == numa_node_id())
-			break;
-	} while (cpu != new_cpu);
-
-	queue_work_on(new_cpu, wq, work);
-}
-
 /* This is called directly from KGD at ISR. */
 void kgd2kfd_interrupt(struct kfd_dev *kfd, const void *ih_ring_entry)
 {
@@ -1098,7 +1091,7 @@ void kgd2kfd_interrupt(struct kfd_dev *kfd, const void *ih_ring_entry)
 			    	patched_ihre, &is_patched)
 		    && enqueue_ih_ring_entry(node,
 			    	is_patched ? patched_ihre : ih_ring_entry)) {
-			kfd_queue_work(node->ih_wq, &node->interrupt_work);
+			queue_work(node->kfd->ih_wq, &node->interrupt_work);
 			spin_unlock_irqrestore(&node->interrupt_lock, flags);
 			return;
 		}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_interrupt.c b/drivers/gpu/drm/amd/amdkfd/kfd_interrupt.c
index 9b6b6e8825934..15b4b70cf1997 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_interrupt.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_interrupt.c
@@ -62,11 +62,14 @@ int kfd_interrupt_init(struct kfd_node *node)
 		return r;
 	}
 
-	node->ih_wq = alloc_workqueue("KFD IH", WQ_HIGHPRI, 1);
-	if (unlikely(!node->ih_wq)) {
-		kfifo_free(&node->ih_fifo);
-		dev_err(node->adev->dev, "Failed to allocate KFD IH workqueue\n");
-		return -ENOMEM;
+	if (!node->kfd->ih_wq) {
+		node->kfd->ih_wq = alloc_workqueue("KFD IH", WQ_HIGHPRI | WQ_UNBOUND,
+						   node->kfd->num_nodes);
+		if (unlikely(!node->kfd->ih_wq)) {
+			kfifo_free(&node->ih_fifo);
+			dev_err(node->adev->dev, "Failed to allocate KFD IH workqueue\n");
+			return -ENOMEM;
+		}
 	}
 	spin_lock_init(&node->interrupt_lock);
 
@@ -96,16 +99,6 @@ void kfd_interrupt_exit(struct kfd_node *node)
 	spin_lock_irqsave(&node->interrupt_lock, flags);
 	node->interrupts_active = false;
 	spin_unlock_irqrestore(&node->interrupt_lock, flags);
-
-	/*
-	 * flush_work ensures that there are no outstanding
-	 * work-queue items that will access interrupt_ring. New work items
-	 * can't be created because we stopped interrupt handling above.
-	 */
-	flush_workqueue(node->ih_wq);
-
-	destroy_workqueue(node->ih_wq);
-
 	kfifo_free(&node->ih_fifo);
 }
 
@@ -162,7 +155,7 @@ static void interrupt_wq(struct work_struct *work)
 			/* If we spent more than a second processing signals,
 			 * reschedule the worker to avoid soft-lockup warnings
 			 */
-			queue_work(dev->ih_wq, &dev->interrupt_work);
+			queue_work(dev->kfd->ih_wq, &dev->interrupt_work);
 			break;
 		}
 	}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
index 26e48fdc87289..75523f30cd38b 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
@@ -273,7 +273,6 @@ struct kfd_node {
 
 	/* Interrupts */
 	struct kfifo ih_fifo;
-	struct workqueue_struct *ih_wq;
 	struct work_struct interrupt_work;
 	spinlock_t interrupt_lock;
 
@@ -366,6 +365,8 @@ struct kfd_dev {
 	struct kfd_node *nodes[MAX_KFD_NODES];
 	unsigned int num_nodes;
 
+	struct workqueue_struct *ih_wq;
+
 	/* Kernel doorbells for KFD device */
 	struct amdgpu_bo *doorbells;
 
-- 
2.39.5

[PATCH AUTOSEL 6.12 29/31] drm/amd/display: Increase sanitizer frame larger than limit when compile testing with clang

[Sasha Levin]

From: Nathan Chancellor <nathan@kernel.org>

[ Upstream commit e4479aecf6581af81bc0908575447878d2a07e01 ]

Commit 24909d9ec7c3 ("drm/amd/display: Overwriting dualDPP UBF values
before usage") added a new warning in dml2/display_mode_core.c when
building allmodconfig with clang:

  drivers/gpu/drm/amd/amdgpu/../display/dc/dml2/display_mode_core.c:6268:13: error: stack frame size (3128) exceeds limit (3072) in 'dml_prefetch_check' [-Werror,-Wframe-larger-than]
   6268 | static void dml_prefetch_check(struct display_mode_lib_st *mode_lib)
        |             ^

