"malloc(): memory corruption" when running benchmark.py

[zhuwenxi]

Hi, there's an error when I was running python test/benchmark.py.

Environment:

• CentOS 7
• GCC 6.3
• LLVM 8.0
• Pytorch 1.4.0 (build from source)

Error message:

Tracing model with JIT
Warming JIT up with 10 runs
Running JIT 100 times
Done benchmarking JIT
Tracing model with TVM
*** Error in `python': malloc(): memory corruption (fast): 0x00007fb16f942920 ***
======= Backtrace: =========
/lib64/libc.so.6(+0x7ab54)[0x7fb167ccfb54]
/lib64/libc.so.6(+0x7ddf7)[0x7fb167cd2df7]
/lib64/libc.so.6(__libc_malloc+0x4c)[0x7fb167cd510c]
/data1/root/anaconda3/envs/tvm/bin/../lib/libstdc++.so.6(_Znwm+0x15)[0x7fb0d6a934e5]
/data1/root/anaconda3/envs/tvm/lib/python3.6/site-packages/tvm-0.6.dev0-py3.6-linux-x86_64.egg/tvm/libtvm.so(_ZN3tvm5relay12ConstantNode4makeENS_7runtime7NDArrayE+0x18)[0x7fb0836646e8]
/data1/root/workspace/tvm_trial/pytorch-tvm/torch_tvm/_torch_tvm.cpython-36m-x86_64-linux-gnu.so(+0x34d5a)[0x7fb1278cad5a]
/data1/root/workspace/tvm_trial/pytorch-tvm/torch_tvm/_torch_tvm.cpython-36m-x86_64-linux-gnu.so(+0x38200)[0x7fb1278ce200]
/data1/root/workspace/tvm_trial/pytorch-tvm/torch_tvm/_torch_tvm.cpython-36m-x86_64-linux-gnu.so(+0x3cdd4)[0x7fb1278d2dd4]
/data1/root/workspace/tvm_trial/pytorch-tvm/torch_tvm/_torch_tvm.cpython-36m-x86_64-linux-gnu.so(+0x9904c)[0x7fb12792f04c]
/root/anaconda3/envs/tvm/lib/python3.6/site-packages/torch/lib/libtorch.so(+0x392a7b6)[0x7fb0bf4717b6]
/root/anaconda3/envs/tvm/lib/python3.6/site-packages/torch/lib/libtorch.so(_ZN5torch3jit16InterpreterState3runERSt6vectorIN3c106IValueESaIS4_EE+0x1c)[0x7fb0bf466a4c]
/root/anaconda3/envs/tvm/lib/python3.6/site-packages/torch/lib/libtorch.so(+0x38fab71)[0x7fb0bf441b71]
/root/anaconda3/envs/tvm/lib/python3.6/site-packages/torch/lib/libtorch.so(_ZN5torch3jit8Function3runERSt6vectorIN3c106IValueESaIS4_EE+0x63)[0x7fb0bf730df3]
/root/anaconda3/envs/tvm/lib/python3.6/site-packages/torch/lib/libtorch_python.so(+0x7871ba)[0x7fb1186701ba]
/root/anaconda3/envs/tvm/lib/python3.6/site-packages/torch/lib/libtorch_python.so(+0x787b1d)[0x7fb118670b1d]
/root/anaconda3/envs/tvm/lib/python3.6/site-packages/torch/lib/libtorch_python.so(+0x746b7d)[0x7fb11862fb7d]
/root/anaconda3/envs/tvm/lib/python3.6/site-packages/torch/lib/libtorch_python.so(+0x2832f2)[0x7fb11816c2f2]
python(_PyCFunction_FastCallDict+0x154)[0x7fb16856a334]
python(_PyObject_FastCallDict+0x2bf)[0x7fb16856a74f]
python(_PyObject_Call_Prepend+0x63)[0x7fb16856f173]
python(PyObject_Call+0x3e)[0x7fb16856a13e]
python(+0x16a101)[0x7fb1685c3101]
python(PyObject_Call+0x3e)[0x7fb16856a13e]
python(_PyEval_EvalFrameDefault+0x1ab0)[0x7fb168615d00]
python(+0x191fae)[0x7fb1685eafae]
python(+0x192be6)[0x7fb1685ebbe6]
python(+0x198a65)[0x7fb1685f1a65]
python(_PyEval_EvalFrameDefault+0x30a)[0x7fb16861455a]
python(PyEval_EvalCodeEx+0x966)[0x7fb1685ecd06]
python(+0x1945e6)[0x7fb1685ed5e6]
python(PyObject_Call+0x3e)[0x7fb16856a13e]
python(_PyEval_EvalFrameDefault+0x1ab0)[0x7fb168615d00]
python(+0x191fae)[0x7fb1685eafae]
python(+0x192be6)[0x7fb1685ebbe6]
python(+0x198a65)[0x7fb1685f1a65]
python(_PyEval_EvalFrameDefault+0x30a)[0x7fb16861455a]
python(+0x191fae)[0x7fb1685eafae]
python(+0x192be6)[0x7fb1685ebbe6]
python(+0x198a65)[0x7fb1685f1a65]
python(_PyEval_EvalFrameDefault+0x30a)[0x7fb16861455a]
python(+0x191b76)[0x7fb1685eab76]
python(+0x192b83)[0x7fb1685ebb83]
python(+0x198a65)[0x7fb1685f1a65]
python(_PyEval_EvalFrameDefault+0x30a)[0x7fb16861455a]
python(+0x191b76)[0x7fb1685eab76]
python(+0x192b83)[0x7fb1685ebb83]
python(+0x198a65)[0x7fb1685f1a65]
python(_PyEval_EvalFrameDefault+0x30a)[0x7fb16861455a]
python(+0x19296b)[0x7fb1685eb96b]
python(+0x198a65)[0x7fb1685f1a65]
python(_PyEval_EvalFrameDefault+0x30a)[0x7fb16861455a]
python(PyEval_EvalCodeEx+0x329)[0x7fb1685ec6c9]
python(PyEval_EvalCode+0x1c)[0x7fb1685ed45c]
python(+0x214d54)[0x7fb16866dd54]
python(PyRun_FileExFlags+0xa1)[0x7fb16866e151]
python(PyRun_SimpleFileExFlags+0x1c3)[0x7fb16866e353]
python(Py_Main+0x613)[0x7fb168671e43]
python(main+0xee)[0x7fb16853c28e]
/lib64/libc.so.6(__libc_start_main+0xf5)[0x7fb167c76c05]
python(+0x1c1fff)[0x7fb16861afff]