Database Users

pre-rooted stock 6.2.1 update

See: http://forum.xda-developers.com/showthread.php?t=1402440
Sorry - don't have enough posts to reply directly in the dev tree, but it would seem to me that if all the updates are just zip files, couldn't they be modified to include root? Then the normal update system would handle the installation? You wouldn't need root then.. Just dump it into the "kindleupdates" folder and let it do its thing. It would just install it.
It looks like the install is just running a script.. one of the first things it does is wipe the system folder.. doesn't mean you can't include additional files and copy them in afterwards with the script does it?

AndrewTL said:
See: http://forum.xda-developers.com/showthread.php?t=1402440
Sorry - don't have enough posts to reply directly in the dev tree, but it would seem to me that if all the updates are just zip files, couldn't they be modified to include root? Then the normal update system would handle the installation? You wouldn't need root then.. Just dump it into the "kindleupdates" folder and let it do its thing. It would just install it.
It looks like the install is just running a script.. one of the first things it does is wipe the system folder.. doesn't mean you can't include additional files and copy them in afterwards with the script does it?
Click to expand...
Click to collapse
You need to go through all the lines of code for that and I believe some of them are binary files as well. Possible but will take much time. For now we must wait until someone finds a better solution.
Sent from my Kindle Fire using Tapatalk

signature issues.

6.2.1 experiment?
transfuntioner said:
signature issues.
Click to expand...
Click to collapse
Maybe I don't follow the process.. but I'm guessing no signatures..
not a signed bootloader.. no signatures required...
Get beyond compare and look at the differences between the OTA updates and the zip that Eldarerathis put together.
The ".bin" files are just ".zip" files renamed.
Look at
META-INF/com/google/android/updater-script in the bin files they release.
It is just a script.
And to do the things that it is doing, it has to be getting root permissions while it is running.
I don't know how to recover if it doesn't work, but I half wonder if you were to take the zip that fellow created, change the extension to ".bin" and just dump it into the update folder if the fire would install it for you?
He took out the image files from the script. He's just updating the system tree directly. He doesn't update the bootloader, recovery or backup trees.
For that matter if those contained something nasty, they could probably be pulled from an earlier update.
== cut ===
package_extract_file("recovery.img", "/dev/block/platform/mmci-omap-hs.1/by-name/recovery");
show_progress(0.100000, 10);
package_extract_file("u-boot.bin", "/dev/block/platform/mmci-omap-hs.1/by-name/bootloader");
format("ext4", "EMMC", "/dev/block/platform/mmci-omap-hs.1/by-name/backup");
mount("ext4", "EMMC", "/dev/block/platform/mmci-omap-hs.1/by-name/backup", "/backup");
package_extract_file("backup_userdata.zip", "/backup/backup_userdata.zip");
== cut ===
Please nobody try this idea that doesn't know enough to recover if it doesn't work, because I've not a clue how.
But could someone explain why wouldn't it work?
Worst case the rev might need to be manually edited to bump it one for the script.
Seems like it might be an easy way to get around headaches with OTA updates.

Andrew,
Are you having issues with getting into recovery so that you can test this? The Devolopers have a copy of TWRP 2 working perfectly on the device so you can get a clean backup of the OS before you test.
You sound like you may know what to do to fix the issue and I just want to help you get a good recovery in case it doesn't pan out or for testing.
~enjoy.

Actually I'm just afraid of messing up what I have working now.
I've got TWRP 2 running with the 6.2.1 image.
I'm setting these up as gifts. I spent a good 12 hours yesterday fighting the darn things. Guess I'm getting gun shy. Had a lot of issues with adb. Kudos to the new utility and setting the second ports up. I'm learning.
The OTA install wipes the system directory. It's in the script.
I was tempted to just comment that line out and see if it would leave the rest alone.
If I brick the thing its going to be painful for me to sort out.
I also didn't want to risk losing root. I also haven't tried loading an older OTA version.
At the time, I couldn't ask the fellow that developed the rooted 6.2.1 zip the question in his thread as I didn't have enough posts.
I'm going to be out of pocket for quite some time. I'm not going to have time to spend like this again for at least a few days.
But I was checking the zip he made versus the .bin/.zip files that the OTA releases using beyond compare. I wanted to see what files had actually changed.
That's when I found that script. It is manipulating permissions directly. It has to be running at a root level. I don't see anything like a signature on it, so why couldn't their own update system be used to tweak their own updates to allow for root?

