Magisk General Support / Discussion

[m0han]

Yeah, he had problem with Basic Integrity because he didn't properly configure DenyList

then i did as per your advice. yet, basic integrity failed.

....Sometimes a combination of MHPC and USNF is needed to pass CTS, sometimes one or the other....

i suppose this held good for me.

 

[pndwal]

I'm not familiar with Pixel and A12, but did you see:

No, I hadn't.

I'm not across all the issues w/ Android 12 either. I was really just moving the OT (non-issue related queries re. Magisk's future) in that issue thread to this General Discussion forum...

But I see you've brought this information to the attention of early A12/Magisk guinea pigs in same issue. Looks like it may be quite relavent too.

Hope this helps them / devs addressing that issue. PW

 

[V0latyle]

Folks:

Please realize that the "failure to load/verify boot images" message is not due to a Magisk issue.

Android 12 introduces Verified Boot on some devices, such as the Pixel 5, which will block a modified boot image from loading.

Two new Verified Bootfeatures implemented in Android 12 will interfere with attempts to root.

Dm-verity (device-mapper-verity) is a method by which an image on block devices (the underlying storage layer of the file system) can be checked to determine if it matches an expected configuration, using a cryptographic hash tree. If the hash doesn't match, dm-verity prevents the stored code from loading.

Vbmeta verification is the other half of this - it provides a cryptographically signed reference hash which is used to verify the integrity of /boot, /system, and /vendor partitions. The vbmeta image is only used to verify /boot, while vbmeta-system is used to verify /system.

This was implemented to prevent persistent rootkits by means of a hardware level security check, to prevent "potentially harmful applications" such as Magisk from evading detection, as such applications residing within the kernel will have higher privileges than the detection applications.

What this means is that with these two enabled, a modified boot image will cause a verification error when flashed to the device, preventing boot. Interestingly, this check is not performed against "live" boot images loaded via ADB, so with dm-verity and vbmeta verification enabled, a modified image can be booted as long as the image in /boot is intact.

So, if you get stuck at bootloader with the failure message, this is because boot verification is not disabled.

You can find instructions on how to disable it in my thread here (this guide is for the Pixel 5; the instructions may be similar for other devices but it is YOUR responsibility to determine the correct procedure)

As far as failing safetynet, I used the Universal Safetynet Fix on 12 Beta 5 with no issues. I haven't updated to 12 Final yet so YMMV.

 

[pndwal]

If I understand you correctly, you got it passing without the use of MHPC...

But you can also keep it enabled - if reset and not used, it's dormant and not affecting...

I stand to be corrected here, but I think MHPC now changes 'sensitive' props that MagiskHide used to handle to 'safe' (undetected) values by default (even if MHPC is not configured).

This may do no harm of course, but it should be understood that MHPC is no longer 'dormant' when reset or left unconfigured.

Also, as USNF now apparently does this also, this function may be duplicated / redundant, although I imagine this wouldn't cause any real conflict, only increase load time marginally. PW

 

[pndwal]

Good old Magisk alpha 31....bulletproof

I made a GPay purchase for days and even a few hours earlier...then bang, gone...root detected....

Remove and reinstall alpha 31...all is sane again....

Guess im just going to stay here for a while again until the dust settles, cant keep up with (currently) 3 forks of Magisk

...Nigel...

...But that version (abfb3bb3-alpha) was already dated... did you ever try later 8276a077-alpha? I haven't had issues yet... PW

 

[zgfg]

I stand to be corrected here, but I think MHPC now changes 'sensitive' props that MagiskHide used to handle to 'safe' (undetected) values by default (even if MHPC is not configured).

This may do no harm of course, but it should be understood that MHPC is no longer 'dormant' when reset or left unconfigured.

Also, as USNF now apparently does this also, this function may be duplicated / redundant, although I imagine this wouldn't cause any real conflict, only increase load time marginally. PW

Didn't know and didn't expect (changes in effect if not activated through the module).
However, my /data/adb/modules/MagiskHidePropsConf/system.prop is empty (besides the header), hence the module makes no props changes through that file for me

Yet, my sensitive props are already ok by Xiaomi.eu ROM and I pass SafetyNet with or without MHPC

Anyway, since I usually experiment (also systemlessly) with changing props directly by manipulating my /data/adb/modules/.../system.prop file, I have now disabled that dormant or not MHPC

 

[Didgeridoohan]

I stand to be corrected here, but I think MHPC now changes 'sensitive' props that MagiskHide used to handle to 'safe' (undetected) values by default (even if MHPC is not configured).

