K5 Backporting a newer version of the Kindle firmware/GUI from Paperwhite to Touch - Page 3

WaseemAlkurdi · 07-23-2019, 03:45 PM

P.S.
1. If not for @geekmaster's U-Boot profiles (under the Select Boot thread) and imx_usb, then both my Kindle and I would have become toast by now.

2. /var/local/logs/* on the third eMMC partition did not change at all between boots.

3. (

) While I was getting my feet deep in the cesspool of "why the hell USB Downloader Mode doesn't blank the e-Ink display and how you should check your host's dmesg instead", a rather amazing thought (is it?) struck my head:
Why not use KVM and qemu to boot Android on the Kindle?
Note that we have U-Boot access, so we can modify ARM ELs however we please. This isn't possible on more "mainstream" consumer devices (like phones and ARM tablets) because U-Boot isn't there.
I know, it would be slow as molasses, but who cares? Isn't that a good first step?

knc1 · 07-23-2019, 04:33 PM

Quote:

Originally Posted by WaseemAlkurdi

Oh .. that makes two of the elite devs ... ixtab and geekmaster ...
------

????
Last I heard, ixtab completed the dissertation for his Phd in computer science.
Which in turn resulted in having to get a job, rather than laze around in school(s) all the time.
geekmaster is "gone", ixtab is just "busy elsewhere".

WaseemAlkurdi · 07-23-2019, 05:23 PM

Quote:

Originally Posted by knc1

????
Last I heard, ixtab completed the dissertation for his Phd in computer science.
Which in turn resulted in having to get a job, rather than laze around in school(s) all the time.
geekmaster is "gone", ixtab is just "busy elsewhere".

Damn!

I thought he had passed away as well because of this reply on this thread to somebody asking about links formerly hosted on ixtab.tk (the person maintaining those links has passed away last summer):

Post #1366:

Quote:

Hi. I need to debrick PW2 but the links is not working

pw_5.2.0-mmcblk0p1+jb+ssh: http://www.mediafire.com/?5w619l68mrcawlb
pw_5.2.0-mmcblk0p1+jb+ssh: http://ixtab.tk/kindle-touch-images/...cblk0p1.img.gz

and

pw_5.2.0-mmcblk0p2: http://www.mediafire.com/?afodh08q5had1z1
pw_5.2.0-mmcblk0p2: http://ixtab.tk/kindle-touch-images/...cblk0p2.img.gz

May be anyone have this files or working link. Thank you.

Post #1376:

Quote:

The maintainer of those Media Fire links passed away a few months ago.

Additionally, ixtab.tk is down, so that sort of reinforced that mistaken thought.

Apologies to you @ixtab, if you ever read this

WaseemAlkurdi · 07-26-2019, 08:07 AM

It's been a while since I last did the unsuccessful transplant ...
Seems I have to get a serial cable after all.
But first, how about transplanting /opt ... would that work?

knc1 · 07-26-2019, 08:55 AM

Quote:

Originally Posted by WaseemAlkurdi

Damn!

I thought he had passed away as well because of this reply on this thread to somebody asking about links formerly hosted on ixtab.tk (the person maintaining those links has passed away last summer):
- - - - - -

That was myself.
Note the domain names shown in the posted (missing) links:
ixtab.tk :: Missing
mediafire.com :: Passed away (not mediafire.com, just that other person's account).

WaseemAlkurdi · 07-26-2019, 05:17 PM

So ... I copied /opt across.
The thing booted, but the progress bar gets stuck at almost-filled (yep, a progress bar, not a Paperwhite spinning circle), before it restarts from scratch.
Now what?
If USB networking can be brought up on boot (which I believe to be the case), then one could monitor system logs and see what's causing the reboots.

NiLuJe · 07-26-2019, 09:27 PM

Depends entirely on what exactly fails to boot.

As-is, USBNet has a hard dependency on volumd, so it starts as soon as possible, right behind volumd.

While bypassing volumd, and as such possibly starting earlier, is theoretically possible, it's in practice more trouble than it's worth, since it causes far too many side-effects.

It should be starting before the Java framework, which is usually where shit happens.

If not, you can always reboot in diags and peek at logs from there.

WaseemAlkurdi · 07-27-2019, 05:32 PM

Quote:

Originally Posted by NiLuJe

Depends entirely on what exactly fails to boot.

As-is, USBNet has a hard dependency on volumd, so it starts as soon as possible, right behind volumd.

I believe that volumd starts without an issue, simply because /mnt/us is exported over USB despite the thing being unable to finish booting.

Side note: The first time I rebooted with the Paperwhite /opt in place, the words *** MKK (something) JB *** were displayed for a moment or two. That means that USBNetwork (at least its daemons) should've started up, doesn't it?

Quote:

While bypassing volumd, and as such possibly starting earlier, is theoretically possible, it's in practice more trouble than it's worth, since it causes far too many side-effects.

I plan to boot with as little modifications as possible ... not only it is -as you mentioned- more trouble than it's worth, it also causes unpredictable failures into the future over and above those in the present.

Quote:

It should be starting before the Java framework, which is usually where shit happens.

If not, you can always reboot in diags and peek at logs from there.

The Java framework seems to have started, actually, as the progress bar fills up until there's only one 'step' left, before it fails and restarts.

I actually haven't tried installing the USBNetwork-on-boot package (I have USBNetwork, but it's not set to start on boot just yet), because files under /opt might override the package or something. If the logs are accessible from diags, then it would be a great boon to our present endeavor.

NiLuJe · 07-27-2019, 08:23 PM

@WaseemAlkurdi:

1/ If you have actual (relevant) content in there, yes. If it's an empty drive, then, no.

2/ No, that's the bridge.

