1 Naming and data format standards for sysfs files
2 ------------------------------------------------
4 The libsensors library offers an interface to the raw sensors data
5 through the sysfs interface. Since lm-sensors 3.0.0, libsensors is
6 completely chip-independent. It assumes that all the kernel drivers
7 implement the standard sysfs interface described in this document.
8 This makes adding or updating support for any given chip very easy, as
9 libsensors, and applications using it, do not need to be modified.
10 This is a major improvement compared to lm-sensors 2.
12 Note that motherboards vary widely in the connections to sensor chips.
13 There is no standard that ensures, for example, that the second
14 temperature sensor is connected to the CPU, or that the second fan is on
15 the CPU. Also, some values reported by the chips need some computation
16 before they make full sense. For example, most chips can only measure
17 voltages between 0 and +4V. Other voltages are scaled back into that
18 range using external resistors. Since the values of these resistors
19 can change from motherboard to motherboard, the conversions cannot be
20 hard coded into the driver and have to be done in user space.
22 For this reason, even if we aim at a chip-independent libsensors, it will
23 still require a configuration file (e.g. /etc/sensors.conf) for proper
24 values conversion, labeling of inputs and hiding of unused inputs.
26 An alternative method that some programs use is to access the sysfs
27 files directly. This document briefly describes the standards that the
28 drivers follow, so that an application program can scan for entries and
29 access this data in a simple and consistent way. That said, such programs
30 will have to implement conversion, labeling and hiding of inputs. For
31 this reason, it is still not recommended to bypass the library.
33 Each chip gets its own directory in the sysfs /sys/devices tree. To
34 find all sensor chips, it is easier to follow the device symlinks from
35 /sys/class/hwmon/hwmon*.
37 Up to lm-sensors 3.0.0, libsensors looks for hardware monitoring attributes
38 in the "physical" device directory. Since lm-sensors 3.0.1, attributes found
39 in the hwmon "class" device directory are also supported. Complex drivers
40 (e.g. drivers for multifunction chips) may want to use this possibility to
41 avoid namespace pollution. The only drawback will be that older versions of
42 libsensors won't support the driver in question.
44 All sysfs values are fixed point numbers.
46 There is only one value per file, unlike the older /proc specification.
47 The common scheme for files naming is: <type><number>_<item>. Usual
48 types for sensor chips are "in" (voltage), "temp" (temperature) and
49 "fan" (fan). Usual items are "input" (measured value), "max" (high
50 threshold, "min" (low threshold). Numbering usually starts from 1,
51 except for voltages which start from 0 (because most data sheets use
52 this). A number is always used for elements that can be present more
53 than once, even if there is a single element of the given type on the
54 specific chip. Other files do not refer to a specific element, so
55 they have a simple name, and no number.
57 Alarms are direct indications read from the chips. The drivers do NOT
58 make comparisons of readings to thresholds. This allows violations
59 between readings to be caught and alarmed. The exact definition of an
60 alarm (for example, whether a threshold must be met or must be exceeded
61 to cause an alarm) is chip-dependent.
63 When setting values of hwmon sysfs attributes, the string representation of
64 the desired value must be written, note that strings which are not a number
65 are interpreted as 0! For more on how written strings are interpreted see the
66 "sysfs attribute writes interpretation" section at the end of this file.
68 -------------------------------------------------------------------------
70 [0-*] denotes any positive number starting from 0
71 [1-*] denotes any positive number starting from 1
76 Read/write values may be read-only for some chips, depending on the
77 hardware implementation.
79 All entries (except name) are optional, and should only be created in a
80 given driver if the chip has the feature.
88 This should be a short, lowercase string, not containing
89 spaces nor dashes, representing the chip name. This is
90 the only mandatory attribute.
91 I2C devices get this attribute created automatically.
99 in[0-*]_min Voltage min value.
103 in[0-*]_max Voltage max value.
107 in[0-*]_input Voltage input value.
110 Voltage measured on the chip pin.
111 Actual voltage depends on the scaling resistors on the
112 motherboard, as recommended in the chip datasheet.
113 This varies by chip and by motherboard.
114 Because of this variation, values are generally NOT scaled
115 by the chip driver, and must be done by the application.
116 However, some drivers (notably lm87 and via686a)
117 do scale, because of internal resistors built into a chip.
118 These drivers will output the actual voltage. Rule of
119 thumb: drivers should report the voltage values at the
122 in[0-*]_label Suggested voltage channel label.
124 Should only be created if the driver has hints about what
125 this voltage channel is being used for, and user-space
126 doesn't. In all other cases, the label is provided by
130 cpu[0-*]_vid CPU core reference voltage.
135 vrm Voltage Regulator Module version number.
136 RW (but changing it should no more be necessary)
137 Originally the VRM standard version multiplied by 10, but now
138 an arbitrary number, as not all standards have a version
140 Affects the way the driver calculates the CPU core reference
141 voltage from the vid pins.
