[Discussion] Magisk Delta - Another unofficial third-party Magisk fork

Search This thread

NippleSauce

Senior Member
Jun 23, 2013
546
373
AdAway is working for some things, but is clearly not for these URLs as there is ad spew all over the pages. These are blocked with vanilla Magisk.

Try:


or

No ads on either of those pages for me. So, something is either messed up with your root access or AdAway configuration. I, however, am only using the three default hosts sources in my AdAway setup and didn't add any other hosts sources.

Perhaps going through your rooting process steps can help to resolve the problem? For example, are you using direct install within Delta or Magisk? For a while now, doing that has not worked as it patches your boot.img instead of your init_boot.img. Perhaps the use of sulist is causing the issue that you are facing - as I am not utilizing that feature. I am only using zygisk, bootloop protection and hide.

Anyway, if you write out the steps taken then we can see if there were any missteps and can chime in to help =). If it was all correct, then perhaps we'll be stumped...but it always helps to work together. Teamwork makes the dreamwork - as the old saying goes =D
 

krakout

Senior Member
Hello! I'm getting errors in the log and I wanted to revert from canary to stable on my P7P. However, when I hit update, I get an "error downloading file".

Any ideas?


Screenshot_20230318-210852.png

Update: ah, unhiding it allowed the download, but updating the app fails? It just closes, next I open it, it's still canary and prompts me to update...
 
Last edited:

_litz

Senior Member
Oct 29, 2011
622
278
Alpharetta, GA
No ads on either of those pages for me. So, something is either messed up with your root access or AdAway configuration. I, however, am only using the three default hosts sources in my AdAway setup and didn't add any other hosts sources.

Perhaps going through your rooting process steps can help to resolve the problem? For example, are you using direct install within Delta or Magisk? For a while now, doing that has not worked as it patches your boot.img instead of your init_boot.img. Perhaps the use of sulist is causing the issue that you are facing - as I am not utilizing that feature. I am only using zygisk, bootloop protection and hide.

Anyway, if you write out the steps taken then we can see if there were any missteps and can chime in to help =). If it was all correct, then perhaps we'll be stumped...but it always helps to work together. Teamwork makes the dreamwork - as the old saying goes =D

Hi, no I'm a very very manual person when it comes to things like this ... I flash the updates directly from ADB/Fastboot, as well as directly patching init_boot from the app (and flashing it as well). Automated processes are nice, but ask anyone who's trusted their Tesla to drive for them how well that worked out ....

Root is working as designed in all aspects except for this. It's as if one of two things is happening ... the hosts file overlay is not present (despite AdAway not throwing any errors) or apps in general not seeing the overlaid hosts file (being presented with the default empty hosts file).

I only have zygisk (to allow for Universal Safetynet to function), bootloop protection, and magiskhide.
 
  • Like
Reactions: pocketrule

zgfg

Senior Member
Oct 10, 2016
8,977
6,716
Redmi K20 / Xiaomi Mi 9T
Xiaomi Mi 11 Ultra
Hi, no I'm a very very manual person when it comes to things like this ... I flash the updates directly from ADB/Fastboot, as well as directly patching init_boot from the app (and flashing it as well). Automated processes are nice, but ask anyone who's trusted their Tesla to drive for them how well that worked out ....

Root is working as designed in all aspects except for this. It's as if one of two things is happening ... the hosts file overlay is not present (despite AdAway not throwing any errors) or apps in general not seeing the overlaid hosts file (being presented with the default empty hosts file).

I only have zygisk (to allow for Universal Safetynet to function), bootloop protection, and magiskhide.
Assuming that AdAway was granted root and that you enabled Systemless hosts and rebooted

If running SuList mode, AdAway must be enabled in the Confgure Sulist (and reboot then)
 

_litz

Senior Member
Oct 29, 2011
622
278
Alpharetta, GA
Assuming that AdAway was granted root and that you enabled Systemless hosts and rebooted

If running SuList mode, AdAway must be enabled in the Confgure Sulist (and reboot then)

Correct, and in both cases yes. AdAway is not reporting any issues, in addition if I view the hosts file I'm seeing the proper number of entries (e.g. thousands of lines).
 
  • Like
Reactions: pocketrule

NippleSauce

Senior Member
Jun 23, 2013
546
373
Hi, no I'm a very very manual person when it comes to things like this ... I flash the updates directly from ADB/Fastboot, as well as directly patching init_boot from the app (and flashing it as well).....

I only have zygisk (to allow for Universal Safetynet to function), bootloop protection, and magiskhide.
Gotcha. I do it all manually as well.

