Merge tag 'amd-drm-fixes-6.7-2024-01-04' of...

Merge tag 'amd-drm-fixes-6.7-2024-01-04' of https://gitlab.freedesktop.org/agd5f/linux into drm-fixes

amdgpu:
- DP MST fix
- SMU 13.0.6 fixes
- Fix displays on macbooks using vega12
- Fix VSC and colorimetry on DP/eDP
Signed-off-by: Dave Airlie <airlied@redhat.com>

From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240104152139.4931-1-alexander.deucher@amd.com

drivers/gpu/drm/amd/amdgpu/amdgpu_device.c

@@ -2188,15 +2188,8 @@ static int amdgpu_device_parse_gpu_info_fw(struct amdgpu_device *adev)
 
 	adev->firmware.gpu_info_fw = NULL;
 
-	if (adev->mman.discovery_bin) {
-		/*
-		 * FIXME: The bounding box is still needed by Navi12, so
-		 * temporarily read it from gpu_info firmware. Should be dropped
-		 * when DAL no longer needs it.
-		 */
-		if (adev->asic_type != CHIP_NAVI12)
-			return 0;
-	}
+	if (adev->mman.discovery_bin)
+		return 0;
 
 	switch (adev->asic_type) {
 	default:

drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c

@@ -6170,8 +6170,9 @@ create_stream_for_sink(struct amdgpu_dm_connector *aconnector,
 
 	if (stream->signal == SIGNAL_TYPE_HDMI_TYPE_A)
 		mod_build_hf_vsif_infopacket(stream, &stream->vsp_infopacket);
-
-	if (stream->link->psr_settings.psr_feature_enabled || stream->link->replay_settings.replay_feature_enabled) {
+	else if (stream->signal == SIGNAL_TYPE_DISPLAY_PORT ||
+			 stream->signal == SIGNAL_TYPE_DISPLAY_PORT_MST ||
+			 stream->signal == SIGNAL_TYPE_EDP) {
 		//
 		// should decide stream support vsc sdp colorimetry capability
 		// before building vsc info packet
@@ -6187,8 +6188,9 @@ create_stream_for_sink(struct amdgpu_dm_connector *aconnector,
 		if (stream->out_transfer_func->tf == TRANSFER_FUNCTION_GAMMA22)
 			tf = TRANSFER_FUNC_GAMMA_22;
 		mod_build_vsc_infopacket(stream, &stream->vsc_infopacket, stream->output_color_space, tf);
-		aconnector->psr_skip_count = AMDGPU_DM_PSR_ENTRY_DELAY;
 
+		if (stream->link->psr_settings.psr_feature_enabled)
+			aconnector->psr_skip_count = AMDGPU_DM_PSR_ENTRY_DELAY;
 	}
 finish:
 	dc_sink_release(sink);
@@ -6914,8 +6916,7 @@ static int dm_encoder_helper_atomic_check(struct drm_encoder *encoder,
 	if (IS_ERR(mst_state))
 		return PTR_ERR(mst_state);
 
-	if (!mst_state->pbn_div)
-		mst_state->pbn_div = dm_mst_get_pbn_divider(aconnector->mst_root->dc_link);
+	mst_state->pbn_div = dm_mst_get_pbn_divider(aconnector->mst_root->dc_link);
 
 	if (!state->duplicated) {
 		int max_bpc = conn_state->max_requested_bpc;

drivers/gpu/drm/amd/display/dc/dml/dcn20/dcn20_fpu.c

@@ -440,7 +440,115 @@ struct _vcs_dpi_soc_bounding_box_st dcn2_0_nv14_soc = {
 	.use_urgent_burst_bw = 0
 };
 
