Related
This is not a thread to say ‘thanks’ to developers. Only post here if you have some input in this development progress, by either fixing bugs, or coming up with ideas (and ideally, try your idea and report what happens).
Having that said, let’s move on to where we are at the moment. Currently, we have a build which is based upon Ice Cold Sandwich 6.2 Test Edition for the HTC Desire HD. This build boots up by flashing my custom kernel on it, based on Flyer HC Original Kernel Source, with some patches:
-Quota2 Support
-Genlock Support
-Updated KGSL driver by applying Lord Clockans patches from his kernel
The ZIP we are working on now, can be downloaded here:http://w728685.open.ge.tt/1/files/9iI3CbF/0/blob?download
This ZIP contains a build which is mainly Lord Clockans ICS, with essential changes in the ROM to get it booting (we used LeeDroid’s Flyer HC as the donor ROM). This ROM is made for a HoneyComb HBOOT!
After you have flashed that ZIP, this is the kernel you need: http://ge.tt/9iI3CbF/v/1
You MUST flash this kernel to get it up and running, or it’ll just loop in the logcat complaining about unsupported stuff.
This set should do the following:
-Boot up (bootanimation resets after some time, it’s ok, it will boot within 5 minutes).
-Touchscreen works after you did this: open ADB Shell (by entering ‘adb shell’ on your pc,), and send the command “echo ‘1’> /sys/android_touch/event_google”. Now touchscreen should be working. (For some reason the file 90flyerhacks in /system/etc/init.d doesn’t do its job. If anyone has a clue > say so!) Also, the ntrig.c file in /drivers/input/touchscreen/ is modified in the kernel, don’t know if it helped though.
-Full Hardware Accceleration
-Screen is bugged at certain moments (seems to happen especially during things concering wallpapers. Rendering a wallpaper, showing a heavy LWP, etc.). The build.prop file contains some lines such as:
ro.fb.mode=1
debug.composition.type=gpu/mdp
debug.fb.rgb565=0
Perhaps we have to change these lines to get it running without issues (strange thing is, logcat doesn’t seem to make an issue about this). We can also try a recompile of the kernel using the entire driver folder from Lord Clockans kernel (it’s in /drivers/gpu/msm/ in his kernel, and /drivers/video/msm/gpu/kgsl_adreno205_hc in ours, moving the folder to where it should be causes issues, so we keep it like this for the time being).
-Sound is working
-Backcamera seems to be working fine, haven’t tested frontcamera.
-Hardware buttons under the screen aren’t working, and this should be a low
priority for now. It isn’t ideal to use them in ICS anyway (different button functions in comparison with the softbuttons), and I think we shouldn’t focus on this for now.
-GPS switch can be enabled, haven’t tested whether it works.
-Bluetooth switch can be enabled, but turns off automatically after about ten seconds.
-WiFi switch can’t be changed, logcat tells us that it can’t connect with /dev/rfkill, applying this patch in the ramdisk could fix it:
https://github.com/OpenSensation/an...mmit/da09360d70888a7e126871ab2ae5b218514cad18
-Power button and Volume buttons work fine.
-Pen works after we send the event_google command mentioned before.
If the ROM immediately reboots within 10 seconds and it seems like it’s looping, and you can’t even logcat, hold down Power and Volume Down until you get to HBOOT, select FASTBOOT, and enter ‘fastboot erase cache’ in the CMD/Terminal, and reboot, and it should work.
And that’s about it for the current things we know I think. If you manage to fix something, please write down and report what you did to fix it, so we can keep a list. I will work on a vendor folder using the known information, so we can compile a ROM from source, made for our device.
And last, but not least: don’t post here asking for release dates. We will do our utmost best to bring ICS experience to the Flyer, but do not bug us. Thanks.
Now let’s get this up and running guys! I will keep a log of what is fixed and what is not, so we can keep the thread organized. Else we’ll have 50 ZIPs around, not knowing which one is the right one et cetera.
Source code of kernel: http://github.com/Chaosz-X
Ramdisk will be uploaded tomorrow.
Donations are much appreciated, as I'm developing and building atm on a 1.67GHz Dual core...
Donations can me made to [email protected] using Paypal.
i suggest creating two repos... one for ROM and one for kernel... that way single file changes can be tracked even on ROM front, not just kernel...
willing devs can fork those repo and send pull requests to main repo when they get stuff working
I'm not a developer, and I have nothing but love for all the great devs that worked so hard to bring ICS to the Flyer.
Might I suggest, however, that instead of pursuing CM9 builds, the next effort be centered around porting the Android Open Kang Project (AOKP) ROM? In terms of features and customizations, it's considerably ahead of CM9, with numerous supported devices to boot.
I think this ROM, versus CM9, would give users the most flexible and dynamic ICS experience for their Flyer.
Chaosz-X said:
-Touchscreen works after you did this: open ADB Shell (by entering ‘adb shell’ on your pc,), and send the command “echo ‘1’> /sys/android_touch/event_google”. Now touchscreen should be working. (For some reason the file 90flyerhacks in /system/etc/init.d doesn’t do its job. If anyone has a clue > say so!)
Click to expand...
Click to collapse
Does it work if you add
write /sys/android_touch/event_google 1
to the init.rc file?
Could you post the init.rc file where you trigger the init.d scripts?
nickiberli said:
Does it work if you add
write /sys/android_touch/event_google 1
to the init.rc file?
Could you post the init.rc file where you trigger the init.d scripts?
Click to expand...
Click to collapse
yes that should work correctly
No, I didn't add that. I'll try it tomorrow (or one of you, if you have a Linux machine ready. Laptop with Ubuntu is having some issues atm, can fix it tomorrow).
So, assuming that's working, let's get to the display. We know that it causes issues, but how?
-Logcat doesn't error (except for HtcEbdLog, saying something about rotation, which seems to happen when glitches occur. Perhaps we need to swap the files?)
-At the very beginning, it does tell that initOverlay failed. This overlay-thing is baked in the ROM, and can be disabled in the ROM itself by setting an option to 'false' in the vendor setup. We don't know if it fixes the issue, but since the attempts to add this functionality to the kernel failed, it could be a better idea to fix it in the ROM to get rid of the error.
Any other suggestions?
Edit: and before I forget, tomorrow I'll check the ZIP for what is changed in comparison with the ROM it is based upon, so we know what we have to add in to a new build too.
