Skip to contentSkip to conversation

drm/amd/powerplay: support hwmon temperature channel labels

PATCH In review Evan Quan 2 participants 13 messages Apr 18, 2019 – Apr 19, 2019 7 messages not in the archive

Subsystems (2)
lore
Share & tools
lore

Evan Quan

author
#0 lore raw

PATCH 1/6 drm/amd/powerplay: support hwmon temperature channel labels

Expose temp[1-3]_label hwmon interfaces. While temp2_label and temp3_label are visible for SOC15 dGPUs only.

Change-Id I7f1e10c52ec21d272027554cdf6da97103e0be58
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c        | 40 +++++++++++++++++++
 .../gpu/drm/amd/include/kgd_pp_interface.h    |  7 ++++
 2 files changed, 47 insertions(+)
drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c +40 -0
index 552127b74f78..c17eb228417e 100644
@@ -120,6 +120,15 @@ static const struct cg_flag_name clocks[] = {
 	{0, NULL},
 };
 
+static const struct hwmon_temp_label {
+	enum PP_HWMON_TEMP channel;
+	const char *label;
+} temp_label[] = {
+	{PP_TEMP_JUNCTION, "junction"},
+	{PP_TEMP_EDGE, "edge"},
+	{PP_TEMP_MEM, "mem"},
+};
+
 void amdgpu_pm_acpi_event_handler(struct amdgpu_device *adev)
 {
 	if (adev->pm.dpm_enabled) {
@@ -1457,6 +1466,20 @@ static ssize_t amdgpu_hwmon_show_temp_thresh(struct device *dev,
 	return snprintf(buf, PAGE_SIZE, "%d\n", temp);
 }
 
+
+static ssize_t amdgpu_hwmon_show_temp_label(struct device *dev,
+					     struct device_attribute *attr,
+					     char *buf)
+{
+	struct amdgpu_device *adev = dev_get_drvdata(dev);
+	int channel = to_sensor_dev_attr(attr)->index;
+
+	if (channel >= PP_TEMP_MAX)
+		return -EINVAL;
+
+	return snprintf(buf, PAGE_SIZE, "%s\n", temp_label[channel].label);
+}
+
 static ssize_t amdgpu_hwmon_get_pwm1_enable(struct device *dev,
 					    struct device_attribute *attr,
 					    char *buf)
@@ -2026,6 +2049,9 @@ static ssize_t amdgpu_hwmon_show_mclk_label(struct device *dev,
  *
  * hwmon interfaces for GPU temperature:
  *
+ * - temp[1-3]_label: temperature channel label
+ *   - temp2_label and temp3_label are supported on SOC15 dGPUs only
+ *
  * - temp1_input: the on die GPU temperature in millidegrees Celsius
  *
  * - temp1_crit: temperature critical max value in millidegrees Celsius
@@ -2081,6 +2107,9 @@ static ssize_t amdgpu_hwmon_show_mclk_label(struct device *dev,
 static SENSOR_DEVICE_ATTR(temp1_input, S_IRUGO, amdgpu_hwmon_show_temp, NULL, 0);
 static SENSOR_DEVICE_ATTR(temp1_crit, S_IRUGO, amdgpu_hwmon_show_temp_thresh, NULL, 0);
 static SENSOR_DEVICE_ATTR(temp1_crit_hyst, S_IRUGO, amdgpu_hwmon_show_temp_thresh, NULL, 1);
+static SENSOR_DEVICE_ATTR(temp1_label, S_IRUGO, amdgpu_hwmon_show_temp_label, NULL, PP_TEMP_JUNCTION);
+static SENSOR_DEVICE_ATTR(temp2_label, S_IRUGO, amdgpu_hwmon_show_temp_label, NULL, PP_TEMP_EDGE);
+static SENSOR_DEVICE_ATTR(temp3_label, S_IRUGO, amdgpu_hwmon_show_temp_label, NULL, PP_TEMP_MEM);
 static SENSOR_DEVICE_ATTR(pwm1, S_IRUGO | S_IWUSR, amdgpu_hwmon_get_pwm1, amdgpu_hwmon_set_pwm1, 0);
 static SENSOR_DEVICE_ATTR(pwm1_enable, S_IRUGO | S_IWUSR, amdgpu_hwmon_get_pwm1_enable, amdgpu_hwmon_set_pwm1_enable, 0);
 static SENSOR_DEVICE_ATTR(pwm1_min, S_IRUGO, amdgpu_hwmon_get_pwm1_min, NULL, 0);
@@ -2107,6 +2136,9 @@ static struct attribute *hwmon_attributes[] = {
 	&sensor_dev_attr_temp1_input.dev_attr.attr,
 	&sensor_dev_attr_temp1_crit.dev_attr.attr,
 	&sensor_dev_attr_temp1_crit_hyst.dev_attr.attr,
+	&sensor_dev_attr_temp1_label.dev_attr.attr,
+	&sensor_dev_attr_temp2_label.dev_attr.attr,
+	&sensor_dev_attr_temp3_label.dev_attr.attr,
 	&sensor_dev_attr_pwm1.dev_attr.attr,
 	&sensor_dev_attr_pwm1_enable.dev_attr.attr,
 	&sensor_dev_attr_pwm1_min.dev_attr.attr,
@@ -2229,6 +2261,14 @@ static umode_t hwmon_attributes_visible(struct kobject *kobj,
 	     attr == &sensor_dev_attr_freq2_label.dev_attr.attr))
 		return 0;
 
+	/* only SOC15 dGPUs support edge and mem temperatures */
+	if (((adev->flags & AMD_IS_APU) ||
+	     adev->asic_type < CHIP_VEGA10) &&
+	    (attr == &sensor_dev_attr_temp2_label.dev_attr.attr ||
+	     attr == &sensor_dev_attr_temp3_label.dev_attr.attr))
+		return 0;
+
+
 	return effective_mode;
 }
 
drivers/gpu/drm/amd/include/kgd_pp_interface.h +7 -0
index 2b579ba9b685..17324c0d503e 100644
@@ -159,6 +159,13 @@ struct pp_states_info {
 	uint32_t states[16];
 };
 
+enum PP_HWMON_TEMP {
+	PP_TEMP_JUNCTION = 0,
+	PP_TEMP_EDGE,
+	PP_TEMP_MEM,
+	PP_TEMP_MAX
+};
+
 #define PP_GROUP_MASK        0xF0000000
 #define PP_GROUP_SHIFT       28
 
Signature · 6 lines
2.21.0

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): [email protected]

author
#2 lore raw

PATCH 2/6 drm/amd/powerplay: support edge/memory critical limit values

These new interfaces(temp2_crit, temp2_crit_hyst, temp3_crit, temp3_crit_hyst) are supported on SOC15 dGPUs only.

Change-Id Ia87e3f6ad816b51d6680eb74c8f755d6c2b0a6ae
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h       |  8 +++
 drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c        | 51 +++++++++++++++++--
 .../drm/amd/powerplay/hwmgr/hardwaremanager.c | 12 ++++-
 .../drm/amd/powerplay/hwmgr/vega10_hwmgr.c    |  6 +++
 .../drm/amd/powerplay/hwmgr/vega12_hwmgr.c    |  7 +++
 .../drm/amd/powerplay/hwmgr/vega20_hwmgr.c    |  7 +++
 .../gpu/drm/amd/powerplay/inc/power_state.h   |  4 ++
 .../gpu/drm/amd/powerplay/inc/pp_thermal.h    |  8 +--
 8 files changed, 95 insertions(+), 8 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h +8 -0
index dca35407879d..e1492438ae7b 100644
@@ -75,6 +75,14 @@ struct amdgpu_dpm_thermal {
 	int                min_temp;
 	/* high temperature threshold */
 	int                max_temp;
+	/* edge low temperature threshold */
+	int                min_edge_temp;
+	/* edge high temperature critical threshold */
+	int                max_edge_crit_temp;
+	/* memory low temperature threshold */
+	int                min_mem_temp;
+	/* memory high temperature critical threshold */
+	int                max_mem_crit_temp;
 	/* was last interrupt low to high or high to low */
 	bool               high_to_low;
 	/* interrupt source */
drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c +48 -3

82 lines not on this page · raw message

drivers/gpu/drm/amd/powerplay/hwmgr/hardwaremanager.c +11 -1
index 70f7f47a2fcf..c8397b20f71a 100644
@@ -225,7 +225,13 @@ int phm_register_irq_handlers(struct pp_hwmgr *hwmgr)
 int phm_start_thermal_controller(struct pp_hwmgr *hwmgr)
 {
 	int ret = 0;
-	struct PP_TemperatureRange range = {TEMP_RANGE_MIN, TEMP_RANGE_MAX};
+	struct PP_TemperatureRange range = {
+		TEMP_RANGE_MIN,
+		TEMP_RANGE_MAX,
+		TEMP_RANGE_MIN,
+		TEMP_RANGE_MAX,
+		TEMP_RANGE_MIN,
+		TEMP_RANGE_MAX};
 	struct amdgpu_device *adev = hwmgr->adev;
 
 	if (hwmgr->hwmgr_func->get_thermal_temperature_range)
@@ -239,6 +245,10 @@ int phm_start_thermal_controller(struct pp_hwmgr *hwmgr)
 
 	adev->pm.dpm.thermal.min_temp = range.min;
 	adev->pm.dpm.thermal.max_temp = range.max;
+	adev->pm.dpm.thermal.min_edge_temp = range.edge_min;
+	adev->pm.dpm.thermal.max_edge_crit_temp = range.edge_crit_max;
+	adev->pm.dpm.thermal.min_mem_temp = range.mem_min;
+	adev->pm.dpm.thermal.max_mem_crit_temp = range.mem_crit_max;
 
 	return ret;
 }
drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c +6 -0
index 384c37875cd0..65aa7e70d7d9 100644
@@ -4854,11 +4854,17 @@ static int vega10_get_thermal_temperature_range(struct pp_hwmgr *hwmgr,
 {
 	struct phm_ppt_v2_information *table_info =
 			(struct phm_ppt_v2_information *)hwmgr->pptable;
+	struct vega10_hwmgr *data = hwmgr->backend;
+	PPTable_t *pp_table = &(data->smc_state_table.pp_table);
 
 	memcpy(thermal_data, &SMU7ThermalWithDelayPolicy[0], sizeof(struct PP_TemperatureRange));
 
 	thermal_data->max = table_info->tdp_table->usSoftwareShutdownTemp *
 		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+	thermal_data->edge_crit_max = pp_table->TedgeLimit *
+		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+	thermal_data->mem_crit_max = pp_table->ThbmLimit *
+		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
 
 	return 0;
 }
drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c +7 -0
index 707cd4b0357f..c0994851e7c7 100644
@@ -2528,11 +2528,18 @@ static int vega12_get_thermal_temperature_range(struct pp_hwmgr *hwmgr,
 {
 	struct phm_ppt_v3_information *pptable_information =
 		(struct phm_ppt_v3_information *)hwmgr->pptable;
+	struct vega12_hwmgr *data =
+			(struct vega12_hwmgr *)(hwmgr->backend);
+	PPTable_t *pp_table = &(data->smc_state_table.pp_table);
 
 	memcpy(thermal_data, &SMU7ThermalWithDelayPolicy[0], sizeof(struct PP_TemperatureRange));
 
 	thermal_data->max = pptable_information->us_software_shutdown_temp *
 		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+	thermal_data->edge_crit_max = pp_table->TedgeLimit *
+		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+	thermal_data->mem_crit_max = pp_table->ThbmLimit *
+		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
 
 	return 0;
 }
drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c +7 -0
index 39a547084e90..5d9aa0f22c86 100644
@@ -3965,11 +3965,18 @@ static int vega20_get_thermal_temperature_range(struct pp_hwmgr *hwmgr,
 {
 	struct phm_ppt_v3_information *pptable_information =
 		(struct phm_ppt_v3_information *)hwmgr->pptable;
+	struct vega20_hwmgr *data =
+			(struct vega20_hwmgr *)(hwmgr->backend);
+	PPTable_t *pp_table = &(data->smc_state_table.pp_table);
 
 	memcpy(thermal_data, &SMU7ThermalWithDelayPolicy[0], sizeof(struct PP_TemperatureRange));
 
 	thermal_data->max = pptable_information->us_software_shutdown_temp *
 		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+	thermal_data->edge_crit_max = pp_table->TedgeLimit *
+		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+	thermal_data->mem_crit_max = pp_table->ThbmLimit *
+		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
 
 	return 0;
 }
drivers/gpu/drm/amd/powerplay/inc/power_state.h +4 -0
index a99b5cbb113e..c102415ddc98 100644
@@ -124,6 +124,10 @@ struct PP_StateSoftwareAlgorithmBlock {
 struct PP_TemperatureRange {
 	int min;
 	int max;
+	int edge_min;
+	int edge_crit_max;
+	int mem_min;
+	int mem_crit_max;
 };
 
 struct PP_StateValidationBlock {
drivers/gpu/drm/amd/powerplay/inc/pp_thermal.h +4 -4
index 201d2b6329ab..75a0a2f8bea2 100644
@@ -27,14 +27,14 @@
 
 static const struct PP_TemperatureRange SMU7ThermalWithDelayPolicy[] =
 {
-	{-273150,  99000},
-	{ 120000, 120000},
+	{-273150,  99000, -273150, 99000, -273150, 99000},
+	{ 120000, 120000, 120000, 120000, 120000, 120000},
 };
 
 static const struct PP_TemperatureRange SMU7ThermalPolicy[] =
 {
-	{-273150,  99000},
-	{ 120000, 120000},
+	{-273150,  99000, -273150, 99000, -273150, 99000},
+	{ 120000, 120000, 120000, 120000, 120000, 120000},
 };
 
 #endif
Signature · 6 lines
2.21.0

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): [email protected]

Reviewed-by
#4 lore raw

Re: [PATCH 2/6] drm/amd/powerplay: support edge/memory critical limit values

Quoted 6 lines · Evan Quan

These new interfaces(temp2_crit, temp2_crit_hyst, temp3_crit, temp3_crit_hyst) are supported on SOC15 dGPUs only.

