K5 JBPatch

[knc1]

@ixtab: Isn't Java one of those languages where the interface (definition) and the implementation can be in two different files (here, only one of the required two).

It has been a long time since I Java'd -
One would probably be called an "abstract class" and the "class implementation".
Hence two similar error messages, depending on which of the two can not be found.

[ixtab]

Originally Posted by gabe1986

Hi ixtab.

My name is Gabriel, and I have recently installed jbpatch on my kt. Before anything else, I want to tell you Thanks, you have done a great job!

Everything is working very well, except an issue. I'm experiencing problems with the page refresh rate, that I have disabled from the kt menu, and despite this, I can see refreshing every page! :-(, all this happened since I installed jbpatch.

I hope you could give me some help.

Thanks for all.

Gabriel

Please verify that this problem is really caused by JBPatch (it's very unlikely!):

1. Verify that you can reproduce the problem across 1 or 2 reboots. Continue only if the problem appears consistently. (Otherwise, the problem is not related to JBPatch)
2. Uninstall JBPatch and test it again. If the problem doesn't go away, then it's not caused by JBPatch. If the problem actually does go away, continue.
3. Reinstall JBPatch and verify that the problem reappears.

JBPatch can only be the reason for the problem if all of 1-3 pertain. Please do all of this and report back - thanks!

[twobob]

I WOULD LIKE TO POINT OUT THAT THIS SUB-CONVERSATION IS ABOUT THE TOTALLY ALPHA KINDLE 3 VERSION OF JB PATCH, AND DOES NOT REFLECT THE EXCELLENT QUALITY OF THE MAIN OFFERING.

I don't mind being a pig.

am attempting to get DIAGs to boot currently.

will let you know maybe I foofed somewhere.

[ixtab]

Originally Posted by knc1

@ixtab: Isn't Java one of those languages where the interface (definition) and the implementation can be in two different files (here, only one of the required two).

It has been a long time since I Java'd -
One would probably be called an "abstract class" and the "class implementation".
Hence two similar error messages, depending on which of the two can not be found.

Yeah, but that's called "interface"/"abstract class", and I hope that I know the difference.

The thing that I don't understand is why directly referencing an inexistent class gives a ClassNotFoundException (all fine until here), but trying to load an (existing) class which has a reference to an inexistent class gives a "NoSuchClassDefError". The problem is exactly the same - a required class isn't found, so why not throw the same exception?

Answering my own question: it's probably due to an idiosyncrasy of the JVM, namely the difference between checked and unchecked exceptions. In short: trying to load a named class can throw a (checked=must be caught) "ClassNotFoundException". But if that class is actually found, but isn't "initializable" because it itself depends on another non-found class, that produces an unchecked exception, namely "NoClassDefFoundError". Go figure.

[twobob]

I WOULD LIKE TO POINT OUT THAT THIS SUB-CONVERSATION IS ABOUT THE TOTALLY ALPHA KINDLE 3 VERSION OF JB PATCH, AND DOES NOT REFLECT THE EXCELLENT QUALITY OF THE MAIN OFFERING.

err... I will take clues how you get the 3 into DIAGS...

(currently looking at factory reset screen, it likes showing that one )

: ) I cant believe it's not easier to find on the wiki. I'll update it it I can get some clues

EDIT: Heh. I never ever went into DIAGS on the 3, if the /mnt/us/ENABLE trigger is all we have... not very helpful for recovery purposes after the horse has bolted.

Thank goodness for k3flasher.

[qlob]

if you are at the YKNR screen, reboot with the power switch then, as soon as you can, launch myts with launchpad, and do touch /mnt/us/ENABLE_DIAGS and wait for a reboot loop

[twobob]

I WOULD LIKE TO POINT OUT THAT THIS SUB-CONVERSATION IS ABOUT THE TOTALLY ALPHA KINDLE 3 VERSION OF JB PATCH, AND DOES NOT REFLECT THE EXCELLENT QUALITY OF THE MAIN OFFERING.

Originally Posted by qlob

if you are at the YKNR screen, reboot with the power switch then, as soon as you can, launch myts with launchpad, and do touch /mnt/us/ENABLE_DIAGS and wait for a reboot loop

Heh. no. far more borked than that.

