From b36c62ce4f326f523f0cf9c3792073863b2ecc10 Mon Sep 17 00:00:00 2001
From: yf711 <yifanl@microsoft.com>
Date: Fri, 15 Dec 2023 15:20:01 -0800
Subject: [PATCH] fix

---
 .../tensorrt/tensorrt_execution_provider.cc   | 19 +++++++++----------
 1 file changed, 9 insertions(+), 10 deletions(-)

diff --git a/onnxruntime/core/providers/tensorrt/tensorrt_execution_provider.cc b/onnxruntime/core/providers/tensorrt/tensorrt_execution_provider.cc
index 3b0d6997308f3..714876b25262d 100644
--- a/onnxruntime/core/providers/tensorrt/tensorrt_execution_provider.cc
+++ b/onnxruntime/core/providers/tensorrt/tensorrt_execution_provider.cc
@@ -2729,12 +2729,11 @@ common::Status TensorrtExecutionProvider::Compile(const std::vector<FusedNodeAnd
       LOGS_DEFAULT(WARNING) << "[TensorRT EP] Builder heuristics are enabled. For TRT > 8.5, trt_build_heuristics_enable is deprecated, please set builder optimization level as 2 to enable builder heuristics.";
     }
 #elif NV_TENSORRT_MAJOR == 8 && NV_TENSORRT_MINOR > 5 || NV_TENSORRT_MAJOR > 8
-    // for TRT 8.6 onwards, heuristic-based tactic option is automatically enabled by setting builder optimization level 2 
+    // for TRT 8.6 onwards, heuristic-based tactic option is automatically enabled by setting builder optimization level 2
     if (build_heuristics_enable_) {
       if (builder_optimization_level_ == 2) {
-        LOGS_DEFAULT(WARNING) << "[TensorRT EP] Builder heuristics are enabled automatically by builder optimization level 2. trt_build_heuristics_enable is deprecated on TRT 8.6 onwards.";
-      }
-      else {
+        LOGS_DEFAULT(WARNING) << "[TensorRT EP] Builder heuristics are automatically enabled by builder optimization level 2. trt_build_heuristics_enable is deprecated on TRT 8.6 onwards.";
+      }else {
         LOGS_DEFAULT(WARNING) << "[TensorRT EP] trt_build_heuristics_enable is deprecated on TRT 8.6 onwards. Please set builder optimization level as 2 to enable builder heuristics.";
       }
     }
@@ -2814,7 +2813,7 @@ common::Status TensorrtExecutionProvider::Compile(const std::vector<FusedNodeAnd
           engine_file.seekg(0, std::ios::beg);
           std::unique_ptr<char[]> engine_buf{new char[engine_size]};
           engine_file.read((char*)engine_buf.get(), engine_size);
-          trt_engine = std::unique_ptr<nvinfer1::ICudaEngine>(runtime_->deserializeCudaEngine(engine_buf.get(), engine_size, nullptr));
+          trt_engine = std::unique_ptr<nvinfer1::ICudaEngine>(runtime_->deserializeCudaEngine(engine_buf.get(), engine_size));
           LOGS_DEFAULT(VERBOSE) << "[TensorRT EP] DeSerialized " + engine_cache_path;
           if (trt_engine == nullptr) {
             return ORT_MAKE_STATUS(ONNXRUNTIME, EP_FAIL,
@@ -2833,7 +2832,7 @@ common::Status TensorrtExecutionProvider::Compile(const std::vector<FusedNodeAnd
                                    "TensorRT EP could not call engine decryption function decrypt");
           }
           // Deserialize engine
-          trt_engine = std::unique_ptr<nvinfer1::ICudaEngine>(runtime_->deserializeCudaEngine(engine_buf.get(), engine_size, nullptr));
+          trt_engine = std::unique_ptr<nvinfer1::ICudaEngine>(runtime_->deserializeCudaEngine(engine_buf.get(), engine_size));
           LOGS_DEFAULT(VERBOSE) << "[TensorRT EP] Decrypted and DeSerialized " + encrypted_engine_cache_path;
           if (trt_engine == nullptr) {
             return ORT_MAKE_STATUS(ONNXRUNTIME, EP_FAIL,
@@ -2874,7 +2873,7 @@ common::Status TensorrtExecutionProvider::Compile(const std::vector<FusedNodeAnd
             return ORT_MAKE_STATUS(ONNXRUNTIME, EP_FAIL,
                                    "TensorRT EP failed to create engine from network for fused node: " + fused_node.Name());
           }
-          trt_engine = std::unique_ptr<nvinfer1::ICudaEngine>(runtime_->deserializeCudaEngine(serialized_engine->data(), serialized_engine->size(), nullptr));
+          trt_engine = std::unique_ptr<nvinfer1::ICudaEngine>(runtime_->deserializeCudaEngine(serialized_engine->data(), serialized_engine->size()));
           if (trt_engine == nullptr) {
             return ORT_MAKE_STATUS(ONNXRUNTIME, EP_FAIL,
                                    "TensorRT EP failed to deserialize engine for fused node: " + fused_node.Name());
@@ -3078,7 +3077,7 @@ common::Status TensorrtExecutionProvider::Compile(const std::vector<FusedNodeAnd
           // https://docs.nvidia.com/deeplearning/tensorrt/developer-guide/index.html#threading
           trt_state->engine->reset();
           *(trt_state->engine) = std::unique_ptr<nvinfer1::ICudaEngine>(
-              trt_state->runtime->deserializeCudaEngine(engine_buf.get(), engine_size, nullptr));
+              trt_state->runtime->deserializeCudaEngine(engine_buf.get(), engine_size));
           if (!(*(trt_state->engine))) {
             return ORT_MAKE_STATUS(ONNXRUNTIME, EP_FAIL, "TensorRT EP Failed to Build Engine.");
           }
@@ -3103,7 +3102,7 @@ common::Status TensorrtExecutionProvider::Compile(const std::vector<FusedNodeAnd
           // Note: Deserializing an engine from a TensorRT runtime is thread safe per TRT doc
           // https://docs.nvidia.com/deeplearning/tensorrt/developer-guide/index.html#threading
           trt_state->engine->reset();
-          *(trt_state->engine) = std::unique_ptr<nvinfer1::ICudaEngine>(trt_state->runtime->deserializeCudaEngine(engine_buf.get(), engine_size, nullptr));
+          *(trt_state->engine) = std::unique_ptr<nvinfer1::ICudaEngine>(trt_state->runtime->deserializeCudaEngine(engine_buf.get(), engine_size));
           if (!(*(trt_state->engine))) {
             return ORT_MAKE_STATUS(ONNXRUNTIME, EP_FAIL,
                                    "TensorRT EP could not deserialize engine from encrypted cache: " + encrypted_engine_cache_path);
@@ -3234,7 +3233,7 @@ common::Status TensorrtExecutionProvider::Compile(const std::vector<FusedNodeAnd
             return ORT_MAKE_STATUS(ONNXRUNTIME, EP_FAIL, "TensorRT EP failed to create engine from network.");
           }
           *(trt_state->engine) = std::unique_ptr<nvinfer1::ICudaEngine>(
-              trt_state->runtime->deserializeCudaEngine(serialized_engine->data(), serialized_engine->size(), nullptr));
+              trt_state->runtime->deserializeCudaEngine(serialized_engine->data(), serialized_engine->size()));
           if (!(*(trt_state->engine))) {
             return ORT_MAKE_STATUS(ONNXRUNTIME, EP_FAIL, "TensorRT EP failed to deserialize engine.");
           }