Commit be4e3509314a ("drm/amd/display: DML21 Reintegration For Various
Fixes") introduced one in dml2_core/dml2_core_dcn4_calcs.c with the same
configuration:

  drivers/gpu/drm/amd/amdgpu/../display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c:7236:13: error: stack frame size (3256) exceeds limit (3072) in 'dml_core_mode_support' [-Werror,-Wframe-larger-than]
   7236 | static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out_params)
        |             ^

In the case of the first warning, the stack usage was already at the
limit at the parent change, so the offending change was rather
innocuous. In the case of the second warning, there was a rather
dramatic increase in stack usage compared to the parent:

  drivers/gpu/drm/amd/amdgpu/../display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c:7032:13: error: stack frame size (2696) exceeds limit (2048) in 'dml_core_mode_support' [-Werror,-Wframe-larger-than]
   7032 | static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out_params)
        |             ^

This is an unfortunate interaction between an issue with stack slot
reuse in LLVM that gets exacerbated by sanitization (which gets enabled
with all{mod,yes}config) and function calls using a much higher number
of parameters than is typical in the kernel, necessitating passing most
of these values on the stack.

While it is possible that there should be source code changes to address
these warnings, this code is difficult to modify for various reasons, as
has been noted in other changes that have occurred for similar reasons,
such as commit 6740ec97bcdb ("drm/amd/display: Increase frame warning
limit with KASAN or KCSAN in dml2").

Increase the frame larger than limit when compile testing with clang and
the sanitizers enabled to avoid this breakage in all{mod,yes}config, as
they are commonly used and valuable testing targets. While it is not the
best to hide this issue, it is not really relevant when compile testing,
as the sanitizers are commonly stressful on optimizations and they are
only truly useful at runtime, which COMPILE_TEST states will not occur
with the current build.

Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202412121748.chuX4sap-lkp@intel.com/
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 drivers/gpu/drm/amd/display/dc/dml2/Makefile | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/drivers/gpu/drm/amd/display/dc/dml2/Makefile b/drivers/gpu/drm/amd/display/dc/dml2/Makefile
index c4378e620cbf9..986a69c5bd4bc 100644
--- a/drivers/gpu/drm/amd/display/dc/dml2/Makefile
+++ b/drivers/gpu/drm/amd/display/dc/dml2/Makefile
@@ -29,7 +29,11 @@ dml2_rcflags := $(CC_FLAGS_NO_FPU)
 
 ifneq ($(CONFIG_FRAME_WARN),0)
 ifeq ($(filter y,$(CONFIG_KASAN)$(CONFIG_KCSAN)),y)
+ifeq ($(CONFIG_CC_IS_CLANG)$(CONFIG_COMPILE_TEST),yy)
+frame_warn_flag := -Wframe-larger-than=4096
+else
 frame_warn_flag := -Wframe-larger-than=3072
+endif
 else
 frame_warn_flag := -Wframe-larger-than=2048
 endif
-- 
2.39.5

[PATCH AUTOSEL 6.12 30/31] drm/amd/display: Limit Scaling Ratio on DCN3.01

[Sasha Levin]

From: Gabe Teeger <Gabe.Teeger@amd.com>

[ Upstream commit abc0ad6d08440761b199988c329ad7ac83f41c9b ]

[why]
Underflow and flickering was occuring due to high scaling ratios
when resizing videos.

[how]
Limit the scaling ratios by increasing the max scaling factor

Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Gabe Teeger <Gabe.Teeger@amd.com>
Signed-off-by: Tom Chung <chiahsuan.chung@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 .../drm/amd/display/dc/resource/dcn301/dcn301_resource.c  | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn301/dcn301_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn301/dcn301_resource.c
index 7d04739c3ba14..4bbbe07ecde7d 100644
--- a/drivers/gpu/drm/amd/display/dc/resource/dcn301/dcn301_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/resource/dcn301/dcn301_resource.c
@@ -671,9 +671,9 @@ static const struct dc_plane_cap plane_cap = {
 
 	/* 6:1 downscaling ratio: 1000/6 = 166.666 */
 	.max_downscale_factor = {
-			.argb8888 = 167,
-			.nv12 = 167,
-			.fp16 = 167 
+			.argb8888 = 358,
+			.nv12 = 358,
+			.fp16 = 358
 	},
 	64,
 	64
@@ -694,7 +694,7 @@ static const struct dc_debug_options debug_defaults_drv = {
 	.disable_dcc = DCC_ENABLE,
 	.vsr_support = true,
 	.performance_trace = false,
-	.max_downscale_src_width = 7680,/*upto 8K*/
+	.max_downscale_src_width = 4096,/*upto true 4k*/
 	.scl_reset_length10 = true,
 	.sanity_checks = false,
 	.underflow_assert_delay_us = 0xFFFFFFFF,
-- 
2.39.5