I have a choice of RESET TO FACTORY DEFAULTS or RING CUSTOMER SERVICES depending on my key mashings !!!

HAHAHA Classic death. I wish I could screenie the pics ho hum.

Anyways. k3flasher to the rescue, the device was at 100% charge as it goes anyway. Kernel rolled back to 3.0.2 - rootfs now following suit : )

Enough thread hijacking. First thing I will do is upgrade to 3.3 then try this again. just in case some of my customisations borked it. don't see why but it's not really a fair test unless I give it at least a go with a clean machine.

Don't recall how long it took to flash the rootfs last time but it wasn't forever. : ) Thanks for the tips mate. but "No, it's fubarred"

[gabe1986]

Hi again ixtab

Thank you for your quick response!

I did what you told me, rebooting my kindle a few times, and everything went perfect!, no more refreshing when changing pages ...a strange issue, I don't know what it could have happened.

...but a couple of ideas came to my mind, I don't know if this is the correct place for suggestions, but here I go...

I was thinking that it would be nice to have the possibility of choosing the refreshing rate, for example, change the default number of pages (6 pages) to maybe 7, 8, 9 or 10 .

And another idea: It would be great being able to do a google search from the main search bar of the kindle touch, by now the only available search options are:1)My content 2)Kindle Store 3)Wikipedia 4)Dictionary

Thanks for all!!

Gabe

[Daveychan]

Hello,

First, i'd like to say THANK YOU for all the hard work and time put into these patches, hacks, and tweaks. i'm very grateful!

I've just JB my KPW (v5.2) thanks to all the information here, and i've also SSHed in, etc. I'd now like to install the JBPatch, and i've read that others have done it (but not HOW they've done it)...

Is there a way to manually install the patch since the upgrade function is greyed out? Is there a way to manually force an update of a .bin file? (i've tried the "lipc-set-prop com.lab126.ota startUpdate 1" command with the .bin in my /mnt/us directory, but nothing happens)

My final goal is to play around with the various tweaks that JBPatch offers, but until i can get it into my system, i cannot play with the patches of course...

Please and Thanks!

[NiLuJe]

@Daveychan: The current release doesn't really/completely support 5.2.0/PaperWhite. Unless you intend to help with porting it/patches to the PW, there's really not much interest in installing it on the PW yet.

(But to answer the question anyway: check the sources, and read the install script, it's pretty straight-forward).

TL;DR: For now, there's a reason the 'Update Your Kindle' link is grayed on the PaperWhite for the average Joe .

[jscris]

Originally Posted by NiLuJe

@Daveychan: The current release doesn't really/completely support 5.2.0/PaperWhite. Unless you intend to help with porting it/patches to the PW, there's really not much interest in installing it on the PW yet.

(But to answer the question anyway: check the sources, and read the install script, it's pretty straight-forward).

TL;DR: For now, there's a reason the 'Update Your Kindle' link is grayed on the PaperWhite for the average Joe .

Consider this a show of interest in installing on the PW!

[twobob]

Originally Posted by ixtab

snip...

Note: the attached file *is* the jbpatch.jar, but mobileread won't allow to attach jar files. Just rename jbpatch.zip to jbpatch.jar (don't unzip it).

So, yeah. I am now in a position to test this again.

Utterly clean kindle 3.4

renamed zip to jar

put to device with mntroot rw
sftp://root@k/opt/amazon/ebook/lib/jbpatch.jar

edit /opt/amazon/ebook/bin/start.sh

to say:

## remove old bootstrap
## --  app=com.lab126.linux.arm.LuigiServiceProvider

## ++ new bootstrap
app=com.mobileread.ixtab.jbpatch.bootstrap.K3Startup
##

that's the complete steps correct?

thanks. I have the file in place just want to double check my work this time before another 4 hour hose down

EDIT:

what about??