Change-Id Ia87e3f6ad816b51d6680eb74c8f755d6c2b0a6ae
Quoted 260 lines
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h       |  8 +++
 drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c        | 51 +++++++++++++++++--
 .../drm/amd/powerplay/hwmgr/hardwaremanager.c | 12 ++++-
 .../drm/amd/powerplay/hwmgr/vega10_hwmgr.c    |  6 +++
 .../drm/amd/powerplay/hwmgr/vega12_hwmgr.c    |  7 +++
 .../drm/amd/powerplay/hwmgr/vega20_hwmgr.c    |  7 +++
 .../gpu/drm/amd/powerplay/inc/power_state.h   |  4 ++
 .../gpu/drm/amd/powerplay/inc/pp_thermal.h    |  8 +--
 8 files changed, 95 insertions(+), 8 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h +8 -0
index dca35407879d..e1492438ae7b 100644
@@ -75,6 +75,14 @@ struct amdgpu_dpm_thermal {
        int                min_temp;
        /* high temperature threshold */
        int                max_temp;
+       /* edge low temperature threshold */
+       int                min_edge_temp;
+       /* edge high temperature critical threshold */
+       int                max_edge_crit_temp;
+       /* memory low temperature threshold */
+       int                min_mem_temp;
+       /* memory high temperature critical threshold */
+       int                max_mem_crit_temp;
        /* was last interrupt low to high or high to low */
        bool               high_to_low;
        /* interrupt source */
drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c +48 -3
index c17eb228417e..3f6b5b5bb0c6 100644
@@ -1466,6 +1466,37 @@ static ssize_t amdgpu_hwmon_show_temp_thresh(struct device *dev,
        return snprintf(buf, PAGE_SIZE, "%d\n", temp);
 }
 
+static ssize_t amdgpu_hwmon_show_edge_temp_thresh(struct device *dev,
+                                            struct device_attribute *attr,
+                                            char *buf)
+{
+       struct amdgpu_device *adev = dev_get_drvdata(dev);
+       int hyst = to_sensor_dev_attr(attr)->index;
+       int temp;
+
+       if (hyst)
+               temp = adev->pm.dpm.thermal.min_edge_temp;
+       else
+               temp = adev->pm.dpm.thermal.max_edge_crit_temp;
+
+       return snprintf(buf, PAGE_SIZE, "%d\n", temp);
+}
+
+static ssize_t amdgpu_hwmon_show_mem_temp_thresh(struct device *dev,
+                                            struct device_attribute *attr,
+                                            char *buf)
+{
+       struct amdgpu_device *adev = dev_get_drvdata(dev);
+       int hyst = to_sensor_dev_attr(attr)->index;
+       int temp;
+
+       if (hyst)
+               temp = adev->pm.dpm.thermal.min_mem_temp;
+       else
+               temp = adev->pm.dpm.thermal.max_mem_crit_temp;
+
+       return snprintf(buf, PAGE_SIZE, "%d\n", temp);
+}
 
 static ssize_t amdgpu_hwmon_show_temp_label(struct device *dev,
                                             struct device_attribute *attr,
@@ -2054,9 +2085,11 @@ static ssize_t amdgpu_hwmon_show_mclk_label(struct device *dev,
  *
  * - temp1_input: the on die GPU temperature in millidegrees Celsius
  *
- * - temp1_crit: temperature critical max value in millidegrees Celsius
+ * - temp[1-3]_crit: temperature critical max value in millidegrees Celsius
+ *   - temp2_crit and temp3_crit are supported on SOC15 dGPUs only
  *
- * - temp1_crit_hyst: temperature hysteresis for critical limit in millidegrees Celsius
+ * - temp[1-3]_crit_hyst: temperature hysteresis for critical limit in millidegrees Celsius
+ *   - temp2_crit_hyst and temp3_crit_hyst are supported on SOC15 dGPUs only
  *
  * hwmon interfaces for GPU voltage:
  *
@@ -2107,6 +2140,10 @@ static ssize_t amdgpu_hwmon_show_mclk_label(struct device *dev,
 static SENSOR_DEVICE_ATTR(temp1_input, S_IRUGO, amdgpu_hwmon_show_temp, NULL, 0);
 static SENSOR_DEVICE_ATTR(temp1_crit, S_IRUGO, amdgpu_hwmon_show_temp_thresh, NULL, 0);
 static SENSOR_DEVICE_ATTR(temp1_crit_hyst, S_IRUGO, amdgpu_hwmon_show_temp_thresh, NULL, 1);
+static SENSOR_DEVICE_ATTR(temp2_crit, S_IRUGO, amdgpu_hwmon_show_edge_temp_thresh, NULL, 0);
+static SENSOR_DEVICE_ATTR(temp2_crit_hyst, S_IRUGO, amdgpu_hwmon_show_edge_temp_thresh, NULL, 1);
+static SENSOR_DEVICE_ATTR(temp3_crit, S_IRUGO, amdgpu_hwmon_show_mem_temp_thresh, NULL, 0);
+static SENSOR_DEVICE_ATTR(temp3_crit_hyst, S_IRUGO, amdgpu_hwmon_show_mem_temp_thresh, NULL, 1);
 static SENSOR_DEVICE_ATTR(temp1_label, S_IRUGO, amdgpu_hwmon_show_temp_label, NULL, PP_TEMP_JUNCTION);
 static SENSOR_DEVICE_ATTR(temp2_label, S_IRUGO, amdgpu_hwmon_show_temp_label, NULL, PP_TEMP_EDGE);
 static SENSOR_DEVICE_ATTR(temp3_label, S_IRUGO, amdgpu_hwmon_show_temp_label, NULL, PP_TEMP_MEM);
@@ -2136,6 +2173,10 @@ static struct attribute *hwmon_attributes[] = {
        &sensor_dev_attr_temp1_input.dev_attr.attr,
        &sensor_dev_attr_temp1_crit.dev_attr.attr,
        &sensor_dev_attr_temp1_crit_hyst.dev_attr.attr,
+       &sensor_dev_attr_temp2_crit.dev_attr.attr,
+       &sensor_dev_attr_temp2_crit_hyst.dev_attr.attr,
+       &sensor_dev_attr_temp3_crit.dev_attr.attr,
+       &sensor_dev_attr_temp3_crit_hyst.dev_attr.attr,
        &sensor_dev_attr_temp1_label.dev_attr.attr,
        &sensor_dev_attr_temp2_label.dev_attr.attr,
        &sensor_dev_attr_temp3_label.dev_attr.attr,
@@ -2265,7 +2306,11 @@ static umode_t hwmon_attributes_visible(struct kobject *kobj,
        if (((adev->flags & AMD_IS_APU) ||
             adev->asic_type < CHIP_VEGA10) &&
            (attr == &sensor_dev_attr_temp2_label.dev_attr.attr ||
-            attr == &sensor_dev_attr_temp3_label.dev_attr.attr))
+            attr == &sensor_dev_attr_temp3_label.dev_attr.attr ||
+            attr == &sensor_dev_attr_temp2_crit.dev_attr.attr ||
+            attr == &sensor_dev_attr_temp2_crit_hyst.dev_attr.attr ||
+            attr == &sensor_dev_attr_temp3_crit.dev_attr.attr ||
+            attr == &sensor_dev_attr_temp3_crit_hyst.dev_attr.attr))
                return 0;
 
 
drivers/gpu/drm/amd/powerplay/hwmgr/hardwaremanager.c +11 -1
index 70f7f47a2fcf..c8397b20f71a 100644
@@ -225,7 +225,13 @@ int phm_register_irq_handlers(struct pp_hwmgr *hwmgr)
 int phm_start_thermal_controller(struct pp_hwmgr *hwmgr)
 {
        int ret = 0;
-       struct PP_TemperatureRange range = {TEMP_RANGE_MIN, TEMP_RANGE_MAX};
+       struct PP_TemperatureRange range = {
+               TEMP_RANGE_MIN,
+               TEMP_RANGE_MAX,
+               TEMP_RANGE_MIN,
+               TEMP_RANGE_MAX,
+               TEMP_RANGE_MIN,
+               TEMP_RANGE_MAX};
        struct amdgpu_device *adev = hwmgr->adev;
 
        if (hwmgr->hwmgr_func->get_thermal_temperature_range)
@@ -239,6 +245,10 @@ int phm_start_thermal_controller(struct pp_hwmgr *hwmgr)
 
        adev->pm.dpm.thermal.min_temp = range.min;
        adev->pm.dpm.thermal.max_temp = range.max;
+       adev->pm.dpm.thermal.min_edge_temp = range.edge_min;
+       adev->pm.dpm.thermal.max_edge_crit_temp = range.edge_crit_max;
+       adev->pm.dpm.thermal.min_mem_temp = range.mem_min;
+       adev->pm.dpm.thermal.max_mem_crit_temp = range.mem_crit_max;
 
        return ret;
 }
drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c +6 -0
index 384c37875cd0..65aa7e70d7d9 100644
@@ -4854,11 +4854,17 @@ static int vega10_get_thermal_temperature_range(struct pp_hwmgr *hwmgr,
 {
        struct phm_ppt_v2_information *table_info =
                        (struct phm_ppt_v2_information *)hwmgr->pptable;
+       struct vega10_hwmgr *data = hwmgr->backend;
+       PPTable_t *pp_table = &(data->smc_state_table.pp_table);
 
        memcpy(thermal_data, &SMU7ThermalWithDelayPolicy[0], sizeof(struct PP_TemperatureRange));
 
        thermal_data->max = table_info->tdp_table->usSoftwareShutdownTemp *
                PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+       thermal_data->edge_crit_max = pp_table->TedgeLimit *
+               PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+       thermal_data->mem_crit_max = pp_table->ThbmLimit *
+               PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
 
        return 0;
 }
drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c +7 -0
index 707cd4b0357f..c0994851e7c7 100644
@@ -2528,11 +2528,18 @@ static int vega12_get_thermal_temperature_range(struct pp_hwmgr *hwmgr,
 {
        struct phm_ppt_v3_information *pptable_information =
                (struct phm_ppt_v3_information *)hwmgr->pptable;
+       struct vega12_hwmgr *data =
+                       (struct vega12_hwmgr *)(hwmgr->backend);
+       PPTable_t *pp_table = &(data->smc_state_table.pp_table);
 
        memcpy(thermal_data, &SMU7ThermalWithDelayPolicy[0], sizeof(struct PP_TemperatureRange));
 
        thermal_data->max = pptable_information->us_software_shutdown_temp *
                PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+       thermal_data->edge_crit_max = pp_table->TedgeLimit *
+               PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+       thermal_data->mem_crit_max = pp_table->ThbmLimit *
+               PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
 
        return 0;
 }
drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c +7 -0
index 39a547084e90..5d9aa0f22c86 100644
@@ -3965,11 +3965,18 @@ static int vega20_get_thermal_temperature_range(struct pp_hwmgr *hwmgr,
 {
        struct phm_ppt_v3_information *pptable_information =
                (struct phm_ppt_v3_information *)hwmgr->pptable;
+       struct vega20_hwmgr *data =
+                       (struct vega20_hwmgr *)(hwmgr->backend);
+       PPTable_t *pp_table = &(data->smc_state_table.pp_table);
 
        memcpy(thermal_data, &SMU7ThermalWithDelayPolicy[0], sizeof(struct PP_TemperatureRange));
 
        thermal_data->max = pptable_information->us_software_shutdown_temp *
                PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+       thermal_data->edge_crit_max = pp_table->TedgeLimit *
+               PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+       thermal_data->mem_crit_max = pp_table->ThbmLimit *
+               PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
 
        return 0;
 }
drivers/gpu/drm/amd/powerplay/inc/power_state.h +4 -0
index a99b5cbb113e..c102415ddc98 100644
@@ -124,6 +124,10 @@ struct PP_StateSoftwareAlgorithmBlock {
 struct PP_TemperatureRange {
        int min;
        int max;
+       int edge_min;
+       int edge_crit_max;
+       int mem_min;
+       int mem_crit_max;
 };
 
 struct PP_StateValidationBlock {
drivers/gpu/drm/amd/powerplay/inc/pp_thermal.h +4 -5
index 201d2b6329ab..75a0a2f8bea2 100644
@@ -27,14 +27,14 @@
 
 static const struct PP_TemperatureRange SMU7ThermalWithDelayPolicy[] =
 {
-       {-273150,  99000},
-       { 120000, 120000},
+       {-273150,  99000, -273150, 99000, -273150, 99000},
+       { 120000, 120000, 120000, 120000, 120000, 120000},
 };
 
 static const struct PP_TemperatureRange SMU7ThermalPolicy[] =
 {
-       {-273150,  99000},
-       { 120000, 120000},
+       {-273150,  99000, -273150, 99000, -273150, 99000},
+       { 120000, 120000, 120000, 120000, 120000, 120000},
 };
 
 #endif
--

2.21.0

Quoted 4 lines

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

author
#5 lore raw

PATCH 3/6 drm/amd/powerplay: support temperature emergency max values

In reply to a message not in the archive

These new interfaces(temp1_emergency, temp2_emergency, temp3_emergency) are supported on SOC15 dGPUs only.

