drm/i915/hwmon: Show device level energy usage

Use i915 HWMON to display device level energy input.

v2: Updated the date and kernel version in feature description
v3:
  - Cleaned up hwm_energy function and removed unused function
    i915_hwmon_energy_status_get (Ashutosh)
v4: KernelVersion: 6.2, Date: February 2023 in doc (Tvrtko)
v5: Change contact to intel-gfx (Rodrigo)
    Change return type of hwm_energy to void (Andi)
Signed-off-by: Dale B Stimson <dale.b.stimson@intel.com>
Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Signed-off-by: Riana Tauro <riana.tauro@intel.com>
Signed-off-by: Badal Nilawar <badal.nilawar@intel.com>
Acked-by: Guenter Roeck <linux@roeck-us.net>
Reviewed-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Reviewed-by: Anshuman Gupta <anshuman.gupta@intel.com>
Signed-off-by: Anshuman Gupta <anshuman.gupta@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20221013154526.2105579-5-ashutosh.dixit@intel.com

Documentation/ABI/testing/sysfs-driver-intel-i915-hwmon

@@ -25,3 +25,11 @@ Contact:	intel-gfx@lists.freedesktop.org
 Description:	RO. Card default power limit (default TDP setting).
 
 		Only supported for particular Intel i915 graphics platforms.
+
+What:		/sys/devices/.../hwmon/hwmon<i>/energy1_input
+Date:		February 2023
+KernelVersion:	6.2
+Contact:	intel-gfx@lists.freedesktop.org
+Description:	RO. Energy input of device in microjoules.
+
+		Only supported for particular Intel i915 graphics platforms.

drivers/gpu/drm/i915/i915_hwmon.c

@@ -17,21 +17,30 @@
  * SF_* - scale factors for particular quantities according to hwmon spec.
  * - voltage  - millivolts
  * - power  - microwatts
+ * - energy - microjoules
  */
 #define SF_VOLTAGE	1000
 #define SF_POWER	1000000
+#define SF_ENERGY	1000000
 
 struct hwm_reg {
 	i915_reg_t gt_perf_status;
 	i915_reg_t pkg_power_sku_unit;
 	i915_reg_t pkg_power_sku;
 	i915_reg_t pkg_rapl_limit;
+	i915_reg_t energy_status_all;
+};
+
+struct hwm_energy_info {
+	u32 reg_val_prev;
+	long accum_energy;			/* Accumulated energy for energy1_input */
 };
 
 struct hwm_drvdata {
 	struct i915_hwmon *hwmon;
 	struct intel_uncore *uncore;
 	struct device *hwmon_dev;
+	struct hwm_energy_info ei;		/*  Energy info for energy1_input */
 	char name[12];
 };
 
@@ -40,6 +49,7 @@ struct i915_hwmon {
 	struct mutex hwmon_lock;		/* counter overflow logic and rmw */
 	struct hwm_reg rg;
 	int scl_shift_power;
+	int scl_shift_energy;
 };
 
 static void
@@ -98,9 +108,58 @@ hwm_field_scale_and_write(struct hwm_drvdata *ddat, i915_reg_t rgadr,
 					    bits_to_clear, bits_to_set);
 }
 
+/*
+ * hwm_energy - Obtain energy value
+ *
+ * The underlying energy hardware register is 32-bits and is subject to
+ * overflow. How long before overflow? For example, with an example
+ * scaling bit shift of 14 bits (see register *PACKAGE_POWER_SKU_UNIT) and
+ * a power draw of 1000 watts, the 32-bit counter will overflow in
+ * approximately 4.36 minutes.
+ *
+ * Examples:
+ *    1 watt:  (2^32 >> 14) /    1 W / (60 * 60 * 24) secs/day -> 3 days
+ * 1000 watts: (2^32 >> 14) / 1000 W / 60             secs/min -> 4.36 minutes
+ *
+ * The function significantly increases overflow duration (from 4.36
+ * minutes) by accumulating the energy register into a 'long' as allowed by
+ * the hwmon API. Using x86_64 128 bit arithmetic (see mul_u64_u32_shr()),
+ * a 'long' of 63 bits, SF_ENERGY of 1e6 (~20 bits) and
+ * hwmon->scl_shift_energy of 14 bits we have 57 (63 - 20 + 14) bits before
+ * energy1_input overflows. This at 1000 W is an overflow duration of 278 years.
+ */
+static void
+hwm_energy(struct hwm_drvdata *ddat, long *energy)
+{
+	struct intel_uncore *uncore = ddat->uncore;
+	struct i915_hwmon *hwmon = ddat->hwmon;
+	struct hwm_energy_info *ei = &ddat->ei;
+	intel_wakeref_t wakeref;
+	i915_reg_t rgaddr;
+	u32 reg_val;
+
+	rgaddr = hwmon->rg.energy_status_all;
+
+	mutex_lock(&hwmon->hwmon_lock);
+
+	with_intel_runtime_pm(uncore->rpm, wakeref)
+		reg_val = intel_uncore_read(uncore, rgaddr);
+
+	if (reg_val >= ei->reg_val_prev)
+		ei->accum_energy += reg_val - ei->reg_val_prev;
+	else
+		ei->accum_energy += UINT_MAX - ei->reg_val_prev + reg_val;
+	ei->reg_val_prev = reg_val;
+
+	*energy = mul_u64_u32_shr(ei->accum_energy, SF_ENERGY,
+				  hwmon->scl_shift_energy);
+	mutex_unlock(&hwmon->hwmon_lock);
+}
+
 static const struct hwmon_channel_info *hwm_info[] = {
 	HWMON_CHANNEL_INFO(in, HWMON_I_INPUT),
 	HWMON_CHANNEL_INFO(power, HWMON_P_MAX | HWMON_P_RATED_MAX),
+	HWMON_CHANNEL_INFO(energy, HWMON_E_INPUT),
 	NULL
 };
 