## conditionally swap to old bootstrap
## 
if [ -n "${FRAMEWORK_CRASH_FILE}" ]; then
	app=com.lab126.linux.arm.LuigiServiceProvider
    echo "## NOTE: Framework is using default loader as crash detected. ##"
    echo "## To revert back to the jbpath loader, remove file: 

 ## touch /mnt/us/ENABLE_DIAGS  ?? Maybe?
${FRAMEWORK_CRASH_FILE} ##"
	else
## ++ new bootstrap
	app=com.mobileread.ixtab.jbpatch.bootstrap.K3Startup
##
fi

taking our lead from the header script bits..

DEBUG_CVM_CRASH_FILE=/mnt/us/DEBUG_CVM_CRASH
# Legacy file, will be removed shortly
DONT_RESTART_FRAMEWORK_FILE_OLD=/mnt/us/system/dont-restart-framework

which are the files that ultimately get created on total foof.
(and referenced as $FRAMEWORK_CRASH_FILE)

Cheers.

[twobob]

Weeeeeeeeeeeeee IT WORKS! YAY.

Call me paranoid but I used the second version of the start.sh ...

EDIT:

Originally Posted by arikfunke

...

@twobob - I realise this is too late now but it is safer to make a copy of the "start.sh" file which you then modify. Leave the original as it was so that the kindle always starts. To then test the framework do the following:

killall start.sh
killall cvm
./start_new.sh

This way, if anything goes wrong it is enough to hit the power switch to hard-reset the device to its non-modified state.

So that's also an option for a less "permanent" or testing solution.

END EDIT:
Will dig up some logs. The custom configuration I was using may have done something to make it "not work" last time. It was hardly an OOTB example device...

You can remove the warning mate

LOGS:

Log start timestamp: Wed Oct 24 21:39:59 GMT 2012
Bootstrap OK, PatchingClassLoader instantiated
   Packages still handled by original ClassLoader:
   - com.mobileread.ixtab.jbpatch.bootstrap

JBPatch version 2.4.1: initializing patches

I: Directory synchronization thread started
I: Wed Oct 24 21:40:00 GMT 2012: synchronized 
com.mobileread.arikfunke.patch.K3BrowserGoogleMobileSearchPatch.class

I: (builtin) DeviceInfo registered for 
com.amazon.kindle.settings.dialog.DeviceInfoDialog

I: Patch policy initialized
I: com.mobileread.arikfunke.patch.K3BrowserGoogleMobileSearchPatch 
was not found in configuration, adding entry and enabling patch

I: updated configuration file

I: File /var/local/jbpatch/com.mobileread.arikfunke.patch.
K3BrowserGoogleMobileSearchPatch-en.txt successfully initialized

I: com.mobileread.arikfunke.patch.K3BrowserGoogleMobileSearchPatch 
registered for com.amazon.ebook.booklet.browser.resources.BrowserResources

search in browser for test get abcXYZtest - success report.

extra little detail:

I: com.mobileread.arikfunke.patch.K3BrowserGoogleMobi leSearchPatch applied to com.amazon.ebook.booklet.browser.resources.Browser Resources (6c8c6e226e2f60e134fa7ccec784c4ce -> 96d802f5b0cef88cdf54a2407e263af8)
W: com.mobileread.arikfunke.patch.K3BrowserGoogleMobi leSearchPatch produced MD5 "96d802f5b0cef88cdf54a2407e263af8", but declared "?"

I recall the warning from the 5 but wasn't sure about the ? declaration so I thought I would report it.

[Daveychan]

@NiLuJe - LOL - Thanks for your direct reply! Looking forward to the progress. Much appreciated!

[QUOTE=NiLuJe;2277254]@Daveychan: The current release doesn't really/completely support 5.2.0/PaperWhite. Unless you intend to help with porting it/patches to the PW, there's really not much interest in installing it on the PW yet.

(But to answer the question anyway: check the sources, and read the install script, it's pretty straight-forward).

[arikfunke]

@ixtab - Sorry I only just saw your post. It would appear that twobob has already reported success. Shout if there is anything else I can do.

@twobob - I realise this is too late now but it is safer to make a copy of the "start.sh" file which you then modify. Leave the original as it was so that the kindle always starts. To then test the framework do the following:

killall start.sh
killall cvm
./start_new.sh

This way, if anything goes wrong it is enough to hit the power switch to hard-reset the device to its non-modified state.