Merge tag 'amd-drm-fixes-6.11-2024-07-27' of...

Merge tag 'amd-drm-fixes-6.11-2024-07-27' of https://gitlab.freedesktop.org/agd5f/linux into drm-fixes

amd-drm-fixes-6.11-2024-07-27:

amdgpu:
- SMU 14.x update
- Fix contiguous VRAM handling for IB parsing
- GFX 12 fix
- Regression fix for old APUs
Signed-off-by: Dave Airlie <airlied@redhat.com>

From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240728025407.2115881-1-alexander.deucher@amd.com

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

@@ -1778,7 +1778,7 @@ int amdgpu_cs_find_mapping(struct amdgpu_cs_parser *parser,
 	struct ttm_operation_ctx ctx = { false, false };
 	struct amdgpu_vm *vm = &fpriv->vm;
 	struct amdgpu_bo_va_mapping *mapping;
-	int r;
+	int i, r;
 
 	addr /= AMDGPU_GPU_PAGE_SIZE;
 
@@ -1793,13 +1793,13 @@ int amdgpu_cs_find_mapping(struct amdgpu_cs_parser *parser,
 	if (dma_resv_locking_ctx((*bo)->tbo.base.resv) != &parser->exec.ticket)
 		return -EINVAL;
 
-	if (!((*bo)->flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS)) {
-		(*bo)->flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
-		amdgpu_bo_placement_from_domain(*bo, (*bo)->allowed_domains);
-		r = ttm_bo_validate(&(*bo)->tbo, &(*bo)->placement, &ctx);
-		if (r)
-			return r;
-	}
+	(*bo)->flags |= AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS;
+	amdgpu_bo_placement_from_domain(*bo, (*bo)->allowed_domains);
+	for (i = 0; i < (*bo)->placement.num_placement; i++)
+		(*bo)->placements[i].flags |= TTM_PL_FLAG_CONTIGUOUS;
+	r = ttm_bo_validate(&(*bo)->tbo, &(*bo)->placement, &ctx);
+	if (r)
+		return r;
 
 	return amdgpu_ttm_alloc_gart(&(*bo)->tbo);
 }

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

@@ -103,7 +103,7 @@ static int amdgpu_mes_event_log_init(struct amdgpu_device *adev)
 	if (!amdgpu_mes_log_enable)
 		return 0;
 
-	r = amdgpu_bo_create_kernel(adev, AMDGPU_MES_LOG_BUFFER_SIZE, PAGE_SIZE,
+	r = amdgpu_bo_create_kernel(adev, adev->mes.event_log_size, PAGE_SIZE,
 				    AMDGPU_GEM_DOMAIN_GTT,
 				    &adev->mes.event_log_gpu_obj,
 				    &adev->mes.event_log_gpu_addr,
@@ -113,7 +113,7 @@ static int amdgpu_mes_event_log_init(struct amdgpu_device *adev)
 		return r;
 	}
 
-	memset(adev->mes.event_log_cpu_addr, 0, PAGE_SIZE);
+	memset(adev->mes.event_log_cpu_addr, 0, adev->mes.event_log_size);
 
 	return  0;
 
@@ -1573,7 +1573,7 @@ static int amdgpu_debugfs_mes_event_log_show(struct seq_file *m, void *unused)
 	uint32_t *mem = (uint32_t *)(adev->mes.event_log_cpu_addr);
 
 	seq_hex_dump(m, "", DUMP_PREFIX_OFFSET, 32, 4,
-		     mem, AMDGPU_MES_LOG_BUFFER_SIZE, false);
+		     mem, adev->mes.event_log_size, false);
 
 	return 0;
 }

drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h

@@ -52,7 +52,6 @@ enum amdgpu_mes_priority_level {
 
 #define AMDGPU_MES_PROC_CTX_SIZE 0x1000 /* one page area */
 #define AMDGPU_MES_GANG_CTX_SIZE 0x1000 /* one page area */
-#define AMDGPU_MES_LOG_BUFFER_SIZE 0x4000 /* Maximu log buffer size for MES */
 
 struct amdgpu_mes_funcs;
 