Chaosz-X said:
No, I didn't add that. I'll try it tomorrow (or one of you, if you have a Linux machine ready. Laptop with Ubuntu is having some issues atm, can fix it tomorrow).
So, assuming that's working, let's get to the display. We know that it causes issues, but how?
-Logcat doesn't error (except for HtcEbdLog, saying something about rotation, which seems to happen when glitches occur. Perhaps we need to swap the files?)
-At the very beginning, it does tell that initOverlay failed. This overlay-thing is baked in the ROM, and can be disabled in the ROM itself by setting an option to 'false' in the vendor setup. We don't know if it fixes the issue, but since the attempts to add this functionality to the kernel failed, it could be a better idea to fix it in the ROM to get rid of the error.
Any other suggestions?
Edit: and before I forget, tomorrow I'll check the ZIP for what is changed in comparison with the ROM it is based upon, so we know what we have to add in to a new build too.
Click to expand...
Click to collapse
u want me to add stuff to that ramdisk?
start a github repo for the ramdisk and ROM too... trust me... its VERY helpful...
I'll start installing Ubuntu within 30 minutes, and will add everything to a GitHub by tonight within about 3 hours I think.
Chaosz-X said:
I'll start installing Ubuntu within 30 minutes, and will add everything to a GitHub by tonight within about 3 hours I think.
Click to expand...
Click to collapse
thats great... it should make keeping track of changes much easier and other interested devs can contribute if they want...
Hello, all.
Attempting to help by working on the build.prop settings. I'm S-OFF with revolutionary and have my device fully configured with the Honeycomb HBOOT/radio/bootloader/etc.
After flashing the ROM and the kernel, the device would immediately reboot upon attempting to return to the main recovery menu. I had to fastboot oem rebootRUU and reflash Globatron's PG41IMG-WIFI-light-B10 zip to get recovery back.
The device never proceeded to the boot animation.
Thracks said:
Hello, all.
Attempting to help by working on the build.prop settings. I'm S-OFF with revolutionary and have my device fully configured with the Honeycomb HBOOT/radio/bootloader/etc.
After flashing the ROM and the kernel, the device would immediately reboot upon attempting to return to the main recovery menu. I had to fastboot oem rebootRUU and reflash Globatron's PG41IMG-WIFI-light-B10 zip to get recovery back.
The device never proceeded to the boot animation.
Click to expand...
Click to collapse
I can confirm this happened to me also. The only difference is that I was using the kernel for a port of mine, and it did the same thing.
Yes, it is known that for some reason, Go Back already reboots the device. It's no problem though, it works fine anyway. If it is really looping the bootanimation every ten seconds, keep Vol Down and Power pressed, until you get to HBOOT. Select FASTBOOT, and enter 'fastboot erase cache' in the terminal. Then reboot on the device, and it should boot up fine.
Also, kernel won't work on Honeycomb, haven't checked why it doesn't though (and I won't btw, our goal is not to make a new kernel for Honeycomb, but for ICS, where it does its job fine).
Kernel source is uploaded to my Github: http://github.com/Chaosz-X
Edit the CROSS_COMPILE for your toolchain, and it should compile fine (it does here).
Will add the link to the OP within 5 minutes, ramdisk will be uploaded tomorrow.
If chaos has a device tree set up for the Flyer/View, no build prop editing should be necessary, can just submit requests - but I don't notice it on the git.
Thanks for uploading, I'll have a look and see what has changed so far. It's a little difficult to keep track of the changes since there are no commits that document the patches applied over time, but I'll try to make do with what you've got.
Is this straight from flyer HC kernel source?
I may be doing this wrong, but the touch screen commands do not seem to be working for me
Cimer said:
I may be doing this wrong, but the touch screen commands do not seem to be working for me
Click to expand...
Click to collapse
You have to type it in manually. Copying and pasting does not work, at least it did not for me. I think the issue is with the '
lawguy said:
You have to type it in manually. Copying and pasting does not work, at least it did not for me. I think the issue is with the '
Click to expand...
Click to collapse
It may just be me, I tried manual too. I had to use http://forum.xda-developers.com/showthread.php?t=1379875 since I'm stuck on Windows 8..maybe something is up with that. I guess I will just need to sit and wait =/
Cimer said:
It may just be me, I tried manual too. I had to use http://forum.xda-developers.com/showthread.php?t=1379875 since I'm stuck on Windows 8..maybe something is up with that. I guess I will just need to sit and wait =/
Click to expand...
Click to collapse
Hang tight for now. Wifi Does not work and the screen freaks out. I'm sure that these things will be fixed soon.
Going to upload ramdisk source within an hour, and try to apply a fix for WiFi, and try to fix the touchscreen by using the command in init.rc mentioned here earlier.
The new modem arrived, I have Internet again, so we have remote access to our Linux machine and can start compiling etc again
Indeed, we're grabbing AOKP source code now. Within about 2 hours we think, we'll start compiling the ROM for the HTC Flyer. Don't know if it will work out, but we have to start somewhere
Ever since I've moved to AOSP based roms (SlimKat, CyanogenMod, OmniRom, Resurrection-Remix,...) , I've noticed that in the DDMS tool (a tool for developers) that it shows all processes, and not just debuggable processes, including all third party apps' processes and OS's processes.
This never occurred for me on stock roms, no matter which device or stock rom I've looked at.
This issue annoys me as a developer, since I usually run only a single app that I develop and when I go to DDMS to choose the app's process, I need to find it amongst many processes (even of WhatsApp) , and, provided it hasn't changed its position on the list, click it and debug it.
So I asked about this on some kernels forums, and usually people didn't mind about it (and for good reason - it can bother only few people).
I've found the next tweak which can do it, but since I'm a complete newb in creating/modifying kernels, I have no idea how to patch it:
setprop ro.debuggable=0
Click to expand...
Click to collapse
Found from here .
I've contacted a few kernel developers and one of them agreed to check it out. I've tested it on CM and it worked just as it was intended :
Only debuggable apps' processes were visible and ready to be debugged.
Anyway, I wish to ask whoever is an expert in developing/modifying kernels those questions:
1. Why isn't this tweak enabled by default?
I mean, this is not the normal way roms work. It's not like this on any stock rom I've seen, not even on Nexus stock roms...
I also don't get why it's this way even for "final/stable" version roms (meaning not nightlies or experimental ones), since you'd probably not need to be able to debug them (users won't give you their device just for that...) .