-struct _vcs_dpi_soc_bounding_box_st dcn2_0_nv12_soc = { 0 };
+struct _vcs_dpi_soc_bounding_box_st dcn2_0_nv12_soc = {
+	.clock_limits = {
+		{
+			.state = 0,
+			.dcfclk_mhz = 560.0,
+			.fabricclk_mhz = 560.0,
+			.dispclk_mhz = 513.0,
+			.dppclk_mhz = 513.0,
+			.phyclk_mhz = 540.0,
+			.socclk_mhz = 560.0,
+			.dscclk_mhz = 171.0,
+			.dram_speed_mts = 1069.0,
+		},
+		{
+			.state = 1,
+			.dcfclk_mhz = 694.0,
+			.fabricclk_mhz = 694.0,
+			.dispclk_mhz = 642.0,
+			.dppclk_mhz = 642.0,
+			.phyclk_mhz = 600.0,
+			.socclk_mhz = 694.0,
+			.dscclk_mhz = 214.0,
+			.dram_speed_mts = 1324.0,
+		},
+		{
+			.state = 2,
+			.dcfclk_mhz = 875.0,
+			.fabricclk_mhz = 875.0,
+			.dispclk_mhz = 734.0,
+			.dppclk_mhz = 734.0,
+			.phyclk_mhz = 810.0,
+			.socclk_mhz = 875.0,
+			.dscclk_mhz = 245.0,
+			.dram_speed_mts = 1670.0,
+		},
+		{
+			.state = 3,
+			.dcfclk_mhz = 1000.0,
+			.fabricclk_mhz = 1000.0,
+			.dispclk_mhz = 1100.0,
+			.dppclk_mhz = 1100.0,
+			.phyclk_mhz = 810.0,
+			.socclk_mhz = 1000.0,
+			.dscclk_mhz = 367.0,
+			.dram_speed_mts = 2000.0,
+		},
+		{
+			.state = 4,
+			.dcfclk_mhz = 1200.0,
+			.fabricclk_mhz = 1200.0,
+			.dispclk_mhz = 1284.0,
+			.dppclk_mhz = 1284.0,
+			.phyclk_mhz = 810.0,
+			.socclk_mhz = 1200.0,
+			.dscclk_mhz = 428.0,
+			.dram_speed_mts = 2000.0,
+		},
+		{
+			.state = 5,
+			.dcfclk_mhz = 1200.0,
+			.fabricclk_mhz = 1200.0,
+			.dispclk_mhz = 1284.0,
+			.dppclk_mhz = 1284.0,
+			.phyclk_mhz = 810.0,
+			.socclk_mhz = 1200.0,
+			.dscclk_mhz = 428.0,
+			.dram_speed_mts = 2000.0,
+		},
+	},
+
+	.num_states = 5,
+	.sr_exit_time_us = 1.9,
+	.sr_enter_plus_exit_time_us = 4.4,
+	.urgent_latency_us = 3.0,
+	.urgent_latency_pixel_data_only_us = 4.0,
+	.urgent_latency_pixel_mixed_with_vm_data_us = 4.0,
+	.urgent_latency_vm_data_only_us = 4.0,
+	.urgent_out_of_order_return_per_channel_pixel_only_bytes = 4096,
+	.urgent_out_of_order_return_per_channel_pixel_and_vm_bytes = 4096,
+	.urgent_out_of_order_return_per_channel_vm_only_bytes = 4096,
+	.pct_ideal_dram_sdp_bw_after_urgent_pixel_only = 40.0,
+	.pct_ideal_dram_sdp_bw_after_urgent_pixel_and_vm = 40.0,
+	.pct_ideal_dram_sdp_bw_after_urgent_vm_only = 40.0,
+	.max_avg_sdp_bw_use_normal_percent = 40.0,
+	.max_avg_dram_bw_use_normal_percent = 40.0,
+	.writeback_latency_us = 12.0,
+	.ideal_dram_bw_after_urgent_percent = 40.0,
+	.max_request_size_bytes = 256,
+	.dram_channel_width_bytes = 16,
+	.fabric_datapath_to_dcn_data_return_bytes = 64,
+	.dcn_downspread_percent = 0.5,
+	.downspread_percent = 0.5,
+	.dram_page_open_time_ns = 50.0,
+	.dram_rw_turnaround_time_ns = 17.5,
+	.dram_return_buffer_per_channel_bytes = 8192,
+	.round_trip_ping_latency_dcfclk_cycles = 131,
+	.urgent_out_of_order_return_per_channel_bytes = 4096,
+	.channel_interleave_bytes = 256,
+	.num_banks = 8,
+	.num_chans = 16,
+	.vmm_page_size_bytes = 4096,
+	.dram_clock_change_latency_us = 45.0,
+	.writeback_dram_clock_change_latency_us = 23.0,
+	.return_bus_width_bytes = 64,
+	.dispclk_dppclk_vco_speed_mhz = 3850,
+	.xfc_bus_transport_time_us = 20,
+	.xfc_xbuf_latency_tolerance_us = 50,
+	.use_urgent_burst_bw = 0,
+};
 
