Segmentation Fault (Core Dumped) When Adding Tensor and nn.Linear Module in Eager Mode

[LiSsHhUuAaIi]

Summary

I encountered a segmentation fault (core dumped) when running a model in OneFlow's eager mode. The error occurs when attempting to add a tensor (flow.relu(value)) and an nn.Linear module (self.linear) in the forward method. Instead of raising a proper Python exception for this type mismatch, OneFlow crashes with a segmentation fault.

Code to reproduce bug

import oneflow as flow
import oneflow.nn as nn

class TestModel(nn.Module):
    def __init__(self):
        super(TestModel, self).__init__()
        self.linear = nn.Linear(10, 5)

    def forward(self, value):
        # BUG: Attempting to add tensor (flow.relu(value)) and module (self.linear)
        # This should raise TypeError but causes segmentation fault
        return flow.relu(value) + self.linear

# Test data
input_tensor = flow.randn(1, 10)
model = TestModel()

try:
    # This should raise TypeError but instead causes segfault
    result = model(input_tensor)
    print("Success:", result)
except Exception as e:
    print("Caught exception:", e)

print("This line is unreachable due to segmentation fault")

Error Logs

Stack trace (most recent call last):
   Object "/home/miniconda3/envs/oneflow-test/lib/python3.10/site-packages/oneflow/_oneflow_internal.cpython-310-x86_64-linux-gnu.so", at 0x7f500ff21562, in 
   Object "/home/miniconda3/envs/oneflow-test/lib/python3.10/site-packages/oneflow/_oneflow_internal.cpython-310-x86_64-linux-gnu.so", at 0x7f500fd4f14f, in functional::add(_object*, _object*, _object*)
   Object "/home/miniconda3/envs/oneflow-test/lib/python3.10/site-packages/oneflow/../oneflow.libs/liboneflow-efdae59a.so", at 0x7f50004aa579, in functional::Add(TensorTuple const&, bool)
    ...
    ...
Segmentation fault (Address not mapped to object [0x79])
Segmentation fault (core dumped)

System Information

• OS: Ubuntu 20.04.6 LTS
• OneFlow version: '1.0.0.dev20251101+cpu'
• Python version: Python 3.10.19