2. Can you please explain to me, step by step (but don't assume I know anything about kernels development), how to take a rom's zip, and change its kernel so that this tweak will be enabled?
Maybe I could make an automatic tool that does it, either in Java for desktop or for Android.
3. Is there any way to enable this tweak without modifying the kernel? Maybe by using a tool that works with it?
Or maybe, provided we got root, change something to enable it?
Lenovo Tab 3 8/TB3-850F
Question & Answer Forum
Link & General Resource Guide
RE: Stock & Custom ROM(s)
INTRODUCTION & PURPOSE:
I have started this thread as a Q&A forum and general info guide for my Stock 6.0 ROM for the Lenovo Tab 3 8, and as a forum for my stock-based custom ROM which is nearing completion. In addition, I have posted some useful threads for this device below. It has been brought to my attention by XDA Senior Member @pndwal, as well as other members, that topics regarding systemless rooting, dm-verity, force encryption, Trusted Execution Environment (/tee1 & /tee2), etc., need a dedicated forum for open discussion and brain-storming, all while adhering to XDA's policy of staying on topic in the device threads. In addition to the topics outlined, this forum welcomes any questions, concerns and colloquy regarding ROMs, kernels, recoveries, & other development for this device.
LENOVO TAB 3 8/TB3-850F LINKS:
Stock Android 6.0 ROM Thread: https://forum.xda-developers.com/android/general/rom-lenovo-tab-3-8-tb3-850f-t3617594
Bootloader, TWRP & Rooting: https://forum.xda-developers.com/android/general/guide-lenovo-tab3-8-tb3-850f-t3559786/page17
Unbricking/Restoration Tutorial: https://forum.xda-developers.com/android/help/lenovo-tab-3-8-tb3-850f-unbrick-root-t3598727
Stock Firmware Partition Images: https://forum.xda-developers.com/android/general/rom-lenovo-tab-3-8-tb3-850f-android-6-0-t3593043
Lenovo Tab 3 8" User Manual (PDF)
for TB3-850F & TB3-850M: English: https://drive.google.com/file/d/18gCTfuZecJnlB0ddBIN02YPaDpuvmzov/view?usp=drivesdk
Lenovo File Manager (APK): https://forum.xda-developers.com/android/general/app-lenovo-file-manager-lenovo-tab-3-8-t3706161
RULES:
Rule 1: Be respectful to XDA policies and to one another. We are all here to learn and to grow. The only stupid question is a question not asked;
Rule 2: Read Rule 1.
MotoJunkie01 said:
It has been brought to my attention by XDA Senior Member @pndwal, as well as other members, that topics regarding systemless rooting, dm-verity, force encryption, Trusted Execution Environment (/tee1 & /tee2), etc., need a dedicated forum for open discussion and brain-storming, all while adhering to XDA's policy of staying on topic in the device threads.
Click to expand...
Click to collapse
MotoJunkie01 said:
@pndwal I would like to invite you, first & foremost, to the ROM Q&A for this tablet. I'm hopeful that the Q&A will provide answers to relevant questions like yours, and provide input, user experiences, & ideas for future ROM development for the Tab 3 8.
Click to expand...
Click to collapse
A generous interpretation... Thanks for invitation. I'll be sure to come up with a curly one or two.
For now, could you give a rough idea of what needed changing to bypass force encryption? And also, reasons for pre-rooting? Thanks, PW.
pndwal said:
A generous interpretation... Thanks for invitation. I'll be sure to come up with a curly one or two.
For now, could you give a rough idea of what needed changing to bypass force encryption? And also, reasons for pre-rooting? Thanks, PW.
Click to expand...
Click to collapse
Reasons for pre-rooting are merely as a convenience to the user. Many users will use the stock build I compiled and root themselves, while others will take the more convenient route and flash the ROM which is pre-rooted. The advantages of the rooted ROM are that it will be moderately debloated, deodexed, zipaligned, and will of course have BusyBox binaries pre-injected and the Magisk Systemless User Interface installed.
To bypass force encryption, I first needed to bypass dm-veriity. Both are enabled by default on this tablet within the ramdisk/fstab. So the boot image needed to be unpacked, the values of "1" enabling both dm-verity & force encryption needed to be changed to values of "0", thus disabling both features. The boot image was then repacked and then archived within my ROM installation package (zip).
I'd also like to take a poll to ask whether members want Viper4AndroidFX v2.5.0.5 (with NEON audio drivers) on the custom ROM, in lieu of the Dolby Digital Sound audio package that comes pre-installed on the tablet. Any input would be appreciated.
MotoJunkie01 said:
I'd also like to take a poll to ask whether members want Viper4AndroidFX v2.5.0.5 (with NEON audio drivers) on the custom ROM, in lieu of the Dolby Digital Sound audio package that comes pre-installed on the tablet. Any input would be appreciated.
Click to expand...
Click to collapse
Viper can be installed as a module in magisk. PW
pndwal said:
Viper can be installed as a module in magisk. PW
Click to expand...
Click to collapse
Yes it can, but it breaks the stock camera completely, due to the SELinux policy mod which is installed by the module. If included as a feature of the ROM, I have a workaround which sets SELinux to permissive on boot, instead of permanently disabling the policy from within the kernel. This prevents breaking of the stock camera.
MotoJunkie01 said:
Yes it can, but it breaks the stock camera completely, due to the SELinux policy mod which is installed by the module. If included as a feature of the ROM, I have a workaround which sets SELinux to permissive on boot, instead of permanently disabling the policy from within the kernel. This prevents breaking of the stock camera.
Click to expand...
Click to collapse
I did that in past too. Don't think its needed now.
Viper working fine for me as Magisk module (phone and TB3-850f tablet). SELinux enforcing, camera working fine (incl. video), 2.5.0.4 driver, 2.5.0.5 app.
Seems SELinux policy change no longer required as V4A module for Magisk installs AML (Audio Modification Library) as Magisk framework which no longer requires permissive. Let me know if I'm missing something. PW.
pndwal said:
I did that in past too. Don't think its needed now.
Viper working fine for me as Magisk module (phone and TB3-850f tablet). SELinux enforcing, camera working fine (incl. video), 2.5.0.4 driver, 2.5.0.5 app.
Seems SELinux policy change no longer required as V4A module for Magisk installs AML (Audio Modification Library) as Magisk framework which no longer requires permissive. Let me know if I'm missing something. PW.
Click to expand...
