net5.0 app crash on Apple M1 under Rosetta 2 emulation

[k15tfu]

Hi! Sometimes my self-contained app published for osx-x64 crashes on M1 as follows:

Process:               Mandelbrot [46749]
Path:                  /Users/USER/*/Mandelbrot
Identifier:            Mandelbrot
Version:               0
Code Type:             X86-64 (Translated)
Parent Process:        zsh [45679]
User ID:               502

Date/Time:             2021-02-27 19:33:29.473 +0300
OS Version:            macOS 11.2.1 (20D74)
Report Version:        12
Anonymous UUID:        99CA9E8A-60BC-03E4-C062-DCBF45315D56

Sleep/Wake UUID:       4230E9B0-E302-4819-826A-DC28A40EC099

Time Awake Since Boot: 520000 seconds
Time Since Wake:       96000 seconds

System Integrity Protection: enabled

Crashed Thread:        13

Exception Type:        EXC_BAD_ACCESS (SIGSEGV)
Exception Codes:       KERN_INVALID_ADDRESS at 0x0000000100000000
Exception Note:        EXC_CORPSE_NOTIFY

Termination Signal:    Segmentation fault: 11
Termination Reason:    Namespace SIGNAL, Code 0xb
Terminating Process:   exc handler [46749]

VM Regions Near 0x100000000:
--> 
    __TEXT                      102708000-102716000    [   56K] r-x/r-x SM=COW  /Users/*

[...]
Thread 13 Crashed:
0   <translation info unavailable>	0x0000000102ad0718 ???
1   libsystem_kernel.dylib        	0x00007fff20325f86 swtch_pri + 10
2   libcoreclr.dylib              	0x000000010ac27185 PAL_DispatchException + 357
3   libcoreclr.dylib              	0x000000010ac26d37 PAL_DispatchExceptionWrapper + 10
4   ???                           	0x0000000100000000 0 + 4294967296
5   ???                           	0x0000000111a74ee9 ???
6   ???                           	0x0000000111a74738 ???
7   ???                           	0x0000000111a749e8 ???
8   ???                           	0x0000000111a71fc7 ???
9   ???                           	0x0000000111a70bb1 ???
10  ???                           	0x00000001115d03e2 ???
11  ???                           	0x00000001115d97e1 ???
12  ???                           	0x00000001115d04fe ???
13  libcoreclr.dylib              	0x000000010af113b9 CallDescrWorkerInternal + 124
14  libcoreclr.dylib              	0x000000010ad6609f MethodDescCallSite::CallTargetWorker(unsigned long const*, unsigned long*, int) + 1519
15  libcoreclr.dylib              	0x000000010ad76a04 ThreadNative::KickOffThread_Worker(void*) + 404
16  libcoreclr.dylib              	0x000000010ad30cdb ManagedThreadBase_DispatchOuter(ManagedThreadCallState*) + 315
17  libcoreclr.dylib              	0x000000010ad31290 ManagedThreadBase::KickOff(void (*)(void*), void*) + 32
18  libcoreclr.dylib              	0x000000010ad76b7f ThreadNative::KickOffThread(void*) + 191
19  libcoreclr.dylib              	0x000000010ac24ffa CorUnix::CPalThread::ThreadEntry(void*) + 426
20  libsystem_pthread.dylib       	0x00007fff2035a950 _pthread_start + 224
21  libsystem_pthread.dylib       	0x00007fff2035647b thread_start + 15

Full log is here Mandelbrot_2021-02-27-193331.crash.txt. I was unable to reproduce the crash last few hours, but you can try to do it using Mandelbrot.zip from #13041:

1. Add net5.0 as TargetFrameworks in .csproj
2. $DOTNET_SDK/dotnet publish --self-contained=true -c Release -f net5.0 -r osx-x64 Mandelbrot.csproj -o tmp
3. for i in {1..1000}; do echo $i; ./tmp/Mandelbrot --width=1024 --height=768 || break; done

macOS Big Sur 11.2.1
.NET SDK 5.0.103 x64

Linked issues: #44897

[dotnet-issue-labeler]

I couldn't figure out the best area label to add to this issue. If you have write-permissions please help me learn by adding exactly one area label.

[k15tfu]

@sdmaclea @janvorli Do you need any more information I can provide?

[janvorli]

@k15tfu I am sorry, I was OOF for a month from February 22 and I've missed this issue after getting back. I'll try to repro it locally running under lldb to see if I can get more details on it.

[mangod9]

Hi @k15tfu, are you still able to repro this on 11.3+? There have been fixes in that release which resolved Rosetta issues.

[k15tfu]

@mangod9 Yes, it still fails on macOS 11.4:

Process:               Mandelbrot [21856]
Path:                  /opt/teamcity-agent/*/Mandelbrot
Identifier:            Mandelbrot
Version:               0
Code Type:             X86-64 (Translated)
Parent Process:        mono-sgen [21046]
Responsible:           bash [553]
User ID:               502

Date/Time:             2021-07-13 06:14:08.585 +0300
OS Version:            macOS 11.4 (20F71)
Report Version:        12
Anonymous UUID:        3BC01934-B29E-74C0-87B8-B558D0C364C1


Time Awake Since Boot: 3900000 seconds

System Integrity Protection: enabled

Crashed Thread:        26

Exception Type:        EXC_BAD_ACCESS (SIGSEGV)
Exception Codes:       KERN_INVALID_ADDRESS at 0xbfeb8b8c76c0fc02 -> 0xffff8b8c76c0fc02 (possible pointer authentication failure)
Exception Note:        EXC_CORPSE_NOTIFY

Termination Signal:    Segmentation fault: 11
Termination Reason:    Namespace SIGNAL, Code 0xb
Terminating Process:   exc handler [21856]

VM Regions Near 0xffff8b8c76c0fc02:
--> mapped file              7fffe0000000-7fffe2e88000 [ 46.5M] r--/r-- SM=COW  Object_id=ca6c06b5
    
[...]
Thread 26 Crashed:
0   ???                           	0xbfeb8b8c76c0fc02 0 + 13829300516175281154
1   libsystem_kernel.dylib        	0x00007fff203293c2 swtch_pri + 10
2   libcoreclr.dylib              	0x000000010c937f95 PAL_DispatchException + 357
3   libcoreclr.dylib              	0x000000010c937b47 PAL_DispatchExceptionWrapper + 10
4   ???                           	0xffff8b8c76c0fc02 0 + 18446616034136882178
5   ???                           	0x00000001137a46e1 ???
6   ???                           	0x00000001137a645d ???
7   ???                           	0x00000001137a3a8e ???
8   ???                           	0x00000001137a5f3c ???
9   ???                           	0x0000000113798972 ???
10  ???                           	0x0000000113797d8e ???
11  ???                           	0x000000011379775d ???
12  ???                           	0x00000001137975bd ???
13  ???                           	0x0000000113796ff6 ???
14  libcoreclr.dylib              	0x000000010cc223c9 CallDescrWorkerInternal + 124
15  libcoreclr.dylib              	0x000000010ca76ebf MethodDescCallSite::CallTargetWorker(unsigned long const*, unsigned long*, int) + 1519
16  libcoreclr.dylib              	0x000000010ca87754 ThreadNative::KickOffThread_Worker(void*) + 404
17  libcoreclr.dylib              	0x000000010ca41b0b ManagedThreadBase_DispatchOuter(ManagedThreadCallState*) + 315
18  libcoreclr.dylib              	0x000000010ca420c0 ManagedThreadBase::KickOff(void (*)(void*), void*) + 32
19  libcoreclr.dylib              	0x000000010ca878cf ThreadNative::KickOffThread(void*) + 191
20  libcoreclr.dylib              	0x000000010c935e0a CorUnix::CPalThread::ThreadEntry(void*) + 426
21  libsystem_pthread.dylib       	0x00007fff2035e8fc _pthread_start + 224
22  libsystem_pthread.dylib       	0x00007fff2035a443 thread_start + 15

[deeprobin]

@mangod9 Is this still relevant for the .NET 5 milestone? (The EOS for this version is already on May 8, 2022).

[mangod9]

@k15tfu is this still blocking you? Given that .net 6 has been now released perhaps we could close this?

[deeprobin]

@mangod9 I think we can close this

.NET 6 is the first fully supported SDK for Rosetta 2 emulation.

See dotnet/sdk#22380 for details.

For reference: #44897 (comment)

/cc @jkotas

[mangod9]

cool, thanks