SIGABRT on throw instruction in C++ code file of VisionOS project in Xcode 15 beta

App & System Services Core OS
Izhido_XR OP
Created Jul ’23
Replies 7
Boosts 0
Views 1.2k
Participants 4

I'm getting a weird crash from the template VisionOS app (the one that appears upon creating a new VisionOS project). I went and modified the code to use two additional C++ files, one of which throws an exception upon a specific circumstance, and another one that catches and handles the exception.

If I attempt to run the app with the new code, instead of the code catching the exception, I get a SIGABRT signal, as if C++ exceptions were not enabled at project level.

The following gist contains a minimal example with the weird behavior:

https://gist.github.com/Izhido/100a92f45aaf8bacffe73893d6109077

Replace the contents of the template VisionOS project with these files, run the project, and press the "Do sum" button. Xcode 15 beta will report a SIGABRT signal at sumofnumbers_impl.cpp, line 8.

What am I missing here?

(Incidentally, the same code in a MacOS project runs just fine - I can share the project upon request. Also, for some reason I cannot share screenshots or files in this forum - that's why I provided the gist.)

Replies  7
Boosts  0
Views  1.2k
Participants  4
DTS Engineer OP
Apple
Jul ’23

That certainly seems weird. We just released new visionOS stuff yesterday. Were you testing with the new stuff?

Share and Enjoy
—
Quinn “The Eskimo!” @ Developer Technical Support @ Apple
let myEmail = "eskimo" + "1" + "@" + "apple.com"

0
Izhido_XR OP
Jul ’23

Yeah, I noticed last night Beta 5 was released along with a new simulator. I removed Beta 4 and reinstalled the new one.

Unfortunately, the weird issue persists.

Do you reckon I need to do something special to the project to make sure I'm really using the latest version of everything? Or should it be already working as it is?

0
DTS Engineer OP
Apple
Jul ’23

Thanks for confirming that.

I tried this myself and got the same result. Here’s the backtrace of the crash:

(lldb)  bt
* thread #1, queue = 'com.apple.main-thread', stop reason = signal SIGABRT
    … #0:  … libsystem_kernel.dylib`__pthread_kill + 10
    … #1:  … libsystem_pthread.dylib`pthread_kill + 263
    … #2:  … libsystem_c.dylib`abort + 133
    … #3:  … libc++abi.dylib`__cxxabiv1::readEncodedPointer(unsigned char const**, unsigned char, unsigned long) + 269
    … #4:  … libc++abi.dylib`__gxx_personality_v0 + 399
    … #5:  … libunwind.dylib`_Unwind_RaiseException + 401
    … #6:  … libc++abi.dylib`__cxa_throw + 108
  * … #7:  … Test734484`sum_numbers_impl(top=15) at sumofnumbers_impl.cpp:8:3
    … #8:  … Test734484`sum_numbers(top=15) at sumofnumbers.cpp:8:10
    … #9:  … Test734484`closure #2 in closure #2 in ContentView.body.getter at ContentView.swift:29:30
    …

Frame 9 is your SwiftUI button handler, which calls your C++ code which throws an exception at frame 7. The exception runtime is busily unwinding the stack, a process that involves reading C++ exception handler tables. In frame 3 it gets a value it doesn’t understand and calls abort.

This stuff should just work out of the box, as it did when you tested the same code in a macOS project. I don’t know enough about the C++ runtime to say why it’s not working in this case, but it’s definitely something that we need to look into.

I’d appreciate you filing a bug about this, making sure to include a copy of your test project. And please post your bug number, just for the record.

Share and Enjoy
—
Quinn “The Eskimo!” @ Developer Technical Support @ Apple
let myEmail = "eskimo" + "1" + "@" + "apple.com"

0
Izhido_XR OP
Jul ’23

I was not aware there was a direct way to post bug reports. Thank you so much for sharing!

The feedback has been posted. Bug number FB12790129.

Appreciated!

0
Vision Pro Engineer OP
Apple
Aug ’23

Hi @Izhido_XR , I ran this on an M1 MacBook and did not see any crashes. I used Xcode 15 beta 5 from the developer site as well, so there was no difference on that. This seems to be an Intel problem with the simulator

1
Izhido_XR OP
Dec ’23

Finally got an answer from the posted feedback. 4 months later.

The answer? "Please know that visionOS simulator isn’t supported on Intel machines. Developing for visionOS requires a Mac with Apple silicon."

sigh

  1. That's obviously what they were going to say;
  2. It took 1+ month after they disabled the simulator in Intel-based machines, and introduce the other one, before answering;
  3. Despite what I said earlier, I just don´t have the money to get a new machine.

Have you guys, ever wished, so hard, to give a collective "f****k you" to a whole company, but had to resist to do it on grounds of not wishing to have your reputation and professionalism level brought directly into the ground?

That's me right now.

0
endecotp OP
Dec ’23

4 months ago:

I was not aware there was a direct way to post bug reports. Thank you so much for sharing!

Today:

Have you guys, ever wished, so hard, to give a collective "f****k you" to a whole company

Thanks for your feedback about the "Feedback Assistant" experience. So that we can be sure that your feedback reaches the right people, I’d appreciate you filing a bug about this. And please post your bug number, just for the record.

0
SIGABRT on throw instruction in C++ code file of VisionOS project in Xcode 15 beta
First post date Last post date
 
 
Q