TensorRT Segmentation Fault

[vmazalov]

Describe the issue

This code throws Segmentation Fault. Is there a recommended way to investigate?

CheckStatus(m_ort->SessionOptionsAppendExecutionProvider_TensorRT(m_sessionOptions, &trtOptions));

To reproduce

m_ort = OrtGetApiBase()->GetApi(14);
CheckStatus(m_ort->CreateSessionOptions(&m_sessionOptions));
CheckStatus(m_ort->SetIntraOpNumThreads(m_sessionOptions, 1));
CheckStatus(m_ort->SetSessionGraphOptimizationLevel(m_sessionOptions, sessionSetting.m_bDisableOptimizations ? ORT_ENABLE_BASIC : ORT_ENABLE_ALL));
int logSeverityLevel = 3;
CheckStatus(m_ort->SetSessionLogSeverityLevel(m_sessionOptions, logSeverityLevel));
CheckStatus(m_ort->DisableMemPattern(m_sessionOptions));
OrtTensorRTProviderOptions trtOptions;
CheckStatus(m_ort->SessionOptionsAppendExecutionProvider_TensorRT(m_sessionOptions, &trtOptions)); // This line is causing segmentation fault

Urgency

No response

Platform

Linux

OS Version

Ubuntu 20.04.6 LTS

ONNX Runtime Installation

Built from Source

ONNX Runtime Version or Commit ID

1.15.1

ONNX Runtime API

C++

Architecture

X64

Execution Provider

TensorRT

Execution Provider Library Version

https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/secure/8.6.1/tars/TensorRT-8.6.1.6.Linux.x86_64-gnu.cuda-11.8.tar.gz

[vmazalov]

This was fixed by using the _V2 TRT binding and most importantly ensuring that LD_LIBRARY_PATH includes the TRT path /usr/lib/TensorRT-8.6.1.6/targets/x86_64-linux-gnu/lib