Change-Id I2552df63f9c8c50294b3940bb2a402217673c2bc
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h       |  6 +++
 drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c        | 40 ++++++++++++++++++-
 .../drm/amd/powerplay/hwmgr/hardwaremanager.c |  6 +++
 .../drm/amd/powerplay/hwmgr/vega10_hwmgr.c    |  6 +++
 .../drm/amd/powerplay/hwmgr/vega12_hwmgr.c    |  6 +++
 .../drm/amd/powerplay/hwmgr/vega20_hwmgr.c    |  6 +++
 .../gpu/drm/amd/powerplay/inc/power_state.h   |  3 ++
 .../gpu/drm/amd/powerplay/inc/pp_thermal.h    | 12 ++++--
 8 files changed, 80 insertions(+), 5 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h +6 -0
index e1492438ae7b..32e2def42f30 100644
@@ -75,14 +75,20 @@ struct amdgpu_dpm_thermal {
 	int                min_temp;
 	/* high temperature threshold */
 	int                max_temp;
+	/* hotspot max emergency(shutdown) temp */
+	int                max_hotspot_emergency_temp;
 	/* edge low temperature threshold */
 	int                min_edge_temp;
 	/* edge high temperature critical threshold */
 	int                max_edge_crit_temp;
+	/* edge max emergency(shutdown) temp */
+	int                max_edge_emergency_temp;
 	/* memory low temperature threshold */
 	int                min_mem_temp;
 	/* memory high temperature critical threshold */
 	int                max_mem_crit_temp;
+	/* memory max emergency(shutdown) temp */
+	int                max_mem_emergency_temp;
 	/* was last interrupt low to high or high to low */
 	bool               high_to_low;
 	/* interrupt source */
drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c +39 -1
index 3f6b5b5bb0c6..be33144e2dca 100644
@@ -1511,6 +1511,32 @@ static ssize_t amdgpu_hwmon_show_temp_label(struct device *dev,
 	return snprintf(buf, PAGE_SIZE, "%s\n", temp_label[channel].label);
 }
 
+static ssize_t amdgpu_hwmon_show_temp_emergency(struct device *dev,
+					     struct device_attribute *attr,
+					     char *buf)
+{
+	struct amdgpu_device *adev = dev_get_drvdata(dev);
+	int channel = to_sensor_dev_attr(attr)->index;
+	int temp;
+
+	if (channel >= PP_TEMP_MAX)
+		return -EINVAL;
+
+	switch (channel) {
+	case PP_TEMP_JUNCTION:
+		temp = adev->pm.dpm.thermal.max_hotspot_emergency_temp;
+		break;
+	case PP_TEMP_EDGE:
+		temp = adev->pm.dpm.thermal.max_edge_emergency_temp;
+		break;
+	case PP_TEMP_MEM:
+		temp = adev->pm.dpm.thermal.max_mem_emergency_temp;
+		break;
+	}
+
+	return snprintf(buf, PAGE_SIZE, "%d\n", temp);
+}
+
 static ssize_t amdgpu_hwmon_get_pwm1_enable(struct device *dev,
 					    struct device_attribute *attr,
 					    char *buf)
@@ -2091,6 +2117,9 @@ static ssize_t amdgpu_hwmon_show_mclk_label(struct device *dev,
  * - temp[1-3]_crit_hyst: temperature hysteresis for critical limit in millidegrees Celsius
  *   - temp2_crit_hyst and temp3_crit_hyst are supported on SOC15 dGPUs only
  *
+ * - temp[1-3]_emergency: temperature emergency max value(asic shutdown) in millidegrees Celsius
+ *   - these are supported on SOC15 dGPUs only
+ *
  * hwmon interfaces for GPU voltage:
  *
  * - in0_input: the voltage on the GPU in millivolts
@@ -2140,10 +2169,13 @@ static ssize_t amdgpu_hwmon_show_mclk_label(struct device *dev,
 static SENSOR_DEVICE_ATTR(temp1_input, S_IRUGO, amdgpu_hwmon_show_temp, NULL, 0);
 static SENSOR_DEVICE_ATTR(temp1_crit, S_IRUGO, amdgpu_hwmon_show_temp_thresh, NULL, 0);
 static SENSOR_DEVICE_ATTR(temp1_crit_hyst, S_IRUGO, amdgpu_hwmon_show_temp_thresh, NULL, 1);
+static SENSOR_DEVICE_ATTR(temp1_emergency, S_IRUGO, amdgpu_hwmon_show_temp_emergency, NULL, PP_TEMP_JUNCTION);
 static SENSOR_DEVICE_ATTR(temp2_crit, S_IRUGO, amdgpu_hwmon_show_edge_temp_thresh, NULL, 0);
 static SENSOR_DEVICE_ATTR(temp2_crit_hyst, S_IRUGO, amdgpu_hwmon_show_edge_temp_thresh, NULL, 1);
+static SENSOR_DEVICE_ATTR(temp2_emergency, S_IRUGO, amdgpu_hwmon_show_temp_emergency, NULL, PP_TEMP_EDGE);
 static SENSOR_DEVICE_ATTR(temp3_crit, S_IRUGO, amdgpu_hwmon_show_mem_temp_thresh, NULL, 0);
 static SENSOR_DEVICE_ATTR(temp3_crit_hyst, S_IRUGO, amdgpu_hwmon_show_mem_temp_thresh, NULL, 1);
+static SENSOR_DEVICE_ATTR(temp3_emergency, S_IRUGO, amdgpu_hwmon_show_temp_emergency, NULL, PP_TEMP_MEM);
 static SENSOR_DEVICE_ATTR(temp1_label, S_IRUGO, amdgpu_hwmon_show_temp_label, NULL, PP_TEMP_JUNCTION);
 static SENSOR_DEVICE_ATTR(temp2_label, S_IRUGO, amdgpu_hwmon_show_temp_label, NULL, PP_TEMP_EDGE);
 static SENSOR_DEVICE_ATTR(temp3_label, S_IRUGO, amdgpu_hwmon_show_temp_label, NULL, PP_TEMP_MEM);
@@ -2180,6 +2212,9 @@ static struct attribute *hwmon_attributes[] = {
 	&sensor_dev_attr_temp1_label.dev_attr.attr,
 	&sensor_dev_attr_temp2_label.dev_attr.attr,
 	&sensor_dev_attr_temp3_label.dev_attr.attr,
+	&sensor_dev_attr_temp1_emergency.dev_attr.attr,
+	&sensor_dev_attr_temp2_emergency.dev_attr.attr,
+	&sensor_dev_attr_temp3_emergency.dev_attr.attr,
 	&sensor_dev_attr_pwm1.dev_attr.attr,
 	&sensor_dev_attr_pwm1_enable.dev_attr.attr,
 	&sensor_dev_attr_pwm1_min.dev_attr.attr,
@@ -2310,7 +2345,10 @@ static umode_t hwmon_attributes_visible(struct kobject *kobj,
 	     attr == &sensor_dev_attr_temp2_crit.dev_attr.attr ||
 	     attr == &sensor_dev_attr_temp2_crit_hyst.dev_attr.attr ||
 	     attr == &sensor_dev_attr_temp3_crit.dev_attr.attr ||
-	     attr == &sensor_dev_attr_temp3_crit_hyst.dev_attr.attr))
+	     attr == &sensor_dev_attr_temp3_crit_hyst.dev_attr.attr ||
+	     attr == &sensor_dev_attr_temp1_emergency.dev_attr.attr ||
+	     attr == &sensor_dev_attr_temp2_emergency.dev_attr.attr ||
+	     attr == &sensor_dev_attr_temp3_emergency.dev_attr.attr))
 		return 0;
 
 
drivers/gpu/drm/amd/powerplay/hwmgr/hardwaremanager.c +6 -0
index c8397b20f71a..12027b194a4a 100644
@@ -228,9 +228,12 @@ int phm_start_thermal_controller(struct pp_hwmgr *hwmgr)
 	struct PP_TemperatureRange range = {
 		TEMP_RANGE_MIN,
 		TEMP_RANGE_MAX,
+		TEMP_RANGE_MAX,
 		TEMP_RANGE_MIN,
 		TEMP_RANGE_MAX,
+		TEMP_RANGE_MAX,
 		TEMP_RANGE_MIN,
+		TEMP_RANGE_MAX,
 		TEMP_RANGE_MAX};
 	struct amdgpu_device *adev = hwmgr->adev;
 
@@ -245,10 +248,13 @@ int phm_start_thermal_controller(struct pp_hwmgr *hwmgr)
 
 	adev->pm.dpm.thermal.min_temp = range.min;
 	adev->pm.dpm.thermal.max_temp = range.max;
+	adev->pm.dpm.thermal.max_hotspot_emergency_temp = range.hotspot_emergency_max;
 	adev->pm.dpm.thermal.min_edge_temp = range.edge_min;
 	adev->pm.dpm.thermal.max_edge_crit_temp = range.edge_crit_max;
+	adev->pm.dpm.thermal.max_edge_emergency_temp = range.edge_emergency_max;
 	adev->pm.dpm.thermal.min_mem_temp = range.mem_min;
 	adev->pm.dpm.thermal.max_mem_crit_temp = range.mem_crit_max;
+	adev->pm.dpm.thermal.max_mem_emergency_temp = range.mem_emergency_max;
 
 	return ret;
 }
drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c +6 -0
index 65aa7e70d7d9..1d78a5ee9523 100644
@@ -4861,10 +4861,16 @@ static int vega10_get_thermal_temperature_range(struct pp_hwmgr *hwmgr,
 
 	thermal_data->max = table_info->tdp_table->usSoftwareShutdownTemp *
 		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+	thermal_data->hotspot_emergency_max = (pp_table->ThotspotLimit + CTF_OFFSET_HOTSPOT) *
+		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
 	thermal_data->edge_crit_max = pp_table->TedgeLimit *
 		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+	thermal_data->edge_emergency_max = (pp_table->TedgeLimit + CTF_OFFSET_EDGE) *
+		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
 	thermal_data->mem_crit_max = pp_table->ThbmLimit *
 		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+	thermal_data->mem_emergency_max = (pp_table->ThbmLimit + CTF_OFFSET_HBM)*
+		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
 
 	return 0;
 }
drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c +6 -0
index c0994851e7c7..a283046347c9 100644
@@ -2536,10 +2536,16 @@ static int vega12_get_thermal_temperature_range(struct pp_hwmgr *hwmgr,
 
 	thermal_data->max = pptable_information->us_software_shutdown_temp *
 		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+	thermal_data->hotspot_emergency_max = (pp_table->ThotspotLimit + CTF_OFFSET_HOTSPOT) *
+		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
 	thermal_data->edge_crit_max = pp_table->TedgeLimit *
 		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+	thermal_data->edge_emergency_max = (pp_table->TedgeLimit + CTF_OFFSET_EDGE) *
+		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
 	thermal_data->mem_crit_max = pp_table->ThbmLimit *
 		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+	thermal_data->mem_emergency_max = (pp_table->ThbmLimit + CTF_OFFSET_HBM)*
+		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
 
 	return 0;
 }
drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c +6 -0
index 5d9aa0f22c86..0c0714862eb8 100644
@@ -3973,10 +3973,16 @@ static int vega20_get_thermal_temperature_range(struct pp_hwmgr *hwmgr,
 
 	thermal_data->max = pptable_information->us_software_shutdown_temp *
 		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+	thermal_data->hotspot_emergency_max = (pp_table->ThotspotLimit + CTF_OFFSET_HOTSPOT) *
+		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
 	thermal_data->edge_crit_max = pp_table->TedgeLimit *
 		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+	thermal_data->edge_emergency_max = (pp_table->TedgeLimit + CTF_OFFSET_EDGE) *
+		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
 	thermal_data->mem_crit_max = pp_table->ThbmLimit *
 		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+	thermal_data->mem_emergency_max = (pp_table->ThbmLimit + CTF_OFFSET_HBM)*
+		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
 
 	return 0;
 }
drivers/gpu/drm/amd/powerplay/inc/power_state.h +3 -0
index c102415ddc98..48d6c4b9fa61 100644
@@ -124,10 +124,13 @@ struct PP_StateSoftwareAlgorithmBlock {
 struct PP_TemperatureRange {
 	int min;
 	int max;
+	int hotspot_emergency_max;
 	int edge_min;
 	int edge_crit_max;
+	int edge_emergency_max;
 	int mem_min;
 	int mem_crit_max;
+	int mem_emergency_max;
 };
 
 struct PP_StateValidationBlock {
drivers/gpu/drm/amd/powerplay/inc/pp_thermal.h +8 -4
index 75a0a2f8bea2..3e30768f9e1c 100644
@@ -27,14 +27,18 @@
 
 static const struct PP_TemperatureRange SMU7ThermalWithDelayPolicy[] =
 {
-	{-273150,  99000, -273150, 99000, -273150, 99000},
-	{ 120000, 120000, 120000, 120000, 120000, 120000},
+	{-273150,  99000, 99000, -273150, 99000, 99000, -273150, 99000, 99000},
+	{ 120000, 120000, 120000, 120000, 120000, 120000, 120000, 120000, 120000},
 };
 
 static const struct PP_TemperatureRange SMU7ThermalPolicy[] =
 {
-	{-273150,  99000, -273150, 99000, -273150, 99000},
-	{ 120000, 120000, 120000, 120000, 120000, 120000},
+	{-273150,  99000, 99000, -273150, 99000, 99000, -273150, 99000, 99000},
+	{ 120000, 120000, 120000, 120000, 120000, 120000, 120000, 120000, 120000},
 };
 
+#define CTF_OFFSET_EDGE			5
+#define CTF_OFFSET_HOTSPOT		5
+#define CTF_OFFSET_HBM			5
+
 #endif
Signature · 6 lines
2.21.0

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): [email protected]

Reviewed-by
#7 lore raw

Re: [PATCH 3/6] drm/amd/powerplay: support temperature emergency max values

Quoted 6 lines · Evan Quan

These new interfaces(temp1_emergency, temp2_emergency, temp3_emergency) are supported on SOC15 dGPUs only.

