[ltp] getting hdaps up and running on the w530

[Саша Питосин]

On 4/14/14 3:19 PM, Raider Sail wrote:

> 1. How do we interface with the APS hardware via the new embedded controller on the W530?

Hello,

I think I have made some progress with latest ThinkPad accelerometer, 
observing I/O port communications between the Windows Driver and hardware.

Only ports 0x1610 and 0x1611 are used, using an "indirect" addressing mode:
  - writing to 0x1610 selects the register to be accessed
  - reading (or writing) 0x1611 reads (or writes) the chosen register.

It appears that some operations can take a while, and you have to poll 
register 0x00 until it return zero.

Readings are delivered through a FIFO (registers 0x10-0x2f), each entry 
containing:
  - X axis (2 bytes, little-endian)
  - Y axis (2 bytes, little-endian)
  - One extra byte (maybe related to temperature?)

Before accessing the FIFO, 0xff is always written to register 0x01 
(latch?), and 0x00 is written to the same register, maybe to release it.

Getting new readings involves waiting for the accelerometer to be ready, 
locking the FIFO, writing 0x11 to register 0x00 and waiting again. 
Register 0x10 will then contain the number of available FIFO entries 
(between 0 and 5), followed by the 5-byte records described above in 
registers 0x11 to (0x11 + 5 * num. of records)-1. Then register 0x2f 
returns zero and the FIFO can be released.

An example of a reading sequence:
     0x1610 <-- 0x00
     0x1611 --> 0x00 # register 0x00 returns 0x00
     0x1610 <-- 0x01
     0x1611 <-- 0xff # register 0x01 is written 0xff
     0x1610 <-- 0x00
     0x1611 <-- 0x11 # register 0x00 is written 0x11
     0x1610 <-- 0x00
     0x1611 --> 0x11 # register 0x00 returns 0x11 (not ready yet - polling!)
     0x1611 --> 0x00 # register 0x00 returns 0x00
     0x1610 <-- 0x10
     0x1611 --> 0x01 # register 0x10 returns 0x01
     0x1610 <-- 0x11
     0x1611 --> 0x11 # first data byte
     0x1610 <-- 0x12
     0x1611 --> 0x02 # second data byte
     0x1610 <-- 0x13
     0x1611 --> 0x9a # ...
     0x1610 <-- 0x14
     0x1611 --> 0x02
     0x1610 <-- 0x2a
     0x1611 --> 0x00
     0x1610 <-- 0x2b
     0x1611 --> 0x80
     0x1610 <-- 0x2f
     0x1611 --> 0x00 # register 0x2f returns 0x00
     0x1610 <-- 0x01
     0x1611 <-- 0x00 # register 0x01 is written 0x00

At boot, the following registers are configured:
     0x1610 <-- 0x00
     0x1611 --> 0x00 # register 0x00 returns 0x00
     0x1610 <-- 0x01
     0x1611 <-- 0xff # register 0x01 is written 0xff
     0x1610 <-- 0x10
     0x1611 <-- 0x81 # register 0x10 is written 0x81
     0x1610 <-- 0x00
     0x1611 <-- 0x17
     0x1610 <-- 0x00
     0x1611 --> 0x00
     0x1610 <-- 0x10
     0x1611 --> 0x01
     0x1610 <-- 0x11
     0x1611 --> 0x00
     0x1610 <-- 0x12
     0x1611 --> 0x00
     0x1610 <-- 0x13
     0x1611 --> 0x00
     0x1610 <-- 0x2f
     0x1611 --> 0x00
     0x1610 <-- 0x01
     0x1611 <-- 0x00 # register 0x01 is written 0x00
     0x1610 <-- 0x00
     0x1611 --> 0x00 # register 0x00 returns 0x00
     0x1610 <-- 0x01
     0x1611 <-- 0xff # register 0x01 is written 0xff
     0x1610 <-- 0x10
     0x1611 <-- 0x01 # register 0x10 is written 0x01
     0x1610 <-- 0x00
     0x1611 <-- 0x14
     0x1610 <-- 0x00
     0x1611 --> 0x00
     0x1610 <-- 0x10
     0x1611 --> 0x01
     0x1610 <-- 0x2f
     0x1611 --> 0x00
     0x1610 <-- 0x01
     0x1611 <-- 0x00 # register 0x01 is written 0x00
     0x1610 <-- 0x00
     0x1611 --> 0x00 # register 0x00 returns 0x00
     0x1610 <-- 0x01
     0x1611 <-- 0xff # register 0x01 is written 0xff
     0x1610 <-- 0x10
     0x1611 <-- 0xc8 # register 0x10 is written 0xc8
     0x1610 <-- 0x11
     0x1611 <-- 0x00
     0x1610 <-- 0x12
     0x1611 <-- 0x02
     0x1610 <-- 0x00
     0x1611 <-- 0x10 # register 0x00 is written 0x10
     0x1610 <-- 0x00
     0x1611 --> 0x10 # register 0x00 returns 0x10 (not ready yet - polling!)
     0x1611 --> 0x00 # register 0x00 returns 0x00
     0x1610 <-- 0x10
     0x1611 --> 0xc8
     0x1610 <-- 0x2f
     0x1611 --> 0x00
     0x1610 <-- 0x01
     0x1611 <-- 0x00 # register 0x01 is written 0x00


Regards,

Sasha