 struct _vcs_dpi_ip_params_st dcn2_1_ip = {
 	.odm_capable = 1,

drivers/gpu/drm/amd/display/modules/info_packet/info_packet.c

@@ -147,12 +147,15 @@ void mod_build_vsc_infopacket(const struct dc_stream_state *stream,
 	}
 
 	/* VSC packet set to 4 for PSR-SU, or 2 for PSR1 */
-	if (stream->link->psr_settings.psr_version == DC_PSR_VERSION_SU_1)
-		vsc_packet_revision = vsc_packet_rev4;
-	else if (stream->link->replay_settings.config.replay_supported)
+	if (stream->link->psr_settings.psr_feature_enabled) {
+		if (stream->link->psr_settings.psr_version == DC_PSR_VERSION_SU_1)
+			vsc_packet_revision = vsc_packet_rev4;
+		else if (stream->link->psr_settings.psr_version == DC_PSR_VERSION_1)
+			vsc_packet_revision = vsc_packet_rev2;
+	}
+
+	if (stream->link->replay_settings.config.replay_supported)
 		vsc_packet_revision = vsc_packet_rev4;
-	else if (stream->link->psr_settings.psr_version == DC_PSR_VERSION_1)
-		vsc_packet_revision = vsc_packet_rev2;
 
 	/* Update to revision 5 for extended colorimetry support */
 	if (stream->use_vsc_sdp_for_colorimetry)

drivers/gpu/drm/amd/include/kgd_pp_interface.h

@@ -318,6 +318,7 @@ enum pp_xgmi_plpd_mode {
 #define MAX_GFX_CLKS 8
 #define MAX_CLKS 4
 #define NUM_VCN 4
+#define NUM_JPEG_ENG 32
 
 struct seq_file;
 enum amd_pp_clock_type;
@@ -774,6 +775,85 @@ struct gpu_metrics_v1_4 {
 	uint16_t			padding;
 };
 