Click to collapse
You are pretty well correct. But, in order to enjoy the full spectrum of the NEON audio drivers, SELinux must go permissive. The Viper version I am speaking of was compiled by Deuteronomy Sound Technologies (and Arise) and is known as Viper4 AriseFX (a modded, themeable, and undoubtedly the most feature packed Viper package available for Android). The Viper module available on Magisk is a bare bones package, and when used in combination with my patched boot image (due to SELinux policy) it breaks the stock camera and causes other instabilities within the ROM.
I'm working also on a custom kernel for the TB3-850F, which will have some audio tweaks available from the kernel itself, as well as some tuneable governors, preset TCP congestion algorithms, etc.
MotoJunkie01 said:
You are pretty well correct. But, in order to enjoy the full spectrum of the NEON audio drivers, SELinux must go permissive. The Viper version I am speaking of was compiled by Deuteronomy Sound Technologies (and Arise) and is known as Viper4 AriseFX (a modded, themeable, and undoubtedly the most feature packed Viper package available for Android). The Viper module available on Magisk is a bare bones package, and when used in combination with my patched boot image (due to SELinux policy) it breaks the stock camera and causes other instabilities within the ROM.
I'm working also on a custom kernel for the TB3-850F, which will have some audio tweaks available from the kernel itself, as well as some tuneable governors, preset TCP congestion algorithms, etc.
Click to expand...
Click to collapse
I see. Perhaps you could explain the benefits of neon audio - I wasn't aware.
Also, would permissive SELinux not break SafetyNet check? Thanks, PW.
pndwal said:
I see. Perhaps you could explain the benefits of neon audio - I wasn't aware.
Also, would permissive SELinux not break Safety net check? Thanks, PW.
Click to expand...
Click to collapse
Yes and no. Depending on how you set the SELinux policy. I've found that setting SELinux to permissive on boot only, by way of a third party app like Kernel Adiutor-Mod or The SELinux Toggler, does not break SafetyNet. However, permanently disabling SELinux as enforcing, by way of modding the kernel itself, has been reported to cause a SafetyNet fail on both custom and stock ROMs.
You raise a good question though, and it is a factor to which I'll be paying close attention during development for this tablet.
I think I've decided to include Viper4AriseFX in my ROM as optional, and available by flashing a separate zip installer subsequent to installing the ROM itself.
By the way @pndwal, you seem to know your way around Android pretty well. What are some other features you would like to see in a custom built ROM for the Tab 3 8?
MotoJunkie01 said:
Yes and no. Depending on how you set the SELinux policy. I've found that setting SELinux to permissive on boot only, by way of a third party app like Kernel Adiutor-Mod or The SELinux Toggler, does not break SafetyNet. However, permanently disabling SELinux as enforcing, by way of modding the kernel itself, has been reported to cause a SafetyNet fail on both custom and stock ROMs.
You raise a good question though, and it is a factor to which I'll be paying close attention during development for this tablet.
I think I've decided to include Viper4AriseFX in my ROM as optional, and available by flashing a separate zip installer subsequent to installing the ROM itself.
Click to expand...
Click to collapse
Optional install sounds like a good idea, especially if SELinux permissive is optional to. I'm hesitant to use permissive environment as many apps etc require enforcing, and attempts to circumvent this, eg Magisk's ability to hide permissive etc, are reportedly not foolproof.
Not sure about attraction with Neon and ARISE, but seems permissive SELinux requirement to use these may be short-lived anyway. See post from today, at https://forum.xda-developers.com/an...d-systems-auditory-research-t3379709/page3125
problem:
NEON Enabled: No
Enabled: No
Status: Abnormal
ETC...
How solve it please?
Magisk 14.3 thats why. You need permissive. ARISE hasnt been updated for the new changes to how magisk handles selinux. Just for now you'll need to switch to permissive to get viper to work. Hopefully we'll have a build up soon to make it work, I know that ZackPTG and Ghost started to update it although i'm not sure on progress.
Hope it helps, PW.
pndwal said:
Optional install sounds like a good idea, especially if SELinux permissive is optional to. I'm hesitant to use permissive environment as many apps etc require enforcing, and attempts to circumvent this, eg Magisk's ability to hide permissive etc, are reportedly not foolproof.
Not sure about attraction with Neon and ARISE, but seems permissive SELinux requirement to use these may be short-lived anyway. See post from today, at https://forum.xda-developers.com/an...d-systems-auditory-research-t3379709/page3125
problem:
NEON Enabled: No
Enabled: No
Status: Abnormal
ETC...
How solve it please?
Magisk 14.3 thats why. You need permissive. ARISE hasnt been updated for the new changes to how magisk handles selinux. Just for now you'll need to switch to permissive to get viper to work. Hopefully we'll have a build up soon to make it work, I know that ZackPTG and Ghost started to update it although i'm not sure on progress.
Hope it helps, PW.
Click to expand...
Click to collapse
I've got Viper4AriseFX fully functional with Magisk v14.3. SafetyNet pass. I'm using a method which does not permanently set SELinux to permissive, but which toggles it to permissive only upon boot.
My pre-patched boot image is probably key to the successful installation as well. I'll list a complete change log of the exact mods once my beta release is ready. From what I'm gathering on logcat, Viper4AriseFX is "seeing" SELinux as permissive, while other system components are seeing the policy as enforcing. I believe I've stumbled upon the key to this SELinux policy dilemma.
MotoJunkie01 said:
By the way @pndwal, you seem to know your way around Android pretty well. What are some other features you would like to see in a custom built ROM for the Tab 3 8?
Click to expand...
Click to collapse
No really unique ideas, but interested in improved performance (speed and battery), battery being pretty good already.
MT8161p CPU specs say Quad-Core, 1.3 GHz, but TB3-850f is limited to 1.0 GHz, so a kernel modified to allow overclocking should achieve 30% boost easily (and CPU can usually go ~30% higher than specs, so perhaps 1.7 GHz or so would be achievable) unless I'm missing something.
Could improve Adoptive Memory (SD) handling, but may have to wait for port to N or O for this as anomalies with handling Dev. manifest values to make moveable apps automatically go to SD (as well as not allowing some even to be moved manually that should move fine) seem to an Android M limitation. (Works beautifully/ as expected on my phone with lineage N, and if Dev. hasn't enabled 'move apps to SD', can 'Force allow apps on external' in Developer Options [Makes any app eligible to be written to external storage, regardless of manifest values].) Guess N / O may be a way off, but would be nice.