@@ -135,8 +134,9 @@ struct amdgpu_mes {
 	unsigned long			*doorbell_bitmap;
 
 	/* MES event log buffer */
-	struct amdgpu_bo		*event_log_gpu_obj;
-	uint64_t                        event_log_gpu_addr;
+	uint32_t			event_log_size;
+	struct amdgpu_bo	*event_log_gpu_obj;
+	uint64_t			event_log_gpu_addr;
 	void				*event_log_cpu_addr;
 
 	/* ip specific functions */

drivers/gpu/drm/amd/amdgpu/mes_v11_0.c

@@ -1163,6 +1163,8 @@ static int mes_v11_0_sw_init(void *handle)
 	adev->mes.kiq_hw_init = &mes_v11_0_kiq_hw_init;
 	adev->mes.kiq_hw_fini = &mes_v11_0_kiq_hw_fini;
 
+	adev->mes.event_log_size = AMDGPU_MES_LOG_BUFFER_SIZE;
+
 	r = amdgpu_mes_init(adev);
 	if (r)
 		return r;

drivers/gpu/drm/amd/amdgpu/mes_v12_0.c

@@ -551,8 +551,10 @@ static int mes_v12_0_set_hw_resources(struct amdgpu_mes *mes)
 	mes_set_hw_res_pkt.oversubscription_timer = 50;
 	mes_set_hw_res_pkt.unmapped_doorbell_handling = 1;
 
-	mes_set_hw_res_pkt.enable_mes_event_int_logging = 0;
-	mes_set_hw_res_pkt.event_intr_history_gpu_mc_ptr = mes->event_log_gpu_addr;
+	if (amdgpu_mes_log_enable) {
+		mes_set_hw_res_pkt.enable_mes_event_int_logging = 1;
+		mes_set_hw_res_pkt.event_intr_history_gpu_mc_ptr = mes->event_log_gpu_addr;
+	}
 
 	return mes_v12_0_submit_pkt_and_poll_completion(mes,
 			&mes_set_hw_res_pkt, sizeof(mes_set_hw_res_pkt),
@@ -1237,6 +1239,8 @@ static int mes_v12_0_sw_init(void *handle)
 	adev->mes.kiq_hw_init = &mes_v12_0_kiq_hw_init;
 	adev->mes.kiq_hw_fini = &mes_v12_0_kiq_hw_fini;
 
+	adev->mes.event_log_size = AMDGPU_MES_LOG_BUFFER_SIZE;
+
 	r = amdgpu_mes_init(adev);
 	if (r)
 		return r;

drivers/gpu/drm/amd/include/mes_v11_api_def.h

@@ -28,6 +28,9 @@
 
 #define MES_API_VERSION 1
 
+/* Maximum log buffer size for MES. Needs to be updated if MES expands MES_EVT_INTR_HIST_LOG */
+#define  AMDGPU_MES_LOG_BUFFER_SIZE  0x4000
+
 /* Driver submits one API(cmd) as a single Frame and this command size is same
  * for all API to ease the debugging and parsing of ring buffer.
  */

drivers/gpu/drm/amd/include/mes_v12_api_def.h

@@ -28,6 +28,9 @@
 
 #define MES_API_VERSION 0x14
 
+/* Maximum log buffer size for MES. Needs to be updated if MES expands MES_EVT_INTR_HIST_LOG_12 */
+#define  AMDGPU_MES_LOG_BUFFER_SIZE  0xC000
+
 /* Driver submits one API(cmd) as a single Frame and this command size is same for all API
  * to ease the debugging and parsing of ring buffer.
  */

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

@@ -618,7 +618,8 @@ int amdgpu_pm_load_smu_firmware(struct amdgpu_device *adev, uint32_t *smu_versio
 	const struct amd_pm_funcs *pp_funcs = adev->powerplay.pp_funcs;
 	int r = 0;
 