Change-Id I2552df63f9c8c50294b3940bb2a402217673c2bc
Quoted 261 lines
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h       |  6 +++
 drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c        | 40 ++++++++++++++++++-
 .../drm/amd/powerplay/hwmgr/hardwaremanager.c |  6 +++
 .../drm/amd/powerplay/hwmgr/vega10_hwmgr.c    |  6 +++
 .../drm/amd/powerplay/hwmgr/vega12_hwmgr.c    |  6 +++
 .../drm/amd/powerplay/hwmgr/vega20_hwmgr.c    |  6 +++
 .../gpu/drm/amd/powerplay/inc/power_state.h   |  3 ++
 .../gpu/drm/amd/powerplay/inc/pp_thermal.h    | 12 ++++--
 8 files changed, 80 insertions(+), 5 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_dpm.h +6 -0
index e1492438ae7b..32e2def42f30 100644
@@ -75,14 +75,20 @@ struct amdgpu_dpm_thermal {
        int                min_temp;
        /* high temperature threshold */
        int                max_temp;
+       /* hotspot max emergency(shutdown) temp */
+       int                max_hotspot_emergency_temp;
        /* edge low temperature threshold */
        int                min_edge_temp;
        /* edge high temperature critical threshold */
        int                max_edge_crit_temp;
+       /* edge max emergency(shutdown) temp */
+       int                max_edge_emergency_temp;
        /* memory low temperature threshold */
        int                min_mem_temp;
        /* memory high temperature critical threshold */
        int                max_mem_crit_temp;
+       /* memory max emergency(shutdown) temp */
+       int                max_mem_emergency_temp;
        /* was last interrupt low to high or high to low */
        bool               high_to_low;
        /* interrupt source */
drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c +39 -1
index 3f6b5b5bb0c6..be33144e2dca 100644
@@ -1511,6 +1511,32 @@ static ssize_t amdgpu_hwmon_show_temp_label(struct device *dev,
        return snprintf(buf, PAGE_SIZE, "%s\n", temp_label[channel].label);
 }
 
+static ssize_t amdgpu_hwmon_show_temp_emergency(struct device *dev,
+                                            struct device_attribute *attr,
+                                            char *buf)
+{
+       struct amdgpu_device *adev = dev_get_drvdata(dev);
+       int channel = to_sensor_dev_attr(attr)->index;
+       int temp;
+
+       if (channel >= PP_TEMP_MAX)
+               return -EINVAL;
+
+       switch (channel) {
+       case PP_TEMP_JUNCTION:
+               temp = adev->pm.dpm.thermal.max_hotspot_emergency_temp;
+               break;
+       case PP_TEMP_EDGE:
+               temp = adev->pm.dpm.thermal.max_edge_emergency_temp;
+               break;
+       case PP_TEMP_MEM:
+               temp = adev->pm.dpm.thermal.max_mem_emergency_temp;
+               break;
+       }
+
+       return snprintf(buf, PAGE_SIZE, "%d\n", temp);
+}
+
 static ssize_t amdgpu_hwmon_get_pwm1_enable(struct device *dev,
                                            struct device_attribute *attr,
                                            char *buf)
@@ -2091,6 +2117,9 @@ static ssize_t amdgpu_hwmon_show_mclk_label(struct device *dev,
  * - temp[1-3]_crit_hyst: temperature hysteresis for critical limit in millidegrees Celsius
  *   - temp2_crit_hyst and temp3_crit_hyst are supported on SOC15 dGPUs only
  *
+ * - temp[1-3]_emergency: temperature emergency max value(asic shutdown) in millidegrees Celsius
+ *   - these are supported on SOC15 dGPUs only
+ *
  * hwmon interfaces for GPU voltage:
  *
  * - in0_input: the voltage on the GPU in millivolts
@@ -2140,10 +2169,13 @@ static ssize_t amdgpu_hwmon_show_mclk_label(struct device *dev,
 static SENSOR_DEVICE_ATTR(temp1_input, S_IRUGO, amdgpu_hwmon_show_temp, NULL, 0);
 static SENSOR_DEVICE_ATTR(temp1_crit, S_IRUGO, amdgpu_hwmon_show_temp_thresh, NULL, 0);
 static SENSOR_DEVICE_ATTR(temp1_crit_hyst, S_IRUGO, amdgpu_hwmon_show_temp_thresh, NULL, 1);
+static SENSOR_DEVICE_ATTR(temp1_emergency, S_IRUGO, amdgpu_hwmon_show_temp_emergency, NULL, PP_TEMP_JUNCTION);
 static SENSOR_DEVICE_ATTR(temp2_crit, S_IRUGO, amdgpu_hwmon_show_edge_temp_thresh, NULL, 0);
 static SENSOR_DEVICE_ATTR(temp2_crit_hyst, S_IRUGO, amdgpu_hwmon_show_edge_temp_thresh, NULL, 1);
+static SENSOR_DEVICE_ATTR(temp2_emergency, S_IRUGO, amdgpu_hwmon_show_temp_emergency, NULL, PP_TEMP_EDGE);
 static SENSOR_DEVICE_ATTR(temp3_crit, S_IRUGO, amdgpu_hwmon_show_mem_temp_thresh, NULL, 0);
 static SENSOR_DEVICE_ATTR(temp3_crit_hyst, S_IRUGO, amdgpu_hwmon_show_mem_temp_thresh, NULL, 1);
+static SENSOR_DEVICE_ATTR(temp3_emergency, S_IRUGO, amdgpu_hwmon_show_temp_emergency, NULL, PP_TEMP_MEM);
 static SENSOR_DEVICE_ATTR(temp1_label, S_IRUGO, amdgpu_hwmon_show_temp_label, NULL, PP_TEMP_JUNCTION);
 static SENSOR_DEVICE_ATTR(temp2_label, S_IRUGO, amdgpu_hwmon_show_temp_label, NULL, PP_TEMP_EDGE);
 static SENSOR_DEVICE_ATTR(temp3_label, S_IRUGO, amdgpu_hwmon_show_temp_label, NULL, PP_TEMP_MEM);
@@ -2180,6 +2212,9 @@ static struct attribute *hwmon_attributes[] = {
        &sensor_dev_attr_temp1_label.dev_attr.attr,
        &sensor_dev_attr_temp2_label.dev_attr.attr,
        &sensor_dev_attr_temp3_label.dev_attr.attr,
+       &sensor_dev_attr_temp1_emergency.dev_attr.attr,
+       &sensor_dev_attr_temp2_emergency.dev_attr.attr,
+       &sensor_dev_attr_temp3_emergency.dev_attr.attr,
        &sensor_dev_attr_pwm1.dev_attr.attr,
        &sensor_dev_attr_pwm1_enable.dev_attr.attr,
        &sensor_dev_attr_pwm1_min.dev_attr.attr,
@@ -2310,7 +2345,10 @@ static umode_t hwmon_attributes_visible(struct kobject *kobj,
             attr == &sensor_dev_attr_temp2_crit.dev_attr.attr ||
             attr == &sensor_dev_attr_temp2_crit_hyst.dev_attr.attr ||
             attr == &sensor_dev_attr_temp3_crit.dev_attr.attr ||
-            attr == &sensor_dev_attr_temp3_crit_hyst.dev_attr.attr))
+            attr == &sensor_dev_attr_temp3_crit_hyst.dev_attr.attr ||
+            attr == &sensor_dev_attr_temp1_emergency.dev_attr.attr ||
+            attr == &sensor_dev_attr_temp2_emergency.dev_attr.attr ||
+            attr == &sensor_dev_attr_temp3_emergency.dev_attr.attr))
                return 0;
 
 
drivers/gpu/drm/amd/powerplay/hwmgr/hardwaremanager.c +6 -0
index c8397b20f71a..12027b194a4a 100644
@@ -228,9 +228,12 @@ int phm_start_thermal_controller(struct pp_hwmgr *hwmgr)
        struct PP_TemperatureRange range = {
                TEMP_RANGE_MIN,
                TEMP_RANGE_MAX,
+               TEMP_RANGE_MAX,
                TEMP_RANGE_MIN,
                TEMP_RANGE_MAX,
+               TEMP_RANGE_MAX,
                TEMP_RANGE_MIN,
+               TEMP_RANGE_MAX,
                TEMP_RANGE_MAX};
        struct amdgpu_device *adev = hwmgr->adev;
 
@@ -245,10 +248,13 @@ int phm_start_thermal_controller(struct pp_hwmgr *hwmgr)
 
        adev->pm.dpm.thermal.min_temp = range.min;
        adev->pm.dpm.thermal.max_temp = range.max;
+       adev->pm.dpm.thermal.max_hotspot_emergency_temp = range.hotspot_emergency_max;
        adev->pm.dpm.thermal.min_edge_temp = range.edge_min;
        adev->pm.dpm.thermal.max_edge_crit_temp = range.edge_crit_max;
+       adev->pm.dpm.thermal.max_edge_emergency_temp = range.edge_emergency_max;
        adev->pm.dpm.thermal.min_mem_temp = range.mem_min;
        adev->pm.dpm.thermal.max_mem_crit_temp = range.mem_crit_max;
+       adev->pm.dpm.thermal.max_mem_emergency_temp = range.mem_emergency_max;
 
        return ret;
 }
drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c +6 -0
index 65aa7e70d7d9..1d78a5ee9523 100644
@@ -4861,10 +4861,16 @@ static int vega10_get_thermal_temperature_range(struct pp_hwmgr *hwmgr,
 
        thermal_data->max = table_info->tdp_table->usSoftwareShutdownTemp *
                PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+       thermal_data->hotspot_emergency_max = (pp_table->ThotspotLimit + CTF_OFFSET_HOTSPOT) *
+               PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
        thermal_data->edge_crit_max = pp_table->TedgeLimit *
                PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+       thermal_data->edge_emergency_max = (pp_table->TedgeLimit + CTF_OFFSET_EDGE) *
+               PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
        thermal_data->mem_crit_max = pp_table->ThbmLimit *
                PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+       thermal_data->mem_emergency_max = (pp_table->ThbmLimit + CTF_OFFSET_HBM)*
+               PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
 
        return 0;
 }
drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c +6 -0
index c0994851e7c7..a283046347c9 100644
@@ -2536,10 +2536,16 @@ static int vega12_get_thermal_temperature_range(struct pp_hwmgr *hwmgr,
 
        thermal_data->max = pptable_information->us_software_shutdown_temp *
                PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+       thermal_data->hotspot_emergency_max = (pp_table->ThotspotLimit + CTF_OFFSET_HOTSPOT) *
+               PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
        thermal_data->edge_crit_max = pp_table->TedgeLimit *
                PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+       thermal_data->edge_emergency_max = (pp_table->TedgeLimit + CTF_OFFSET_EDGE) *
+               PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
        thermal_data->mem_crit_max = pp_table->ThbmLimit *
                PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+       thermal_data->mem_emergency_max = (pp_table->ThbmLimit + CTF_OFFSET_HBM)*
+               PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
 
        return 0;
 }
drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c +6 -0
index 5d9aa0f22c86..0c0714862eb8 100644
@@ -3973,10 +3973,16 @@ static int vega20_get_thermal_temperature_range(struct pp_hwmgr *hwmgr,
 
        thermal_data->max = pptable_information->us_software_shutdown_temp *
                PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+       thermal_data->hotspot_emergency_max = (pp_table->ThotspotLimit + CTF_OFFSET_HOTSPOT) *
+               PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
        thermal_data->edge_crit_max = pp_table->TedgeLimit *
                PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+       thermal_data->edge_emergency_max = (pp_table->TedgeLimit + CTF_OFFSET_EDGE) *
+               PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
        thermal_data->mem_crit_max = pp_table->ThbmLimit *
                PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+       thermal_data->mem_emergency_max = (pp_table->ThbmLimit + CTF_OFFSET_HBM)*
+               PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
 
        return 0;
 }
drivers/gpu/drm/amd/powerplay/inc/power_state.h +3 -0
index c102415ddc98..48d6c4b9fa61 100644
@@ -124,10 +124,13 @@ struct PP_StateSoftwareAlgorithmBlock {
 struct PP_TemperatureRange {
        int min;
        int max;
+       int hotspot_emergency_max;
        int edge_min;
        int edge_crit_max;
+       int edge_emergency_max;
        int mem_min;
        int mem_crit_max;
+       int mem_emergency_max;
 };
 
 struct PP_StateValidationBlock {
drivers/gpu/drm/amd/powerplay/inc/pp_thermal.h +8 -5
index 75a0a2f8bea2..3e30768f9e1c 100644
@@ -27,14 +27,18 @@
 
 static const struct PP_TemperatureRange SMU7ThermalWithDelayPolicy[] =
 {
-       {-273150,  99000, -273150, 99000, -273150, 99000},
-       { 120000, 120000, 120000, 120000, 120000, 120000},
+       {-273150,  99000, 99000, -273150, 99000, 99000, -273150, 99000, 99000},
+       { 120000, 120000, 120000, 120000, 120000, 120000, 120000, 120000, 120000},
 };
 
 static const struct PP_TemperatureRange SMU7ThermalPolicy[] =
 {
-       {-273150,  99000, -273150, 99000, -273150, 99000},
-       { 120000, 120000, 120000, 120000, 120000, 120000},
+       {-273150,  99000, 99000, -273150, 99000, 99000, -273150, 99000, 99000},
+       { 120000, 120000, 120000, 120000, 120000, 120000, 120000, 120000, 120000},
 };
 
+#define CTF_OFFSET_EDGE                        5
+#define CTF_OFFSET_HOTSPOT             5
+#define CTF_OFFSET_HBM                 5
+
 #endif
--

2.21.0

Quoted 4 lines

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

author
#8 lore raw

PATCH 4/6 drm/amd/powerplay: support SMU metrics table on Vega12

In reply to a message not in the archive

That should provide some necessary sensor information.

