Magisk General Support / Discussion

[zgfg]

Thanks! I've now updated, all working fine on RN8T.

Latest @vvb2060 Magisk Alpha build (July 29, 2021) changes (since last GitHub release):​

Chinese Translated:

Magisk (1eb83ad8-alpha-29)​

• Fix the default shell path of busybox

Nb. This build is available only on Telegram at present.
https://t.me/magiskalpha/327

I'll post full Changelog here when available on GitHub.

Last GitHub release, notes, links:
https://xdaforums.com/t/magisk-general-support-discussion.3432382/post-85387865

PW

Seems there is some 'sowing' problem with 29 - anyone could maybe decipher what would that mean in the Magisk concept?

Anyway, be warned

Google Translation from the Alpha Telegram channel:

:
Sudden sowization

bl Bloodfang:
Indeed, the sow turned ()

:
If there is no problem, push to the update channel

Magisk alpha discussion group:
The above is the problem

 

[jons99]

Seems there is some 'sowing' problem with 29 - anyone could maybe decipher what would that mean in the Magisk concept?

Anyway, be warned

Google Translation from the Alpha Telegram channel:

:
Sudden sowization

bl Bloodfang:
Indeed, the sow turned ()

:
If there is no problem, push to the update channel

Magisk alpha discussion group:
The above is the problem

30 is out

Magisk cannot handle oplus.fstab · Issue #3663 · topjohnwu/Magisk

After Installation 21.2. or Canary phone won't boot. There are several fstabs in boot.img. fstab. default, fstab.emmc and oplus.fstab. In cmdline androidboot.fstab_suffix = default. I managed to in...

github.com

 

[pndwal]

Latest @vvb2060 Magisk Alpha build (July 29, 2021) changelog:​

Chinese Translated:

alpha update log​

Magisk (1eb83ad8-alpha-30)​

• Based on 1eb83ad8, please refer to the upstream update log for related modifications
• Properly process any data from magiskd
• Support SharedUserId
• Delete the backup file after restoring the boot image
• Built-in current version update log
• Use the local version when the stub cannot be downloaded, now Magisk can be used completely offline
• Support bootimg v4 format
• Support bootconfig
• Detect /data/adb/magisk/ not updated and prompt to repair
• Remove the disabled and pending deletion marks when upgrading the module, and it is not allowed to mark as pending deletion before restarting
• Fix that it cannot be flashed in some TWRP
• Listen to the PACKAGE_FULLY_REMOVED broadcast to remove completely uninstalled apps from magiskhide
• Show modal waiting pop-up window when hiding/restore Magisk application
• App adapted to Android 12
• Android 12+ super user hides the screen overlay when the window pops up
• For devices supported by the kernel, MagiskSU uses proprietary devpts to bypass some application detection
• Rewrite MagiskSU's logic to open pty and delete extra sepolicy rules
• Fix incorrect signal sending in MagiskHide
• Add untrusted_app_30
• Prompt that some devices do not load the module sepolicy problem
• Add mount information to the log file
• Fix the default shell path of busybox *
• Support oplus.fstab for oppo series devices

https://github.com/vvb2060/Magisk/b...0d886c64224/app/src/main/res/raw/changelog.md

*Nb. Change since last build on GitHub 1eb83ad8-alpha-27

My earlier notes:
https://xdaforums.com/t/magisk-general-support-discussion.3432382/post-85098469

Current notes:

Check 'earlier notes' for installation method / details.

Alpha build users won't yet see Update button notice in Magisk App for this build as VersionCode has (again) NOT changed from 23001.

I'm not sure if this behaviour will be addressed, but users will observe the 'Latest' and 'Installed' version suffix now differs.

Latest in GitHub magisk_files repo is 1eb83ad8-alpha-30

Latest available from Telegram group is
1eb83ad8-alpha-30

https://xdaforums.com/t/magisk-general-support-discussion.3432382/post-85229385

Nb. In Magisk Documentation, @vvb2060 has updated "Internal Details" and "Deployment" since John did.
https://github.com/vvb2060/Magisk/tree/alpha/docs

PW

 

[pndwal]

Seems there is some 'sowing' problem with 29 - anyone could maybe decipher what would that mean in the Magisk concept?

Anyway, be warned

Google Translation from the Alpha Telegram channel:

:
Sudden sowization

bl Bloodfang:
Indeed, the sow turned ()

:
If there is no problem, push to the update channel

Magisk alpha discussion group:
The above is the problem

I think it's just a follow-on comment on "Recently high production " and your comment, which may not have been properly understood.