This may do no harm of course, but it should be understood that MHPC is no longer 'dormant' when reset or left unconfigured.

Also, as USNF now apparently does this also, this function may be duplicated / redundant, although I imagine this wouldn't cause any real conflict, only increase load time marginally. PW

The sensitive props changing is active by default, yes, but only ever does anything if a prop is set to a sensitive value.

 

[pndwal]

Didn't know and didn't expect (changes if not activated in the module).
However, my /data/adb/modules/MagiskHidePropsConf/system.prop is empty (besides the header), hence the module makes no props changes through that file for me

Yet, my sensitive props are already ok by Xiaomi.eu ROM and I pass SafetyNet with or without MHPC.
Anyway, since I usually experiment (also systemlessly) with props changes by directly manipulating my /data/adb/modules/.../system.prop file, I have now disabled the dormant or not MHPC

So seems no effect where prop already changed, which explains your result.

Ref for active setting of 'trigering' values to 'safe' is here:
https://github.com/Magisk-Modules-R...README.md#setreset-magiskhide-sensitive-props PW

 

[73sydney]

...But that version (abfb3bb3-alpha) was already dated... did you ever try later 8276a077-alpha? I haven't had issues yet... PW

wasnt dated by much

i tend not to invite the possibility of more of the same, when i can go right back to stable ...im getting too old for much testing these days Marvin

 

[pndwal]

wasnt dated by much

i tend not to invite the possibility of more of the same, when i can go right back to stable ...im getting too old for much testing these days Marvin

Yup, only one more TJW commit:
https://github.com/topjohnwu/Magisk/commit/fe41df87bb34926cf2d6b448dd5a2878d90d4dc1
fixing this race condition:
https://github.com/topjohnwu/Magisk/pull/4759
Check the best minds on that: reopened, and just closed again...

Clearly some vvb2060 commit(s) too (esp. removing Check SafetyNet).

The fact that vvb2060 pushed fix so quickly should say something. Others reported abfb3bb3-alpha issues gone too.

8276a077-alpha is all good for me ATM... We'll see...

However, I always try to remember 'you can lead a mouse to water, but you cannot make it think'... ... Perhaps life on the bleeding edge does require too much of that... so just don't talk to me about life... Marvin.

 

[pndwal]

This'll be fun:

@MishaalRahman (late of XDA) posted (on twitter) he just got WSA (Windows Subsystem for Android) up and running with Windows 11 (yes, you can sideload APKs!).

Of course, John is now pulling out all stops to 'root' it with Magisk! - Magisk for Windows; what a treat!

There's no ramdisk, but he's located 4 stage init (last 2 are Android 2SI)

He says "I found the init, the kernel binary, and initrd.img. Let's get things cooking baby!"... Go John! PW

 

[arik100]

unfortunately search doesn't work properly for me.

i get error installed N/A
poco X3 pro
installed magisk,
patched boot via magisk
flashed patched boot via fastboot and then via magisk.

i have root but i get this "status"
any ideas?
i run arrow os android 12 (arrowos recovery)

 

[pndwal]

unfortunately search doesn't work properly for me.

i get error installed N/A
poco X3 pro
installed magisk,
patched boot via magisk
flashed patched boot via fastboot and then via magisk.

i have root but i get this "status"
any ideas?
i run arrow os android 12 (arrowos recovery)

Did you have Magisk installed / working on that device previously w/ App hidden? If so, see:
https://xdaforums.com/t/magisk-general-support-discussion.3432382/post-85761971

If you didn't update with Magisk App hidden, there are current issues with this symptom, eg:
1) https://github.com/topjohnwu/Magisk/issues/4319
- SU present but not working. Could try this workaround:
https://github.com/topjohnwu/Magisk/issues/4319#issuecomment-851840533

2) https://github.com/topjohnwu/Magisk/issues/4174
- SU working, but this affects devices with Android 6.x, 7.x.

3) FWIW, here's an old one that affected ArrowOS:
https://github.com/topjohnwu/Magisk/issues/4005

I suspect you have hidden app conflict, but there have been very few trials / issues reported w/ Android 12 as yet. PW

 

[pndwal]

This'll be fun:

@MishaalRahman (late of XDA) posted (on twitter) he just got WSA (Windows Subsystem for Android) up and running with Windows 11 (yes, you can sideload APKs!).

Of course, John is now pulling out all stops to 'root' it with Magisk! - Magisk for Windows; what a treat!

There's no ramdisk, but he's located 4 stage init (last 2 are Android 2SI)

He says "I found the init, the kernel binary, and initrd.img. Let's get things cooking baby!"... Go John! PW

LSPosed guys have Windows 'Root' w/ Magisk! ... and Windows Xposed!