Change-Id I898371cef06795c5369a14c4dd3fe8717959d81a
---
 .../drm/amd/powerplay/hwmgr/vega12_hwmgr.c    | 21 +++++++++++++++++++
 .../drm/amd/powerplay/hwmgr/vega12_hwmgr.h    |  3 +++
 .../drm/amd/powerplay/smumgr/vega12_smumgr.c  | 21 +++++++++++++++++++
 3 files changed, 45 insertions(+)
drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c +21 -0
index a283046347c9..695ac2875540 100644
@@ -1237,6 +1237,27 @@ static uint32_t vega12_dpm_get_mclk(struct pp_hwmgr *hwmgr, bool low)
 	return (mem_clk * 100);
 }
 
+static int vega12_get_metrics_table(struct pp_hwmgr *hwmgr, SmuMetrics_t *metrics_table)
+{
+	struct vega12_hwmgr *data =
+			(struct vega12_hwmgr *)(hwmgr->backend);
+	int ret = 0;
+
+	if (!data->metrics_time || time_after(jiffies, data->metrics_time + HZ / 2)) {
+		ret = smum_smc_table_manager(hwmgr, (uint8_t *)metrics_table,
+				TABLE_SMU_METRICS, true);
+		if (ret) {
+			pr_info("Failed to export SMU metrics table!\n");
+			return ret;
+		}
+		memcpy(&data->metrics_table, metrics_table, sizeof(SmuMetrics_t));
+		data->metrics_time = jiffies;
+	} else
+		memcpy(metrics_table, &data->metrics_table, sizeof(SmuMetrics_t));
+
+	return ret;
+}
+
 static int vega12_get_gpu_power(struct pp_hwmgr *hwmgr, uint32_t *query)
 {
 #if 0
drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.h +3 -0
index b3e424d28994..73875399666a 100644
@@ -396,6 +396,9 @@ struct vega12_hwmgr {
 
 	/* ---- Gfxoff ---- */
 	bool                           gfxoff_controlled_by_driver;
+
+	unsigned long                  metrics_time;
+	SmuMetrics_t                   metrics_table;
 };
 
 #define VEGA12_DPM2_NEAR_TDP_DEC                      10
drivers/gpu/drm/amd/powerplay/smumgr/vega12_smumgr.c +21 -0
index ddb801517667..1eaf0fa28ef7 100644
@@ -287,8 +287,26 @@ static int vega12_smu_init(struct pp_hwmgr *hwmgr)
 	priv->smu_tables.entry[TABLE_OVERDRIVE].version = 0x01;
 	priv->smu_tables.entry[TABLE_OVERDRIVE].size = sizeof(OverDriveTable_t);
 
+	/* allocate space for SMU_METRICS table */
+	ret = amdgpu_bo_create_kernel((struct amdgpu_device *)hwmgr->adev,
+				      sizeof(SmuMetrics_t),
+				      PAGE_SIZE,
+				      AMDGPU_GEM_DOMAIN_VRAM,
+				      &priv->smu_tables.entry[TABLE_SMU_METRICS].handle,
+				      &priv->smu_tables.entry[TABLE_SMU_METRICS].mc_addr,
+				      &priv->smu_tables.entry[TABLE_SMU_METRICS].table);
+	if (ret)
+		goto err4;
+
+	priv->smu_tables.entry[TABLE_SMU_METRICS].version = 0x01;
+	priv->smu_tables.entry[TABLE_SMU_METRICS].size = sizeof(SmuMetrics_t);
+
 	return 0;
 
+err4:
+	amdgpu_bo_free_kernel(&priv->smu_tables.entry[TABLE_OVERDRIVE].handle,
+				&priv->smu_tables.entry[TABLE_OVERDRIVE].mc_addr,
+				&priv->smu_tables.entry[TABLE_OVERDRIVE].table);
 err3:
 	amdgpu_bo_free_kernel(&priv->smu_tables.entry[TABLE_AVFS_FUSE_OVERRIDE].handle,
 				&priv->smu_tables.entry[TABLE_AVFS_FUSE_OVERRIDE].mc_addr,
@@ -334,6 +352,9 @@ static int vega12_smu_fini(struct pp_hwmgr *hwmgr)
 		amdgpu_bo_free_kernel(&priv->smu_tables.entry[TABLE_OVERDRIVE].handle,
 				      &priv->smu_tables.entry[TABLE_OVERDRIVE].mc_addr,
 				      &priv->smu_tables.entry[TABLE_OVERDRIVE].table);
+		amdgpu_bo_free_kernel(&priv->smu_tables.entry[TABLE_SMU_METRICS].handle,
+				      &priv->smu_tables.entry[TABLE_SMU_METRICS].mc_addr,
+				      &priv->smu_tables.entry[TABLE_SMU_METRICS].table);
 		kfree(hwmgr->smu_backend);
 		hwmgr->smu_backend = NULL;
 	}
Signature · 6 lines
2.21.0

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): [email protected]

Reviewed-by
#10 lore raw

Re: [PATCH 4/6] drm/amd/powerplay: support SMU metrics table on Vega12

Quoted 5 lines · Evan Quan

That should provide some necessary sensor information.

Change-Id I898371cef06795c5369a14c4dd3fe8717959d81a
Quoted 96 lines
---
 .../drm/amd/powerplay/hwmgr/vega12_hwmgr.c    | 21 +++++++++++++++++++
 .../drm/amd/powerplay/hwmgr/vega12_hwmgr.h    |  3 +++
 .../drm/amd/powerplay/smumgr/vega12_smumgr.c  | 21 +++++++++++++++++++
 3 files changed, 45 insertions(+)
drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c +21 -0
index a283046347c9..695ac2875540 100644
@@ -1237,6 +1237,27 @@ static uint32_t vega12_dpm_get_mclk(struct pp_hwmgr *hwmgr, bool low)
        return (mem_clk * 100);
 }
 
+static int vega12_get_metrics_table(struct pp_hwmgr *hwmgr, SmuMetrics_t *metrics_table)
+{
+       struct vega12_hwmgr *data =
+                       (struct vega12_hwmgr *)(hwmgr->backend);
+       int ret = 0;
+
+       if (!data->metrics_time || time_after(jiffies, data->metrics_time + HZ / 2)) {
+               ret = smum_smc_table_manager(hwmgr, (uint8_t *)metrics_table,
+                               TABLE_SMU_METRICS, true);
+               if (ret) {
+                       pr_info("Failed to export SMU metrics table!\n");
+                       return ret;
+               }
+               memcpy(&data->metrics_table, metrics_table, sizeof(SmuMetrics_t));
+               data->metrics_time = jiffies;
+       } else
+               memcpy(metrics_table, &data->metrics_table, sizeof(SmuMetrics_t));
+
+       return ret;
+}
+
 static int vega12_get_gpu_power(struct pp_hwmgr *hwmgr, uint32_t *query)
 {
 #if 0
drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.h +3 -0
index b3e424d28994..73875399666a 100644
@@ -396,6 +396,9 @@ struct vega12_hwmgr {
 
        /* ---- Gfxoff ---- */
        bool                           gfxoff_controlled_by_driver;
+
+       unsigned long                  metrics_time;
+       SmuMetrics_t                   metrics_table;
 };
 
 #define VEGA12_DPM2_NEAR_TDP_DEC                      10
drivers/gpu/drm/amd/powerplay/smumgr/vega12_smumgr.c +21 -1
index ddb801517667..1eaf0fa28ef7 100644
@@ -287,8 +287,26 @@ static int vega12_smu_init(struct pp_hwmgr *hwmgr)
        priv->smu_tables.entry[TABLE_OVERDRIVE].version = 0x01;
        priv->smu_tables.entry[TABLE_OVERDRIVE].size = sizeof(OverDriveTable_t);
 
+       /* allocate space for SMU_METRICS table */
+       ret = amdgpu_bo_create_kernel((struct amdgpu_device *)hwmgr->adev,
+                                     sizeof(SmuMetrics_t),
+                                     PAGE_SIZE,
+                                     AMDGPU_GEM_DOMAIN_VRAM,
+                                     &priv->smu_tables.entry[TABLE_SMU_METRICS].handle,
+                                     &priv->smu_tables.entry[TABLE_SMU_METRICS].mc_addr,
+                                     &priv->smu_tables.entry[TABLE_SMU_METRICS].table);
+       if (ret)
+               goto err4;
+
+       priv->smu_tables.entry[TABLE_SMU_METRICS].version = 0x01;
+       priv->smu_tables.entry[TABLE_SMU_METRICS].size = sizeof(SmuMetrics_t);
+
        return 0;
 
+err4:
+       amdgpu_bo_free_kernel(&priv->smu_tables.entry[TABLE_OVERDRIVE].handle,
+                               &priv->smu_tables.entry[TABLE_OVERDRIVE].mc_addr,
+                               &priv->smu_tables.entry[TABLE_OVERDRIVE].table);
 err3:
        amdgpu_bo_free_kernel(&priv->smu_tables.entry[TABLE_AVFS_FUSE_OVERRIDE].handle,
                                &priv->smu_tables.entry[TABLE_AVFS_FUSE_OVERRIDE].mc_addr,
@@ -334,6 +352,9 @@ static int vega12_smu_fini(struct pp_hwmgr *hwmgr)
                amdgpu_bo_free_kernel(&priv->smu_tables.entry[TABLE_OVERDRIVE].handle,
                                      &priv->smu_tables.entry[TABLE_OVERDRIVE].mc_addr,
                                      &priv->smu_tables.entry[TABLE_OVERDRIVE].table);
+               amdgpu_bo_free_kernel(&priv->smu_tables.entry[TABLE_SMU_METRICS].handle,
+                                     &priv->smu_tables.entry[TABLE_SMU_METRICS].mc_addr,
+                                     &priv->smu_tables.entry[TABLE_SMU_METRICS].table);
                kfree(hwmgr->smu_backend);
                hwmgr->smu_backend = NULL;
        }
--

2.21.0

Quoted 4 lines

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

author
#11 lore raw

PATCH 5/6 drm/amd/powerplay: expose current edge and memory temperatures

In reply to a message not in the archive

Two new hwmon interfaces(temp2_input and temp3_input) are added. They are supported on SOC15 dGPUs only.

