Boot loop - what's wrong?

[geekmaster]

On another note, the dead battery I removed from the PW3 now has its amazon label peeled off. Inside is a plastic and metal frame, holding a small circuit board (with the tabs that connect to the kindle battery socket). The PCB has circuitry that can be reused. The foil-pouch encased LiPo battery taped inside the frame is labelled "Sony 27W" (and more text). It only was two wires (labelled + and -) and any standard 1S (single cell) LiPo should work connected in its place, I think -- even one swiped from a cheap rechargeable device like a remote control toy. And the PCB should still identify to the kindle as a genuine amazon battery. I will see if that is so when I find a LiPo battery to stuff in the battery pack. But for now, the good (identical) PW2 battery pack is in my PW3, and the PW2 still acts stone cold dead. It may be worth swapping screens too, for testing. The two devices look identical in all ways except minor labelling (which does not indicate which generation device it is).

[geekmaster]

Originally Posted by shalym

Yeah...this is definitely NOT how an unregistered PW behaves (nor does any other Kindle behave like that, as far as I know...)

Shari

Now that it has a good battery installed, I need to capture boot messages from the serial port. It boots consistently to the (borked) desktop now, so the logs should certainly be different. I suspect that it has some wonky flag set in /var/local, and that needs to be examined and possibly repaired (but I thought a factory reset did exactly that).

Another search for a USB serial adapter is imminent (after dinner)...

Though I probably COULD configure another kindle (with a custom app) as a TTL serial adapter...

[knc1]

The PW-2 and PW-3 have different resolutions and different dimensions (in pixels).
The PW-3 has the same screen as the KV.

[geekmaster]

Easier to swap kindle motherboards than to swap screens, though almost the same effect. Interestingly, the PW3 boot image on a PW2 screen puts the boy in the lower right corner at first, then it centers the tree (chopping the top of the "kindle" logo), then lower right again, and now "Your kindle needs repair". Interestingly, the grayscales were good (at least for the kindle logo word).

I know that startup scripts make assumptions about screen resolution based on serial number prefix (model). So even though the screen works right and displays good (though cropped) text and images, the firmware does not attempt to support the lower resolution and aborts.

But this test told me that the eink display was good (other than a very bright pinprick where there is a small damage spot to the "frontlight" panel.

I hope this "your kindle needs repair" status fixes itself with its original screen restored. It would be tought to reset that flag in its current condition. I *really* need to figure out how to use that "USB BOOT" switch on the mobo. Now to test the PW2 with PW3 screen (probably still dead).

It should be possible to hack a firmware update to support different screen resolutions -- the only obvious difference between the PW2 and PW3 seems to be screen resolution.

[geekmaster]

PW2 still dead. display still shows last PW3 screen saver image. The only thing that works on the PW2 is the LED blinking when no battery installed.

And these boards have a ton of numbered test points on the back (no useful labels though). The probably sat on a pin bed fixture for testing...

PW3 boots up fine with its original screen. YKNR just meant it was unsatisfied with not finding enough pixels to satisfy its GUI layout...

And SHAZAM! Swapping screens made the Home menu appear! It pushed the "Cloud | On Device ..." menu down to make room for the home menu. It is much "more better" now -- EXCEPT the touchscreen is not working... That was a worthwhile experiment, with unexpectedly positive results, but now I have to get the touchscreen working again...

EDIT: Touch not working at all, even after several hard reboots. Even after reseating the touch cable several times (with battery removed). It worked fairly well, then intermittently, then not at all, then after swapping boards still not at all. But I got my menus back -- though I cannot touch and use them now... Not sure if that is progress or not...

[geekmaster]

Okay, a drop of DO_FACTORY_RESTORE, hard power reboot, at langauge select screen, touchscreen works again. Why all these factory resets needed? And... now we go direct to the screen "In order to use your Kindle ... you need to connect to a Wi-Fi network and register your device. No menu, no wayt out, only a "connect to wi-fi" link... But it DOES show our neighborhood access points, and touchscreen works. Another step forward, but still in that amazon registration sandbox that prevents doing ANYTHING on the kindle.

Ahh, AFTER the wifi scan, it gave me a "comple setup later" link in the lower right corner (as you described). No option to set up later until after scanning for wifi...