Lenovo is now pushing ~30MB OTA update to TB3-850F_S100031_171010_ROW, so you'll probably want to capture / incorporate this in your ROM. (Is it possible to modify OTA update to allow flashing on rooted devices without restoring stock recovery? [Edit: Seems this would require merging differential update with complete files updated, and likely should only be flashed as complete ROM unless stock restored. Saw this: https://twrp.me/faq/officialota.html]) This is not available as a complete downloadable ROM on Russian Websites (lenovo-forums.ru or 4pda.ru) or others as yet as far as I can see.
Hope ideas are helpful, PW.
pndwal said:
No really unique ideas, but interested in improved performance (speed and battery), battery being pretty good already.
MT8161p CPU specs say Quad-Core, 1.3 GHz, but TB3-850f is limited to 1.0 GHz, so a kernel modified to allow overclocking should achieve 30% boost easily (and CPU can usually go ~30% higher than specs, so perhaps 1.7 GHz or so would be achievable) unless I'm missing something.
Could improve Adoptive Memory (SD) handling, but may have to wait for port to N or O for this as anomalies with handling Dev. manifest values to make moveable apps automatically go to SD (as well as not allowing some even to be moved manually that should move fine) seem to an Android M limitation. (Works beautifully/ as expected on my phone with lineage N, and if Dev. hasn't enabled 'move apps to SD', can 'Force allow apps on external' in Developer Options [Makes any app eligible to be written to external storage, regardless of manifest values].) Guess N / O may be a way off, but would be nice.
Lenovo is now pushing ~30MB OTA update to TB3-850F_S100031_171010_ROW, so you'll probably want to capture / incorporate this in your ROM. (Is it possible to modify OTA update to allow flashing on rooted devices without restoring stock recovery? [Edit: Seems this would require merging differential update with complete files updated, and likely should only be flashed as complete ROM unless stock restored. Saw this: https://twrp.me/faq/officialota.html]) This is not available as a complete downloadable ROM on Russian Websites (lenovo-forums.ru or 4pda.ru) or others as yet as far as I can see.
Hope ideas are helpful, PW.
Click to expand...
Click to collapse
You may have read the wrong specs. There has been much confusion between the TB3-850M (which runs on the MT8161p), and the TB3-850F (which runs on the MT6735m). A lot of online sources have misstated the board platforms on these two tablets.
My TB3-850F build.prop clearly lists the board platform as MT6735m, as does the hardware reading given by the SP Flash Tool when I sync the device on my PC. I wish mine did have the MT8161 (1.3GHz) versus the 1.0GHz of my MT6735.
But just to be certain there are not variants of the TB3-850F, read your build.prop via file explorer or build prop editor and let me know if yours is in fact the MT8161p.
Thanks for the heads up on the OTA. I wasn't aware of it and I'll definitely be updating my stock ROM thread accordingly.
Oh to answer your question on the OTAs, yes you can flash an OTA to a rooted/modified device by editing the updater-script and omitting the crypto-hash checks performed during the typical OTA installation, and by editing the incremental target and source lines (or omitting them entirely).
Just got the 33mb OTA captured. Looks like there are the usual bug fixes & stability improvements, but also the KRACK exploit has been patched, the partition index updated, and kernel updates. I'm currently compiling an up-to-date stock ROM with TWRP flashable installer.
MotoJunkie01 said:
You may have read the wrong specs. There has been much confusion between the TB3-850M (which runs on the MT8161p), and the TB3-850F (which runs on the MT6735m). A lot of online sources have misstated the board platforms on these two tablets.
My TB3-850F build.prop clearly lists the board platform as MT6735m, as does the hardware reading given by the SP Flash Tool when I sync the device on my PC. I wish mine did have the MT8161 (1.3GHz) versus the 1.0GHz of my MT6735.
But just to be certain there are not variants of the TB3-850F, read your build.prop via file explorer or build prop editor and let me know if yours is in fact the MT8161p.
Thanks for the heads up on the OTA. I wasn't aware of it and I'll definitely be updating my stock ROM thread accordingly.
Oh to answer your question on the OTAs, yes you can flash an OTA to a rooted/modified device by editing the updater-script and omitting the crypto-hash checks performed during the typical OTA installation, and by editing the incremental target and source lines (or omitting them entirely).
Click to expand...
Click to collapse
Yes, I'd forgotten this confusion.
Actually, didn't check online sources, but assumed Phone Tester was reporting correctly. It gives CPU Info: MT8161p.
Just checked Kernel Auditor which gives MT8161p as vendor, but MT6735 as hardware.
CPU-Z gives MT6735 as SoC.
My build.prop also gives ro.board.platform=mt6735m, but also ro.lenovo.cpuinfo=MT8735P.
Russian 4PDA forum gives 850f Processor Type: MT8161, with MT8735 for 850M variant. (http://4pda.ru/devdb/lenovo_tab3_8:850f)
So it's hard to know what or who's correct, but looks to me that newer CPUs have likely been installed on boards originally designed for MT6735. (My CPU could actually be MT8735 as given by build.prop if Lenovo had excess chips from 850M, or simply decided to use these in both models. - I guess they may even have started with MT6735 for 850f before progressively using MT8161 and MT8735.)
So seems to me to be in the realms of possibility that a kernel allowing overclocking may just render spectacular results (as long as later CPUs were in fact used, and these are not crippled by an older board chipset.) But then, I may be way off base . . . Let me know your thoughts. PW.
pndwal said:
Yes, I'd forgotten this confusion.
Actually, didn't check online sources, but assumed Phone Tester was reporting correctly. It gives CPU Info: MT8161p.
Just checked Kernel Auditor which gives MT8161p as vendor, but MT6735 as hardware.
CPU-Z gives MT6735 as SoC.
My build.prop also gives ro.board.platform=mt6735m, but also ro.lenovo.cpuinfo=MT8735P.
Russian 4PDA forum gives 850f Processor Type: MT8161, with MT8735 for 850M variant. (http://4pda.ru/devdb/lenovo_tab3_8:850f)
So it's hard to know what or who's correct, but looks to me that newer CPUs have likely been installed on boards originally designed for MT6735. (My CPU could actually be MT8735 as given by build.prop if Lenovo had excess chips from 850M, or simply decided to use these in both models. - I guess they may even have started with MT6735 for 850f before progressively using MT8161 and MT8735.)