+struct gpu_metrics_v1_5 {
+	struct metrics_table_header	common_header;
+
+	/* Temperature (Celsius) */
+	uint16_t			temperature_hotspot;
+	uint16_t			temperature_mem;
+	uint16_t			temperature_vrsoc;
+
+	/* Power (Watts) */
+	uint16_t			curr_socket_power;
+
+	/* Utilization (%) */
+	uint16_t			average_gfx_activity;
+	uint16_t			average_umc_activity; // memory controller
+	uint16_t			vcn_activity[NUM_VCN];
+	uint16_t			jpeg_activity[NUM_JPEG_ENG];
+
+	/* Energy (15.259uJ (2^-16) units) */
+	uint64_t			energy_accumulator;
+
+	/* Driver attached timestamp (in ns) */
+	uint64_t			system_clock_counter;
+
+	/* Throttle status */
+	uint32_t			throttle_status;
+
+	/* Clock Lock Status. Each bit corresponds to clock instance */
+	uint32_t			gfxclk_lock_status;
+
+	/* Link width (number of lanes) and speed (in 0.1 GT/s) */
+	uint16_t			pcie_link_width;
+	uint16_t			pcie_link_speed;
+
+	/* XGMI bus width and bitrate (in Gbps) */
+	uint16_t			xgmi_link_width;
+	uint16_t			xgmi_link_speed;
+
+	/* Utilization Accumulated (%) */
+	uint32_t			gfx_activity_acc;
+	uint32_t			mem_activity_acc;
+
+	/*PCIE accumulated bandwidth (GB/sec) */
+	uint64_t			pcie_bandwidth_acc;
+
+	/*PCIE instantaneous bandwidth (GB/sec) */
+	uint64_t			pcie_bandwidth_inst;
+
+	/* PCIE L0 to recovery state transition accumulated count */
+	uint64_t			pcie_l0_to_recov_count_acc;
+
+	/* PCIE replay accumulated count */
+	uint64_t			pcie_replay_count_acc;
+
+	/* PCIE replay rollover accumulated count */
+	uint64_t			pcie_replay_rover_count_acc;
+
+	/* PCIE NAK sent  accumulated count */
+	uint32_t			pcie_nak_sent_count_acc;
+
+	/* PCIE NAK received accumulated count */
+	uint32_t			pcie_nak_rcvd_count_acc;
+
+	/* XGMI accumulated data transfer size(KiloBytes) */
+	uint64_t			xgmi_read_data_acc[NUM_XGMI_LINKS];
+	uint64_t			xgmi_write_data_acc[NUM_XGMI_LINKS];
+
+	/* PMFW attached timestamp (10ns resolution) */
+	uint64_t			firmware_timestamp;
+
+	/* Current clocks (Mhz) */
+	uint16_t			current_gfxclk[MAX_GFX_CLKS];
+	uint16_t			current_socclk[MAX_CLKS];
+	uint16_t			current_vclk0[MAX_CLKS];
+	uint16_t			current_dclk0[MAX_CLKS];
+	uint16_t			current_uclk;
+
+	uint16_t			padding;
+};
+
 /*
  * gpu_metrics_v2_0 is not recommended as it's not naturally aligned.
  * Use gpu_metrics_v2_1 or later instead.

drivers/gpu/drm/amd/pm/amdgpu_pm.c

@@ -2128,7 +2128,9 @@ static int default_attr_update(struct amdgpu_device *adev, struct amdgpu_device_
 		if (amdgpu_dpm_is_overdrive_supported(adev))
 			*states = ATTR_STATE_SUPPORTED;
 	} else if (DEVICE_ATTR_IS(mem_busy_percent)) {
-		if (adev->flags & AMD_IS_APU || gc_ver == IP_VERSION(9, 0, 1))
+		if ((adev->flags & AMD_IS_APU &&
+		     gc_ver != IP_VERSION(9, 4, 3)) ||
+		    gc_ver == IP_VERSION(9, 0, 1))
 			*states = ATTR_STATE_UNSUPPORTED;
 	} else if (DEVICE_ATTR_IS(pcie_bw)) {
 		/* PCIe Perf counters won't work on APU nodes */

drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_6_pmfw.h

@@ -123,7 +123,7 @@ typedef enum {
   VOLTAGE_GUARDBAND_COUNT
 } GFX_GUARDBAND_e;
 