And now it all works. Menus, touchscreen, grayscale on eink. Time to update to a jailbreakable 5.6.5, and do the deed(s)...

If my time had any value, it would be cheaper to buy a new one. But no learning there. And learning keeps you young (or so I am trying to convince myself)...

[geekmaster]

Meh.. after going to USB drive mode, then back to kindle GUI, the touchscreen is dead again. Another factory restore flag file, and hard reboot. Language screen, touchscreen works. This seems like perhaps damaged firmware. Hopefully FW 5.6.5 will fix it. After al, the eink waveform was hosed when I first got it...

[knc1]

One of the PW-1 devices I have (purchased as: "broke - for parts") is totally bonkers at less that 80% battery charge.

It is only "normal" full charged - I have learned: Leave it plugged in to a USB hub (external power) - then it is always happy.

It almost seems like (something's) operating voltage tolerance and the voltage output tolerance of one of the voltage regulators (there are quite a few of them) are at opposite ends of there specifications.

The sort of thing that can happen when you toss a million "normal" parts in this bin and match them at random from another million "normal" parts in that other bin.

They just don't play nicely together, although both are 'in spec'.

[geekmaster]

I seem to be detecting a pattern here. The touchscreen works. I plug in USB and access the USB drive. I eject USB. The touchscreen does not work.

This time I will try a hard reset WITHOUT the factory restore flag file. I have the jailbreak files on it, but no way to get to experimental browser now...

Hmm.. this time a simple power off to screen saver, then back on made touchscreen work, though not at first. It took about 10 seconds to see my touch and react to it (when I was just about to hard reboot). Then responsive after that... Already better after firmware update, but still, touchscreen lockups after returning from USB storage mode are a regular occurance -- maybe a voltage reg thing as you suggested knc1?

[geekmaster]

I just did the 5.6.5 jailbreak from Branch Delay. Then I copied the hotfix to the USB drive. Then touchscreen dead. This time I am waiting to see if it wakes up. YES! It takes about a minute or so for touchscreen to work after ejecting USB. Need to sort that little problem out (annoying, but somewhat tolerable)...

EDIT: No, touchscreen not working again, before I could do UYK in settings... I will wait a bit, testing periodically, then hard reset if needed. BTW, those empty button pads on the mobo (power, reset, "magic" key) are connected to empty pads (a 3-pin device labelled CRnnn (nnn=unique number). Perhaps they do not route all the way to the SoC, with not only the switch missing. Perhaps the magic key is one of the other pins on that missing 'CR' device? --- Oh yeah, like before, no reset needed -- touchscreen works again after screensaver mode. Quick hack is after eject USB, touch power twice (in and out of screensaver mode)...

[geekmaster]

Hmm... Locked up again (or touchscreen dead, not sure which) while activating backdoorlock. Not just USB eject. Hmm... Screensaver and back woke it up again though...

Plenty of touchscreen lockups. Sometimes it returns. Sometimes tap power button twice. Not sure. But it works good enough for some purposes.

I added my "Geekmaster Video" extension to KUAL on my PW3, and Big Buck Bunny is playing just fine on my PW3. Awesome, considering it was developed for the K3, and then fighting the incompatible eink headers for the K4 and K5 (even support negative video on K4). I was a bit surprised to see it work with no changes on my PW1 (before it walked off), and I saw reports of it working on a PW2, but now I can confirm it works on PW3 as well (but only uses about a quarter of the pixels. This display sure has lots of extra pixels to work with -- even more than the DXG ... And BTW, I need to repack my videos. They are .gz files. Testing showed that repacking 60MB bunny.gmv.gz to bunny.gmv.bz2 makes it 30MB, and repacking to bunny.gmv.xz makes it 15MB. I can play .bz2 files on anything including my K1, but I cannot yet successfully build xz (or even a hello.c) for the k1. Nothing I tried works, not even using my K1 lib folder in my chroot... I see that I could do pixel doubling (double width and height of pre-dithered video) on the PW3 with very little lost at the right and left edge (well within overscan safe zones).