-	if (!pp_funcs || !pp_funcs->load_firmware || adev->flags & AMD_IS_APU)
+	if (!pp_funcs || !pp_funcs->load_firmware ||
+	    (is_support_sw_smu(adev) && (adev->flags & AMD_IS_APU)))
 		return 0;
 
 	mutex_lock(&adev->pm.mutex);

drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c

@@ -66,6 +66,7 @@
 
 #define MP0_MP1_DATA_REGION_SIZE_COMBOPPTABLE	0x4000
 #define DEBUGSMC_MSG_Mode1Reset        2
+#define LINK_SPEED_MAX					3
 
 static struct cmn2asic_msg_mapping smu_v14_0_2_message_map[SMU_MSG_MAX_COUNT] = {
 	MSG_MAP(TestMessage,			PPSMC_MSG_TestMessage,                 1),
@@ -221,7 +222,6 @@ static struct cmn2asic_mapping smu_v14_0_2_workload_map[PP_SMC_POWER_PROFILE_COU
 	WORKLOAD_MAP(PP_SMC_POWER_PROFILE_WINDOW3D,		WORKLOAD_PPLIB_WINDOW_3D_BIT),
 };
 
-#if 0
 static const uint8_t smu_v14_0_2_throttler_map[] = {
 	[THROTTLER_PPT0_BIT]		= (SMU_THROTTLER_PPT0_BIT),
 	[THROTTLER_PPT1_BIT]		= (SMU_THROTTLER_PPT1_BIT),
@@ -241,7 +241,6 @@ static const uint8_t smu_v14_0_2_throttler_map[] = {
 	[THROTTLER_GFX_APCC_PLUS_BIT]	= (SMU_THROTTLER_APCC_BIT),
 	[THROTTLER_FIT_BIT]		= (SMU_THROTTLER_FIT_BIT),
 };
-#endif
 
 static int
 smu_v14_0_2_get_allowed_feature_mask(struct smu_context *smu,
@@ -1869,6 +1868,88 @@ static ssize_t smu_v14_0_2_get_ecc_info(struct smu_context *smu,
 	return ret;
 }
 