Alternative thoughts on preventing OTA update

I don't think I have seen any mention of this idea yet. Sorry if I missed it...
In a recent thread about the 6.2.2 update and people wanting to prevent it, I thought I read that someone saw the file show up in the update directory. I'm assuming this means the same 'kindleupdates' directory you could manually drop the update into -- but if not, the idea is the same. Why not just take some step to prevent access to this directory?
The exact step to take would depend on how smart the developers were about dealing with problems in the update process
The easiest step would be to chmod 555 it. But of course if the update process is running as root it is under no requirement to honor those permissions! (My experience in the unix world tells me that about half the time, programs running as root do honor the permissions even though technically root overrides them).
Another easy step would be to delete it altogether. But they probably thought of that (if it's /mnt/sdcard/kindleupdates where someone could easily accidentally delete it) and recreate it if it's missing.
One trick that is often done is to replace the directory with a file. Some programmers do not think to check this kind of condition - they see there is something there, but they get an error opening it as a directory, and they just declare it's an error.
A more subtle trick would be to replace the directory with a symlink that points to a read-only directory (such as /system). In this case, they could open it as a directory, and just fail to write there. The programmer probably would not have thought to check whether it's a link vs. a real directory. One possible gotcha is if you point to /system, and /system is r/w, then the update could screw something up under /system. So maybe mount /system r/w, mkdir /system/kindleupdates, remount /system r/o, then link the update dir to /system/kindleupdates.
And finally, I don't know if Android has any kind of loopback filesystem capability, but loopback-mounting something read/only on that directory would certainly fake the OS into thinking there was a directory there; it would definitely be read/only, and I don't think they would ever think to check whether there is actually some filesystem mounted there! (and if there was, all you need is an app that constantly accesses some file you put there, which would make it busy so that it couldn't be unmounted).

The first method won't work because the sdcard partition is fat32 and doesn't accept unix permissions.

it downloads to the /cache folder - this folder is also used for other things like market downloads, logs from twrp and i don't know what else
btw. there are a lot of threads about this from the 6.2.1 update
make a short search for "prevent ota update" - you'll have a lot to read ...

well, i just deregistered my kindle acount and i'm still in 6.2.1...

b63 said:
it downloads to the /cache folder - this folder is also used for other things like market downloads, logs from twrp and i don't know what else
Click to expand...
Click to collapse
Ah, that makes this less practical. Still, perhaps when the next update comes out I can try a variation on this but it requires the filename to be known.
If the update is downloaded as a single file to /cache, which is named the same as the file you can manually grab, then someone who hasn't gotten 6.2.2 (and is not averse to this failing) can try this in a root shell:
mkdir /cache/update-kindle-6.2.2_D01E_3205220.bin
mkdir /cache/update-kindle-6.2.2_D01E_3205220.bin/blah
The purpose here is to put something unremovable in the way of the file it wants to download. Most likely if the update sees something with the existing name there it would probably want to blow it away (after determining it's incomplete) - and since any update there would normally be a regular file, they probably would do nothing more complicated than a simple unlink syscall to delete it before re-downloading. However, since it's a directory with something in it, that unlink will fail. In actuality, making the subdirectory (second command above) should be unnecessary because the unlink should not work for directories; there's a special rmdir syscall for them.
btw. there are a lot of threads about this from the 6.2.1 update
make a short search for "prevent ota update" - you'll have a lot to read ...
Click to expand...
Click to collapse
I did read a lot of that last time and I don't think I actually saw a definitively successful method. If there is one it should be stickied
My interest in this is a little different from most of you guys - I have very limited satellite internet and I don't like these unscheduled 185-meg downloads so I want to be able to update only when I want mostly to control that. This kind of means looking for the least-intrusive way to accomplish this.

