drm/amdgpu: add the fan abnormal detection feature

Update the SW CTF limit from existing register
when there's a fan failure detected via SMU interrupt.

Signed-off-by: lyndonli <[email protected]>
Reviewed-by: Hawking Zhang <[email protected]>
Reviewed-by: Kenneth Feng <[email protected]>
Reviewed-by: Evan Quan <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>

Author: lyndonli

drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h

@@ -168,6 +168,7 @@ struct smu_temperature_range {
 	int mem_crit_max;
 	int mem_emergency_max;
 	int software_shutdown_temp;
+	int software_shutdown_temp_offset;
 };
 
 struct smu_state_validation_block {

drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0.c

@@ -1376,6 +1376,7 @@ static int smu_v13_0_irq_process(struct amdgpu_device *adev,
 	 */
 	uint32_t ctxid = entry->src_data[0];
 	uint32_t data;
+	uint32_t high;
 
 	if (client_id == SOC15_IH_CLIENTID_THM) {
 		switch (src_id) {
@@ -1432,6 +1433,33 @@ static int smu_v13_0_irq_process(struct amdgpu_device *adev,
 					schedule_work(&smu->throttling_logging_work);
 
 				break;
+			case 0x8:
+				high = smu->thermal_range.software_shutdown_temp +
+					smu->thermal_range.software_shutdown_temp_offset;
+				high = min(SMU_THERMAL_MAXIMUM_ALERT_TEMP, high);
+				dev_emerg(adev->dev, "Reduce soft CTF limit to %d (by an offset %d)\n",
+							high,
+							smu->thermal_range.software_shutdown_temp_offset);
+
+				data = RREG32_SOC15(THM, 0, regTHM_THERMAL_INT_CTRL);
+				data = REG_SET_FIELD(data, THM_THERMAL_INT_CTRL,
+							DIG_THERM_INTH,
+							(high & 0xff));
+				data = data & (~THM_THERMAL_INT_CTRL__THERM_TRIGGER_MASK_MASK);
+				WREG32_SOC15(THM, 0, regTHM_THERMAL_INT_CTRL, data);
+				break;
+			case 0x9:
+				high = min(SMU_THERMAL_MAXIMUM_ALERT_TEMP,
+					smu->thermal_range.software_shutdown_temp);
+				dev_emerg(adev->dev, "Recover soft CTF limit to %d\n", high);
+
+				data = RREG32_SOC15(THM, 0, regTHM_THERMAL_INT_CTRL);
+				data = REG_SET_FIELD(data, THM_THERMAL_INT_CTRL,
+							DIG_THERM_INTH,
+							(high & 0xff));
+				data = data & (~THM_THERMAL_INT_CTRL__THERM_TRIGGER_MASK_MASK);
+				WREG32_SOC15(THM, 0, regTHM_THERMAL_INT_CTRL, data);
+				break;
 			}
 		}
 	}

drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c

@@ -1223,6 +1223,7 @@ static int smu_v13_0_7_get_thermal_temperature_range(struct smu_context *smu,
 	range->mem_emergency_max = (pptable->SkuTable.TemperatureLimit[TEMP_MEM] + CTF_OFFSET_MEM)*
 		SMU_TEMPERATURE_UNITS_PER_CENTIGRADES;
 	range->software_shutdown_temp = powerplay_table->software_shutdown_temp;
+	range->software_shutdown_temp_offset = pptable->SkuTable.FanAbnormalTempLimitOffset;
 
 	return 0;
 }