+static ssize_t smu_v14_0_2_get_gpu_metrics(struct smu_context *smu,
+					   void **table)
+{
+	struct smu_table_context *smu_table = &smu->smu_table;
+	struct gpu_metrics_v1_3 *gpu_metrics =
+		(struct gpu_metrics_v1_3 *)smu_table->gpu_metrics_table;
+	SmuMetricsExternal_t metrics_ext;
+	SmuMetrics_t *metrics = &metrics_ext.SmuMetrics;
+	int ret = 0;
+
+	ret = smu_cmn_get_metrics_table(smu,
+					&metrics_ext,
+					true);
+	if (ret)
+		return ret;
+
+	smu_cmn_init_soft_gpu_metrics(gpu_metrics, 1, 3);
+
+	gpu_metrics->temperature_edge = metrics->AvgTemperature[TEMP_EDGE];
+	gpu_metrics->temperature_hotspot = metrics->AvgTemperature[TEMP_HOTSPOT];
+	gpu_metrics->temperature_mem = metrics->AvgTemperature[TEMP_MEM];
+	gpu_metrics->temperature_vrgfx = metrics->AvgTemperature[TEMP_VR_GFX];
+	gpu_metrics->temperature_vrsoc = metrics->AvgTemperature[TEMP_VR_SOC];
+	gpu_metrics->temperature_vrmem = max(metrics->AvgTemperature[TEMP_VR_MEM0],
+					     metrics->AvgTemperature[TEMP_VR_MEM1]);
+
+	gpu_metrics->average_gfx_activity = metrics->AverageGfxActivity;
+	gpu_metrics->average_umc_activity = metrics->AverageUclkActivity;
+	gpu_metrics->average_mm_activity = max(metrics->Vcn0ActivityPercentage,
+					       metrics->Vcn1ActivityPercentage);
+
+	gpu_metrics->average_socket_power = metrics->AverageSocketPower;
+	gpu_metrics->energy_accumulator = metrics->EnergyAccumulator;
+
+	if (metrics->AverageGfxActivity <= SMU_14_0_2_BUSY_THRESHOLD)
+		gpu_metrics->average_gfxclk_frequency = metrics->AverageGfxclkFrequencyPostDs;
+	else
+		gpu_metrics->average_gfxclk_frequency = metrics->AverageGfxclkFrequencyPreDs;
+
+	if (metrics->AverageUclkActivity <= SMU_14_0_2_BUSY_THRESHOLD)
+		gpu_metrics->average_uclk_frequency = metrics->AverageMemclkFrequencyPostDs;
+	else
+		gpu_metrics->average_uclk_frequency = metrics->AverageMemclkFrequencyPreDs;
+
+	gpu_metrics->average_vclk0_frequency = metrics->AverageVclk0Frequency;
+	gpu_metrics->average_dclk0_frequency = metrics->AverageDclk0Frequency;
+	gpu_metrics->average_vclk1_frequency = metrics->AverageVclk1Frequency;
+	gpu_metrics->average_dclk1_frequency = metrics->AverageDclk1Frequency;
+
+	gpu_metrics->current_gfxclk = gpu_metrics->average_gfxclk_frequency;
+	gpu_metrics->current_socclk = metrics->CurrClock[PPCLK_SOCCLK];
+	gpu_metrics->current_uclk = metrics->CurrClock[PPCLK_UCLK];
+	gpu_metrics->current_vclk0 = metrics->CurrClock[PPCLK_VCLK_0];
+	gpu_metrics->current_dclk0 = metrics->CurrClock[PPCLK_DCLK_0];
+	gpu_metrics->current_vclk1 = metrics->CurrClock[PPCLK_VCLK_0];
+	gpu_metrics->current_dclk1 = metrics->CurrClock[PPCLK_DCLK_0];
+
+	gpu_metrics->throttle_status =
+			smu_v14_0_2_get_throttler_status(metrics);
+	gpu_metrics->indep_throttle_status =
+			smu_cmn_get_indep_throttler_status(gpu_metrics->throttle_status,
+							   smu_v14_0_2_throttler_map);
+
+	gpu_metrics->current_fan_speed = metrics->AvgFanRpm;
+
+	gpu_metrics->pcie_link_width = metrics->PcieWidth;
+	if ((metrics->PcieRate - 1) > LINK_SPEED_MAX)
+		gpu_metrics->pcie_link_speed = pcie_gen_to_speed(1);
+	else
+		gpu_metrics->pcie_link_speed = pcie_gen_to_speed(metrics->PcieRate);
+
+	gpu_metrics->system_clock_counter = ktime_get_boottime_ns();
+
+	gpu_metrics->voltage_gfx = metrics->AvgVoltage[SVI_PLANE_VDD_GFX];
+	gpu_metrics->voltage_soc = metrics->AvgVoltage[SVI_PLANE_VDD_SOC];
+	gpu_metrics->voltage_mem = metrics->AvgVoltage[SVI_PLANE_VDDIO_MEM];
+
+	*table = (void *)gpu_metrics;
+
+	return sizeof(struct gpu_metrics_v1_3);
+}
+
 static const struct pptable_funcs smu_v14_0_2_ppt_funcs = {
 	.get_allowed_feature_mask = smu_v14_0_2_get_allowed_feature_mask,
 	.set_default_dpm_table = smu_v14_0_2_set_default_dpm_table,
@@ -1905,6 +1986,7 @@ static const struct pptable_funcs smu_v14_0_2_ppt_funcs = {
 	.enable_thermal_alert = smu_v14_0_enable_thermal_alert,
 	.disable_thermal_alert = smu_v14_0_disable_thermal_alert,
 	.notify_memory_pool_location = smu_v14_0_notify_memory_pool_location,
+	.get_gpu_metrics = smu_v14_0_2_get_gpu_metrics,
 	.set_soft_freq_limited_range = smu_v14_0_set_soft_freq_limited_range,
 	.init_pptable_microcode = smu_v14_0_init_pptable_microcode,
 	.populate_umd_state_clk = smu_v14_0_2_populate_umd_state_clk,