Still seems to be struggling with this Magisk issue:
https://github.com/topjohnwu/Magisk/issues/3663#issuecomment-888875725
Re '• Support oplus.fstab for oppo series devices' as we speak however...

Said "I tried my best, MTK asked for more blessings" (translated) on Telegram. PW

 

[m0han]

....I have got a test version thats simplified and removes some of the script duplication
(namely sepolicy) compared to the original, moved to Zacks extended module format,
with correct service and uninstall scripts, and sepolicy.rule. I also added an easy way to
switch between enforcing and permissive.....by default enforcing, need permissive? just
create a blank file named permissive on /sdcard and reboot

I just havent released that yet due to a really unpleasant experience with another member.
Put me back a ways in testing after that shock, so need to test it a bit further.

would be nice to check it out. waiting for it to be ready.

 

[pndwal]

would be nice to check it out. waiting for it to be ready.

& @73sydney

Also, could someone please explain to me what is the advantage of changing sepolicy for viper? It's still not clear to me.

Also, is Legacy mode still needed with this fix? And can someone say if Legacy mode has any disadvantage?

FWIW, I'm running Viper4Android FX 2.7.2.1 from Magisk repo w/ 2.5.0.4 driver on stock MIUI Android 10.

After installing and setting Legacy mode (Compatibility mode does nothing for me), Driver Status is Abnormal, but driver is always reset properly / working (for Legacy mode) simply by disabling module and rebooting, then enabling and rebooting again (2 reboots).

Hope someone knows! PW

 

[m0han]

... could someone please explain to me what is the advantage of changing sepolicy for viper? It's still not clear to me.

Also, is Legacy mode still needed with this fix? And can someone say if Legacy mode has any disadvantage?...

Just sharing my setup: ViPER4Android FX 2.7.1.6 (2716) w/ 2.5.0.4 driver on custom rom AICP 16.1 of 7-july (Android 11).

I don't see the setting for Legacy mode. Driver Status is Normal. Compatibility mode does nothing for me as well. My rom has selinux permissive atm.

 

[zgfg]

Just sharing my setup: ViPER4Android FX 2.7.1.6 (2716) w/ 2.5.0.4 driver on custom rom AICP 16.1 of 7-july (Android 11).

I don't see the setting for Legacy mode. Driver Status is Normal. Compatibility mode does nothing for me as well. My rom has selinux permissive atm.

@pndwal , @sydney73

I use Viper4AndroidFX 2.7.1.2, the latest available module from Repo, and with simply adding SEPolicy rules to the script, everything works perfect on ny A11 (Status Normal, Processing Yes when playing) without the need for SEKinux Permissive mode.
Only the Legacy mode for Driver must be enabled

The same solution has been posted elswhere (incl. the main V4A FX thread on XDA by some user - IDK was that the origin), here is one my post from another thread with the how-to:

There is actually an easier way to fix Viper4Android, follow the original post where the fix was described:

https://xdaforums.com/t/app-all-root-solutions-6-0-viper4android-fx-2-7.3774651/post-84121895

In other words, replace (use TWRP or root explorer like MiXplorer)
/data/adb/modules/ViPER4AndroidFX/post-fs-data.sh
with the attached file (download and unzip) and reboot

Still make sure that Legacy mode in V4A Settings is enabled

---

In case of installing V4A from scratch:

- install ViPER4Android_FX-v2.7.2.1.zip from Repo (through Magisk) and reboot
- open Viper4Android app and let it install the driver, it will reboot
- replace post-fs-data.sh as above and reboot once again
- open Viper4Android app and enable Legacy mode in its Setrings
- also, don't forget to enable Master limiter for e.g. speaker or headphones

Check the driver status, it must show Status Normal (and Processing Yes when playing nusic)

PS: The same was also discussed in this Magisk thread few months ago, actually there was another user who mentioned that solution here and brought me attention (unfortunately, cannot find his post now)

 

[pndwal]

@pndwal , @sydney73

I use Viper4AndroidFX 2.7.1.2, latest available module from Repo, and with simply adding SEPolicy rules to the script, everything works perfect, without the need for Permissive mode.
Only theLegacy mode must be selected

The same solution has been posted elswhere (incl. the main V4A FX thread on XDA by some user - IDK was that the origin ), here is one my post from another thread with the how-to:

PS: The same was also duscussed in this Magisk thread few months ago, actually there was another user who mentioned that solution (unfortunately, cannot find his post now)

mmm. But it works out-of-the-box for me after setting Legacy mode, disabling then re-enabling module (2 reboots) as described...

Seems everyone needs Legacy mode though, but also seems to me no SELinux policy change is needed to reset the driver for this mode...

Has anyone else here tried resetting / fixing just by turning module off, then on again? ... Perhaps I'm missing something. PW

 

[zgfg]

mmm. But it works out-of-the-box for me after setting Legacy mode, disabling then re-enabling module (2 reboots) as described...

Seems everyone needs Legacy mode though, but also seems to me no SELinux policy change is needed to reset the driver for this mode...

Has anyone else here tried resetting / fixing just by turning module off, then on again? ... Perhaps I'm missing something. PW

This way with SEPolicy rules in the script has advantage for me when updating my Xiaomi.eu weekly ROMs:

- Without, I had to go through the procedure with disabling, re-enabling V4A, maybe also AML (it was described in my post I mentioned), then reconfiguring again V4A app (its settings)

- However, upon I added SEPolicy rules section to the script, I can simply install the zip with new ROM, flash its boot.img (prepared in advance) patched by Magisk (flashing one after another in TWRP) , then I reboot and everything (incl. Viper4Android) works out of the box

 

[73sydney]

mmm. But it works out-of-the-box for me after setting Legacy mode, disabling then re-enabling module (2 reboots) as described...

Seems everyone needs Legacy mode though, but also seems to me no SELinux policy change is needed to reset the driver for this mode...

Has anyone else here tried resetting / fixing just by turning module off, then on again? ... Perhaps I'm missing something. PW

Theres not even legacy mode as an option for the version im using ...theres a compatibiliy mode switch, but thats it

Reason i and a lot of others use the Ichigo Viper zip is because its been the only one to work on A11 for us

Im not up to speed on all things Viper, as ive said before (and perhaps you dont understand my function here, lol) i just tend to repackage other peoples shizz to make it easier to use. Im all ears if theres better ways to do stuff

 

[73sydney]

This way with SEPolicy rules in the script has advantage for me when updating my Xiaomi.eu weekly ROMs:

- Without, I had to go through the procedure with disabling, re-enabling V4A, maybe also AML (it was described in my post I mentioned), then reconfiguring again V4A app (its settings)

- However, upon I added SEPolicy rules section to the script, I can simply install the zip with new ROM, flash its boot.img (prepared in advance) patched by Magisk (flashing one after another in TWRP) , then I reboot and everything (incl. Viper4Android) works out of the box

Hehe, theres a better explanation of the sepolicy stuff

 

[zgfg]

Hehe, theres a better explanation of the sepolicy stuff

Since I have also A11, you can try that nethod with V4A 2.7.1.2 instead of your older 2.7.1.6

SEPolicy rules I mention do not include SELinux Permissive, just:

allow audioserver audioserver_tmpfs file { read write execute }
allow audioserver system_file file { execmod }
allow mediaserver mediaserver_tmpfs file { read write execute }
allow mediaserver system_file file { execmod }
allow audioserver unlabeled file { read write execute open getattr }
allow hal_audio_default hal_audio_default process { execmem }
allow hal_audio_default hal_audio_default_tmpfs file { execute }
allow hal_audio_default audio_data_file dir { search }
allow app app_data_file file { execute_no_trans }
allow mtk_hal_audio mtk_hal_audio_tmpfs file { execute }

The point (follow-up to my previous post) is that sepolicy.rule file seems to be executed only when module was installed (and therefore overrulled to default when I later update the ROM)

Whereas the same SEPolicy rules commands put to post-fs-data.sh script are executed on each reboot (if Magisk is installed and V4A module enabled) - hence the V4A works again out-of-the-box even upon updating the ROM

 

[J.Michael]

Seems there is some 'sowing' problem with 29 - anyone could maybe decipher what would that mean in the Magisk concept?

Anyway, be warned

Google Translation from the Alpha Telegram channel:

:
Sudden sowization

bl Bloodfang:
Indeed, the sow turned ()

:
If there is no problem, push to the update channel

Magisk alpha discussion group:
The above is the problem

sow -- female pig that has reproduced?
sow -- plant seeds?

 

[zgfg]

sow -- female pig that has reproduced?
sow -- plant seeds?

Maybe (translation) what we called 'rapid fire' (day after day updates).
But then I don't understand where was the 'sow' problem with 29

 

[pndwal]

Theres not even legacy mode as an option for the version im using ...theres a compatibiliy mode switch, but thats it

Reason i and a lot of others use the Ichigo Viper zip is because its been the only one to work on A11 for us

Im not up to speed on all things Viper, as ive said before (and perhaps you dont understand my function here, lol) i just tend to repackage other peoples shizz to make it easier to use. Im all ears if theres better ways to do stuff

Yup, I'm getting that adding script can save reboots (X2) needed to reset driver on initial installation. ...

However I had wondered if SELinux changes were actually a necessity on some devices. It seems simple reboots / reset will still do the trick for the less time-conscious users, at least on latest 2.7.2.1 (Magisk repo).

Reason i and a lot of others use the Ichigo Viper zip is because its been the only one to work on A11 for us

I think @zgfg has 2.7.2.1 working on Android MIUI.

Also, I just installed it on my Mi Pad 4, LOS 18.1 (ie. Android 11, Alpha Magisk), and was surprised to find it needed only mandatory reboot after module Installation to work!

I did this: installed Viper4Android FX from Magisk repo but instead of reboot immediately, opened Viper4Android FX App (it's already installed), let it install missing Neon driver then reboot, open Viper4Android FX, check driver status; it's Abnormal, select Legacy mode in settings, check driver status again and it's Normal / working! (No reset needed.)

Not sure if it's Android 11, AOSP standard ROM or other that allows swapping to Legacy mode to work immediately, but my RN8T on stock MIUI, Android 10 & Alpha Magisk needs driver reset incl. reboots to work. PW

 

[73sydney]

Since I have also A11, you can try that nethod with V4A 2.7.1.2 instead of your older 2.7.1.6

SEPolicy rules I mention do not include SELinux Permissive, just:

allow audioserver audioserver_tmpfs file { read write execute }
allow audioserver system_file file { execmod }
allow mediaserver mediaserver_tmpfs file { read write execute }
allow mediaserver system_file file { execmod }
allow audioserver unlabeled file { read write execute open getattr }
allow hal_audio_default hal_audio_default process { execmem }
allow hal_audio_default hal_audio_default_tmpfs file { execute }
allow hal_audio_default audio_data_file dir { search }
allow app app_data_file file { execute_no_trans }
allow mtk_hal_audio mtk_hal_audio_tmpfs file { execute }

The point (follow-up to my previous post) is that sepolicy.rule file seems to be executed only when module was installed (and therefore overrulled to default when I later update the ROM)

Whereas the same SEPolicy rules commands put to post-fs-data.sh script are executed on each reboot (if Magisk is installed and V4A module enabled) - hence the V4A works again out-of-the-box even upon updating the ROM

Well now im confused, because the module im working on had as one of its "clean up" goals to remove the duplication of sepolicy.rule AND post-fs-data.sh

Adding to the mix is Johns last changelog for Magisk 23 including:

[General] Several sepolicy.rule copy/installation fixes

I would assume by that it was fucntioning properly to apply any sepolicy.rule files in a modules folder...but maybe not?

So i guess Ichigo was hedging his bets with having both (sepolicy.rule & post-fs-data.sh), but maybe that was trying to account for different magisk versions. Think its safe to say choosing one way would be best going forwards as most should be on 23 at least by now, one would hope. So we think post-fs-data is preferable, from the sound of it...

fwiw, riru-momohider does the same thing - doubles down with sepolicy.rule & post-fs-data....

Maybe i need to reevaluate messing with viper

 

[73sydney]

Yup, I'm getting that adding script can save reboots (X2) needed to reset driver on initial installation. ...

However I had wondered if SELinux changes were actually a necessity on some devices. It seems simple reboots / reset will still do the trick for the less time-conscious users, at least on latest 2.7.2.1 (Magisk repo).


I think @zgfg has 2.7.2.1 working on Android MIUI.

Also, I just installed it on my Mi Pad 4, LOS 18.1 (ie. Android 11, Alpha Magisk), and was surprised to find it needed only mandatory reboot after module Installation to work!

I did this: installed Viper4Android FX from Magisk repo but instead of reboot immediately, opened Viper4Android FX App (it's already installed), let it install missing Neon driver then reboot, open Viper4Android FX, check driver status; it's Abnormal, select Legacy mode in settings, check driver status again and it's Normal / working! (No reset needed.)

Not sure if it's Android 11, AOSP standard ROM or other that allows swapping to Legacy mode to work immediately, but my RN8T on stock MIUI, Android 10 & Alpha Magisk needs driver reset incl. reboots to work. PW

Imma try your method, but last time i tried with the repo version, it did not end well

edit:

wont even download the apk lol

Log: line 214: curl not found

time for Zacks Cross Compiled Binaries it seems

...still think my ways quicker...jesus....

So, yeah that didnt go so well either - my device couldnt be more connected to the internet btw

So yeah, the repo way is looking super efficent

+ echo '- Testing internet connection'
- Testing internet connection
+ /data/adb/modules_update/ccbins/curl -Ls --connect-timeout 3 -I https://www.google.com
/data/user/0/ym.uovnvdolycep/cache/flash/update-binary: /data/adb/modules_update/ccbins/customize.sh: line 4: /data/adb/modules_update/ccbins/curl: not found
+ /data/adb/modules_update/ccbins/busybox- grep -q 'HTTP/.* 200'
/data/user/0/ym.uovnvdolycep/cache/flash/update-binary: /data/adb/modules_update/ccbins/customize.sh: line 4: /data/adb/modules_update/ccbins/busybox-: not found
+ /data/adb/modules_update/ccbins/busybox- ping -q -c 1 -W 1 google.com
+ /data/adb/modules_update/ccbins/busybox- grep -q 'HTTP/.* 200'
/data/user/0/ym.uovnvdolycep/cache/flash/update-binary: /data/adb/modules_update/ccbins/customize.sh: line 4: /data/adb/modules_update/ccbins/busybox-: not found
+ /data/adb/modules_update/ccbins/curl -Ls --connect-timeout 3 -I https://www.baidu.com
/data/user/0/ym.uovnvdolycep/cache/flash/update-binary: /data/adb/modules_update/ccbins/customize.sh: line 4: /data/adb/modules_update/ccbins/curl: not found
+ /data/adb/modules_update/ccbins/busybox- ping -q -c 1 -W 1 baidu.com
+ return 1

Nope, theres currently no way my device can download the apk through the repo module

I mean i can download any other module i like and have it work, just not that one....

I can download the file from the same path in the browser (path from install script): https://zackptg5.com/downloads/v4afx.apk/v4afx.apk

Saving so much time folks

update:

So installed the apk
Opened
Installed driver
Rebooted
Nope...Abnormal
Enabled Legacy...working

Same driver version btw

And then it hit me that the reason i hated the new viper gui was it changed the older easier to use presets, because of course they changed everything to make it harder to use your years old presets and to (try) monetize DDC profiles etc...monetize what was once free...and put analytics in there. Pox on that.

Will be reverting to the Ichigo one thanks

 

[zgfg]

Well now im confused, because the module im working on had as one of its "clean up" goals to remove the duplication of sepolicy.rule AND post-fs-data.sh

Adding to the mix is Johns last changelog for Magisk 23 including:

[General] Several sepolicy.rule copy/installation fixes

I would assume by that it was fucntioning properly to apply any sepolicy.rule files in a modules folder...but maybe not?

So i guess Ichigo was hedging his bets with having both (sepolicy.rule & post-fs-data.sh), but maybe that was trying to account for different magisk versions. Think its safe to say choosing one way would be best going forwards as most should be on 23 at least by now, one would hope. So we think post-fs-data is preferable, from the sound of it...

fwiw, riru-momohider does the same thing - doubles down with sepolicy.rule & post-fs-data....

Maybe i need to reevaluate messing with viper

According to my post I quoted, I started with that procedure at the times of Magisk 21.6 (?)
It was a game changer for me at that point (simple weekly ROM updates)

If processing of SEPolicy rules changed with v23, I didn't care - I continued with the same simple ROM update procedures and did not play with the V4A sh script anymore

Anyway, I still don't understand why do you insist on the older V4A 2.7.1.6 (time ago I thought that going to an older V4A version would void the need for Legacy mode, but apparently you still need the Legacy driver mode)

 

[pndwal]

Maybe (translation) what we called 'rapid fire' (day after day updates).
But then I don't understand where was the 'sow' problem with 29

Seriously guys and gals, the issue is self evident:

What with so many Alpha sprouts sown (or propagated) from one Magic branch, we need to fork the ground or the sows will mistake the root for breakfast and the problem is xposed. What's more, the solution is burried in the code!

But we have help to decipher it! "Indeed, the sow turned"... Yes! "Sudden sowization"! ...

To crush this a little, the inference has got to be that a sow could easily uproot the problem and reproduce it!... With 29. other. sows... And pigs might fly! ... just like Canaries?!

It's so obvious! "Magisk alpha discussion group:
The above is the problem", even if the problem is above me!

Really hope nothing is lost in translation.