So seems to me in the realms of possibility that a kernel allowing overclocking may just render spectacular results (as long as later CPUs were in fact used, and these are not crippled by an older board chipset.) But then, I may be way off base . . . Let me know your thoughts. PW.
Click to expand...
Click to collapse
Yeah I agree with you wholeheartedly that overclocking - at least moderately - could provide some benefit. I'm seeing that our current 1.0GHz maximum clock could safely be overclocked to about 1.33GHz. Definitely something I will be considering. I've been able to optimize this device's meager 1GB RAM by zipaligning the apk files, and by setting a maximum background process limit to build.prop. I've also found that setting the stock kernel's Adaptive Low Memory Killer to Very Aggressive helps as well.
I've installed the OTA and it seems to improve general stability of the device, and provides us with much needed security patches. I'll be updating my stock ROM to the most recent version later today. Here is a link to the captured OTA if anyone is interested in exploring it. However, in order to flash it to a rooted/modified device, the updater-script first needs to be modified. But again, I'll have the updated build posted later today.
OTA_TB3-850F_S100031_171010_ROW: https://drive.google.com/file/d/11xo-7X06ST1RV8X5TJHjM5o2MHfcsNhl/view?usp=drivesdk
MotoJunkie01 said:
Yeah I agree with you wholeheartedly that overclocking - at least moderately - could provide some benefit. I'm seeing that our current 1.0GHz maximum clock could safely be overclocked to about 1.33GHz. Definitely something I will be considering. I've been able to optimize this device's meager 1GB RAM by zipaligning the apk files, and by setting a maximum background process limit to build.prop. I've also found that setting the stock kernel's Adaptive Low Memory Killer to Very Aggressive helps as well.
I've installed the OTA and it seems to improve general stability of the device, and provides us with much needed security patches. I'll be updating my stock ROM to the most recent version later today. Here is a link to the captured OTA if anyone is interested in exploring it. However, in order to flash it to a rooted/modified device, the updater-script first needs to be modified. But again, I'll have the updated build posted later today.
OTA_TB3-850F_S100031_171010_ROW: https://drive.google.com/file/d/11xo-7X06ST1RV8X5TJHjM5o2MHfcsNhl/view?usp=drivesdk
Click to expand...
Click to collapse
Great. What do you make of build.prop entry: ro.lenovo.cpuinfo=MT8735P? PW.
pndwal said:
Great. What do you make of build.prop entry: ro.lenovo.cpuinfo=MT8735P? PW.
Click to expand...
Click to collapse
It's simply a typo that was made in build.prop I believe. Going by ro.board.platform, they seem to have it correct with the 6735 entry. I was just messing with some components of the unpacked boot image from this device, and it seems that except for the SoC differences, the 850M and 850F are otherwise identical. Of course, the 850M, by way of its mt8161p, has full SIM & 4G/LTE data support. In our variant, the SIM slot (next to the micro SD card slot) is blocked off, whereas it's fully accessible in the 850M.
P.S, if you're rooted, add this line to build.prop and let me know if you see a difference in RAM optimization: ro.config.low_ram=true
I can't believe I am still struggling with this.....
I've flashed an AOSP 7.1 build by @amaces from the Unlegacy Android site. Based on what has been said about rooting AOSP builds, I flashed addonsu-phh-arm-new.zip and then installed me.phh.superuser.1033.apk. After a reboot, I get the dreaded "su binary is out of date", the SU app does not respond to any changes and I don't have root.
I've gone round and round on this, first treating it like the AOSP 7.0 I am running on another Nook (which only required the apk installation) and then trying the double barrel assault.
I am stumped.
nmyshkin said:
Based on what has been said about rooting AOSP builds, I flashed addonsu-phh-arm-new.zip and then installed me.phh.superuser.1033.apk. After a reboot, I get the dreaded "su binary is out of date", the SU app does not respond to any changes and I don't have root.
Click to expand...
Click to collapse
Are you sure the control APK is equivalent to the Play Store version (i.e., looks for the same SU binary signatures)? It's possible PHH did update his SU backend; I'll investigate.
Something that I didn't really mention in the FAQ, with current AOSP builds, you can also install the CM backend (addonsu-14.1-arm-new.zip, or the official version at https://download.lineageos.org/extras), which doesn't need any extra UI APK. I had adapted some of the CM14 patches such that the CM/Lineage blurb applies to AOSP as well.
On another note, wouldn't it be more productive to post these questions in the respective threads? Visibility is probably much lower when creating these independent question threads...
Edit: I did look at PHH's SU packages, and they all have the exact same SU backend included in addonsu-phh-arm-new.zip, so I don't have any explanation other than incorrect APK.
amaces said:
Are you sure the control APK is equivalent to the Play Store version (i.e., looks for the same SU binary signatures)? It's possible PHH did update his SU backend; I'll investigate.
Something that I didn't really mention in the FAQ, with current AOSP builds, you can also install the CM backend (addonsu-14.1-arm-new.zip, or the official version at https://download.lineageos.org/extras), which doesn't need any extra UI APK. I had adapted some of the CM14 patches such that the CM/Lineage blurb applies to AOSP as well.
On another note, wouldn't it be more productive to post these questions in the respective threads? Visibility is probably much lower when creating these independent question threads...
Edit: I did look at PHH's SU packages, and they all have the exact same SU backend included in addonsu-phh-arm-new.zip, so I don't have any explanation other than incorrect APK.
Click to expand...
Click to collapse
Mmm....I'll keep plugging away at it. Must be missing something.
Sorry for the new post, but I wasn't sure it would be appreciated to insert questions about the Nook Tablet into the HD/HD+ thread. In retrospect I guess the issue is more general than tablet-specific since it refers to ROMs which have similar rooting issues.
nmyshkin said:
I wasn't sure it would be appreciated to insert questions about the Nook Tablet into the HD/HD+ thread. In retrospect I guess the issue is more general than tablet-specific since it refers to ROMs which have similar rooting issues.
Click to expand...
