-
Notifications
You must be signed in to change notification settings - Fork 187
Bogus Control Flow crash #124
Comments
Okay, I guess it doesn't work well with nlohmann json. Also had the same issue when using ThemIDA, always crashed using nlohmann. |
Good, bug report without any reproducer. But this crash seems like an backend bug, and should be reduced by an expert and submitted to LLVM upstream.
You mean this themida? If the program works fine when leaving compiler but crashes when another protector, it's (hopefully obviously) that protector's bug |
Also, as a life-long professional Windows Hater, LLVM has sub-par support on Windows, esp when using Windows-specific language features like SEH / MSVC Intrinscis / MSFT C++ Extensions/ etc |
Never had issues using it on Windows except intrinsics ^^. It started to crash as soon as I had this included: It only happend when having bogus control flow enabled, without everything was fine ^^ Works fine now, except ConstantEncryption leads to crash (i will not further investigate why) |
In recent versions it got better, but still bugs like llvm/llvm-project#60914 remains unfixed
I'm aware, that was a suggestion. From experience the biggest offender is usually indirectbranch, if you have that pass enabled.
Yeah it's inherently flawed and nobody should be actually using it |
Does someone know a fix to the extreme RAM Usage when using indirect branches? |
Hello,
I tried version 15 and 17 and both are crashing when using Bogus Control Flow.
Doing Post-Run Cleanup
Hikari Out
Spend Time: 1.3452160s
Assertion failed: unsigned(ObjectIdx + NumFixedObjects) < Objects.size() && "Invalid Object Idx!", file C:\Users\PC\Hikari-LLVM15\llvm\include\llvm/CodeGen/MachineFrameInfo.h, line 485
The text was updated successfully, but these errors were encountered: