Tracer script stuck for x64 program in windows

[heheda123123]

the binary is default hello_world.exe built by vs2022 x64.

[heheda123123]

the same source code build for x86 will work

[myr-syn]

Hi,

Thank you for your feedback. It seems that the native tracer for x64 might have an issue with your binary. Could you try to run the tracer in slow mode (no native callbacks) with -s or --slow option flag?

[heheda123123]

I tried it and had the same problem

[fredfrance-oss]

The same issue happens on Linux Ubuntu with a trivial binary:

$ uname -a
Linux user 6.14.0-36-generic #36~24.04.1-Ubuntu SMP PREEMPT_DYNAMIC Wed Oct 15 15:45:17 UTC 2 x86_64 x86_64 x86_64 GNU/Linux

#include <stdio.h>
int main(int argc, char *argv[]) {
	printf("Done\n");
}

$ objdump -x a.out | grep main
0000000000000000       F *UND*	0000000000000000              __libc_start_main@GLIBC_2.34
0000000000001149 g     F .text	0000000000000029              main

$ python3 frinet/trace.py -t "a.out" spawn ./a.out a.out 0x1149
2025-12-04 22:40:27,744     INFO | Using local device
2025-12-04 22:40:27,749     INFO | Connected to device (x64 linux Ubuntu 24.04).
2025-12-04 22:40:27,756     INFO | Spawning process 
2025-12-04 22:40:27,837     INFO | Loading JS module...
2025-12-04 22:40:27,845     INFO | Loading C module...
2025-12-04 22:40:27,846     INFO | Tracing...
2025-12-04 22:40:27,847     INFO | Received ASLR slide: 0x555555554000.
2025-12-04 22:40:27,847     INFO | CTRL+C to interrupt trace
Entering function
Leaving function
2025-12-04 22:40:27,850     INFO | Received trace data for new thread 556587.
2025-12-04 22:40:27,850     INFO | Writing trace data of 10548 bytes...
2025-12-04 22:40:27,851     INFO | Creating file xxx/frinet/tracer/traces/a.out_1764848427_556587.tenet
on_output data:  b'Done\x00\x00\x00\x00\x01\x1b\x03;0\x00\n'
STDOUT : Done0
STDOUT : 
on_output data:  b''
STDERR : 
on_output data:  b''
STDOUT : 

While debugging i noticed a different bug too.
I modified trace.py to output the actual data it received, interestingly the data it receives is incorrect.
When I use -s slow mode, then the output from on_output data is correct.
this output issue only appears when compiling with gcc version 13.3.0 (Ubuntu 13.3.0-6ubuntu2~24.04), but not when compiling with clang Ubuntu clang version 18.1.3 (1ubuntu1).

The getting-stuck bug exists when compiling both with gcc/clang and when running slow-mode/normal-mode

I was not able to test with a x86 (32bit) binary, as this gives the error when tracing:

2025-12-04 22:55:13,320     INFO | Using local device
2025-12-04 22:55:13,324    DEBUG | sysenv: {'arch': 'x64', 'os': {'version': '24.04', 'id': 'ubuntu', 'name': 'Ubuntu'}, 'platform': 'linux', 'name': 'fred', 'access': 'full'}
2025-12-04 22:55:13,324     INFO | Connected to device (x64 linux Ubuntu 24.04).
2025-12-04 22:55:13,336     INFO | Spawning process 
2025-12-04 22:55:13,404 CRITICAL | Connection to remote frida server closed: `the connection is closed`.