143 Also see the Alarms section for status flags associated with voltages.
150 fan[1-*]_min Fan minimum value
151 Unit: revolution/min (RPM)
154 fan[1-*]_max Fan maximum value
155 Unit: revolution/min (RPM)
156 Only rarely supported by the hardware.
159 fan[1-*]_input Fan input value.
160 Unit: revolution/min (RPM)
163 fan[1-*]_div Fan divisor.
164 Integer value in powers of two (1, 2, 4, 8, 16, 32, 64, 128).
166 Some chips only support values 1, 2, 4 and 8.
167 Note that this is actually an internal clock divisor, which
168 affects the measurable speed range, not the read value.
172 Unit: revolution/min (RPM)
174 Only makes sense if the chip supports closed-loop fan speed
175 control based on the measured fan speed.
177 fan[1-*]_label Suggested fan channel label.
179 Should only be created if the driver has hints about what
180 this fan channel is being used for, and user-space doesn't.
181 In all other cases, the label is provided by user-space.
184 Also see the Alarms section for status flags associated with fans.
191 pwm[1-*] Pulse width modulation fan control.
192 Integer value in the range 0 to 255
197 Fan speed control method:
198 0: no fan speed control (i.e. fan at full speed)
199 1: manual fan speed control enabled (using pwm[1-*])
200 2+: automatic fan speed control enabled
201 Check individual chip documentation files for automatic mode
205 pwm[1-*]_mode 0: DC mode (direct current)
206 1: PWM mode (pulse-width modulation)
209 pwm[1-*]_freq Base PWM frequency in Hz.
210 Only possibly available when pwmN_mode is PWM, but not always
214 pwm[1-*]_auto_channels_temp
215 Select which temperature channels affect this PWM output in
216 auto mode. Bitfield, 1 is temp1, 2 is temp2, 4 is temp3 etc...
217 Which values are possible depend on the chip used.
220 pwm[1-*]_auto_point[1-*]_pwm
221 pwm[1-*]_auto_point[1-*]_temp
222 pwm[1-*]_auto_point[1-*]_temp_hyst
223 Define the PWM vs temperature curve. Number of trip points is
224 chip-dependent. Use this for chips which associate trip points
225 to PWM output channels.
230 temp[1-*]_auto_point[1-*]_pwm
231 temp[1-*]_auto_point[1-*]_temp
232 temp[1-*]_auto_point[1-*]_temp_hyst
233 Define the PWM vs temperature curve. Number of trip points is
234 chip-dependent. Use this for chips which associate trip points
235 to temperature channels.
243 temp[1-*]_type Sensor type selection.
252 Not all types are supported by all chips
254 temp[1-*]_max Temperature max value.
255 Unit: millidegree Celsius (or millivolt, see below)
258 temp[1-*]_min Temperature min value.
259 Unit: millidegree Celsius
263 Temperature hysteresis value for max limit.
264 Unit: millidegree Celsius
265 Must be reported as an absolute temperature, NOT a delta
269 temp[1-*]_input Temperature input value.
270 Unit: millidegree Celsius
273 temp[1-*]_crit Temperature critical value, typically greater than
274 corresponding temp_max values.
275 Unit: millidegree Celsius
279 Temperature hysteresis value for critical limit.
280 Unit: millidegree Celsius
281 Must be reported as an absolute temperature, NOT a delta
282 from the critical value.
286 Temperature offset which is added to the temperature reading
288 Unit: millidegree Celsius
291 temp[1-*]_label Suggested temperature channel label.
293 Should only be created if the driver has hints about what
294 this temperature channel is being used for, and user-space
295 doesn't. In all other cases, the label is provided by
300 Historical minimum temperature
301 Unit: millidegree Celsius
305 Historical maximum temperature
306 Unit: millidegree Celsius
309 temp[1-*]_reset_history
310 Reset temp_lowest and temp_highest
314 Reset temp_lowest and temp_highest for all sensors
317 Some chips measure temperature using external thermistors and an ADC, and
318 report the temperature measurement as a voltage. Converting this voltage
319 back to a temperature (or the other way around for limits) requires
320 mathematical functions not available in the kernel, so the conversion
321 must occur in user space. For these chips, all temp* files described
322 above should contain values expressed in millivolt instead of millidegree
323 Celsius. In other words, such temperature channels are handled as voltage
324 channels by the driver.
326 Also see the Alarms section for status flags associated with temperatures.
333 Note that no known chip provides current measurements as of writing,
334 so this part is theoretical, so to say.
336 curr[1-*]_max Current max value
340 curr[1-*]_min Current min value.
344 curr[1-*]_input Current input value
352 power[1-*]_average Average power use
356 power[1-*]_average_interval Power use averaging interval. A poll
357 notification is sent to this file if the
358 hardware changes the averaging interval.
362 power[1-*]_average_interval_max Maximum power use averaging interval
366 power[1-*]_average_interval_min Minimum power use averaging interval
370 power[1-*]_average_highest Historical average maximum power use
374 power[1-*]_average_lowest Historical average minimum power use
378 power[1-*]_average_max A poll notification is sent to
379 power[1-*]_average when power use
380 rises above this value.