Well... Magisk, Riru and LSPosed are now working on WSA!

Brief steps (note this is just a tmp solution)
1. Enable dev mode in win, extract WSA package, disable its signature verification and remove private permissions, and sideload it.
2. Download kernel sources of WSA, write a kernel module to provide su, compile it and relpace the kernel.
3. With su, compile magisk and run its emulator.sh script, then grant su to adb.
4. Compile Riru commenting out codes that restart zygote and install it.
5. Download and install LSPosed.
6. adb reboot userspace.
7. Enjoy

LSPosed Telegram group.

PW

 

[pndwal]

LSPosed guys have Windows 'Root' w/ Magisk! ... and Windows Xposed!

LSPosed Telegram group.

PW

Drooling?

John Wu, 50m

Yup, we managed to get kernel assisted SU up and running in WSA, which bootstraps Magisk using my emulator scripts, and the rest is business as usual.

(Also by "we", I mean Magisk + Riru + LSPosed folks as we are in close collaboration)

Proper Magisk installation would require some additional research and investigation. But it's always fun to get the screenshots out and have people drool over it, doesn't it

To drool or not to drool... That is the question. PW

 

[wilsonhlacerda]

Drooling?


To drool or not to drool... That is the question. PW

Now they should try installng BitGapps systemless on it, probably it will be not that hard ( https://xdaforums.com/t/custom-gapps-bitgapps-for-android.4012165/ )

Amazon, Microsoft, Google will "love" that.

 

[73sydney]

Now they should try installng BitGapps systemless on it, probably it will be not that hard ( https://xdaforums.com/t/custom-gapps-bitgapps-for-android.4012165/ )

Amazon, Microsoft, Google will "love" that.

It cant run any Google Services or any apps that reply on them...personally i dont see the point in Android on Windows

 

[pndwal]

It cant run any Google Services or any apps that reply on them...personally i dont see the point in Android on Windows

Perhaps it's just the drool factor. ...

But I used to live in the same town as Sir Edmund Hillary, and what did he do?

Spoiler:
„˙ǝɹǝɥʇ sɐʍ ʇı ǝsnɐɔǝq„ - ¿ʎɥM
¡ƃuıɥʇ ǝɥʇ pǝqɯıןɔ puɐ 'uıɐʇunoɯ ǝɥʇ oʇ ʇuǝʍ ɥʇoq ǝɥ ˙˙˙ ,ʇuıod ǝɥʇ ǝǝs ʇ,uop I, pıɐs ɹǝʌǝu ǝH

I can't really see the point of Magisk here either (except for some modules?), but I'll watch the proceedings for sure ... as long as top Magisk / XPosed / Riru Devs are excited about it, why not? (Some watch cricket!)

What do I know... Perhaps they can add GMS / get GApps going!

Perhaps Google will even add Windows compatibility if this thing flies! ...

Or perhaps pigs might...

... Until then, Windows users will at least be instantly able to run a plethora of favourite Android apps on their PC's it seems. - That's gotta be Value Added functionality for many users and a selling point for MS! PW

{Mod edit: ISPOILER added to upside-down text - Oswald Boelcke, Moderator}

My Edit: Thanks @Oswald Boelcke - good tip!

 

[DiamondJohn]

For devs, virtual android on windows is instrumental, and with easy root is brilliant. There are countless very strong reason for it. Even for non devs, having an environment to test that is not your daily is priceless in itself. Do you honestly think Mircro$oft went down this difficult route solely for bragging rights?

 

[arik100]

Did you have Magisk installed / working on that device previously w/ App hidden? If so, see:
https://xdaforums.com/t/magisk-general-support-discussion.3432382/post-85761971

If you didn't update with Magisk App hidden, there are current issues with this symptom, eg:
1) https://github.com/topjohnwu/Magisk/issues/4319
- SU present but not working. Could try this workaround:
https://github.com/topjohnwu/Magisk/issues/4319#issuecomment-851840533

2) https://github.com/topjohnwu/Magisk/issues/4174
- SU working, but this affects devices with Android 6.x, 7.x.

3) FWIW, here's an old one that affected ArrowOS:
https://github.com/topjohnwu/Magisk/issues/4005

I suspect you have hidden app conflict, but there have been very few trials / issues reported w/ Android 12 as yet. PW

1. what do you mean with app hidden?
2. about other options - i don't see any solution... just mentions, but again, i do have root , just that i get N/A and i can't install modules (which is annoying...)