Skip to contentSkip to conversation

hwmon: (coretemp) Return -ENODATA for invalid temperature readings

PATCH Applied Ricardo Neri 3 participants 3 messages Oct 6 – Oct 18

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 hwmon: (coretemp) Return -ENODATA for invalid temperature readings

IA32_THERM_STATUS and IA32_PACKAGE_THERM_STATUS have a valid bit, bit 31. The Intel SDM documents it only for the former, but the package register implements it too: x86_pkg_temp_thermal has required it since commit f1a18a105660 ("Thermal: CPU Package temperature thermal").

This driver checked the bit for both registers as well, until commit bf6ea084ebb5 ("hwmon: (coretemp) Do not return -EAGAIN for low temperatures") made it ignore the bit because some CPUs clear it while the temperature is too low to be measured. Since then, temp%d_input reports such a reading as a real temperature. Userspace cannot tell it apart from a genuinely low one.

Fail the read with -ENODATA while the CPU reports the reading as invalid. On a CPU that clears the bit, reading temp%d_input now fails where it used to return a low value. sensors(1) shows N/A in its place.

---
Changes in v2:
- Patches 1 and 2 of v1 have been applied. This is a rework of patch 3.
- Drop temp%d_fault and fail temp%d_input with -ENODATA instead. (Guenter)
- Link to v1: https://lore.kernel.org/r/[email protected]

---
 Documentation/hwmon/coretemp.rst |  3 +++
 drivers/hwmon/coretemp.c         | 21 +++++++++++++++------
 2 files changed, 18 insertions(+), 6 deletions(-)
Documentation/hwmon/coretemp.rst +3 -0
index 349301683381..09f885a847e3 100644
@@ -50,6 +50,9 @@ All sysfs entries are named with their core_id (represented here by 'X').
 
 ================= ========================================================
 tempX_input	  Core temperature (in millidegrees Celsius).
+		  Reading fails with -ENODATA while the CPU reports the
+		  temperature as invalid. Some CPUs do this when the
+		  temperature is too low to be measured.
 tempX_max	  Maximum recommended operating temperature (Tcontrol).
 		  All cooling devices should be turned on.
 tempX_crit	  Maximum junction temperature (in millidegrees Celsius).
drivers/hwmon/coretemp.c +15 -6
index 0ab6bbff5637..a9edbb4924a7 100644
@@ -60,6 +60,8 @@ enum coretemp_attr_index {
 
 /*
  * Per-Core Temperature Data
+ * @temp_valid: Whether the CPU reported @temp as valid. @temp is not
+ *		reported when this is false.
  * @tjmax: The static tjmax value when tjmax cannot be retrieved from
  *		IA32_TEMPERATURE_TARGET MSR.
  * @last_updated: The time when the current temperature value was updated
@@ -72,6 +74,7 @@ enum coretemp_attr_index {
  */
 struct temp_data {
 	int temp;
+	bool temp_valid;
 	int tjmax;
 	unsigned long last_updated;
 	unsigned int cpu;
@@ -398,7 +401,8 @@ static ssize_t show_temp(struct device *dev,
 {
 	struct msr val;
 	struct temp_data *tdata = container_of(devattr, struct temp_data, sd_attrs[ATTR_TEMP]);
-	int tjmax;
+	int tjmax, temp;
+	bool valid;
 
 	mutex_lock(&tdata->update_lock);
 
@@ -407,17 +411,22 @@ static ssize_t show_temp(struct device *dev,
 		tjmax = get_tjmax(tdata, dev);
 		rdmsrq_on_cpu(tdata->cpu, tdata->status_reg, &val.q);
 		/*
-		 * Ignore the valid bit. In all observed cases the register
-		 * value is either low or zero if the valid bit is 0.
-		 * Return it instead of reporting an error which doesn't
-		 * really help at all.
+		 * Bit 31 indicates whether the reading is valid. Some CPUs
+		 * clear it while the temperature is too low to be measured.
+		 * Such a reading is not a temperature. Do not report it.
 		 */
+		tdata->temp_valid = (val.l >> 31) & 1;
 		tdata->temp = tjmax - ((val.l >> 16) & 0xff) * 1000;
 		tdata->last_updated = jiffies;
 	}
 
+	valid = tdata->temp_valid;
+	temp = tdata->temp;
 	mutex_unlock(&tdata->update_lock);
-	return sprintf(buf, "%d\n", tdata->temp);
+
+	if (!valid)
+		return -ENODATA;
+	return sprintf(buf, "%d\n", temp);
 }
 
 static int create_core_attrs(struct temp_data *tdata, struct device *dev)

---
base-commit: 4781ca52761e666cf18b591e6bb0478396c90320
change-id: 20261003-coretemp-enodata-e424540e3b4c

Best regards,

Signature · 1 line
Ricardo Neri <[email protected]>
#1 lore raw
Sashiko has reviewed this patch and found no issues. It looks great!
Quoted 12 lines

IA32_THERM_STATUS and IA32_PACKAGE_THERM_STATUS have a valid bit, bit 31. The Intel SDM documents it only for the former, but the package register implements it too: x86_pkg_temp_thermal has required it since commit f1a18a105660 ("Thermal: CPU Package temperature thermal").