Click to collapse
Sorry, I think I misread the initial question; you were asking about UA acclaim builds, whereas I referred to my personal builds. While I maintain both, only my personal repositories are patched in such a way to utilize CM14.1 SU. The goals of the UA project are different (more about purity), but I can experiment with various things in my own builds. About PHH root, I suspect the builds you installed are missing the persist.sys.root_access property, so the SU backend never starts up. You might be able to get PHH root going on the UA builds by adding
Code:
persist.sys.root_access=3
to your build.prop (remount system, or in recovery).
amaces said:
Sorry, I think I misread the initial question; you were asking about UA acclaim builds, whereas I referred to my personal builds. While I maintain both, only my personal repositories are patched in such a way to utilize CM14.1 SU. The goals of the UA project are different (more about purity), but I can experiment with various things in my own builds. About PHH root, I suspect the builds you installed are missing the persist.sys.root_access property, so the SU backend never starts up. You might be able to get PHH root going on the UA builds by adding
Code:
persist.sys.root_access=3
to your build.prop (remount system, or in recovery).
Click to expand...
Click to collapse
Thanks for the lead. I learned something new in the editing of the build.prop file from recovery
Unfortunately, it did not give the desired result. After a little more work with the ROM it appears it is not quite ready for "prime time". It can't remember my WiFi password and does not have a browser. I guess I could always adb push the AOSP browser (which I like), but the WiFi issue is beyond my pay grade. On the plus side, it does shut down properly (and......"restart" reboots into recovery....not quite what I expected)
I'll just keep an eye on future releases.
Edit: Ha! One more reboot and this time it actually connected to WiFi by itself! Maybe worth a fifth look.....
nmyshkin said:
It can't remember my WiFi password and does not have a browser.
Click to expand...
Click to collapse
The deprecated AOSP browser had become a security risk lately, and it was removed in the UA builds. My own builds contain a different browser, albeit from CM/Lineage. Either way, there are many good options that you can sideload (e.g., Lightning) or install from Play Store.
nmyshkin said:
On the plus side, it does shut down properly (and......"restart" reboots into recovery....not quite what I expected)
Click to expand...
Click to collapse
I built things on the CM/Lineage kernel, and the bug was introduced there sometimes last year. After a few bisection attempts, the root cause hasn't been found yet.
The solution is a patched version of Magisk, courtesy of @fddm. See https://forum.xda-developers.com/nook-tablet/help/acclaim-patched-magisk-tester-t3752775
I'm not responsible for any damage to your device of any sort.
By flashing this you take responsibility of anything that happens.
Process at your own risk!
If you want to check firmware - please make backup. Better to store all partitions, at least efs and may be modem.
As we all know - the tablet does not support treble. And we do not have a separate partition, with enough space (about 450mb) to place the whole vendor there.
As a result, I created a solution and tried to make it as much easy as possible for daily use.
I created 2 files - system.img/vendor.img as ext4 images and put it in to the system partition. And then on early init I mounting real system partition as /system_src and then mount system.img as /system and vendor.img as /vendor.
Also I created twrp treble's version for easy use /system and /vendor. It's can wipe new system\vendor partitions, you can restore images to system/vendor in the same way, as we use for boot/recovery.
How to
Prepare - need only once
1. in the current twrp make full wipe (at least we need clean /system, because later we fill all /system by 2 files: system.img/vendor.img)
2. you need flash my twrp with treble support (sm-t825-twrp_3.2.3-0_treble.img): https://drive.google.com/file/d/1GJMrHI4wkePXh8JmF_jgxBSly38B5tu1/view?usp=sharing
3. flash zip with new system.img/vendor.img as ext4 images: https://drive.google.com/file/d/14lwrnM85CoxcpjTUiJ0yfbGHXLa7k0DB/view?usp=sharing
Installation
1. download treble zip from Links and flash it as usual zip. It will flash boot.img and /vendor partition (this is part, specific for our tablet only)
2. download any treble system (64bit System A-only) image from: https://forum.xda-developers.com/project-treble/trebleenabled-device-development, if it not a .img file - extract it, then copy to sdcard/extsd card and then flash image as system_image in twrp (Install - switch to Install Image and flash as System Image)
3. some images require vendor's modification, in oreo some firmware has a 90 rotate, some 270, in my treble zip 90, if you see - image rotated on 180 - flash zip from Additional Links
4. if you wanna back to original lineage\rr without treble - you need flash normal twrp version and make full wipe or just flash stock firmware via odin
Known Issues
sound 2 skeakers only for now
fingerprint
keys light dnw on some firmwares
Links
2. 20180911: fix rotate, back to stock sound (fix bt, quad speakers): https://drive.google.com/file/d/1e_mzosWYL7bv29n613OKZpvSZPXMh26k/view?usp=sharing
1. 20180802: initial treble: https://drive.google.com/file/d/14lwrnM85CoxcpjTUiJ0yfbGHXLa7k0DB/view?usp=sharing
Few notes about item 2: your tablet has wrong image orientation, panel horizontal, sensor - vertical, as result troubles with show right image in right orientation. In 7.0-8.1 this problem was fixed by adding ro.sf.hwrotation=270 in build.prop. For pie this way dnw. Google remove this option and update configstore with new ISurfaceFlingerConfigs. It was added new method: primaryDisplayOrientation. Problem was in adding new configstore sources in the 15.1 sources. Finally I had successfully builded configstore 1.0/1.1 from 16.0 in 15.1.
Additional Links - need only for some oreo images, if image rotated on 180
2. 20180911: https://drive.google.com/file/d/1S4mvXFcX-d2xoZEp9b9dG3ywUhybT6U2/view?usp=sharing
1. 20180802: initial fix rotate: https://drive.google.com/file/d/1JKGu7vudrU0bJPEdkGwBVFkyCJpsHLLs/view?usp=sharing
Magisk
For some reason original Magisk dnw with our treble. It's somehow connect to sepolicy and I'm almost sure about it connected to our treble way with loop0/loop1 mount.
Finally I made small update of the original latest Magisk and it's working. Only one thing was done by me: inside zip in the binary magiskinit file I remove seclabel (seclabel u:r:magisk:s0) for magisk services.
Fixed 17.1 Magisk beta: https://drive.google.com/file/d/1lMhM8ceaJoy5pHRHjCsxxG2C9e5sB5Ot/view?usp=sharing
17.1 Magisk beta manager (this apk also inside Magisk, but cann't be installed automatically during Magisk install), install it as normal apk after install Magisk: https://drive.google.com/file/d/1qCoVPUxvvxopCGfIpoHdAnKueNIxMXal/view?usp=sharing
GApps
https://opengapps.org
Almost all firmware say about wrong vendor version. This is normal, just press OK.