Change-Id I935c512bd38e080fb8b6e3164c5e5294baff4e91
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c        | 45 +++++++++++++++----
 .../gpu/drm/amd/include/kgd_pp_interface.h    |  2 +
 .../drm/amd/powerplay/hwmgr/vega10_hwmgr.c    | 12 +++++
 .../drm/amd/powerplay/hwmgr/vega12_hwmgr.c    | 19 ++++++++
 .../drm/amd/powerplay/hwmgr/vega20_hwmgr.c    | 18 ++++++++
 5 files changed, 88 insertions(+), 8 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c +37 -8
index be33144e2dca..1007307845d8 100644
@@ -1434,6 +1434,7 @@ static ssize_t amdgpu_hwmon_show_temp(struct device *dev,
 {
 	struct amdgpu_device *adev = dev_get_drvdata(dev);
 	struct drm_device *ddev = adev->ddev;
+	int channel = to_sensor_dev_attr(attr)->index;
 	int r, temp, size = sizeof(temp);
 
 	/* Can't get temperature when the card is off */
@@ -1441,11 +1442,32 @@ static ssize_t amdgpu_hwmon_show_temp(struct device *dev,
 	     (ddev->switch_power_state != DRM_SWITCH_POWER_ON))
 		return -EINVAL;
 
-	/* get the temperature */
-	r = amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_GPU_TEMP,
-				   (void *)&temp, &size);
-	if (r)
-		return r;
+	if (channel >= PP_TEMP_MAX)
+		return -EINVAL;
+
+	switch (channel) {
+	case PP_TEMP_JUNCTION:
+		/* get current junction temperature */
+		r = amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_GPU_TEMP,
+					   (void *)&temp, &size);
+		if (r)
+			return r;
+		break;
+	case PP_TEMP_EDGE:
+		/* get current edge temperature */
+		r = amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_EDGE_TEMP,
+					   (void *)&temp, &size);
+		if (r)
+			return r;
+		break;
+	case PP_TEMP_MEM:
+		/* get current memory temperature */
+		r = amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_MEM_TEMP,
+					   (void *)&temp, &size);
+		if (r)
+			return r;
+		break;
+	}
 
 	return snprintf(buf, PAGE_SIZE, "%d\n", temp);
 }
@@ -2109,7 +2131,8 @@ static ssize_t amdgpu_hwmon_show_mclk_label(struct device *dev,
  * - temp[1-3]_label: temperature channel label
  *   - temp2_label and temp3_label are supported on SOC15 dGPUs only
  *
- * - temp1_input: the on die GPU temperature in millidegrees Celsius
+ * - temp[1-3]_input: the on die GPU temperature in millidegrees Celsius
+ *   - temp2_input and temp3_input are supported on SOC15 dGPUs only
  *
  * - temp[1-3]_crit: temperature critical max value in millidegrees Celsius
  *   - temp2_crit and temp3_crit are supported on SOC15 dGPUs only
@@ -2166,13 +2189,15 @@ static ssize_t amdgpu_hwmon_show_mclk_label(struct device *dev,
  *
  */
 
-static SENSOR_DEVICE_ATTR(temp1_input, S_IRUGO, amdgpu_hwmon_show_temp, NULL, 0);
+static SENSOR_DEVICE_ATTR(temp1_input, S_IRUGO, amdgpu_hwmon_show_temp, NULL, PP_TEMP_JUNCTION);
 static SENSOR_DEVICE_ATTR(temp1_crit, S_IRUGO, amdgpu_hwmon_show_temp_thresh, NULL, 0);
 static SENSOR_DEVICE_ATTR(temp1_crit_hyst, S_IRUGO, amdgpu_hwmon_show_temp_thresh, NULL, 1);
 static SENSOR_DEVICE_ATTR(temp1_emergency, S_IRUGO, amdgpu_hwmon_show_temp_emergency, NULL, PP_TEMP_JUNCTION);
+static SENSOR_DEVICE_ATTR(temp2_input, S_IRUGO, amdgpu_hwmon_show_temp, NULL, PP_TEMP_EDGE);
 static SENSOR_DEVICE_ATTR(temp2_crit, S_IRUGO, amdgpu_hwmon_show_edge_temp_thresh, NULL, 0);
 static SENSOR_DEVICE_ATTR(temp2_crit_hyst, S_IRUGO, amdgpu_hwmon_show_edge_temp_thresh, NULL, 1);
 static SENSOR_DEVICE_ATTR(temp2_emergency, S_IRUGO, amdgpu_hwmon_show_temp_emergency, NULL, PP_TEMP_EDGE);
+static SENSOR_DEVICE_ATTR(temp3_input, S_IRUGO, amdgpu_hwmon_show_temp, NULL, PP_TEMP_MEM);
 static SENSOR_DEVICE_ATTR(temp3_crit, S_IRUGO, amdgpu_hwmon_show_mem_temp_thresh, NULL, 0);
 static SENSOR_DEVICE_ATTR(temp3_crit_hyst, S_IRUGO, amdgpu_hwmon_show_mem_temp_thresh, NULL, 1);
 static SENSOR_DEVICE_ATTR(temp3_emergency, S_IRUGO, amdgpu_hwmon_show_temp_emergency, NULL, PP_TEMP_MEM);
@@ -2205,8 +2230,10 @@ static struct attribute *hwmon_attributes[] = {
 	&sensor_dev_attr_temp1_input.dev_attr.attr,
 	&sensor_dev_attr_temp1_crit.dev_attr.attr,
 	&sensor_dev_attr_temp1_crit_hyst.dev_attr.attr,
+	&sensor_dev_attr_temp2_input.dev_attr.attr,
 	&sensor_dev_attr_temp2_crit.dev_attr.attr,
 	&sensor_dev_attr_temp2_crit_hyst.dev_attr.attr,
+	&sensor_dev_attr_temp3_input.dev_attr.attr,
 	&sensor_dev_attr_temp3_crit.dev_attr.attr,
 	&sensor_dev_attr_temp3_crit_hyst.dev_attr.attr,
 	&sensor_dev_attr_temp1_label.dev_attr.attr,
@@ -2348,7 +2375,9 @@ static umode_t hwmon_attributes_visible(struct kobject *kobj,
 	     attr == &sensor_dev_attr_temp3_crit_hyst.dev_attr.attr ||
 	     attr == &sensor_dev_attr_temp1_emergency.dev_attr.attr ||
 	     attr == &sensor_dev_attr_temp2_emergency.dev_attr.attr ||
-	     attr == &sensor_dev_attr_temp3_emergency.dev_attr.attr))
+	     attr == &sensor_dev_attr_temp3_emergency.dev_attr.attr ||
+	     attr == &sensor_dev_attr_temp2_input.dev_attr.attr ||
+	     attr == &sensor_dev_attr_temp3_input.dev_attr.attr))
 		return 0;
 
 
drivers/gpu/drm/amd/include/kgd_pp_interface.h +2 -0
index 17324c0d503e..19713ffdb03e 100644
@@ -111,6 +111,8 @@ enum amd_pp_sensors {
 	AMDGPU_PP_SENSOR_GPU_LOAD,
 	AMDGPU_PP_SENSOR_GFX_MCLK,
 	AMDGPU_PP_SENSOR_GPU_TEMP,
+	AMDGPU_PP_SENSOR_EDGE_TEMP,
+	AMDGPU_PP_SENSOR_MEM_TEMP,
 	AMDGPU_PP_SENSOR_VCE_POWER,
 	AMDGPU_PP_SENSOR_UVD_POWER,
 	AMDGPU_PP_SENSOR_GPU_POWER,
drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c +12 -0
index 1d78a5ee9523..f4ecbbe854ee 100644
@@ -3785,6 +3785,18 @@ static int vega10_read_sensor(struct pp_hwmgr *hwmgr, int idx,
 		*((uint32_t *)value) = vega10_thermal_get_temperature(hwmgr);
 		*size = 4;
 		break;
+	case AMDGPU_PP_SENSOR_EDGE_TEMP:
+		smum_send_msg_to_smc(hwmgr, PPSMC_MSG_GetTemperatureEdge);
+		*((uint32_t *)value) = smum_get_argument(hwmgr) *
+			PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+		*size = 4;
+		break;
+	case AMDGPU_PP_SENSOR_MEM_TEMP:
+		smum_send_msg_to_smc(hwmgr, PPSMC_MSG_GetTemperatureHBM);
+		*((uint32_t *)value) = smum_get_argument(hwmgr) *
+			PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+		*size = 4;
+		break;
 	case AMDGPU_PP_SENSOR_UVD_POWER:
 		*((uint32_t *)value) = data->uvd_power_gated ? 0 : 1;
 		*size = 4;
drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c +19 -0
index 695ac2875540..86c48cb56f6c 100644
@@ -1338,6 +1338,7 @@ static int vega12_read_sensor(struct pp_hwmgr *hwmgr, int idx,
 			      void *value, int *size)
 {
 	struct vega12_hwmgr *data = (struct vega12_hwmgr *)(hwmgr->backend);
+	SmuMetrics_t metrics_table;
 	int ret = 0;
 
 	switch (idx) {
@@ -1360,6 +1361,24 @@ static int vega12_read_sensor(struct pp_hwmgr *hwmgr, int idx,
 		*((uint32_t *)value) = vega12_thermal_get_temperature(hwmgr);
 		*size = 4;
 		break;
+	case AMDGPU_PP_SENSOR_EDGE_TEMP:
+		ret = vega12_get_metrics_table(hwmgr, &metrics_table);
+		if (ret)
+			return ret;
+
+		*((uint32_t *)value) = metrics_table.TemperatureEdge *
+			PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+		*size = 4;
+		break;
+	case AMDGPU_PP_SENSOR_MEM_TEMP:
+		ret = vega12_get_metrics_table(hwmgr, &metrics_table);
+		if (ret)
+			return ret;
+
+		*((uint32_t *)value) = metrics_table.TemperatureHBM *
+			PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+		*size = 4;
+		break;
 	case AMDGPU_PP_SENSOR_UVD_POWER:
 		*((uint32_t *)value) = data->uvd_power_gated ? 0 : 1;
 		*size = 4;
drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c +18 -0
index 0c0714862eb8..72a71a002f0b 100644
@@ -2142,6 +2142,24 @@ static int vega20_read_sensor(struct pp_hwmgr *hwmgr, int idx,
 		*((uint32_t *)value) = vega20_thermal_get_temperature(hwmgr);
 		*size = 4;
 		break;
+	case AMDGPU_PP_SENSOR_EDGE_TEMP:
+		ret = vega20_get_metrics_table(hwmgr, &metrics_table);
+		if (ret)
+			return ret;
+
+		*((uint32_t *)value) = metrics_table.TemperatureEdge *
+			PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+		*size = 4;
+		break;
+	case AMDGPU_PP_SENSOR_MEM_TEMP:
+		ret = vega20_get_metrics_table(hwmgr, &metrics_table);
+		if (ret)
+			return ret;
+
+		*((uint32_t *)value) = metrics_table.TemperatureHBM *
+			PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+		*size = 4;
+		break;
 	case AMDGPU_PP_SENSOR_UVD_POWER:
 		*((uint32_t *)value) = data->uvd_power_gated ? 0 : 1;
 		*size = 4;
Signature · 6 lines
2.21.0

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): [email protected]

#13 lore raw

Re: [PATCH 5/6] drm/amd/powerplay: expose current edge and memory temperatures

Quoted 118 lines · Evan Quan

Two new hwmon interfaces(temp2_input and temp3_input) are added. They are supported on SOC15 dGPUs only.

Change-Id I935c512bd38e080fb8b6e3164c5e5294baff4e91
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c        | 45 +++++++++++++++----
 .../gpu/drm/amd/include/kgd_pp_interface.h    |  2 +
 .../drm/amd/powerplay/hwmgr/vega10_hwmgr.c    | 12 +++++
 .../drm/amd/powerplay/hwmgr/vega12_hwmgr.c    | 19 ++++++++
 .../drm/amd/powerplay/hwmgr/vega20_hwmgr.c    | 18 ++++++++
 5 files changed, 88 insertions(+), 8 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c +37 -8
index be33144e2dca..1007307845d8 100644
@@ -1434,6 +1434,7 @@ static ssize_t amdgpu_hwmon_show_temp(struct device *dev,
 {
        struct amdgpu_device *adev = dev_get_drvdata(dev);
        struct drm_device *ddev = adev->ddev;
+       int channel = to_sensor_dev_attr(attr)->index;
        int r, temp, size = sizeof(temp);
 
        /* Can't get temperature when the card is off */
@@ -1441,11 +1442,32 @@ static ssize_t amdgpu_hwmon_show_temp(struct device *dev,
             (ddev->switch_power_state != DRM_SWITCH_POWER_ON))
                return -EINVAL;
 
-       /* get the temperature */
-       r = amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_GPU_TEMP,
-                                  (void *)&temp, &size);
-       if (r)
-               return r;
+       if (channel >= PP_TEMP_MAX)
+               return -EINVAL;
+
+       switch (channel) {
+       case PP_TEMP_JUNCTION:
+               /* get current junction temperature */
+               r = amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_GPU_TEMP,
+                                          (void *)&temp, &size);
+               if (r)
+                       return r;
+               break;
+       case PP_TEMP_EDGE:
+               /* get current edge temperature */
+               r = amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_EDGE_TEMP,
+                                          (void *)&temp, &size);
+               if (r)
+                       return r;
+               break;
+       case PP_TEMP_MEM:
+               /* get current memory temperature */
+               r = amdgpu_dpm_read_sensor(adev, AMDGPU_PP_SENSOR_MEM_TEMP,
+                                          (void *)&temp, &size);
+               if (r)
+                       return r;
+               break;
+       }
 
        return snprintf(buf, PAGE_SIZE, "%d\n", temp);
 }
@@ -2109,7 +2131,8 @@ static ssize_t amdgpu_hwmon_show_mclk_label(struct device *dev,
  * - temp[1-3]_label: temperature channel label
  *   - temp2_label and temp3_label are supported on SOC15 dGPUs only
  *
- * - temp1_input: the on die GPU temperature in millidegrees Celsius
+ * - temp[1-3]_input: the on die GPU temperature in millidegrees Celsius
+ *   - temp2_input and temp3_input are supported on SOC15 dGPUs only
  *
  * - temp[1-3]_crit: temperature critical max value in millidegrees Celsius
  *   - temp2_crit and temp3_crit are supported on SOC15 dGPUs only
@@ -2166,13 +2189,15 @@ static ssize_t amdgpu_hwmon_show_mclk_label(struct device *dev,
  *
  */
 
