Thread (20 messages) 20 messages, 7 authors, 2022-02-06

Re: Touchpad stickiness on AMD laptops (was Dell Inspiron/XPS)

From: Hans de Goede <hidden>
Date: 2022-01-17 09:08:55
Also in: linux-i2c, platform-driver-x86

Hi,

On 1/17/22 09:39, Miroslav Bendík wrote:
quoted
[AMD Official Use Only]
quoted
Now i am trying to change ASF registers instead of SMBus registers.
I have tried to enable interrupts and set listen address, but it don't
work or
i can't recognize the difference between interrupts generated by
transfers and
interrupts generated from slave.
Try reading the value of SFx0A ASFStatus bit 5 (it's write to clear if it's an interrupt).
quoted
outb_p(0x02, 0x15 + piix4_smba); // SlaveIntrListenEn
outb_p(0x2c << 1 | 0x01, 0x09 + piix4_smba); // ListenAdr | ListenAdrEn
ASFx04 SlaveAddress instead of  ASFx09 ListenAdr
?
Little bit more informations:

Interrupts are generated only if ASFx09 ListenAdr is:

(0x08 << 1) | 0x01
(0x10 << 1) | 0x01

and touchpad is initialized with synaptics_intertouch=1

There is maybe small correlation between frequency and touch, but i am
not 100% sure.
I know very litlle about this, but I believe that when using
host-notify that after receiving the host-notify you are supposed to
do an I2C read from the SMBus Alert Response Address (ARA, 0x0c) to find
out the source of the notify (since multiple devices on the bus may
be notify capable). I guess that the controller may not do that itself
and that as long as you have not done it the touchpad may keep repeating
the notify.

But as said I know very little about this, so take this with a big
grain of salt :)  I guess you may want to read up a bit on how this
is supposed to work at the bus level. I believe that the SMBUS spec
is public.

Regards,

Hans



There are no register changed in interrupt handler except of
ASFx13 DataBankSel. I can't determine if interrupt is generated from
transfer, or from external event.

ASF should be system for remote management. It should have access to
SMBus and data / command registers are identical, this means, that SMBus
should work (except block transfers).

If ASF just mirrors SMBus, then question is, why i can't access to
touchpad using SMBus? One strange thing is, that i2cdetect on standard
SMbus (0xb00), port 0 returns:

     0  1  2  3  4  5  6  7  8  9  a  b  c  d  e  f
00:                         -- -- -- -- -- -- -- --
10: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
20: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
30: -- -- -- -- -- -- 36 37 -- -- -- -- -- -- -- --
40: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
50: 50 -- -- -- -- -- -- -- 58 -- -- -- -- -- -- --
60: -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
70: -- -- -- -- -- -- -- --

Address 0x58 is exactly 0x2c (synaptics) moved 1 bit left, but i2c-piix4
correctly moves address.
  
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help