What can I do with my semi-functional PW3 motherboard to help people out?

[knc1]

That is the "blank disk" mode of the export system.
It just means that the start-up system hasn't completed the USB export (for some reason) -
I.E: It is probably waiting for something to complete that can not with the screen disconnected.

In that case, you will probably have to investigate where systemd is hung up, waiting for something to complete.

[coplate]

Originally Posted by knc1

That is the "blank disk" mode of the export system.
It just means that the start-up system hasn't completed the USB export (for some reason) -
I.E: It is probably waiting for something to complete that can not with the screen disconnected.

In that case, you will probably have to investigate where systemd is hung up, waiting for something to complete.

Yeah, I was adding some edit while you were replying.

i'll get some more useful info when the serial cable gets here hopefully.

I would like to be able to run things headless for research.

[knc1]

Originally Posted by coplate

i'll get some more useful info when the serial cable gets here hopefully.

And on a thread nearby, some of our inexperienced members are writing about replacing the Amazon OS with another OS on a device without an accessible serial port (not sure they have even read the acronym: JTAG).

[coplate]

You know, this pw3 motherboard has 1 pin labelled jtag, but I'm pretty sure that's not enough. But I'm not an expert :-)

[knc1]

Originally Posted by coplate

You know, this pw3 motherboard has 1 pin labelled jtag, but I'm pretty sure that's not enough. But I'm not an expert :-)

geekmaster wrote (somewhere here) that there is a new 'standard' JTAG interface pin-out that minimizes the number of pins.
IIRC, the previous 'standard' was five (or seven or anyway a lot more than one ). *

But just one pin does sound a bit minimal, even if you pick up a ground connection from the somewhere on the board.

Hmm...
But the JTAG port could be wired to a "Dallas one-wire" interface chip, I am pretty sure they (the "Dallas one-wire" people) make such a device.
Then one pin (for anything) makes sense.

Note:
The Linux kernel has supported the "Dallas one-wire" devices for ages now.

Edit: (*)
Just had one of the old JTAG cables fall into my lap - a "JTAG-3 with SPI" cable.
Pins: TMS, TDI, TDO, TCK, GND, VDD (six pins).
And the other end of the cable is for a parallel port.

I don't think I still own a (working) computer with a parallel port.
So don't even ask me why I am keeping this adapter cable.

[Cinisajoy]

Originally Posted by knc1

geekmaster wrote (somewhere here) that there is a new 'standard' JTAG interface pin-out that minimizes the number of pins.
IIRC, the previous 'standard' was five (or seven or anyway a lot more than one ). *

But just one pin does sound a bit minimal, even if you pick up a ground connection from the somewhere on the board.

Hmm...
But the JTAG port could be wired to a "Dallas one-wire" interface chip, I am pretty sure they (the "Dallas one-wire" people) make such a device.
Then one pin (for anything) makes sense.

Note:
The Linux kernel has supported the "Dallas one-wire" devices for ages now.

Edit: (*)
Just had one of the old JTAG cables fall into my lap - a "JTAG-3 with SPI" cable.
Pins: TMS, TDI, TDO, TCK, GND, VDD (six pins).
And the other end of the cable is for a parallel port.

I don't think I still own a (working) computer with a parallel port.
So don't even ask me why I am keeping this adapter cable.

Because someday, you may need it.

[knc1]

Originally Posted by Cinisajoy

Because someday, you may need it.

Same reason you keep every recipe you meet?
I suppose that is reason enough.

[coplate]

I may not be able to figure out wether it is a 1 wire jtag port, but I do find it really interesting:

I think I can tell with good confidence which pads on the device have been used on thier test bed. They have a tiny dot right in the center from the machine.

( several of them are under the RF shields )

Most are labelled 'TP', TX', or 'TM'

[coplate]

Well, I was solderign it up today, and while I was trying to put some elctrical tape in place today, I pulled the pads off of rx and tx.

I'm going to try to see if it works with foam and pins like the solderless method, if I can still connect to the contacts, but I'll have to position very carfully now.