-#define SMU_METRICS_TABLE_VERSION 0x9
+#define SMU_METRICS_TABLE_VERSION 0xB
 
 typedef struct __attribute__((packed, aligned(4))) {
   uint32_t AccumulationCounter;
@@ -219,7 +219,103 @@ typedef struct __attribute__((packed, aligned(4))) {
   uint32_t PCIenReplayARolloverCountAcc;  // The Pcie counter itself is accumulated
   uint32_t PCIeNAKSentCountAcc;           // The Pcie counter itself is accumulated
   uint32_t PCIeNAKReceivedCountAcc;       // The Pcie counter itself is accumulated
-} MetricsTable_t;
+
+  // VCN/JPEG ACTIVITY
+  uint32_t VcnBusy[4];
+  uint32_t JpegBusy[32];
+} MetricsTableX_t;
+
+typedef struct __attribute__((packed, aligned(4))) {
+  uint32_t AccumulationCounter;
+
+  //TEMPERATURE
+  uint32_t MaxSocketTemperature;
+  uint32_t MaxVrTemperature;
+  uint32_t MaxHbmTemperature;
+  uint64_t MaxSocketTemperatureAcc;
+  uint64_t MaxVrTemperatureAcc;
+  uint64_t MaxHbmTemperatureAcc;
+
+  //POWER
+  uint32_t SocketPowerLimit;
+  uint32_t MaxSocketPowerLimit;
+  uint32_t SocketPower;
+
+  //ENERGY
+  uint64_t Timestamp;
+  uint64_t SocketEnergyAcc;
+  uint64_t CcdEnergyAcc;
+  uint64_t XcdEnergyAcc;
+  uint64_t AidEnergyAcc;
+  uint64_t HbmEnergyAcc;
+
+  //FREQUENCY
+  uint32_t CclkFrequencyLimit;
+  uint32_t GfxclkFrequencyLimit;
+  uint32_t FclkFrequency;
+  uint32_t UclkFrequency;
+  uint32_t SocclkFrequency[4];
+  uint32_t VclkFrequency[4];
+  uint32_t DclkFrequency[4];
+  uint32_t LclkFrequency[4];
+  uint64_t GfxclkFrequencyAcc[8];
+  uint64_t CclkFrequencyAcc[96];
+
+  //FREQUENCY RANGE
+  uint32_t MaxCclkFrequency;
+  uint32_t MinCclkFrequency;
+  uint32_t MaxGfxclkFrequency;
+  uint32_t MinGfxclkFrequency;
+  uint32_t FclkFrequencyTable[4];
+  uint32_t UclkFrequencyTable[4];
+  uint32_t SocclkFrequencyTable[4];
+  uint32_t VclkFrequencyTable[4];
+  uint32_t DclkFrequencyTable[4];
+  uint32_t LclkFrequencyTable[4];
+  uint32_t MaxLclkDpmRange;
+  uint32_t MinLclkDpmRange;
+
+  //XGMI
+  uint32_t XgmiWidth;
+  uint32_t XgmiBitrate;
+  uint64_t XgmiReadBandwidthAcc[8];
+  uint64_t XgmiWriteBandwidthAcc[8];
+
+  //ACTIVITY
+  uint32_t SocketC0Residency;
+  uint32_t SocketGfxBusy;
+  uint32_t DramBandwidthUtilization;
+  uint64_t SocketC0ResidencyAcc;
+  uint64_t SocketGfxBusyAcc;
+  uint64_t DramBandwidthAcc;
+  uint32_t MaxDramBandwidth;
+  uint64_t DramBandwidthUtilizationAcc;
+  uint64_t PcieBandwidthAcc[4];
+
+  //THROTTLERS
+  uint32_t ProchotResidencyAcc;
+  uint32_t PptResidencyAcc;
+  uint32_t SocketThmResidencyAcc;
+  uint32_t VrThmResidencyAcc;
+  uint32_t HbmThmResidencyAcc;
+  uint32_t GfxLockXCDMak;
+
+  // New Items at end to maintain driver compatibility
+  uint32_t GfxclkFrequency[8];
+
+  //PSNs
+  uint64_t PublicSerialNumber_AID[4];
+  uint64_t PublicSerialNumber_XCD[8];
+  uint64_t PublicSerialNumber_CCD[12];
+
+  //XGMI Data tranfser size
+  uint64_t XgmiReadDataSizeAcc[8];//in KByte
+  uint64_t XgmiWriteDataSizeAcc[8];//in KByte
+
+  // VCN/JPEG ACTIVITY
+  uint32_t VcnBusy[4];
+  uint32_t JpegBusy[32];
+} MetricsTableA_t;
 
 #define SMU_VF_METRICS_TABLE_VERSION 0x3
 

drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c

@@ -989,6 +989,9 @@ void smu_cmn_init_soft_gpu_metrics(void *table, uint8_t frev, uint8_t crev)
 	case METRICS_VERSION(1, 4):
 		structure_size = sizeof(struct gpu_metrics_v1_4);
 		break;
+	case METRICS_VERSION(1, 5):
+		structure_size = sizeof(struct gpu_metrics_v1_5);
+		break;
 	case METRICS_VERSION(2, 0):
 		structure_size = sizeof(struct gpu_metrics_v2_0);
 		break;