SIGSEGV on use tensorflow

[mbajlo]

Hello, I am have been trying to setup the example mobilenetv3. In my Cargo.toml I have dependecy tensorflow = { version = "0.19.0" }, up to this point I can make cargo build and it works fine. The problem happens if I add any kind of use tensorflow::, after this if I do build, even though I haven't used any object from tensorflow I get this: Stop reason: signal SIGSEGV: invalid address (fault address: 0x0).

So basically I get segmentation fault, which means I want to use some undefined memory even before I used any code from tensorflow crate. Do you guys know what could be wrong here?

[adamcrume]

I'm unable to reproduce the issue, even running under valgrind. Can you verify which TensorFlow library is getting pulled in? You can try building with -v if it's not clear. You might also want to try running the example with gdb and printing a backtrace (bt) when it crashes.

[mbajlo]

Thanks for a response, I will try what you have explained.

Now when I look at the code I see I have forgotten to remove async main function under the tokio framework. Maybe the problem lies there since there is no async support in tensorflow? I just copied the whole example into the async main