Views
No views yet
InterpreterBuilder::ParseTensors SIGSEGV PoC.tflite model file that triggers a deterministic native SIGSEGV when loaded by TensorFlow Lite.allocate_tensors() or invoke(), inside:tflite::impl::InterpreterBuilder::ParseTensors(...)1crash.tflite Proof-of-concept malformed TFLite model
2reproduce_tf.py Minimal TensorFlow reproduction script
3requirements.txt Python dependency for stable TensorFlow reproduction
4SHA256SUMS SHA-256 checksum for the PoC file
5evidence/ GDB, reproducibility, and diff evidenceccc2cee82c0b90147a0ca5babacd3bad67ae93a6a3cbce5bb4ba52e2e34cc510 crash.tflite1tensorflow==2.21.0
2tf-nightly==2.22.0-dev20260525
3ai-edge-litert==2.1.5
4Python 3.121python3.12 -m venv /tmp/tflite-repro
2source /tmp/tflite-repro/bin/activate
3
4python -m pip install --upgrade pip
5python -m pip install -r requirements.txt
6
7ulimit -c 0
8python reproduce_tf.py crash.tflite
9echo "rc=$?".tflite model files should be rejected cleanly with an exception or error status.1[runtime] tensorflow 2.21.0
2[phase] before tf.lite.Interpreter()
3Fatal Python error: Segmentation fault
4rc=1391Program received signal SIGSEGV, Segmentation fault.
2
3#0 tflite::impl::InterpreterBuilder::ParseTensors(...)
4#1 tflite::impl::InterpreterBuilder::operator()(...)
5#2 tflite::interpreter_wrapper::InterpreterWrapper::CreateInterpreterWrapper(...)
6#3 tflite::interpreter_wrapper::InterpreterWrapper::CreateWrapperCPPFromFile(...)1run rc phase
21 139 before_interpreter+sigsegv
32 139 before_interpreter+sigsegv
43 139 before_interpreter+sigsegv1tensorflow 2.22.0-dev20260525
2run 1: rc=139
3run 2: rc=139
4run 3: rc=1391ai-edge-litert==2.1.5
2rc=1391seed_len=952
2case_len=909
3common_prefix=592
4common_suffix=317
5seed_changed_region_len=43
6case_changed_region_len=0
7likely_mutation=delete offset=592 length=43.tflite model file can terminate a process that attempts to load it using TensorFlow Lite / LiteRT. This is a native parser/loader crash, not a handled Python exception.