EDIT: I still never connected to TTL serial. I was not successful locating one of my many serial adapters (a side effect of putting projects away -- I can find anything when it is all laying out in plain sight, but wife likes it out of sight where I cannot see it nor find it). :O Thanks Branch Delay!

[geekmaster]

OMG. I just ran "titty" (tiny tty script) on my PW3 with its amazing resolution. I know that the titty text output on the eink display (with tiny font) is quite readable (at least to me, but on an 800x600 display like a K3 or K5). On this PW3 the text is so amazingly small I will need at least a 10x (jewelers loupe) lens to even read this text -- it is that small.

I needed to type the commands to pause and resume the framework (copies from my "geekmaster kindle video player" startup script, so that the framework would not corrupt my onscreen font with its title bar intrusion... It works great with a launcher script to do that, just like all my eink demo apps.

[geekmaster]

In conclusion, and to summarize this thread, the PW3 that was boot looping (plus a dead PW2) has been transferred to my possession (in exchange for a K5 3G).

The PW3 had a dead (bloated) LiPo battery, so I swapped in the good (identical) battery from the PW2. I experienced problems (partially hardware, partially a learning curve) trying to set up the PW3 now that it has a good battery.

A factory reset removes the home and settings menus until AFTER you scan for wifi networks in a partial attempt to register. After wifi scanning, you can choose "register later", and then the home menu appears. This was all complicated by the fact that the touchscreen on this PW3 periodically quits working (sometimes predicably) and this made getting past initial force registration difficult to understand (when tapping links or buttons did not work).

I since discovered that the touchscreen (almost always) works after going into (and returning from) screensaver mode, so a double-tap of the power button reactivates the touchscreen when it quits working (though may take up to 10 seconds to begin working in some cases). Returning from USB storage mode seems to always lock up the touchscreen (as do some other predictable events). I am now in the habit of tapping the power button when tapping the screen fails to work on this device.

Conclusion: This PW3 device is totally adequate for me to add and test support for this device to my code, and to know what I am talking about based on hands-on first-hand experience when trying to help others who have problems with their PW3 device. Because of the similarity to the PW2 (other than screen resolution), it may also help me support the PW2.

Thanks to all involved in helping me get this device operational, and I will try to get that K5 3G shipped out today (but I wonder if its 3G can be configured to work in Switzerland?)...

[knc1]

That is a change -
The ON/OFF and RESET switch pads are connected on mine, and work as labeled.

Must be an engineering change (I have a first day of release PW-3).

I never recorded the engineering change level information on my board, but may have captured it in one of the photos in my "Exploring the PW-3" thread.

= = = =

Try reading a book with that PW-3 - -
It IS a nice display.

[geekmaster]

Originally Posted by knc1

That is a change -
The ON/OFF and RESET switch pads are connected on mine, and work as labeled.

Must be an engineering change (I have a first day of release PW-3).

I never recorded the engineering change level information on my board, but may have captured it in one of the photos in my "Exploring the PW-3" thread.

= = = =

Try reading a book with that PW-3 - -
It IS a nice display.

I tried all the switch pads on mine, and none of them work, which is why I suspect they are not connected (missing resistor, at least). I wonder what those missing CRxxx devices are?

I was just searching for a mobi copy of the "Linux Kernel Module Programmers Guide" (which I have been reading on occasion on my Galaxy Note 4 phone), to put on this device. I certainly loved my PW1 before it emancipated itself, so a PW3 should provide a great reading pleasure. The touchscreen might be more stable when just reading -- we shall see.

But reading LKMPG will just make me more annoyed that nothing I compile works on my K1 -- it needs some new kernel modules more than any other of my devices (no g_ether.ko). Even linking to the K1 libraries did not work for me, neither static nor dynamic.

FYI, I did get a basic (crude) shell working with only shared files (using a host PC talking to a VM over shared files). I did not port it to my K1 yet because I think making /mnt/us writable from both sides of a USBMS share is asking for file system corruption (both sides writing to the directory structure). Perhaps it would be okay with fixed-size preallocated files though, but my scripts just append new data (at this point). Even working, it would not support interactive things like nano, so I have not spent that much time at it -- I prefer to get working USB networking on it (hence a kernel module). Though I would be immensely happy at this point if I could get get a simple hello.c proggie running on my K1.