[coplate]

OK, update. While I did let the magic smoke out of the 2v uart I bough, I was able to use a 3.3v uart from my other device to read the TX pin.

I'm not prepared to try to hook that up to the RX pin, all the warning about burning out the kindle too for over voltage :-)

But I though it would be fine for reading from the kindle.
So without the screen, it boots up, and the last line before a kernel dump is:

[    0.530360] Bad flash signature: 0x0
Press [ENTER] for recovery menu...       0 /
=== kernel printk buffer (23556 bytes) ===

It then moves into a progress bar type thing that looks like this, but much slower, I only waited about 15 seconds, because I have to manually hold the probe on the TX pin:

---\
----|
-----/

On the boot log with the screen plugged in, this is instead where it asks

Press [ENTER] for recovery menu...       0 //dev/mmcblk0p4:
CHS=4/16/99776 size=3269459968 bytes
flag type      first       last  lba first   lba size
Partition p1:
0x00 0x0b         16 <large>            16    6385648
    CHS: 0/1/1 - <large>
Partition p2:
Partition p3:
Partition p4:
BOOTING DEFAULT.

So my next step if I keep working with this will be to get another 1.8v chip, and see if I can connect to RX and TX at the same time, and actually interrupt one of those boots.

[knc1]

It is detecting what is assumed to be an eMMC error.
The progress thingy is it attempting to reformat and/or repair the file system.

But since the errors being detected are more likely leakage between the BGA pins -
I would not hold out much hope for it to recover.

Clearances under that chip and between the balls are very small, they would have wicked up moisture quickly. Even being under the RFI shields.
It would have been difficult to get the moisture back out, or to displace it with alcohol.

- - - - -

Although I would have bet that this happened to the SoC first.
(More connections, tighter spaces, etc. - maybe it was under a tighter RFI shield.)

[coplate]

Edit to clarify: the device boots fine if I have the screen plugged in, but not if I have the screen unplugged.



The log makes it look like that, but if I plug the screen in, it boots up no problem, and doesn't have that error, it's the darndest thing.

I assumed the bad flash signature was it checking authenticity of the eink panel, there some chips on that, including the waveform storage chip, yeah?

[knc1]

Originally Posted by coplate

Edit to clarify: the device boots fine if I have the screen plugged in, but not if I have the screen unplugged.



The log makes it look like that, but if I plug the screen in, it boots up no problem, and doesn't have that error, it's the darndest thing.

I assumed the bad flash signature was it checking authenticity of the eink panel, there some chips on that, including the waveform storage chip, yeah?

That is another thing it does during boot-up . . . .

Checking and re-programming as required, the waveform stored in that chip, the chip that is usually on the cable but might be on the panel itself on your model.
(I don't have a PW-2 myself.)

One memory error looks like any other when we have the documentation that Amazon provides (none).

[coplate]

Since it was already broken, i went ahead and used my 3.3v uart to connect to it, it didn't explode, but I know it's a bad idea. :-)

So any children listening at home: "Don't try this."

I was able to prop a needle in place for the UART RX pin, but I have to hold the TX pin in place, since the contact really is only the size of a pinhead after I ripped the pads off.

I am able to interrupt this:

Hit any key to stop autoboot:  0
uboot >

But the other option to enter recovery mode still crashes without a screen, though I confirmed it does work with a screen.

Press [ENTER] for recovery menu...       

=== kernel printk buffer (23556 bytes) ===
<5>[    0.000000] Linux version 3.0.35-lab126 (jenkins-official@sjc10-jbuild11) (collect2: error: ld returned 1 exit status) #1 PREEMPT Fri Jun 3 00:56:12 PDT 2016
<4>[    0.000000] CPU: ARMv7 Processor [412fc09a] revision 10 (ARMv7), cr=10c53c7d

etc.

So at this point, it does not appear that I can make this a headless tool to play with, so I am going to be putting the screen back together with the replacement logic board I bought, and putting this with the old water damaged battery, in case I can score a broken screen that has a good firmware on the cable - this weekend probably.

If anyone has anything they really want me to do this week, throw out ideas, and I'll see if I can figure it out.

[encol]

Any progress on the USB Boot switch?