/cache/update-kindle-6.2.2_D01E_3205220.bin is exactly where it downloads
if you find a way to even prevent the download, that would be greatly appreciated

Unfortunately I already got the update so I can't try it this time.

at least you could try your method with a dummy file of an other name and try to overwrite it with adb - if you can't overwrite it there's a good chance

I think I'm about the only one who prevented 6.2.1. I did it by constantly checking the cache folder. Found the update by chance and deleted it before it updated. Waited over a week for it to come back. Never did. An app that watched the cache folder for the updates and then moved/deleted them would work fine
Sent from my SGH-I897 using xda premium

jcase already work a way around this automatic OTA update, so when FIREMOD is ready to replace burrito I think we will have no more problem with this OTA issue. (you can find jcase announcement in the kindle developer section)

Heres what I have done to prevent this.
1) Droidwall (white list only the apps you want to allow internet access)
2) Removed "otacerts.zip" from /system/etc/security/otacerts.zip.
3) I removed "OTASilentInstall.apk" /system/app
4) Installed this 6.2.2 based Rom http://forum.xda-developers.com/showthread.php?t=1439916
Hopefully this eliminates the OTA. I had my Fire rooted on 6.2.1 with twrp and it OTA'd on its own, broke root and twrp. So I rerooted with burritoroot2 and installed CWM based recovery.

[ROOT]Root shw-m250S/K (and others?) through hidden partition in JB

Not revolutionary (especially since zips work on stock recovery), maybe not even new on other variations, but still I thought worth bring up as at least it's new for for this variant in JB:
Now in JB for korean GSII we finally have a hidden partition(for better or worse) and it can be used to get root. I have no idea if this can apply at all to other variants of the gs2
On the SK ROM this partition contains nothing but apps (apks) which are all readable (I think they must be to work, but anyway they are). Most are probably arguably bloatware anyway, but it looks like some might be desirable or even fairly fundamental, I'm not sure yet.
It also turns out that it is possible to execute setuid-root files from this partition but of course it's not writable without flashing it.
So it's easy to copy all the files off the so called "hidden" partition through adb without root access... add an su binary, repack with make_ext4fs and tar and reflash with odin. Then you can adb in, run /preload/su to get root, and then copy/install su/supersu into the more normal place to make it more readily available to apps.
Of course the only thing preventing this method with the /system partition was that a few files in /system were not readable without root access and copying all the file permissions, links etc correctly could be a minor pain using only toolbox or whatever. For the hidden partition, for now at least, the directory layout is very simple and all readable.
If hemidall actually worked right in linux on this device for me I could do this with one linux script.
I have not tested a straight through trial of this because I got root already, but I've tested all steps.
In the past I got stock root without flashing unsigned kernels by hijacking the ROM through KIES (freeze it right after it's decrypted), unpacking the factoryfs, adding su/supersu, repacking and flashing. This allows some other customizations anyway so is at least sort of useful, not sure this hidden partition method has any added value. Maybe it will be a useful idea at point in time though.

those invisible menu options

There is an interesting discussion on the Russian forum the-ebook.org (paste this link into Google and select the translation; item is on page 2) regarding those annoying invisible menu options in many apps. Although the translation is a little rough around the edges, the gist seems to be that the default text and background colors for the app menus are not exactly what the e-ink display has in mind, often resulting in light or near-white text on a white background--hence, invisible text.
The "solution" used is to change some background color settings in framework-res.apk. I have no idea what the outcome looks like but it seems like a really good idea, although beyond my abilities, and the example used is 1.10 firmware. I'm using 1.21. Any attempt I have made to modify apk files has always failed
Does anyone have a fool-proof set of steps for doing this?

I remember reading about what you're talking about somewhere on the forums but I could never get it to work either...
Here's what I'm talking about:
http://forum.xda-developers.com/showthread.php?t=1356514
http://forum.xda-developers.com/showthread.php?t=1512846

OK, well.....I've got the adjusted background images extracted from the framework-res.apk file that I got from the Russian site (I'm working with the lighter background option at this point). And...I've found the setting in WinRar to simply "store" (not compress) the updated png files. So far, so good. My altered apk file is the same size as the original.
The clincher is to get it back on the Nook without disaster ensuing. I'm going to follow Renate's method for pushing back framework-res.apk via ABD (from this thread):
C:\>adb shell
# stop
# mount -o rw,remount /dev/block/mmcblk0p5 /system
# ^C
C:\>adb push framework-res.apk /system/framework
C:\>adb shell
# reboot
Keep your fingers crossed. If it works, I'll report back with step-by-step and files.
[Report: in concept this "works". On reboot I could see the slightly grey background color in menu options and going to a few apps where I knew the menus were invisible, I could see white text on the slightly grey background. BUT...almost no apps will work. The B&N side seems to function OK, but the App drawer is useless. Back to the drawing board. The Russian site has a method for installation using RootExplorer. I'll try that next.]