So long as you only patched the init_boot with Magisk/Delta and didn't flash it with Magisk/Delta, then you should be all good there. But regardless, the systemless host issue that you are facing likely requires a device wipe to be resolved. But, I know doing that really stinks =/.

I tend to update my P7P without the "-w" comment until I run into issues like the one which you are currently experiencing. Once something like that occurs, I tend to make another backup and then re-image my phone with the same factory image but with the "-w" command intact. After restoring from the backup, everything works normally again and I no longer face whatever issue I was running into.

Of course, the issue can probably be resolved with some root file manager by removing a few files or directories....I just don't know which ones would have to be deleted - hence why I wipe my phone and re-image.
 
  • Like
Reactions: pocketrule

_litz

Senior Member
Oct 29, 2011
622
278
Alpharetta, GA
Gotcha. I do it all manually as well.

So long as you only patched the init_boot with Magisk/Delta and didn't flash it with Magisk/Delta, then you should be all good there. But regardless, the systemless host issue that you are facing likely requires a device wipe to be resolved. But, I know doing that really stinks =/.

I tend to update my P7P without the "-w" comment until I run into issues like the one which you are currently experiencing. Once something like that occurs, I tend to make another backup and then re-image my phone with the same factory image but with the "-w" command intact. After restoring from the backup, everything works normally again and I no longer face whatever issue I was running into.

Of course, the issue can probably be resolved with some root file manager by removing a few files or directories....I just don't know which ones would have to be deleted - hence why I wipe my phone and re-image.
It's easier, frankly, to use AdGuard DNS ...

Wiping and reimaging is truly a nasty hassle, not the least of which is having to re-register all the Okta tokens for work.

So probably gonna just chalk this up to "we're doing something the manufacturer doesn't recommend" and move onwards.

Everything else works fine, so not that big a issue. Just weird.
 

samtron6

Member
Mar 10, 2023
21
3
Moto G play 2023. installed, flashed patched boot.img has bootloop. android 12.. reflashed with origianl patched boot.img from MAGISK and it boots.. idk if it didnt patch the .IMG right or what
 
Isn't it.

If I disable AdAway entirely, and DNS filter via AdGuard Public DNS -- using the SAME filters -- all ads vanish as expected.

Instead of troubleshooting more just uninstall AdAway which will restore the default hosts file. Reboot and install again... That's helped me in the past.

AdAway has given me enough problems I now just use adblocking Magisk modules which don't require systemless hosts be enabled
 

Top Liked Posts

  • There are no posts matching your filters.
  • 3
    If official magisk lets you disable zygisk,
    It does...
    then yes. You literally just make sure you are hiding root from starling (on official magisk you can do this with denylist i think, but shamiko won't work without zygisk so be sure to enable magisk's own denylist toggle,
    ...but Denylist requires Zygisk too.
    on delta it's "MagiskHide") and reboot with zygisk disabled. Log into starling then when you're in, reboot with zygisk enabled again. When starling gets to one day left on their timer, redo the process.
    I did have the Firebase url blocked in adaway, unsure if that's strictly required now but it's probably doing no harm to have it blocked.
    👍

    FWIW, I've already tried this with my current Alpha Magisk setup which is a POC for the upcoming native bridge Zygisk loading refactoring. It seems it's not yet enough to hide Zygisk from this app, but even initial work is not yet completed
    IMG_20230521_122937.jpg

    and native bridge compatible Shamiko is also brand new.

    I suspect when the final function above and testing are completed and NB Zygisk hits official Magisk (expected with 27.0) Zygisk may be properly hidable finally just as NB based Riru was.

    Of course Riru had integrated RiruHide, but Zygisk will always rely on third party hiding going to forward so some further compatibility tweaks for Shamiko will probably also be required for proper Zygisk hiding...

    Meanwhile, someone else might be interested in trying official Magisk with firebaseremoteconfig.googleapis.com blocked and @huskydg's p-trace based portable MagiskHide...

    Please say if passing PI deviceIntegrity verdict is also needed... (I suspect it will be, but some banks don't check this.) If it is, @huskydg's (improved) forked Riru build and his Riru-based forked safetynet-fix build could also be applied to TJW Magisk...

    I may try this when time permits, but real Starling users may have more motivation.

