| Rudolf Marek | 61db011 | 2006-12-12 18:18:30 +0100 | [diff] [blame] | 1 | Kernel driver w83793 | 
|  | 2 | ==================== | 
|  | 3 |  | 
|  | 4 | Supported chips: | 
|  | 5 | * Winbond W83793G/W83793R | 
|  | 6 | Prefix: 'w83793' | 
|  | 7 | Addresses scanned: I2C 0x2c - 0x2f | 
|  | 8 | Datasheet: Still not published | 
|  | 9 |  | 
|  | 10 | Authors: | 
|  | 11 | Yuan Mu (Winbond Electronics) | 
|  | 12 | Rudolf Marek <r.marek@assembler.cz> | 
|  | 13 |  | 
|  | 14 |  | 
|  | 15 | Module parameters | 
|  | 16 | ----------------- | 
|  | 17 |  | 
|  | 18 | * reset int | 
|  | 19 | (default 0) | 
|  | 20 | This parameter is not recommended, it will lose motherboard specific | 
|  | 21 | settings. Use 'reset=1' to reset the chip when loading this module. | 
|  | 22 |  | 
|  | 23 | * force_subclients=bus,caddr,saddr1,saddr2 | 
|  | 24 | This is used to force the i2c addresses for subclients of | 
|  | 25 | a certain chip. Typical usage is `force_subclients=0,0x2f,0x4a,0x4b' | 
|  | 26 | to force the subclients of chip 0x2f on bus 0 to i2c addresses | 
|  | 27 | 0x4a and 0x4b. | 
|  | 28 |  | 
|  | 29 |  | 
|  | 30 | Description | 
|  | 31 | ----------- | 
|  | 32 |  | 
|  | 33 | This driver implements support for Winbond W83793G/W83793R chips. | 
|  | 34 |  | 
|  | 35 | * Exported features | 
|  | 36 | This driver exports 10 voltage sensors, up to 12 fan tachometer inputs, | 
|  | 37 | 6 remote temperatures, up to 8 sets of PWM fan controls, SmartFan | 
|  | 38 | (automatic fan speed control) on all temperature/PWM combinations, 2 | 
|  | 39 | sets of 6-pin CPU VID input. | 
|  | 40 |  | 
|  | 41 | * Sensor resolutions | 
|  | 42 | If your motherboard maker used the reference design, the resolution of | 
|  | 43 | voltage0-2 is 2mV, resolution of voltage3/4/5 is 16mV, 8mV for voltage6, | 
|  | 44 | 24mV for voltage7/8. Temp1-4 have a 0.25 degree Celsius resolution, | 
|  | 45 | temp5-6 have a 1 degree Celsiis resolution. | 
|  | 46 |  | 
|  | 47 | * Temperature sensor types | 
| Gong Jun | 46bed4d | 2007-01-18 22:14:24 +0100 | [diff] [blame] | 48 | Temp1-4 have 2 possible types. It can be read from (and written to) | 
| Rudolf Marek | 61db011 | 2006-12-12 18:18:30 +0100 | [diff] [blame] | 49 | temp[1-4]_type. | 
| Rudolf Marek | 61db011 | 2006-12-12 18:18:30 +0100 | [diff] [blame] | 50 | - If the value is 3, it starts monitoring using a remote termal diode | 
|  | 51 | (default). | 
| Rudolf Marek | 61db011 | 2006-12-12 18:18:30 +0100 | [diff] [blame] | 52 | - If the value is 6, it starts monitoring using the temperature sensor | 
|  | 53 | in Intel CPU and get result by PECI. | 
|  | 54 | Temp5-6 can be connected to external thermistors (value of | 
| Gong Jun | 46bed4d | 2007-01-18 22:14:24 +0100 | [diff] [blame] | 55 | temp[5-6]_type is 4). | 
| Rudolf Marek | 61db011 | 2006-12-12 18:18:30 +0100 | [diff] [blame] | 56 |  | 
|  | 57 | * Alarm mechanism | 
|  | 58 | For voltage sensors, an alarm triggers if the measured value is below | 
|  | 59 | the low voltage limit or over the high voltage limit. | 
|  | 60 | For temperature sensors, an alarm triggers if the measured value goes | 
|  | 61 | above the high temperature limit, and wears off only after the measured | 
|  | 62 | value drops below the hysteresis value. | 
|  | 63 | For fan sensors, an alarm triggers if the measured value is below the | 
|  | 64 | low speed limit. | 
|  | 65 |  | 
|  | 66 | * SmartFan/PWM control | 
|  | 67 | If you want to set a pwm fan to manual mode, you just need to make sure it | 
|  | 68 | is not controlled by any temp channel, for example, you want to set fan1 | 
|  | 69 | to manual mode, you need to check the value of temp[1-6]_fan_map, make | 
|  | 70 | sure bit 0 is cleared in the 6 values. And then set the pwm1 value to | 
|  | 71 | control the fan. | 
|  | 72 |  | 
|  | 73 | Each temperature channel can control all the 8 PWM outputs (by setting the | 
|  | 74 | corresponding bit in tempX_fan_map), you can set the temperature channel | 
|  | 75 | mode using temp[1-6]_pwm_enable, 2 is Thermal Cruise mode and 3 | 
|  | 76 | is the SmartFanII mode. Temperature channels will try to speed up or | 
|  | 77 | slow down all controlled fans, this means one fan can receive different | 
|  | 78 | PWM value requests from different temperature channels, but the chip | 
|  | 79 | will always pick the safest (max) PWM value for each fan. | 
|  | 80 |  | 
|  | 81 | In Thermal Cruise mode, the chip attempts to keep the temperature at a | 
|  | 82 | predefined value, within a tolerance margin. So if tempX_input > | 
|  | 83 | thermal_cruiseX + toleranceX, the chip will increase the PWM value, | 
|  | 84 | if tempX_input < thermal_cruiseX - toleranceX, the chip will decrease | 
|  | 85 | the PWM value. If the temperature is within the tolerance range, the PWM | 
|  | 86 | value is left unchanged. | 
|  | 87 |  | 
|  | 88 | SmartFanII works differently, you have to define up to 7 PWM, temperature | 
|  | 89 | trip points, defining a PWM/temperature curve which the chip will follow. | 
|  | 90 | While not fundamentally different from the Thermal Cruise mode, the | 
|  | 91 | implementation is quite different, giving you a finer-grained control. | 
|  | 92 |  | 
|  | 93 | * Chassis | 
|  | 94 | If the case open alarm triggers, it will stay in this state unless cleared | 
| Jean Delvare | a516dc3 | 2011-01-12 21:55:10 +0100 | [diff] [blame] | 95 | by writing 0 to the sysfs file "intrusion0_alarm". | 
| Rudolf Marek | 61db011 | 2006-12-12 18:18:30 +0100 | [diff] [blame] | 96 |  | 
|  | 97 | * VID and VRM | 
|  | 98 | The VRM version is detected automatically, don't modify the it unless you | 
|  | 99 | *do* know the cpu VRM version and it's not properly detected. | 
|  | 100 |  | 
|  | 101 |  | 
|  | 102 | Notes | 
|  | 103 | ----- | 
|  | 104 |  | 
|  | 105 | Only Fan1-5 and PWM1-3 are guaranteed to always exist, other fan inputs and | 
|  | 106 | PWM outputs may or may not exist depending on the chip pin configuration. |