Checked firmwares
AospExtended (need rotate 270 patch): https://forum.xda-developers.com/pr...evelopment/rom-aospextended-rom-v5-5-t3797509
Cosmic-OS: https://forum.xda-developers.com/pr...ment/cosmic-ospulsar8-1-0201805243-2-t3794806
DroidOnTime: https://forum.xda-developers.com/pr...pment/official-droidontime-dotos-2-x-t3794338
Havoc-OS: https://forum.xda-developers.com/pr...-device-development/rom-havoc-os-8-1-t3819050
Lineage: https://forum.xda-developers.com/pr...evice-development/lineage-phh-treble-t3767690
OmniROM Treskmod (need rotate 270 patch): https://forum.xda-developers.com/pr...development/rom-8-1-omnirom-treskmod-t3818188
Pixel Experience (it's requere register custom firmware): https://forum.xda-developers.com/pr...e-development/8-1-0-pixel-experience-t3796011
Resurrection Remix: https://forum.xda-developers.com/pr...opment/resurrection-remix-phh-treble-t3767688
let's experiment together!
Wow! Man, you rock! My Favourite dev! Thanks for Your hard work!
Thanks for this. One question, do we use A/B or A images?
Peeps1513 said:
Thanks for this. One question, do we use A/B or A images?
Click to expand...
Click to collapse
Forgot: system a only
Valera1978 said:
Forgot: system a only
Click to expand...
Click to collapse
And what did You think about the bugs You have mentioned? (Fingerprint, deepsleep, speakers) Fixable or not?
deepsleep, speakers - this is the common tab s3 problems (same in the original lineage), a bit later will back to it.
fingerprint - need to check, in the treble mode shim libs dnw because of linker in the system, also service from vendor don't see libs in system - i.e. some limitations. For camera I made workaround: launch second cameraserver service with my shim lib in environment, but this way dnw for fingerprint. It should be something easy to fix, but I still don't see it.
Valera1978 said:
deepsleep, speakers - this is the common tab s3 problems (same in the original lineage), a bit later will back to it.
fingerprint - need to check, in the treble mode shim libs dnw because of linker in the system, also service from vendor don't see libs in system - i.e. some limitations. For camera I made workaround: launch second cameraserver service with my shim lib in environment, but this way dnw for fingerprint. It should be something easy to fix, but I still don't see it.
Click to expand...
Click to collapse
Really appreciate your work :highfive:
Nice project. Good stuff.
I finally settled on HAVOC after much playing around but am facing an issue trying to install magisk (16.7), no success yet . When I searched [ROM][TREBLE] Havoc-OS 8.1 it seems that magisk has been installed on other devices so could it be a vendor issue? Any help would be appreciated. phh supersu seems to be fine though.
Peeps1513 said:
I finally settled on HAVOC after much playing around but am facing an issue trying to install magisk (16.7), no success yet . When I searched [ROM][TREBLE] Havoc-OS 8.1 it seems that magisk has been installed on other devices so could it be a vendor issue? Any help would be appreciated. phh supersu seems to be fine though.
Click to expand...
Click to collapse
Yep. I've tested treble but no sucess atm for me. Better stay with non-treble lineage, as most of the things work just fine. Maybe in future if Valera fix some things on treble bulid (if ever). Hope Dev focus on oreo lineage than treble project (which is good stuff, but not good enough as daily driver)
I want to leave a feedback after almost a month with treble roms
I have to change my mind: treble is fantastic! Such a great thing, very helpfull for all of us. I was running Valera's lineage 15.1 as daily driver for about 2 months.
In regular non-treble bulid almost everything work. Bad thing were performance (slow animations) and not working deepsleep.
I was happy but not at al. That wasnt this what i was looking for. Then i found again this thread and again give a try (longer) :fingers-crossed:
I am running now treble omnirom for few weeks. Worst things i find is: magisk did not work (very sad, tried multiple versions with no luck, other users report magisk is working for them) and my bluetooth headphones arent working (probably our side bug too)
In plus: no random reboots, performance is very good, no slow downs, deepsleep problem almost gone (drain 6% overnight connected to wifi)
Didnt find any other bugs
I am really happy now. It would be exelent with magisk and bluetooth audio working
Great job, i stick with treble for now :laugh:
Thanks for your great work, Valera!
Have anyone tried this on SM-T820 (Wi-Fi only version)?
imder said:
Thanks for your great work, Valera!
Have anyone tried this on SM-T820 (Wi-Fi only version)?
Click to expand...
Click to collapse
Work only with T825
Any updates Master?
@Valera1978
Could you PLEASE make this for us left out t820 owners???
How does this rom with Lineage compare to your non-treble build? Any reason to switch if I want to stick with Lineage?
robsmith11 said:
How does this rom with Lineage compare to your non-treble build? Any reason to switch if I want to stick with Lineage?
Click to expand...
Click to collapse
You better wait for Pie relase. Valera is working on it.
suzook said:
@Valera1978
Could you PLEASE make this for us left out t820 owners???
Click to expand...
Click to collapse
Without device this is impossible. Look like somehow vendors is differ in 820/825, and I need device to understand, what is wrong, why vendor from 825 dnw on 820 (except ril, may be gps). It should be something simple. Because on my another samsung's tablet on exynos no diffs at all. Treble for wifi version I had made based on lte version in first attempt.
robsmith11 said:
How does this rom with Lineage compare to your non-treble build? Any reason to switch if I want to stick with Lineage?
Click to expand...
Click to collapse
I'll update non-treble build soon. And treble too. During working on treble I had fixed some stuff, like hexedit finger libs to remove shim lib. But I need fix audio in calls again first.
About pie - I dont see any reason to start work on it right now, it's too early, it still has a lot of problems. Better spend this time to make something cool - may be even make samsung port from note 9 or tab s4 with dex and some stuff.
But if you don't want use treble (I agree, it's may be too complicate with special twrp and additional partitions), I think I can prepare integrated pie system images with simple flash without any additional manipulations, like you flash lineage.
Valera1978 said:
Without device this is impossible. Look like somehow vendors is differ in 820/825, and I need device to understand, what is wrong, why vendor from 825 dnw on 820 (except ril, may be gps). It should be something simple. Because on my another samsung's tablet on exynos no diffs at all. Treble for wifi version I had made based on lte version in first attempt.
Click to expand...
Click to collapse
Man, if you would be somehow able to build a unified or a separate treble rom for 820, you would be the greatest man on xda! Cause we have no support for 820 anymore...