OK...day 2. Here's what does NOT work:
1. method in post above using ADB
2. method using RootBorwser as adapted from Russian site and detailed below:
a. Change permissions in /system, /system/etc and /system/framework folders so that all users have write access
(note this is my kludge to get around not knowing how to "mount" the /system partition as r/w--maybe it's wrong?)
b. Use ADB wireless to move modified framework-res.apk into /system/etc
c. Use RootBrowser to check ownership of modified framework-res.apk (should be and was already owner: 0-root, group: 0-root)
d. Use RootBrowser to change permissions on modified framework-res.apk to rw-r--r-- (664)
e. Use RootBrowser to move (cut/paste) modified framework-res.apk into /system/framework (overwrite)
With Superuser permission, this all went off without a hitch.
f. Use RootBrowser to reset permissions of folders listed in (a)
g.Shut down Nook and restart.
The result is the same as the ADB-only method described before. The Nook starts up just fine. You can see that the background color of menus is slightly gray. Those changes have obviously worked. But the vast majority of apps will not run (ADW Launcher is an exception). At one point while I was fiddling with things the Nook spontaneously rebooted.
I guess that's better than spontaneously combusting
SIGH. Clearly, despite my best efforts, something I did in handling the framework-res.apk has damaged it in some subtle way, OR, my inability to properly "remount" the /system as R/W is causing the problem, although using ADB this is accomplished without difficulty and since the result is the same...it must be the modified apk file.
And this is why I have a dedicated SD card backup......

Method 3 that does NOT work:
1. Install Ninjamorph and BusyBox from Market
2. Follow instructions for altering framework-res.apk found here.
Two ways to Finish Project, with zip-align and without. Both yield the same result which is the same as the other methods above, i.e., the B&N stuff mostly works and the desired contrast of the menus is achieved so you can actually see what used to be invisible, but most apps will not run. Really frustrating.
I have to say that while this method seemed promising it is tedious in the extreme as each of the 28 png files must be replaced individually and that means each must be located in a much larger list (which reverts to the top after each replacement....). Ugh.
I simply don't believe anyone who says they can make these modifications with the instructions they have provided. It must be that people who are more familiar with this stuff are leaving out information which is so obvious to them that they don't even think to mention it

framework-res.apk is an apk and therefore it must be signed.
It's a system apk so it must be signed with the system signature.
Modifying a few things doesn't annoy the signature matching, other stuff does.
When you have problems, please quote from logcat because that tells you exactly what the problem is.
Using ADB:
Code:
logcat
Whatever.9.png are special files.
The are usually created thusly:
http://developer.android.com/guide/topics/graphics/2d-graphics.html#nine-patch
When they get packed into an apk they are turned into a PNG graphic with alpha channel.
aapt handles this.
If you take a PNG with sidebars and just zip it, it will not work.

Renate NST said:
When you have problems, please quote from logcat because that tells you exactly what the problem is.
Using ADB:
Code:
logcat
Whatever.9.png are special files.
The are usually created thusly:
http://developer.android.com/guide/topics/graphics/2d-graphics.html#nine-patch
When they get packed into an apk they are turned into a PNG graphic with alpha channel.
aapt handles this.
If you take a PNG with sidebars and just zip it, it will not work.
Click to expand...
Click to collapse
I just tried again and this time after pushing the amended framework-res.apk file (yes, all 28 amendations are nine-patch files) I typed in logcat before rebooting.
Whoosh!!! Lines of information went streaming by faster than I could follow, so much that some of the earliest disappeared from the top of the console window. I have no clue how to extract the text from the console window
The top-most complaint I saw was a reference to system error 1248 in association with the CleanMaster app. Then there seems to be a periodic (15 second) dhcpcd renewal. Eventually there is a section that says "Framework disconnected, eof, failed to read size, closing connection". Then comes a long list of notifications from the Service Manager about all the services that have just died. After that it just continues with the 15-second dhcpcd renewal cycle. Then I gave up and rebooted.
The result is the same as before. The new image backgrounds have been incorporated into the system, but most of the app drawer is just pretty icons. ADW runs--at least the drawer and home page appear. The B&N Home and Library pages load but you can't access any of the books shown. Wi-Fi can be accessed but no apps that use it will run. Occasionally the CleanMaster app throws up an error message.
Here's what I've learned so far:
1. In the original amended framework-res.apk file (for FW 1.10) viewing the archive reveals that the files which have been changed all have an "archive " attribute. None of the original files show any attribute. I don't know whether that is important. I've searched on this topic and have come up with nothing.
2. In moving the amended *.9.png files from the original Russian example for FW 1.10 into a copy of my own framework-res.apk for FW 1.21, those "archive" attributes came along for the ride and the resulting amended file does show "STORE" for the method so I think I got that part right and didn't recompress any files when moving from one apk to another (I dragged the files from one instance of WinRAR to another--I tried 7zip as well...).
But I have no idea why people report that this procedure works just fine. For me, it is predictable, but not successful.
I've just completed yet another restore from backup. I'm going to uninstall CleanMaster and try again. Perhaps it's background activities are driving the process into the ground. I have no clue.

