| Adam Baker | 630300d | 2016-03-05 15:34:56 +0000 | [diff] [blame] | 1 | Kernel driver nsa320_hwmon |
| 2 | ========================== |
| 3 | |
| 4 | Supported chips: |
| 5 | * Holtek HT46R065 microcontroller with onboard firmware that configures |
| 6 | it to act as a hardware monitor. |
| 7 | Prefix: 'nsa320' |
| 8 | Addresses scanned: none |
| 9 | Datasheet: Not available, driver was reverse engineered based upon the |
| 10 | Zyxel kernel source |
| 11 | |
| 12 | Author: |
| 13 | Adam Baker <linux@baker-net.org.uk> |
| 14 | |
| 15 | Description |
| 16 | ----------- |
| 17 | |
| 18 | This chip is known to be used in the Zyxel NSA320 and NSA325 NAS Units and |
| 19 | also in some variants of the NSA310 but the driver has only been tested |
| 20 | on the NSA320. In all of these devices it is connected to the same 3 GPIO |
| 21 | lines which are used to provide chip select, clock and data lines. The |
| 22 | interface behaves similarly to SPI but at much lower speeds than are normally |
| 23 | used for SPI. |
| 24 | |
| 25 | Following each chip select pulse the chip will generate a single 32 bit word |
| 26 | that contains 0x55 as a marker to indicate that data is being read correctly, |
| 27 | followed by an 8 bit fan speed in 100s of RPM and a 16 bit temperature in |
| 28 | tenths of a degree. |
| 29 | |
| 30 | |
| 31 | sysfs-Interface |
| 32 | --------------- |
| 33 | |
| 34 | temp1_input - temperature input |
| 35 | fan1_input - fan speed |
| 36 | |
| 37 | Notes |
| 38 | ----- |
| 39 | |
| 40 | The access timings used in the driver are the same as used in the Zyxel |
| 41 | provided kernel. Testing has shown that if the delay between chip select and |
| 42 | the first clock pulse is reduced from 100 ms to just under 10ms then the chip |
| 43 | will not produce any output. If the duration of either phase of the clock |
| 44 | is reduced from 100 us to less than 15 us then data pulses are likely to be |
| 45 | read twice corrupting the output. The above analysis is based upon a sample |
| 46 | of one unit but suggests that the Zyxel provided delay values include a |
| 47 | reasonable tolerance. |
| 48 | |
| 49 | The driver incorporates a limit that it will not check for updated values |
| 50 | faster than once a second. This is because the hardware takes a relatively long |
| 51 | time to read the data from the device and when it does it reads both temp and |
| 52 | fan speed. As the most likely case for two accesses in quick succession is |
| 53 | to read both of these values avoiding a second read delay is desirable. |