This driver checked the bit for both registers as well, until commit bf6ea084ebb5 ("hwmon: (coretemp) Do not return -EAGAIN for low temperatures") made it ignore the bit because some CPUs clear it while the temperature is too low to be measured. Since then, temp%d_input reports such a reading as a real temperature. Userspace cannot tell it apart from a genuinely low one.

Quoted 7 lines

Fail the read with -ENODATA while the CPU reports the reading as invalid. On a CPU that clears the bit, reading temp%d_input now fails where it used to return a low value. sensors(1) shows N/A in its place.

Sashiko has reviewed this patch and found no issues. It looks great!

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
maintainer Applied
#2 lore raw
In reply to Ricardo Neri
Quoted 12 lines · Ricardo Neri

IA32_THERM_STATUS and IA32_PACKAGE_THERM_STATUS have a valid bit, bit 31. The Intel SDM documents it only for the former, but the package register implements it too: x86_pkg_temp_thermal has required it since commit f1a18a105660 ("Thermal: CPU Package temperature thermal").

This driver checked the bit for both registers as well, until commit bf6ea084ebb5 ("hwmon: (coretemp) Do not return -EAGAIN for low temperatures") made it ignore the bit because some CPUs clear it while the temperature is too low to be measured. Since then, temp%d_input reports such a reading as a real temperature. Userspace cannot tell it apart from a genuinely low one.

Quoted 7 lines

Fail the read with -ENODATA while the CPU reports the reading as invalid. On a CPU that clears the bit, reading temp%d_input now fails where it used to return a low value. sensors(1) shows N/A in its place.

Applied.

Thanks,
Guenter

Quoted 90 lines

---
Changes in v2:
- Patches 1 and 2 of v1 have been applied. This is a rework of patch 3.
- Drop temp%d_fault and fail temp%d_input with -ENODATA instead. (Guenter)
- Link to v1: https://lore.kernel.org/r/[email protected]

---
 Documentation/hwmon/coretemp.rst |  3 +++
 drivers/hwmon/coretemp.c         | 21 +++++++++++++++------
 2 files changed, 18 insertions(+), 6 deletions(-)

---
base-commit: 4781ca52761e666cf18b591e6bb0478396c90320
change-id: 20261003-coretemp-enodata-e424540e3b4c

Best regards,

Documentation/hwmon/coretemp.rst +3 -0
index 349301683381..09f885a847e3 100644
@@ -50,6 +50,9 @@ All sysfs entries are named with their core_id (represented here by 'X').
 
 ================= ========================================================
 tempX_input	  Core temperature (in millidegrees Celsius).
+		  Reading fails with -ENODATA while the CPU reports the
+		  temperature as invalid. Some CPUs do this when the
+		  temperature is too low to be measured.
 tempX_max	  Maximum recommended operating temperature (Tcontrol).
 		  All cooling devices should be turned on.
 tempX_crit	  Maximum junction temperature (in millidegrees Celsius).
drivers/hwmon/coretemp.c +15 -6
index 0ab6bbff5637..a9edbb4924a7 100644
@@ -60,6 +60,8 @@ enum coretemp_attr_index {
 
 /*
  * Per-Core Temperature Data
+ * @temp_valid: Whether the CPU reported @temp as valid. @temp is not
+ *		reported when this is false.
  * @tjmax: The static tjmax value when tjmax cannot be retrieved from
  *		IA32_TEMPERATURE_TARGET MSR.
  * @last_updated: The time when the current temperature value was updated
@@ -72,6 +74,7 @@ enum coretemp_attr_index {
  */
 struct temp_data {
 	int temp;
+	bool temp_valid;
 	int tjmax;
 	unsigned long last_updated;
 	unsigned int cpu;
@@ -398,7 +401,8 @@ static ssize_t show_temp(struct device *dev,
 {
 	struct msr val;
 	struct temp_data *tdata = container_of(devattr, struct temp_data, sd_attrs[ATTR_TEMP]);
-	int tjmax;
+	int tjmax, temp;
+	bool valid;
 
 	mutex_lock(&tdata->update_lock);
 
@@ -407,17 +411,22 @@ static ssize_t show_temp(struct device *dev,
 		tjmax = get_tjmax(tdata, dev);
 		rdmsrq_on_cpu(tdata->cpu, tdata->status_reg, &val.q);
 		/*
-		 * Ignore the valid bit. In all observed cases the register
-		 * value is either low or zero if the valid bit is 0.
-		 * Return it instead of reporting an error which doesn't
-		 * really help at all.
+		 * Bit 31 indicates whether the reading is valid. Some CPUs
+		 * clear it while the temperature is too low to be measured.
+		 * Such a reading is not a temperature. Do not report it.
 		 */
+		tdata->temp_valid = (val.l >> 31) & 1;
 		tdata->temp = tjmax - ((val.l >> 16) & 0xff) * 1000;
 		tdata->last_updated = jiffies;
 	}
 
+	valid = tdata->temp_valid;
+	temp = tdata->temp;
 	mutex_unlock(&tdata->update_lock);
-	return sprintf(buf, "%d\n", tdata->temp);
+
+	if (!valid)
+		return -ENODATA;
+	return sprintf(buf, "%d\n", temp);
 }
 
 static int create_core_attrs(struct temp_data *tdata, struct device *dev)