Here's the logcat session file (learned how to get that done!) after a re-try, having removed CleanMaster first.
No change in the outcome but no bleating from the Nook about CleanMaster errors.

I wonder--is it the modified apk that is causing the system to malfunction, or....is it the way it is being pushed back to the Nook?
So..an experiment: I pulled a copy of the stock framework-res.apk from my Nook. I didn't do anything to it at all. Then I pushed it back to the Nook via wireless ABD:
C:\>adb shell
# stop
# mount -o rw,remount /dev/block/mmcblk0p5 /system
# ^C
C:\>adb push framework-res.apk /system/framework
C:\>adb shell
# reboot
This is supposed to work, yes? It does not. It leaves me in the same condition that all of the other attempts by this and other methods have. The Nook boots normally and displays Home but you can't access the "currently reading" book. I can get to the app drawer via the quicknav buttons, but very few apps will work (including ADB). There is no way to examine the file system because RootBrowser will not work (although ES File Explorer does, but it doesn't have root access).
So....whether the modified apk is OK or not, I would never know because all of the methods I have tried to get the framework-res.apk back onto the Nook have been unsuccessful.

That procedure should work fine.
Have you checked using the stock framework-res.apk ?

Renate NST said:
That procedure should work fine.
Have you checked using the stock framework-res.apk ?
Click to expand...
Click to collapse
Yes, that's what I just tried. Just pulled it via ADB and then pushed it right back. I also installed a copy of Root Explorer (I generally used Root Browser) because that's what was used in the original thread on the Russian site where I got started with the whole project. That also yields the same results. The Home screen loads but you can't access the book currently being read from it or from the little button at the top left. You can access any of the QuickNav options. The Library "functions". But you can't access books from their covers and the double-tap does not work.
In the app drawer, ADW seems to work fine but you can only run a very few apps, and none that require wi-fi (although wi-fi works). Too much fiddling around and the Nook spontaneously reboots.
When I have tried to move in the modified apk with the slightly gray background 9.png files I can see that the new image backgrounds have been used in the drop-down menus. This suggests to me that the problem is not in how the 9.png files have been moved from the FW1.10 apk obtained from the Russian site into my FW1.21 apk but rather in the integration of the modified apk into the system. I've checked permissions, etc. Everything is OK but the system is just screwed up.
I'm running FW1.21 which has been rooted using Nook Manager with Gapps added. I've done the multi-touch modification and have swapped in a modified internal.db file which seems to have solved the confusion of the "reading now" button. I have some apps that run along in the background, like Tasker and Clean Master (probably others that I don't realize). Do I need a completely clean system to make this change?

I saw the logcat and it showed that it's unhappy and killing the Android.
I couldn't see exactly where the problem is.
I think that you are doing too many things at once.
If there are specific things that do not work, a logcat when you do that should show.

Renate NST said:
I saw the logcat and it showed that it's unhappy and killing the Android.
I couldn't see exactly where the problem is.
I think that you are doing too many things at once.
If there are specific things that do not work, a logcat when you do that should show.
Click to expand...
Click to collapse
Yes, I was probably a little unclear. The ONLY thing that I did when I produced the logcat was attempt to push the modified framework-res.apk file back to the Nook. All those other things mentioned have long ago been successfully accomplished and included in my current backup (which I've had to use dozens of time in the last week or two as I struggle with this framework issue).
The ONLY change to the stock framework-res.apk is the overwriting of 28 *.9.png images with ones copied from another framework-res.apk which, unfortunately, is from FW1.10 (or I'd just use it "as is" without the copying). However, even pushing back an unmodified stock apk results in the same mess.
Like I say, based on what functionality remains when the modified apk file is pushed over, it is clear that the new images are being used, but the system function is severely degraded.
I don't know much about the process, but from what I've read I'm wondering if it would be better to use a zip via CM to deliver this modified file? That way the Android system is not even running (right?) during the replacement procedure.