384 power[1-*]_average_min A poll notification is sent to
385 power[1-*]_average when power use
386 sinks below this value.
390 power[1-*]_input Instantaneous power use
394 power[1-*]_input_highest Historical maximum power use
398 power[1-*]_input_lowest Historical minimum power use
402 power[1-*]_reset_history Reset input_highest, input_lowest,
403 average_highest and average_lowest.
406 power[1-*]_accuracy Accuracy of the power meter.
410 power[1-*]_alarm 1 if the system is drawing more power than the
411 cap allows; 0 otherwise. A poll notification is
412 sent to this file when the power use exceeds the
413 cap. This file only appears if the cap is known
414 to be enforced by hardware.
417 power[1-*]_cap If power use rises above this limit, the
418 system should take action to reduce power use.
419 A poll notification is sent to this file if the
420 cap is changed by the hardware. The *_cap
421 files only appear if the cap is known to be
422 enforced by hardware.
426 power[1-*]_cap_hyst Margin of hysteresis built around capping and
431 power[1-*]_cap_max Maximum cap that can be set.
435 power[1-*]_cap_min Minimum cap that can be set.
443 energy[1-*]_input Cumulative energy use
452 Each channel or limit may have an associated alarm file, containing a
453 boolean value. 1 means than an alarm condition exists, 0 means no alarm.
455 Usually a given chip will either use channel-related alarms, or
456 limit-related alarms, not both. The driver should just reflect the hardware
481 Each input channel may have an associated fault file. This can be used
482 to notify open diodes, unconnected fans etc. where the hardware
483 supports it. When this boolean has value 1, the measurement for that
484 channel should not be trusted.
489 Input fault condition
494 Some chips also offer the possibility to get beeped when an alarm occurs:
496 beep_enable Master beep enable
509 In theory, a chip could provide per-limit beep masking, but no such chip
512 Old drivers provided a different, non-standard interface to alarms and
513 beeps. These interface files are deprecated, but will be kept around
514 for compatibility reasons:
516 alarms Alarm bitmask.
518 Integer representation of one to four bytes.
519 A '1' bit means an alarm.
520 Chips should be programmed for 'comparator' mode so that
521 the alarm will 'come back' after you read the register
522 if it is still valid.
523 Generally a direct representation of a chip's internal
524 alarm registers; there is no standard for the position
525 of individual bits. For this reason, the use of this
526 interface file for new drivers is discouraged. Use
527 individual *_alarm and *_fault files instead.
528 Bits are defined in kernel/include/sensors.h.
530 beep_mask Bitmask for beep.
531 Same format as 'alarms' with the same bit locations,
532 use discouraged for the same reason. Use individual
533 *_beep files instead.
537 ***********************
538 * Intrusion detection *
539 ***********************
542 Chassis intrusion detection
544 1: intrusion detected
546 Contrary to regular alarm flags which clear themselves
547 automatically when read, this one sticks until cleared by
548 the user. This is done by writing 0 to the file. Writing
549 other values is unsupported.
552 Chassis intrusion beep
558 sysfs attribute writes interpretation
559 -------------------------------------
561 hwmon sysfs attributes always contain numbers, so the first thing to do is to
562 convert the input to a number, there are 2 ways todo this depending whether
563 the number can be negative or not:
564 unsigned long u = simple_strtoul(buf, NULL, 10);
565 long s = simple_strtol(buf, NULL, 10);
567 With buf being the buffer with the user input being passed by the kernel.
568 Notice that we do not use the second argument of strto[u]l, and thus cannot
569 tell when 0 is returned, if this was really 0 or is caused by invalid input.
570 This is done deliberately as checking this everywhere would add a lot of
573 Notice that it is important to always store the converted value in an
574 unsigned long or long, so that no wrap around can happen before any further
577 After the input string is converted to an (unsigned) long, the value should be
578 checked if its acceptable. Be careful with further conversions on the value
579 before checking it for validity, as these conversions could still cause a wrap
580 around before the check. For example do not multiply the result, and only
581 add/subtract if it has been divided before the add/subtract.
583 What to do if a value is found to be invalid, depends on the type of the
584 sysfs attribute that is being set. If it is a continuous setting like a
585 tempX_max or inX_max attribute, then the value should be clamped to its
586 limits using SENSORS_LIMIT(value, min_limit, max_limit). If it is not
587 continuous like for example a tempX_type, then when an invalid value is
588 written, -EINVAL should be returned.
590 Example1, temp1_max, register is a signed 8 bit value (-128 - 127 degrees):
592 long v = simple_strtol(buf, NULL, 10) / 1000;
593 v = SENSORS_LIMIT(v, -128, 127);
594 /* write v to register */
596 Example2, fan divider setting, valid values 2, 4 and 8:
598 unsigned long v = simple_strtoul(buf, NULL, 10);
601 case 2: v = 1; break;
602 case 4: v = 2; break;
603 case 8: v = 3; break;
607 /* write v to register */