-static SENSOR_DEVICE_ATTR(temp1_input, S_IRUGO, amdgpu_hwmon_show_temp, NULL, 0);
+static SENSOR_DEVICE_ATTR(temp1_input, S_IRUGO, amdgpu_hwmon_show_temp, NULL, PP_TEMP_JUNCTION);
 static SENSOR_DEVICE_ATTR(temp1_crit, S_IRUGO, amdgpu_hwmon_show_temp_thresh, NULL, 0);
 static SENSOR_DEVICE_ATTR(temp1_crit_hyst, S_IRUGO, amdgpu_hwmon_show_temp_thresh, NULL, 1);
 static SENSOR_DEVICE_ATTR(temp1_emergency, S_IRUGO, amdgpu_hwmon_show_temp_emergency, NULL, PP_TEMP_JUNCTION);
+static SENSOR_DEVICE_ATTR(temp2_input, S_IRUGO, amdgpu_hwmon_show_temp, NULL, PP_TEMP_EDGE);
 static SENSOR_DEVICE_ATTR(temp2_crit, S_IRUGO, amdgpu_hwmon_show_edge_temp_thresh, NULL, 0);
 static SENSOR_DEVICE_ATTR(temp2_crit_hyst, S_IRUGO, amdgpu_hwmon_show_edge_temp_thresh, NULL, 1);
 static SENSOR_DEVICE_ATTR(temp2_emergency, S_IRUGO, amdgpu_hwmon_show_temp_emergency, NULL, PP_TEMP_EDGE);
+static SENSOR_DEVICE_ATTR(temp3_input, S_IRUGO, amdgpu_hwmon_show_temp, NULL, PP_TEMP_MEM);
 static SENSOR_DEVICE_ATTR(temp3_crit, S_IRUGO, amdgpu_hwmon_show_mem_temp_thresh, NULL, 0);
 static SENSOR_DEVICE_ATTR(temp3_crit_hyst, S_IRUGO, amdgpu_hwmon_show_mem_temp_thresh, NULL, 1);
 static SENSOR_DEVICE_ATTR(temp3_emergency, S_IRUGO, amdgpu_hwmon_show_temp_emergency, NULL, PP_TEMP_MEM);
@@ -2205,8 +2230,10 @@ static struct attribute *hwmon_attributes[] = {
        &sensor_dev_attr_temp1_input.dev_attr.attr,
        &sensor_dev_attr_temp1_crit.dev_attr.attr,
        &sensor_dev_attr_temp1_crit_hyst.dev_attr.attr,
+       &sensor_dev_attr_temp2_input.dev_attr.attr,
        &sensor_dev_attr_temp2_crit.dev_attr.attr,
        &sensor_dev_attr_temp2_crit_hyst.dev_attr.attr,
+       &sensor_dev_attr_temp3_input.dev_attr.attr,
        &sensor_dev_attr_temp3_crit.dev_attr.attr,
        &sensor_dev_attr_temp3_crit_hyst.dev_attr.attr,
        &sensor_dev_attr_temp1_label.dev_attr.attr,
@@ -2348,7 +2375,9 @@ static umode_t hwmon_attributes_visible(struct kobject *kobj,
             attr == &sensor_dev_attr_temp3_crit_hyst.dev_attr.attr ||
             attr == &sensor_dev_attr_temp1_emergency.dev_attr.attr ||
             attr == &sensor_dev_attr_temp2_emergency.dev_attr.attr ||
-            attr == &sensor_dev_attr_temp3_emergency.dev_attr.attr))
+            attr == &sensor_dev_attr_temp3_emergency.dev_attr.attr ||
+            attr == &sensor_dev_attr_temp2_input.dev_attr.attr ||
+            attr == &sensor_dev_attr_temp3_input.dev_attr.attr))
                return 0;
 
 
drivers/gpu/drm/amd/include/kgd_pp_interface.h +0 -0
index 17324c0d503e..19713ffdb03e 100644

drivers/gpu/drm/amd/include/kgd_pp_interface.h enum amd_pp_sensors { · line 113

Quoted 3 lines
drivers/gpu/drm/amd/include/kgd_pp_interface.h +0 -0
@@ -111,6 +111,8 @@ enum amd_pp_sensors {
        AMDGPU_PP_SENSOR_GPU_LOAD,
        AMDGPU_PP_SENSOR_GFX_MCLK,
        AMDGPU_PP_SENSOR_GPU_TEMP,

Add:
AMDGPU_PP_SENSOR_GPU_JUNCTION_TEMP = AMDGPU_PP_SENSOR_GPU_TEMP, and use that for clarity. That said, existing asics use AMDGPU_PP_SENSOR_GPU_TEMP for the edge temperature, so I'd suggest making
AMDGPU_PP_SENSOR_GPU_EDGE_TEMP = AMDGPU_PP_SENSOR_GPU_TEMP, and then add a new entry for JUNCTION.

drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c static int vega10_read_sensor(struct pp_hwmgr *hwmgr, int idx, · line 3785

Quoted 11 lines
drivers/gpu/drm/amd/include/kgd_pp_interface.h +2 -0
+       AMDGPU_PP_SENSOR_EDGE_TEMP,
+       AMDGPU_PP_SENSOR_MEM_TEMP,
        AMDGPU_PP_SENSOR_VCE_POWER,
        AMDGPU_PP_SENSOR_UVD_POWER,
        AMDGPU_PP_SENSOR_GPU_POWER,
drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c +0 -0
index 1d78a5ee9523..f4ecbbe854ee 100644
@@ -3785,6 +3785,18 @@ static int vega10_read_sensor(struct pp_hwmgr *hwmgr, int idx,
                *((uint32_t *)value) = vega10_thermal_get_temperature(hwmgr);

I think vega10_thermal_get_temperature() returns the edge temperature on vega10. Maybe it would be better to switch to PPSMC_MSG_GetTemperatureHotspot for AMDGPU_PP_SENSOR_GPU_JUNCTION_TEMP or use vega10_thermal_get_temperature() for EDGE.

Quoted 86 lines
drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c +12 -0
                *size = 4;
                break;
+       case AMDGPU_PP_SENSOR_EDGE_TEMP:
+               smum_send_msg_to_smc(hwmgr, PPSMC_MSG_GetTemperatureEdge);
+               *((uint32_t *)value) = smum_get_argument(hwmgr) *
+                       PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+               *size = 4;
+               break;
+       case AMDGPU_PP_SENSOR_MEM_TEMP:
+               smum_send_msg_to_smc(hwmgr, PPSMC_MSG_GetTemperatureHBM);
+               *((uint32_t *)value) = smum_get_argument(hwmgr) *
+                       PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+               *size = 4;
+               break;
        case AMDGPU_PP_SENSOR_UVD_POWER:
                *((uint32_t *)value) = data->uvd_power_gated ? 0 : 1;
                *size = 4;
drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c +19 -0
index 695ac2875540..86c48cb56f6c 100644
@@ -1338,6 +1338,7 @@ static int vega12_read_sensor(struct pp_hwmgr *hwmgr, int idx,
                              void *value, int *size)
 {
        struct vega12_hwmgr *data = (struct vega12_hwmgr *)(hwmgr->backend);
+       SmuMetrics_t metrics_table;
        int ret = 0;
 
        switch (idx) {
@@ -1360,6 +1361,24 @@ static int vega12_read_sensor(struct pp_hwmgr *hwmgr, int idx,
                *((uint32_t *)value) = vega12_thermal_get_temperature(hwmgr);
                *size = 4;
                break;
+       case AMDGPU_PP_SENSOR_EDGE_TEMP:
+               ret = vega12_get_metrics_table(hwmgr, &metrics_table);
+               if (ret)
+                       return ret;
+
+               *((uint32_t *)value) = metrics_table.TemperatureEdge *
+                       PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+               *size = 4;
+               break;
+       case AMDGPU_PP_SENSOR_MEM_TEMP:
+               ret = vega12_get_metrics_table(hwmgr, &metrics_table);
+               if (ret)
+                       return ret;
+
+               *((uint32_t *)value) = metrics_table.TemperatureHBM *
+                       PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+               *size = 4;
+               break;
        case AMDGPU_PP_SENSOR_UVD_POWER:
                *((uint32_t *)value) = data->uvd_power_gated ? 0 : 1;
                *size = 4;
drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c +18 -1
index 0c0714862eb8..72a71a002f0b 100644
@@ -2142,6 +2142,24 @@ static int vega20_read_sensor(struct pp_hwmgr *hwmgr, int idx,
                *((uint32_t *)value) = vega20_thermal_get_temperature(hwmgr);
                *size = 4;
                break;
+       case AMDGPU_PP_SENSOR_EDGE_TEMP:
+               ret = vega20_get_metrics_table(hwmgr, &metrics_table);
+               if (ret)
+                       return ret;
+
+               *((uint32_t *)value) = metrics_table.TemperatureEdge *
+                       PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+               *size = 4;
+               break;
+       case AMDGPU_PP_SENSOR_MEM_TEMP:
+               ret = vega20_get_metrics_table(hwmgr, &metrics_table);
+               if (ret)
+                       return ret;
+
+               *((uint32_t *)value) = metrics_table.TemperatureHBM *
+                       PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
+               *size = 4;
+               break;
        case AMDGPU_PP_SENSOR_UVD_POWER:
                *((uint32_t *)value) = data->uvd_power_gated ? 0 : 1;
                *size = 4;
--

2.21.0

Quoted 4 lines

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

author
#14 lore raw

PATCH 6/6 drm/amd/powerplay: correct SOC15 hotspot temperature critical max

In reply to a message not in the archive

Correct Vega10, Vega12 and Vega20 hotspot temperature critical max values.

Change-Id I77bb77761e8530066ec4f3225f8555cf8f672348
---
 drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c | 2 +-
 drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c | 2 +-
 drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)
drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c +1 -1
index f4ecbbe854ee..efd9947eb723 100644
@@ -4871,7 +4871,7 @@ static int vega10_get_thermal_temperature_range(struct pp_hwmgr *hwmgr,
 
 	memcpy(thermal_data, &SMU7ThermalWithDelayPolicy[0], sizeof(struct PP_TemperatureRange));
 
-	thermal_data->max = table_info->tdp_table->usSoftwareShutdownTemp *
+	thermal_data->max = pp_table->ThotspotLimit *
 		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
 	thermal_data->hotspot_emergency_max = (pp_table->ThotspotLimit + CTF_OFFSET_HOTSPOT) *
 		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c +1 -1
index 86c48cb56f6c..ba35118a35b8 100644
@@ -2574,7 +2574,7 @@ static int vega12_get_thermal_temperature_range(struct pp_hwmgr *hwmgr,
 
 	memcpy(thermal_data, &SMU7ThermalWithDelayPolicy[0], sizeof(struct PP_TemperatureRange));
 
-	thermal_data->max = pptable_information->us_software_shutdown_temp *
+	thermal_data->max = pp_table->ThotspotLimit *
 		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
 	thermal_data->hotspot_emergency_max = (pp_table->ThotspotLimit + CTF_OFFSET_HOTSPOT) *
 		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c +1 -1
index 72a71a002f0b..8dcd04561e8f 100644
@@ -3989,7 +3989,7 @@ static int vega20_get_thermal_temperature_range(struct pp_hwmgr *hwmgr,
 
 	memcpy(thermal_data, &SMU7ThermalWithDelayPolicy[0], sizeof(struct PP_TemperatureRange));
 
-	thermal_data->max = pptable_information->us_software_shutdown_temp *
+	thermal_data->max = pp_table->ThotspotLimit *
 		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
 	thermal_data->hotspot_emergency_max = (pp_table->ThotspotLimit + CTF_OFFSET_HOTSPOT) *
 		PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
Signature · 6 lines
2.21.0

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): [email protected]

Acked-by
#16 lore raw

Re: [PATCH 6/6] drm/amd/powerplay: correct SOC15 hotspot temperature critical max

Quoted 6 lines · Evan Quan

Correct Vega10, Vega12 and Vega20 hotspot temperature critical max values.

Change-Id I77bb77761e8530066ec4f3225f8555cf8f672348
Quoted 48 lines
---
 drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c | 2 +-
 drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c | 2 +-
 drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)
drivers/gpu/drm/amd/powerplay/hwmgr/vega10_hwmgr.c +1 -1
index f4ecbbe854ee..efd9947eb723 100644
@@ -4871,7 +4871,7 @@ static int vega10_get_thermal_temperature_range(struct pp_hwmgr *hwmgr,
 
        memcpy(thermal_data, &SMU7ThermalWithDelayPolicy[0], sizeof(struct PP_TemperatureRange));
 
-       thermal_data->max = table_info->tdp_table->usSoftwareShutdownTemp *
+       thermal_data->max = pp_table->ThotspotLimit *
                PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
        thermal_data->hotspot_emergency_max = (pp_table->ThotspotLimit + CTF_OFFSET_HOTSPOT) *
                PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
drivers/gpu/drm/amd/powerplay/hwmgr/vega12_hwmgr.c +1 -1
index 86c48cb56f6c..ba35118a35b8 100644
@@ -2574,7 +2574,7 @@ static int vega12_get_thermal_temperature_range(struct pp_hwmgr *hwmgr,
 
        memcpy(thermal_data, &SMU7ThermalWithDelayPolicy[0], sizeof(struct PP_TemperatureRange));
 
-       thermal_data->max = pptable_information->us_software_shutdown_temp *
+       thermal_data->max = pp_table->ThotspotLimit *
                PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
        thermal_data->hotspot_emergency_max = (pp_table->ThotspotLimit + CTF_OFFSET_HOTSPOT) *
                PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
drivers/gpu/drm/amd/powerplay/hwmgr/vega20_hwmgr.c +1 -2
index 72a71a002f0b..8dcd04561e8f 100644
@@ -3989,7 +3989,7 @@ static int vega20_get_thermal_temperature_range(struct pp_hwmgr *hwmgr,
 
        memcpy(thermal_data, &SMU7ThermalWithDelayPolicy[0], sizeof(struct PP_TemperatureRange));
 
-       thermal_data->max = pptable_information->us_software_shutdown_temp *
+       thermal_data->max = pp_table->ThotspotLimit *
                PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
        thermal_data->hotspot_emergency_max = (pp_table->ThotspotLimit + CTF_OFFSET_HOTSPOT) *
                PP_TEMPERATURE_UNITS_PER_CENTIGRADES;
--

2.21.0

Quoted 4 lines

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

Reviewed-by
#17 lore raw
In reply to a message not in the archive
Quoted 6 lines · Evan Quan

Expose temp[1-3]_label hwmon interfaces. While temp2_label and temp3_label are visible for SOC15 dGPUs only.