When you say "stop" Android is no longer running.
Try clearing the cache:
Code:
rm -R /data/dalvik-cache/*

Renate NST said:
When you say "stop" Android is no longer running.
Try clearing the cache:
Code:
rm -R /data/dalvik-cache/*
Click to expand...
Click to collapse
Before pushing? After, but before rebooting?

Anytime that Android is stopped you can clear the cache.

I tried this with a copy of the stock apk:
Code:
C:\>adb shell
# stop
# rm -R /data/dalvik-cache/*
# mount -o rw,remount /dev/block/mmcblk0p5 /system
# ^C
C:\>adb push framework-res.apk /system/framework
C:\>adb shell
# reboot
It put me in an very long chase of the black and white dots that I almost thought would be endless but eventually it finished booting and is in the same condition as all the previous methods. Very crippled.
I can't figure it out

First, get the 1.2.1 update off B&N's website and unzip.
Get framework-res.apk out of that and push it.
The stuff in /system/framework should all be chmod 644.
An ADB push probably leaves it with wider access.
None of this should make any difference.
I'd guess that you are either missing a resource in your fw-r or else you modified something else.
Find an app that crashes. Get a logcat of just that crashing.

Renate NST said:
First, get the 1.2.1 update off B&N's website and unzip.
Get framework-res.apk out of that and push it.
The stuff in /system/framework should all be chmod 644.
An ADB push probably leaves it with wider access.
None of this should make any difference.
I'd guess that you are either missing a resource in your fw-r or else you modified something else.
Find an app that crashes. Get a logcat of just that crashing.
Click to expand...
Click to collapse
I got a copy of the FW 1.21 zip from B&N and extracted the framework-res.apk. I didn't try to push it yet. Instead I got the checksum for the current file on the Nook and compared it to the newly minted file. They were identical. I don't think there is any point in trying yet again with the same procedure. If my reasoning is incorrect, I'm certainly game to try anyway.
All files in /system/framework are chmod 644. The Framework folder itself (and the system folder) are something else, but the contents are all 644.
The only modifications I have made after rooting with Nook Manager are the installation of Google Apps, the implementation of multi-touch (I have to go back and look that up to see what all I did, but surely nothing with framework-res.apk or it never would have worked....), editing settings.db to relabel the QuickNav buttons after programming with NTMM, and pushing a modified copy of internal.db to fix the schizo "reading now" button. As i say, these are all long-established changes and the Nook has been stable with them. I think Google apps may modify framework.jar, but I'm not sure.
[I checked on multi-touch. I flashed a new kernal image and added one line to /etc/permissions/required_hardware.xml in order to enable multi-touch...have no idea what "flashed a new kernal" actually did, but it worked]
Two really noob questions: 1) how can I get a logcat of an app crashing when ADB will not run once a copy of framework-res.apk has been pushed? 2) if the Android system is actually stopped when I type "stop"in ADB shell, how does ADB continue to function?
And, actually, apps do not so much "crash" once I've attempted to put in a "new" framework-res.apk--most just refuse to run. But maybe there is something going on in the droid brain while the screen flickers a little and nothing else happens.

ADB runs under Linux, not the Android subsystem.
You should always be able to access ADB.
If ADB isn't running continuously and reliably you have problems.
If some app does not run, give the specific section in logcat where it doesn't run.

NST sleep but retain last screen image

Forgive me if I have a gross misunderstanding about the hardware in the NST, but in my eyes it should be possible for the Nook to sleep but retain the screen, perhaps with a small banner indicating it's in sleep state and you need to press the 'n' to wake. It just seems silly having the benefit of an e-ink display only for the screensaver to kick it when it sleeps. If it didn't do this, I could (for example) leave a map open in OsmAnd and refer to it still while the Nook is sleeping. The only 'solutions' to this I've found involved plain and simply keeping the Nook awake! A hack I can think of would be to somehow bodge something together to take a screenshot just before the nook sleeps and set it as the screensaver image. Seems silly though... I've had a search around and can't find anything related to this specifically, more either people wanting to keep the nook awake or disable the slide to unlock.
Any thoughts welcome!