@@ -194,6 +253,33 @@ hwm_power_write(struct hwm_drvdata *ddat, u32 attr, int chan, long val)
 	}
 }
 
+static umode_t
+hwm_energy_is_visible(const struct hwm_drvdata *ddat, u32 attr)
+{
+	struct i915_hwmon *hwmon = ddat->hwmon;
+	i915_reg_t rgaddr;
+
+	switch (attr) {
+	case hwmon_energy_input:
+		rgaddr = hwmon->rg.energy_status_all;
+		return i915_mmio_reg_valid(rgaddr) ? 0444 : 0;
+	default:
+		return 0;
+	}
+}
+
+static int
+hwm_energy_read(struct hwm_drvdata *ddat, u32 attr, long *val)
+{
+	switch (attr) {
+	case hwmon_energy_input:
+		hwm_energy(ddat, val);
+		return 0;
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
 static umode_t
 hwm_is_visible(const void *drvdata, enum hwmon_sensor_types type,
 	       u32 attr, int channel)
@@ -205,6 +291,8 @@ hwm_is_visible(const void *drvdata, enum hwmon_sensor_types type,
 		return hwm_in_is_visible(ddat, attr);
 	case hwmon_power:
 		return hwm_power_is_visible(ddat, attr, channel);
+	case hwmon_energy:
+		return hwm_energy_is_visible(ddat, attr);
 	default:
 		return 0;
 	}
@@ -221,6 +309,8 @@ hwm_read(struct device *dev, enum hwmon_sensor_types type, u32 attr,
 		return hwm_in_read(ddat, attr, val);
 	case hwmon_power:
 		return hwm_power_read(ddat, attr, channel, val);
+	case hwmon_energy:
+		return hwm_energy_read(ddat, attr, val);
 	default:
 		return -EOPNOTSUPP;
 	}
@@ -256,8 +346,10 @@ hwm_get_preregistration_info(struct drm_i915_private *i915)
 {
 	struct i915_hwmon *hwmon = i915->hwmon;
 	struct intel_uncore *uncore = &i915->uncore;
+	struct hwm_drvdata *ddat = &hwmon->ddat;
 	intel_wakeref_t wakeref;
 	u32 val_sku_unit = 0;
+	long energy;
 
 	/* Available for all Gen12+/dGfx */
 	hwmon->rg.gt_perf_status = GEN12_RPSTAT1;
@@ -266,10 +358,12 @@ hwm_get_preregistration_info(struct drm_i915_private *i915)
 		hwmon->rg.pkg_power_sku_unit = PCU_PACKAGE_POWER_SKU_UNIT;
 		hwmon->rg.pkg_power_sku = PCU_PACKAGE_POWER_SKU;
 		hwmon->rg.pkg_rapl_limit = PCU_PACKAGE_RAPL_LIMIT;
+		hwmon->rg.energy_status_all = PCU_PACKAGE_ENERGY_STATUS;
 	} else {
 		hwmon->rg.pkg_power_sku_unit = INVALID_MMIO_REG;
 		hwmon->rg.pkg_power_sku = INVALID_MMIO_REG;
 		hwmon->rg.pkg_rapl_limit = INVALID_MMIO_REG;
+		hwmon->rg.energy_status_all = INVALID_MMIO_REG;
 	}
 
 	with_intel_runtime_pm(uncore->rpm, wakeref) {
@@ -280,9 +374,17 @@ hwm_get_preregistration_info(struct drm_i915_private *i915)
 		if (i915_mmio_reg_valid(hwmon->rg.pkg_power_sku_unit))
 			val_sku_unit = intel_uncore_read(uncore,
 							 hwmon->rg.pkg_power_sku_unit);
-
-		hwmon->scl_shift_power = REG_FIELD_GET(PKG_PWR_UNIT, val_sku_unit);
 	}
+
+	hwmon->scl_shift_power = REG_FIELD_GET(PKG_PWR_UNIT, val_sku_unit);
+	hwmon->scl_shift_energy = REG_FIELD_GET(PKG_ENERGY_UNIT, val_sku_unit);
+
+	/*
+	 * Initialize 'struct hwm_energy_info', i.e. set fields to the
+	 * first value of the energy register read
+	 */
+	if (i915_mmio_reg_valid(hwmon->rg.energy_status_all))
+		hwm_energy(ddat, &energy);
 }
 
 void i915_hwmon_register(struct drm_i915_private *i915)

drivers/gpu/drm/i915/intel_mchbar_regs.h

@@ -197,7 +197,9 @@
 
 #define PCU_PACKAGE_POWER_SKU_UNIT		_MMIO(MCHBAR_MIRROR_BASE_SNB + 0x5938)
 #define   PKG_PWR_UNIT				REG_GENMASK(3, 0)
+#define   PKG_ENERGY_UNIT			REG_GENMASK(12, 8)
 #define   PKG_TIME_UNIT				REG_GENMASK(19, 16)
+#define PCU_PACKAGE_ENERGY_STATUS              _MMIO(MCHBAR_MIRROR_BASE_SNB + 0x593c)
 
 #define GEN6_GT_PERF_STATUS			_MMIO(MCHBAR_MIRROR_BASE_SNB + 0x5948)
 #define GEN6_RP_STATE_LIMITS			_MMIO(MCHBAR_MIRROR_BASE_SNB + 0x5994)