Change-Id I7f1e10c52ec21d272027554cdf6da97103e0be58

I'd suggest making this one last in the series since otherwise we'll have labels without temps for a few commits.

Quoted 112 lines
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c        | 40 +++++++++++++++++++
 .../gpu/drm/amd/include/kgd_pp_interface.h    |  7 ++++
 2 files changed, 47 insertions(+)
drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c +40 -0
index 552127b74f78..c17eb228417e 100644
@@ -120,6 +120,15 @@ static const struct cg_flag_name clocks[] = {
        {0, NULL},
 };
 
+static const struct hwmon_temp_label {
+       enum PP_HWMON_TEMP channel;
+       const char *label;
+} temp_label[] = {
+       {PP_TEMP_JUNCTION, "junction"},
+       {PP_TEMP_EDGE, "edge"},
+       {PP_TEMP_MEM, "mem"},
+};
+
 void amdgpu_pm_acpi_event_handler(struct amdgpu_device *adev)
 {
        if (adev->pm.dpm_enabled) {
@@ -1457,6 +1466,20 @@ static ssize_t amdgpu_hwmon_show_temp_thresh(struct device *dev,
        return snprintf(buf, PAGE_SIZE, "%d\n", temp);
 }
 
+
+static ssize_t amdgpu_hwmon_show_temp_label(struct device *dev,
+                                            struct device_attribute *attr,
+                                            char *buf)
+{
+       struct amdgpu_device *adev = dev_get_drvdata(dev);
+       int channel = to_sensor_dev_attr(attr)->index;
+
+       if (channel >= PP_TEMP_MAX)
+               return -EINVAL;
+
+       return snprintf(buf, PAGE_SIZE, "%s\n", temp_label[channel].label);
+}
+
 static ssize_t amdgpu_hwmon_get_pwm1_enable(struct device *dev,
                                            struct device_attribute *attr,
                                            char *buf)
@@ -2026,6 +2049,9 @@ static ssize_t amdgpu_hwmon_show_mclk_label(struct device *dev,
  *
  * hwmon interfaces for GPU temperature:
  *
+ * - temp[1-3]_label: temperature channel label
+ *   - temp2_label and temp3_label are supported on SOC15 dGPUs only
+ *
  * - temp1_input: the on die GPU temperature in millidegrees Celsius
  *
  * - temp1_crit: temperature critical max value in millidegrees Celsius
@@ -2081,6 +2107,9 @@ static ssize_t amdgpu_hwmon_show_mclk_label(struct device *dev,
 static SENSOR_DEVICE_ATTR(temp1_input, S_IRUGO, amdgpu_hwmon_show_temp, NULL, 0);
 static SENSOR_DEVICE_ATTR(temp1_crit, S_IRUGO, amdgpu_hwmon_show_temp_thresh, NULL, 0);
 static SENSOR_DEVICE_ATTR(temp1_crit_hyst, S_IRUGO, amdgpu_hwmon_show_temp_thresh, NULL, 1);
+static SENSOR_DEVICE_ATTR(temp1_label, S_IRUGO, amdgpu_hwmon_show_temp_label, NULL, PP_TEMP_JUNCTION);
+static SENSOR_DEVICE_ATTR(temp2_label, S_IRUGO, amdgpu_hwmon_show_temp_label, NULL, PP_TEMP_EDGE);
+static SENSOR_DEVICE_ATTR(temp3_label, S_IRUGO, amdgpu_hwmon_show_temp_label, NULL, PP_TEMP_MEM);
 static SENSOR_DEVICE_ATTR(pwm1, S_IRUGO | S_IWUSR, amdgpu_hwmon_get_pwm1, amdgpu_hwmon_set_pwm1, 0);
 static SENSOR_DEVICE_ATTR(pwm1_enable, S_IRUGO | S_IWUSR, amdgpu_hwmon_get_pwm1_enable, amdgpu_hwmon_set_pwm1_enable, 0);
 static SENSOR_DEVICE_ATTR(pwm1_min, S_IRUGO, amdgpu_hwmon_get_pwm1_min, NULL, 0);
@@ -2107,6 +2136,9 @@ static struct attribute *hwmon_attributes[] = {
        &sensor_dev_attr_temp1_input.dev_attr.attr,
        &sensor_dev_attr_temp1_crit.dev_attr.attr,
        &sensor_dev_attr_temp1_crit_hyst.dev_attr.attr,
+       &sensor_dev_attr_temp1_label.dev_attr.attr,
+       &sensor_dev_attr_temp2_label.dev_attr.attr,
+       &sensor_dev_attr_temp3_label.dev_attr.attr,
        &sensor_dev_attr_pwm1.dev_attr.attr,
        &sensor_dev_attr_pwm1_enable.dev_attr.attr,
        &sensor_dev_attr_pwm1_min.dev_attr.attr,
@@ -2229,6 +2261,14 @@ static umode_t hwmon_attributes_visible(struct kobject *kobj,
             attr == &sensor_dev_attr_freq2_label.dev_attr.attr))
                return 0;
 
+       /* only SOC15 dGPUs support edge and mem temperatures */
+       if (((adev->flags & AMD_IS_APU) ||
+            adev->asic_type < CHIP_VEGA10) &&
+           (attr == &sensor_dev_attr_temp2_label.dev_attr.attr ||
+            attr == &sensor_dev_attr_temp3_label.dev_attr.attr))
+               return 0;
+
+
        return effective_mode;
 }
 
drivers/gpu/drm/amd/include/kgd_pp_interface.h +7 -1
index 2b579ba9b685..17324c0d503e 100644
@@ -159,6 +159,13 @@ struct pp_states_info {
        uint32_t states[16];
 };
 
+enum PP_HWMON_TEMP {
+       PP_TEMP_JUNCTION = 0,
+       PP_TEMP_EDGE,
+       PP_TEMP_MEM,
+       PP_TEMP_MAX
+};
+
 #define PP_GROUP_MASK        0xF0000000
 #define PP_GROUP_SHIFT       28
 
--

2.21.0

Quoted 4 lines

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): CADnq5_PU-E_bgMNq5_ZokpXrKjxxE33uuhNYfqY2ggBOeePkhg-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org

#19 lore raw
Quoted 26 lines · Alex Deucher
Quoted 6 lines · Evan Quan

Expose temp[1-3]_label hwmon interfaces. While temp2_label and temp3_label are visible for SOC15 dGPUs only.

Change-Id I7f1e10c52ec21d272027554cdf6da97103e0be58

I'd suggest making this one last in the series since otherwise we'll have labels without temps for a few commits.

Quoted 13 lines
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c        | 40 +++++++++++++++++++
 .../gpu/drm/amd/include/kgd_pp_interface.h    |  7 ++++
 2 files changed, 47 insertions(+)
drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c +0 -0
index 552127b74f78..c17eb228417e 100644
@@ -120,6 +120,15 @@ static const struct cg_flag_name clocks[] = {
        {0, NULL},
 };
 
Quoted 6 lines

drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c static const struct cg_flag_name clocks[] = { · line 128

Quoted 6 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c +6 -0
+static const struct hwmon_temp_label {
+       enum PP_HWMON_TEMP channel;
+       const char *label;
+} temp_label[] = {
+       {PP_TEMP_JUNCTION, "junction"},
+       {PP_TEMP_EDGE, "edge"},

Actually switch the order here. Other than vega20, existing asics expose edge today, so lets make temp1 be edge and then temp2 be junction and temp3 be memory.

Alex

Quoted 90 lines
Quoted 90 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c +34 -0
+       {PP_TEMP_MEM, "mem"},
+};
+
 void amdgpu_pm_acpi_event_handler(struct amdgpu_device *adev)
 {
        if (adev->pm.dpm_enabled) {
@@ -1457,6 +1466,20 @@ static ssize_t amdgpu_hwmon_show_temp_thresh(struct device *dev,
        return snprintf(buf, PAGE_SIZE, "%d\n", temp);
 }
 
+
+static ssize_t amdgpu_hwmon_show_temp_label(struct device *dev,
+                                            struct device_attribute *attr,
+                                            char *buf)
+{
+       struct amdgpu_device *adev = dev_get_drvdata(dev);
+       int channel = to_sensor_dev_attr(attr)->index;
+
+       if (channel >= PP_TEMP_MAX)
+               return -EINVAL;
+
+       return snprintf(buf, PAGE_SIZE, "%s\n", temp_label[channel].label);
+}
+
 static ssize_t amdgpu_hwmon_get_pwm1_enable(struct device *dev,
                                            struct device_attribute *attr,
                                            char *buf)
@@ -2026,6 +2049,9 @@ static ssize_t amdgpu_hwmon_show_mclk_label(struct device *dev,
  *
  * hwmon interfaces for GPU temperature:
  *
+ * - temp[1-3]_label: temperature channel label
+ *   - temp2_label and temp3_label are supported on SOC15 dGPUs only
+ *
  * - temp1_input: the on die GPU temperature in millidegrees Celsius
  *
  * - temp1_crit: temperature critical max value in millidegrees Celsius
@@ -2081,6 +2107,9 @@ static ssize_t amdgpu_hwmon_show_mclk_label(struct device *dev,
 static SENSOR_DEVICE_ATTR(temp1_input, S_IRUGO, amdgpu_hwmon_show_temp, NULL, 0);
 static SENSOR_DEVICE_ATTR(temp1_crit, S_IRUGO, amdgpu_hwmon_show_temp_thresh, NULL, 0);
 static SENSOR_DEVICE_ATTR(temp1_crit_hyst, S_IRUGO, amdgpu_hwmon_show_temp_thresh, NULL, 1);
+static SENSOR_DEVICE_ATTR(temp1_label, S_IRUGO, amdgpu_hwmon_show_temp_label, NULL, PP_TEMP_JUNCTION);
+static SENSOR_DEVICE_ATTR(temp2_label, S_IRUGO, amdgpu_hwmon_show_temp_label, NULL, PP_TEMP_EDGE);
+static SENSOR_DEVICE_ATTR(temp3_label, S_IRUGO, amdgpu_hwmon_show_temp_label, NULL, PP_TEMP_MEM);
 static SENSOR_DEVICE_ATTR(pwm1, S_IRUGO | S_IWUSR, amdgpu_hwmon_get_pwm1, amdgpu_hwmon_set_pwm1, 0);
 static SENSOR_DEVICE_ATTR(pwm1_enable, S_IRUGO | S_IWUSR, amdgpu_hwmon_get_pwm1_enable, amdgpu_hwmon_set_pwm1_enable, 0);
 static SENSOR_DEVICE_ATTR(pwm1_min, S_IRUGO, amdgpu_hwmon_get_pwm1_min, NULL, 0);
@@ -2107,6 +2136,9 @@ static struct attribute *hwmon_attributes[] = {
        &sensor_dev_attr_temp1_input.dev_attr.attr,
        &sensor_dev_attr_temp1_crit.dev_attr.attr,
        &sensor_dev_attr_temp1_crit_hyst.dev_attr.attr,
+       &sensor_dev_attr_temp1_label.dev_attr.attr,
+       &sensor_dev_attr_temp2_label.dev_attr.attr,
+       &sensor_dev_attr_temp3_label.dev_attr.attr,
        &sensor_dev_attr_pwm1.dev_attr.attr,
        &sensor_dev_attr_pwm1_enable.dev_attr.attr,
        &sensor_dev_attr_pwm1_min.dev_attr.attr,
@@ -2229,6 +2261,14 @@ static umode_t hwmon_attributes_visible(struct kobject *kobj,
             attr == &sensor_dev_attr_freq2_label.dev_attr.attr))
                return 0;
 
+       /* only SOC15 dGPUs support edge and mem temperatures */
+       if (((adev->flags & AMD_IS_APU) ||
+            adev->asic_type < CHIP_VEGA10) &&
+           (attr == &sensor_dev_attr_temp2_label.dev_attr.attr ||
+            attr == &sensor_dev_attr_temp3_label.dev_attr.attr))
+               return 0;
+
+
        return effective_mode;
 }
 
drivers/gpu/drm/amd/include/kgd_pp_interface.h +7 -0
index 2b579ba9b685..17324c0d503e 100644
@@ -159,6 +159,13 @@ struct pp_states_info {
        uint32_t states[16];
 };
 
+enum PP_HWMON_TEMP {
+       PP_TEMP_JUNCTION = 0,
+       PP_TEMP_EDGE,
+       PP_TEMP_MEM,
+       PP_TEMP_MAX
+};
+
 #define PP_GROUP_MASK        0xF0000000
 #define PP_GROUP_SHIFT       28
 
Quoted 7 lines
Quoted 7 lines

--
2.21.0

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx