A certain module is crashing an app #7
Copy link
Copy link
Closed
Labels
questionFurther information is requestedFurther information is requested
Description
Steps to reproduce/复现步骤
- Install Android Faker module
- Select instagram under scope and reboot.
- Open instagram.
Expected behaviour/预期行为
- Instagram should open.
Actual behaviour/实际行为
- App crashes with crash information with lsposed logs complaining about something.
Xposed Module List/Xposed 模块列表
Android FakerMagisk Module List/Magisk 模块列表
Shamiko, Zygisk-Next, Zygisk-Assitant, Zygisk-Lsposed and some audio modules.LSPosed version/LSPosed 版本
1.10.0 (7082)
Android version/Android 版本
14
Magisk version/Magisk 版本
Magisk alpha 27006
Riru version/Riru 版本
N/A
Version requirement/版本要求
- I am using latest debug CI version of LSPosed and enable verbose log/我正在使用最新 CI 调试版本且启用详细日志
Logs/日志
LSPosed_2024-08-29T19_35_47.817727.zip
LSPosed_2024-08-29T19_35_52.300657.zip
Reactions are currently unavailable
Metadata
Metadata
Assignees
Labels
questionFurther information is requestedFurther information is requested
Activity
[-]A certain module is crashing the app[/-][+]A certain module is crashing an app[/+]Rtizer-9 commentedon Aug 29, 2024
Some info that might be helpful for you:
I've frequently faced this app crash error with android faker, geergit and ssl-killer modules with instagram or whatever is under scope with almost all releases of mywalkb. The pumpcin fork worked perfectly with all these modules except the last release which was incompatible with waenhancer.
I'm telling you this because I think all these problems are linked to a common thing for which pumpcin already had patched his fork.
PS: Thanks for picking up the development of LSPosed, really appreciate it.
Dev4Mod commentedon Aug 29, 2024
On Android Faker, there is an option to fake the device (model name and Android version). From Android 14, it is necessary to specify RECEIVER_EXPORT or RECEIVER_NOT_EXPORT. If Android Faker is faking for a version lower than 14 and the App's target SDK (Instagram) is 14, it will throw this exception.
Dev4Mod commentedon Aug 29, 2024
more info here
Rtizer-9 commentedon Aug 29, 2024
@Dev4Mod Working after turning off that option. Should I close the issue now. From what I understand lsposed can't/shouldn't do anything about and android faker itself should have prop available for a14 first (I checked it only has a13 and below).
Rtizer-9 commentedon Aug 29, 2024
@JingMatrix ssl killer module is still crashing.
JingMatrix commentedon Aug 29, 2024
Please provide some necessary logs.
It is not an open source App, which further requires root permission to run.
Hence, I won't test it on my device, and you should provide the crashing logs for analysis.
Dev4Mod commentedon Aug 30, 2024
I know the Dev of this module, apparently it seems to be a problem in a hook of the module. But I did a shallow investigation, but this is not a problem of LSPosed. Besides, there's a strange lib on Instagram "libaborthooks.so" this might be a protection for hooks, I'm not sure. But I repeat, it's not a problem of LSPosed.
Rtizer-9 commentedon Aug 30, 2024
@JingMatrix fyi, the latest build in actions allows usage of ssl killer without any app crash. It's a different thing if it works on an application or not, and whether it's completely crashing the app.
The latest build doesn't crash the apps outright on launch.
13 remaining items
Rtizer-9 commentedon Sep 2, 2024
I completely agree with and understand everything you said above.
The reason I'm using this setup is only because of its stability and best root detection bypass and nothing else. A normal user only cares about that irrespective of how big the error log becomes because all the apps/modules are working fine for me.
For the npm thing if you look at my comment edit history, I'd actually written for you to collaborate with these guys but from what it looks like they've just merged the hiding patches and that new parasitic manager so obviously you're the guy doing main development so it's stupid to ask you to take some "inspiration" from them and that's why I deleted that. So that's all I meant by mentioning their fork.
I guess I'll trouble you the day about that core patch issue when I'll be left with no choice because I can't lose the root detection bypass alpha and shamiko provides because my bank apps do need that.
As I already mentioned I did try to do what you suggested but shamiko is incompatible (even to me it's surprising) so I'm left with no choice other than use zygisk-next and alpha.
Thanks for assisting. I'll keep testing your latest builds meanwhile 👍
JingMatrix commentedon Sep 3, 2024
@Rtizer-9 Did you use the debug version of LSPosed when you sent the logs? If not, please do it.
Rtizer-9 commentedon Sep 3, 2024
I installed both the latest debug build of corepatch and lsposed and phone is stuck on boot; it's not a bootloop but it's just stuck there. I'll update if I manage to boot or if you've some advice to make it boot then I'm all ears.
Update: No luck, it won't boot.
JingMatrix commentedon Sep 3, 2024
Okay, next time, do use the debug version. As you have said in the issue,
, otherwise it was wasting my time since I supposed that you already knew it.
If you cannot boot your phone at all, try enter the safe mode.
For debugging the issue, I need logs for that, you may try run the following command from a computer since the moment that your screen is on:
I guess there were conflicts between modules.
Rtizer-9 commentedon Sep 3, 2024
Sure. I recovered the phone and booted somehow and will update with debug logs if I manage to boot it in debug when I get some time.
Qing0731 commentedon Sep 9, 2024
JingMatrix commentedon Sep 9, 2024
Qing0731 commentedon Sep 9, 2024
I am now using the debug version of the core path
LSPosed_2024-09-09T23_41_43.952034.zip
JingMatrix commentedon Sep 9, 2024
@Nimpip Core Patch seems working well on your device, which functions of it doesn't work as expected ?
Qing0731 commentedon Sep 9, 2024
Applications with different signatures cannot be overwritten
Qing0731 commentedon Sep 9, 2024
And when I used the core of the Lucky cracker to crack the magik module, the function worked, so I guess it's not a problem with my system
Merge pull request #7 from LucisUrbe/master
Decode a jfieldID before dereferencing it as an ArtField
art: a locality probe for the BoxReturn path, and the chain it explains