[Arm-netbook] [EOMA68] RS232 on SoCs

[joem]

On Tue, 2013-09-17 at 07:11 +0000, joem wrote:
> On Mon, 2013-09-16 at 21:43 +0200, Henrik Nordström wrote:
> > mån 2013-09-16 klockan 15:47 +0000 skrev joem:
> > 
> > > I can verify something like that is happening.
> > > Haven't found the exact cause in hardware yet.
> > > The symptoms seem to be spinning itself around the mmc card.
> > 
> > It's not only SD card. Actually I haven't had much issues with SD cards
> > related to this, only AXP, resulting in the AXP powerig up but not
> > responding to i2c requests.
> > 
> > > The simplest suggestion I can make is to put 330R between the +3.3V
> > > rail and gnd of the mmc. But I have not tried it yet to
> > > know whether it can work.
> > 
> > Maybe. Guess it depends on the resistance in the UART dongle and also
> > UART0 RX -> VCC3_3 internal resistance at power off. But will cost a bit
> > of additional power drain during normal usage.
> 
> The power drain is 10mA - about same as having an extra LED on.
> It should be enough to kill any voltages that are outstanding
> if its being fed from 3 pins of RS232 interface.
> 
> I'll get a chance to try out a few hardware checks this evening and
> hopefully get more clues.


Did more checking last night and its not the MMC from an electrical
point of view.

The problem is like this.

Power on - the device boots ok.
Power off and then power on - mmc can't be read
Power off and then power on - mmc can be read no problem 
Power off and then power on - mmc can't be read
Power off and then power on - mmc can be read no problem
Power off and then power on - mmc can't be read

A long delay between power off and on makes no difference to problem.

It appears may be something gets written somewhere so that
every alternative boot attempt fails because of remembering something in
flash or EEPROM?