3/ USBNet has zero interaction with /opt, one way or another.

4/ That doesn't necessarily mean much, as the progress bar is more of a timer bar than an actual progress bar (i.e., it tracks elapsed time, not progress).
Generally, the framework should be starting between the 40-60% mark (there's generally a slight hitch to the bar, because, again, timer vs. massive CPU load).

5/ Logs are in /var/local, so, yeah, you can access them from both diags or main.

WaseemAlkurdi · 07-28-2019, 04:51 PM

@NiLuJe
I took copies of /var/log/messages and other log files, booting with the stock and Paperwhite /opt directories.
I have sent them to you in a private message. Hope you can shed a light about what can be possibly wrong. (My quick skimming can't find anything wrong, except that the framework finishes starting in five seconds instead of 27 on stock).

07-23-2019, 03:45 PM	#31
WaseemAlkurdi Zealot Posts: 147 Karma: 20 Join Date: Jul 2019 Device: Kindle Touch (K5) Wi-Fi x 2, Kindle (7th Gen, KT2), Paperwhite 3rd Gen	P.S. 1. If not for @geekmaster's U-Boot profiles (under the Select Boot thread) and imx_usb, then both my Kindle and I would have become toast by now. 2. /var/local/logs/* on the third eMMC partition did not change at all between boots. 3. () While I was getting my feet deep in the cesspool of "why the hell USB Downloader Mode doesn't blank the e-Ink display and how you should check your host's dmesg instead", a rather amazing thought (is it?) struck my head: Why not use KVM and qemu to boot Android on the Kindle? Note that we have U-Boot access, so we can modify ARM ELs however we please. This isn't possible on more "mainstream" consumer devices (like phones and ARM tablets) because U-Boot isn't there. I know, it would be slow as molasses, but who cares? Isn't that a good first step?

07-27-2019, 08:23 PM	#39
NiLuJe BLAM! Posts: 13,506 Karma: 26047202 Join Date: Jun 2010 Location: Paris, France Device: Kindle 2i, 3g, 4, 5w, PW, PW2, PW5; Kobo H2O, Forma, Elipsa, Sage, C2E	@WaseemAlkurdi: 1/ If you have actual (relevant) content in there, yes. If it's an empty drive, then, no. 2/ No, that's the bridge. 3/ USBNet has zero interaction with /opt, one way or another. 4/ That doesn't necessarily mean much, as the progress bar is more of a timer bar than an actual progress bar (i.e., it tracks elapsed time, not progress). Generally, the framework should be starting between the 40-60% mark (there's generally a slight hitch to the bar, because, again, timer vs. massive CPU load). 5/ Logs are in /var/local, so, yeah, you can access them from both diags or main. Last edited by NiLuJe; 07-27-2019 at 08:25 PM.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Paperwhite w/ 3G firmware version question	cspittell	Amazon Kindle	7	06-15-2018 07:29 PM
Current Firmware Version of New Paperwhite or Voyage	victoryhollow	Kindle Developer's Corner	17	06-11-2015 08:08 AM
Paperwhite 2: what firmware version does it come with?	ixtab	Amazon Kindle	48	11-08-2013 10:57 PM
Backporting firmware 4.0.1 on Kindle 3 - in progress	giorgio130	Kindle Developer's Corner	49	04-02-2012 04:18 AM
How to install prior version of Calibre when newer version installed?	SilentSeven	Calibre	3	04-13-2011 12:46 PM

07-26-2019, 08:07 AM	#34
WaseemAlkurdi Zealot Posts: 147 Karma: 20 Join Date: Jul 2019 Device: Kindle Touch (K5) Wi-Fi x 2, Kindle (7th Gen, KT2), Paperwhite 3rd Gen	It's been a while since I last did the unsuccessful transplant ... Seems I have to get a serial cable after all. But first, how about transplanting /opt ... would that work?

07-26-2019, 05:17 PM	#36
WaseemAlkurdi Zealot Posts: 147 Karma: 20 Join Date: Jul 2019 Device: Kindle Touch (K5) Wi-Fi x 2, Kindle (7th Gen, KT2), Paperwhite 3rd Gen	So ... I copied /opt across. The thing booted, but the progress bar gets stuck at almost-filled (yep, a progress bar, not a Paperwhite spinning circle), before it restarts from scratch. Now what? If USB networking can be brought up on boot (which I believe to be the case), then one could monitor system logs and see what's causing the reboots.

07-26-2019, 09:27 PM	#37
NiLuJe BLAM! Posts: 13,506 Karma: 26047202 Join Date: Jun 2010 Location: Paris, France Device: Kindle 2i, 3g, 4, 5w, PW, PW2, PW5; Kobo H2O, Forma, Elipsa, Sage, C2E	Depends entirely on what exactly fails to boot. As-is, USBNet has a hard dependency on volumd, so it starts as soon as possible, right behind volumd. While bypassing volumd, and as such possibly starting earlier, is theoretically possible, it's in practice more trouble than it's worth, since it causes far too many side-effects. It should be starting before the Java framework, which is usually where shit happens. If not, you can always reboot in diags and peek at logs from there.

07-28-2019, 04:51 PM	#40
WaseemAlkurdi Zealot Posts: 147 Karma: 20 Join Date: Jul 2019 Device: Kindle Touch (K5) Wi-Fi x 2, Kindle (7th Gen, KT2), Paperwhite 3rd Gen	@NiLuJe I took copies of /var/log/messages and other log files, booting with the stock and Paperwhite /opt directories. I have sent them to you in a private message. Hope you can shed a light about what can be possibly wrong. (My quick skimming can't find anything wrong, except that the framework finishes starting in five seconds instead of 27 on stock).

Advert

Advert