    👍 PW

    Edit:

    Tried this with current Alpha w/ MagiskHide instead of Denylist (Zygisk disabled), and voilà!:
    IMG_20230521_133635~2.jpg

    ... I can confirm PI deviceIntegrity is not checked as it is failing and blocking firebaseremoteconfig.googleapis.com was not required either... (12 May 2023 update.)
    2
    Is magisk delta dead?
    This on TG Channel dt. 23-May-2023
    1
    I'm fairly new to MD, I have fdac22ba installed on Pixel 7 Pro. Its currently set up using Magisk Hide but I want to change it and use SU List instead but when I tried it I had trouble with 2 modules which showed as installed but I couldn't open them, BCR & AOSPMods. Anyone have similar issue with these apps. Or is there a specific way it has to be done. I tried reinstalling, rebooting etc.
    Install Su-List Auto-WhiteList module (scroll up in this thread to find about), or manually add module's apps to SuList (and reboot)
    1
    Starling bank app seems to be detecting root now regardless of Delta/Alpha,modules, or setting combos. Super frustrating.
    1
    Try this, it's per-app. Maybe it could help. (Lsposed)

    No luck with this either

    Edit:
    Got it working by disabling Zygisk entirely. App gives you 7 days from root detection once logged in so I guess I'll just turn zygisk off for a boot cycle once a week to keep the app going. Pain in the ass or what.
  • 54
    This is not an officially supported topjohnwu project.
    If you are looking for official Magisk source, please go to this page


    Introduction

    Custom Magisk fork by HuskyDG. Sync with official Magisk adding back MagiskHide.
    http://huskydg.github.io/magisk-files

    REPORT BUG IS STILL ACCEPTED ON XDA THREAD BUT PLEASE INCLUDE ADDITIONAL INFORMATION SUCH AS LOGCAT AND MAGISK LOG
    7
    Revive MagiskHide from Magisk v23.0
    • The implementation of MagiskHide is ptrace Zygote process, every forks of Zygote will be notified and traced also.
    • MagiskHide from Magisk v23.0 which monitors every thread spawn event of Zygote fork (app process is heavily a multithreads process which will spawn threads to trigger MagiskHide to check UID and cmdline)
    • There is an exception that app zygote does not spawn threads and thus it won’t trigger MagiskHide to unmount Magisk and detach. (MagiskDetector and Momo take advantages of this fact to detect MagiskHide through ptrace)
    • To fix this problem, we trace the syscalls prctl() instead of thread spawn event of Zygote fork like MagiskHide in Magisk v23.0
    • After processes has been forked from zygote, there will be atleast prctl() is called to change the process name. For normal app process and isolated process, the process name will be changed as followed:
      • zygote -> (unknown name) -> <pre-initialized> -> (process name). So the key is <pre-initialized>, after that we can guess it is target process or not.
    • For app zygote, there is only once prctl() is called to change process name: zygote -> package.name_zygote.
    • The changing process name happens before apk is being loaded so we can detach it from ptrace, do unmount all Magisk files and nearly there is no traces left after that.
    6
    Thread re-opened
    6
    And now Bingo, with today's Magisk-017cca27-delta(25205), Momo does no more detect Zygisk

    - Zygisk and MagiskHide enabled, Shamiko disabled
    👍
    Btw, maybe I should start thinking of Livin' by Mandiri 😁
    Interesting analysis from @5ec1cff here:
    https://github.com/5ec1cff/my-notes/blob/master/analyse-livin.md

    Seems Livin / others may use smap for detection... This study discusses TJW's sanitize environment commit not working properly and LSP fixes in Shamiko as well as efforts needed to hide zygisk hooks, esp for functions of libandroid_runtime.so:
    https://github.com/5ec1cff/my-notes/blob/master/new-idea-detect-zygisk.md
    ... I note dev refers to @canye's (Magisk Bravo) analysis of environment variables too...

    This article analyses more recent Zygisk changes and Dev's attempts to hide zygisk /proc/self/attr/current exposure etc:
    https://github.com/5ec1cff/my-notes/blob/master/zygisk-new-start-mode.md
    ... it ends with this conjecture:
    "Maybe it's time to consider loading zygisk with native bridge?"
    More on this and comparison w/ riru hiding / methods and plans for hiding Zygisk here:
    https://github.com/5ec1cff/my-notes/blob/master/maru.md

    Many of the ideas revealed in Dev's notes above are implemented in Magisk Maru fork (and adopted for now in Delta)... See top 4 commits here:
    https://github.com/5ec1cff/Magisk/commits/maru
    ... use of complete native/src/zygisk/elf_util.cpp file from LSPosed in inject with native bridge Is interesting...

    ... Many other interesting notes here:
    https://github.com/5ec1cff/my-notes
    incl. "build-magisk-on-windows" etc...

    ... I've only taken a cursory glance at notes I mentioned...

    Very credible Zygisk hiding efforts for a fork like Delta however... 👍 👀 PW