I had the exact same thought,
so I'm making an app that takes a screenshot every 1:50 minutes and saves it at a screensaver,
to be displayed when the nook goes to sleep.
hopefully it will be ready this weekend so stay tuned

It should be be simple enough.
Modify /system/framework/android.policy.jar
The class is com.android.internal.policy.impl.LockScreen
What puts up the "screensaver" is updateBackgroundImage()
The layout is in /system/framework/framework-res.apk,
res/layout/keyguard_screen_gossamer_unlock.xml
I'm not sure if changing the layout to transparent and not updating the image should be enough.

Renate NST said:
It should be be simple enough.
Modify /system/framework/android.policy.jar
The class is com.android.internal.policy.impl.LockScreen
What puts up the "screensaver" is updateBackgroundImage()
The layout is in /system/framework/framework-res.apk,
res/layout/keyguard_screen_gossamer_unlock.xml
I'm not sure if changing the layout to transparent and not updating the image should be enough.
Click to expand...
Click to collapse
Oh I wish you would have said that two days ago...
It seems a far better solution than my app, maybe I'll try playing with that later.

I tried replacing updateBackgroundImage()
with a stub, but the screensaver still comes up as usual,
Only difference is the "slide to unlock" screen now has black background, so that's what this function is doing.
Time for plan B (or C, really):
If I were to modify framework-res.apk as you suggest, Will I need to resign the whole system, as here?

Update -
The function that sets the screensaver is createScreensaver(), in class com.android.server.PowerManagerService, found in /system/framework/services.jar
I modified it to make the screensaver invisible, but unfortunately it also means there's no indication that the nook sleeps.
Anyway, it's better than constantly taking screenshots, so I'll stick with that.

nivieru said:
Update -
The function that sets the screensaver is createScreensaver(), in class com.android.server.PowerManagerService, found in /system/framework/services.jar
I modified it to make the screensaver invisible, but unfortunately it also means there's no indication that the nook sleeps.
Anyway, it's better than constantly taking screenshots, so I'll stick with that.
Click to expand...
Click to collapse
How did you do that? I'm using the screenshot sreensaver app for that and I find it very useful. But the above mentioned way seems to be less power consuming. Could you explain it for somebody with very little knowledge about manipulating apk, though I know how to manipulate apk with xdaAutotool.

I don't know xdaAutotool, I use apktool but you can probably use whatever tool you like.
First you need a patched /system/framework/services.jar
the one I attached here is for firmware 1.2.1 rooted with NookManager, so it also includes the NookManager patches.
if this is your setup as well, skip to 4
if you run a different firmware or don't want the NookManager patches you will need to patch it yourself
patching /system/framework/services.jar - some vauge instructions:
you will need apktool (or XdaAutotool or whatever) and the android-sdk.
1) use apktool to decompile services.jar
2) modify createScreensaver() in file smali/com/android/server/PowerManagerService.smali according to the attached patch-services.txt
notice - this patch is for firmware 1.2.1 with NookManager patches, it might not be ready for use with other versions - not only the line numbers, also the register v6 might not be a good choice if it is used later in the code without being assigned a new value first.
3) recompile with apktool.
replacing services.jar with patched version - complete instructions:
4) making a full backup before messing with the system is good practice, although the nook is notoriously hard to brick.
you could use NookManager to do the backup.
5) connect to your device with adb
6) apply these commands:
Code:
adb push pathced-services.jar /media
adb shell stop
adb shell mount -o remount,rw /dev/block/mmcblk0p5 /system
adb shell cp /system/framework/services.jar /system/framework/services.jar.backup
adb shell cp /media/pathced-services.jar /system/framework/services.jar
adb shell chmod 644 /system/framework/services.jar
adb shell reboot
7) relax, first boot takes a few minutes, this is normal.
with this patch the screensaver is invisible, so there is no indication at all that the device is asleep.
also, there's no magic button to change behavior - if you want to the screensaver back you need to replace services.jar with original version

Thanks for that extensive description. I'll try the next days to implement it in my system which differs from yours, since I'm using the custom rom from pinguy1982.

