[Openec] Ladies and Gents, we have a "brick with a monitor"

[Frieder Ferlemann]

Hi Richard,

Richard A. Smith schrieb:
> Frieder Ferlemann wrote:
> 
>> - the UART RX is not working here if WLAN is not powered.
>>   RX/BAT_LO (CN24,3 CON2,45) seems clamped at about 2V
>>   if WLAN is powered off.
>>   Is that known and is CON2,45 within absolute maximum ratings?
> 
> On your B1 BAT_L0 goes to an input on the WLAN module.  If wlan is not 
> powered then you are driving an un-powered pin.  So your results are 
> somewhat expected.  That signal moved in later revs.

Good:)

> You will need to enable WLAN or you won't get very far.  WLAN enables 
> the source for the 3.3V rail.  Most of the other voltage rails come from 
> the 3.3V rail.

Yes, seems either enable or leaving the pins untouched will do.


There is another one like that too. Citing from line 86 of uart.c

  /*                           Warning KEY_OUT_3 is also used for ISP_CLK
     so if the serial debricking adapter is connected output drivers work
     against each others with a duty cycle of 1/6 */

I'm ignoring this:)   It spoils reading the cursor keys while
the debricking adapter is connected. (Well that is always here:)
(If the debricking adapter is to be used together with the cursor keys
one could probably spend a condensator and a ferrite to have an
AC path to the external clock and a rather DC path to the keyboard
column. But it's probably not worth it.


>> - the ec-dump.fth like dump locks near the output of the
>>   content of 0xff50.
> 
> Interesting.  0xff50 is not documented as anything.  Its in the general 
> purpose wakeup section.  Perhaps you are tickling some undocumented 
> register that crashes things?

I'm not sure it hangs on something specific to 0xff50.
0xff50 reads fine with mff50 and then <RETURN> also it hangs after it
has output the first 0 (so a little while after the access has happened)
GPWUPS ff50:  0

> On my C1 (KB3700B) and a B4 I can read 0xff50 with indexed IO via OFW 
> and the EC still responds to battery insertion/removal.  Not sure how 
> similar that is to a read via the 8051.
> 
>> With the monitor it might be possible to power up the XO?
> 
> I think so.  When you get it right you should see some post codes.  The 
> EC can decode post codes so you can watch for those.

I'm currently hesitating to apply power (I did so not by manually poking
the registers but by the code in power.c).
Got some strange noises from the switching regulators. So I decided
to bring the code into a somewhat useful state and then risk the XO
again.

Greetings,

Frieder