Got it.
I compared your PowerManagerService.smali with the one in my services.jar and made the changes.
Recompiled the services.jar.
Exchanged the original classes.dex with the new compiled classes.dex in 7zip.
Signed the jar file.
Had a lot of problems with adb, which didn't recognize my device.
Lost my temper, packed it into a ZIP-file and replaced it via CWM.
Everything is working now. Thanks!

nhedgehog said:
Got it.
I compared your PowerManagerService.smali with the one in my services.jar and made the changes.
Recompiled the services.jar.
Exchanged the original classes.dex with the new compiled classes.dex in 7zip.
Signed the jar file.
Had a lot of problems with adb, which didn't recognize my device.
Lost my temper, packed it into a ZIP-file and replaced it via CWM.
Everything is working now. Thanks!
Click to expand...
Click to collapse
Hi @nhedgehog
I also use the custom rom from pinguy1982.
Can you share your zip file to flash with CWM
Thanks

tebra said:
Hi @nhedgehog
I also use the custom rom from pinguy1982.
Can you share your zip file to flash with CWM
Click to expand...
Click to collapse
Here you are:
Patch (nivieru's method) via ZIP, tested only with modded ROM from pinguy1982 and installed Nooter-Part1.zip. Install via CWM, to be safe make a backup before you do the patch.
Services-org.zip=original services.jar
Services-patch.zip=patched services.jar

Hi everyone!
I've patched the jar and now enjoy the last image in sleep mode. But, I'd like to wake app the Nook every few hours, update the page in dolphin and go back sleep. Page in dolphin has a meta refresh. But I can't wake up the Nook on interval. Is it impossible or I missing smth very important on this topic?
If anybody has some experience on topic, please comment.
Update, my solution:
Re-signed the system with personal cert and implemented an app instead of dolphin+web page.
App wakes up every few hours, updates view and throws device to deep sleep. Hope to get an uptime up to 10-15 days.

nivieru said:
I
replacing services.jar with patched version - complete instructions:
4) making a full backup before messing with the system is good practice, although the nook is notoriously hard to brick.
you could use NookManager to do the backup.
5) connect to your device with adb
6) apply these commands:
Code:
adb push pathced-services.jar /media
adb shell stop
adb shell mount -o remount,rw /dev/block/mmcblk0p5 /system
adb shell cp /system/framework/services.jar /system/framework/services.jar.backup
adb shell cp /media/pathced-services.jar /system/framework/services.jar
adb shell chmod 644 /system/framework/services.jar
adb shell reboot
7) relax, first boot takes a few minutes, this is normal.
with this patch the screensaver is invisible, so there is no indication at all that the device is asleep.
also, there's no magic button to change behavior - if you want to the screensaver back you need to replace services.jar with original version
Click to expand...
Click to collapse
I've done exactly as described in points 4-7 on NST firmware 1.2.1 , rooted with Nook Manager.
I used Root explorer for replacement and rename of patched-services.jar.
After rebooting of NST, it works as intended - the screensaver is unseen, but it appears other problem.
When i click at random .apk, nothing happens, only the screen is flickering once and that's all.
I cannot install apk-s, nor uninstall them....? WTF ?
Someone with explanation...?

gsms said:
I've done exactly as described in points 4-7 on NST firmware 1.2.1 , rooted with Nook Manager.
I used Root explorer for replacement and rename of patched-services.jar.
After rebooting of NST, it works as intended - the screensaver is unseen, but it appears other problem.
When i click at random .apk, nothing happens, only the screen is flickering once and that's all.
I cannot install apk-s, nor uninstall them....? WTF ?
Someone with explanation...?
Click to expand...
Click to collapse
Replacing services.jar while the android system is running can cause problems, that's why I suggest doing it through adb after issuing the "stop" command, which stops android while leaving the underlying linux (with adb) running.
You should try that and see if it helps.

I will try, but before that another question from me.
Is there a chance the file "patched-services.jar" to be remade as .zip file and installed thru CWM.
If this can be done, i will be glad somebody to share the .zip file, because i think it is more comfortable for installation than using adb shell....
thanks
P.S. Adb don't work correctly for me, so the only available option is obviously flashable .zip file...

...

Has anyone tried this with FW 1.2.2 yet?

jptiger said:
Has anyone tried this with FW 1.2.2 yet?
Click to expand...
Click to collapse
If the original services.jar was patched, it will break NTMM since that relies on patches. But if you can discern what the additional patches are you could patch the 1.2.2 file used in NookManager.
Edit: but if you just want to see how it might work, you can probably use the 1.2.1 jar without any problems. When I first started working on the update to NookManager I used the original 1.2.0 jars that were provided with FW 1.2.2 and saw no evidence of problems. A diff showed very minor changes. Depending on how extensive the patching is for this mod, it might be easier--if you still want the use of NTMM--to patch the file in post #8 for NTMM since the patches are well documented on github. Otherwise you have a lot of diffs to look at and sort out.