Fix post merge jobs pipeline build issues (#14346)

- Fix debug node inputs outputs nullptr dereference with ONNX optional types.
- Fix model test memory leak.
- Convert jobs to stages in post-merge-jobs.yml to allow a subset of builds to be enabled when running manually.
- Fix buffer overrun in CumSum op exposed by Mimalloc build.

onnxruntime/core/framework/debug_node_inputs_outputs_utils.cc

@@ -478,23 +478,26 @@ void DumpNodeInputs(
 
       if (type) {
         if (type->IsTensorType()) {
-          const auto& tensor = *context.Input<Tensor>(i);
-          const auto& shape = tensor.Shape();
+          if (const auto* tensor = context.Input<Tensor>(i); tensor != nullptr) {
+            const auto& shape = tensor->Shape();
 
-          const bool is_shape_set = (dump_options.dump_flags & NodeDumpOptions::DumpFlags::Shape) != 0;
-          PrintIf(is_shape_set, MakeString(" Shape: ", shape, "\n"));
+            const bool is_shape_set = (dump_options.dump_flags & NodeDumpOptions::DumpFlags::Shape) != 0;
+            PrintIf(is_shape_set, MakeString(" Shape: ", shape, "\n"));
 
-          if ((dump_options.dump_flags & NodeDumpOptions::DumpFlags::InputData) != 0) {
-            tensor_metadata.name = input_defs[i]->Name();
-            tensor_metadata.step = dump_context.iteration;
-            tensor_metadata.consumer = node.Name() + ":" + std::to_string(i);
-            DumpTensor(dump_options, tensor, tensor_metadata, session_state);
+            if ((dump_options.dump_flags & NodeDumpOptions::DumpFlags::InputData) != 0) {
+              tensor_metadata.name = input_defs[i]->Name();
+              tensor_metadata.step = dump_context.iteration;
+              tensor_metadata.consumer = node.Name() + ":" + std::to_string(i);
+              DumpTensor(dump_options, *tensor, tensor_metadata, session_state);
+            }
+          } else {
+            std::cout << " is empty optional tensor.\n";
           }
         } else {
           std::cout << " is non-tensor type.\n";
         }
       } else {
-        // should never happen...
+        // this could happen with an empty Optional input
         std::cout << " was missing data type\n";
       }
     } else {
@@ -546,23 +549,26 @@ void DumpNodeOutputs(
       const auto* type = context.OutputType(i);
       if (type) {
         if (type->IsTensorType()) {
-          const auto& tensor = *context.Output<Tensor>(i);
-          const auto& shape = tensor.Shape();
+          if (const auto* tensor = context.Output<Tensor>(i); tensor != nullptr) {
+            const auto& shape = tensor->Shape();
 
-          const bool is_shape_set = (dump_options.dump_flags & NodeDumpOptions::DumpFlags::Shape) != 0;
-          PrintIf(is_shape_set, MakeString(" Shape: ", shape, "\n"));
+            const bool is_shape_set = (dump_options.dump_flags & NodeDumpOptions::DumpFlags::Shape) != 0;
+            PrintIf(is_shape_set, MakeString(" Shape: ", shape, "\n"));
 
-          if ((dump_options.dump_flags & NodeDumpOptions::DumpFlags::OutputData) != 0) {
-            tensor_metadata.name = output_defs[i]->Name();
-            tensor_metadata.step = dump_context.iteration;
-            tensor_metadata.producer = node.Name() + ":" + std::to_string(i);
-            DumpTensor(dump_options, tensor, tensor_metadata, session_state);
+            if ((dump_options.dump_flags & NodeDumpOptions::DumpFlags::OutputData) != 0) {
+              tensor_metadata.name = output_defs[i]->Name();
+              tensor_metadata.step = dump_context.iteration;
+              tensor_metadata.producer = node.Name() + ":" + std::to_string(i);
+              DumpTensor(dump_options, *tensor, tensor_metadata, session_state);
+            }
+          } else {
+            std::cout << " is empty optional tensor.\n";
           }
         } else {
           std::cout << " is non-tensor type.\n";
         }
       } else {
-        // should never happen...
+        // should never happen in a successful run
         std::cout << "missing data type\n";
       }
     } else {

onnxruntime/core/platform/windows/debug_alloc.cc

@@ -170,6 +170,13 @@ Memory_LeakCheck::Memory_LeakCheck() noexcept {
   g_heap = HeapCreate(0, 0, 0);
 }
 
+// print message to debug output and stdout
+// no trailing newline will be added
+static void DebugPrint(const char* message) {
+  OutputDebugStringA(message);
+  std::cout << "memleakdbg: " << message;
+}
+
 Memory_LeakCheck::~Memory_LeakCheck() {
   SymbolHelper symbols;
 
@@ -219,17 +226,17 @@ Memory_LeakCheck::~Memory_LeakCheck() {
         string.find("testing::internal::ThreadLocalRegistryImpl::GetThreadLocalsMapLocked") == std::string::npos &&
         string.find("testing::internal::ThreadLocalRegistryImpl::GetValueOnCurrentThread") == std::string::npos) {
       if (leaked_bytes == 0)
-        OutputDebugStringA("\n-----Starting Heap Trace-----\n\n");
+        DebugPrint("\n-----Starting Heap Trace-----\n\n");
 
       leak_count++;
       leaked_bytes += entry.cbData - sizeof(MemoryBlock);
-      OutputDebugStringA(string.c_str());
-      OutputDebugStringA("\n");
+      DebugPrint(string.c_str());
+      DebugPrint("\n");
     }
   }
 
   if (leaked_bytes) {
-    OutputDebugStringA("-----Ending Heap Trace-----\n\n");
+    DebugPrint("-----Ending Heap Trace-----\n\n");
 
     std::string string;
     char buffer[1024];
@@ -242,7 +249,7 @@ Memory_LeakCheck::~Memory_LeakCheck() {
     }
 
   } else {
-    OutputDebugStringA("\n----- No memory leaks detected -----\n\n");
+    DebugPrint("\n----- No memory leaks detected -----\n\n");
   }
 
   HeapDestroy(heap);

onnxruntime/core/providers/cpu/math/cumsum.cc

@@ -215,7 +215,8 @@ Status CumSum<T>::Compute(OpKernelContext* ctx) const {
       ::ZeroOutSliceAtIndex<T>(output_tensor, rank, axis, index, slice_dims, steps, slice_size);
       ++index;
     }
-    {
+
+    if (index < dim) {
       // The next slice is a copy of the input (if exclusive == false then this is the first slice)
       auto input_starts(::GetStarts(rank, axis, 0));
       auto output_starts(::GetStarts(rank, axis, index));

onnxruntime/test/providers/cpu/model_tests.cc

@@ -673,25 +673,24 @@ TEST_P(ModelTest, Run) {
 
   for (bool is_single_thread : use_single_thread) {
     for (ExecutionMode execution_mode : execution_modes) {
-      OrtSessionOptions* ortso;
-      ASSERT_ORT_STATUS_OK(OrtApis::CreateSessionOptions(&ortso));
+      Ort::SessionOptions ortso{};
       if (!is_single_thread) {
-        ASSERT_ORT_STATUS_OK(OrtApis::DisablePerSessionThreads(ortso));
+        ortso.DisablePerSessionThreads();
       } else {
-        ASSERT_ORT_STATUS_OK(OrtApis::SetIntraOpNumThreads(ortso, 1));
+        ortso.SetIntraOpNumThreads(1);
       }
-      ASSERT_ORT_STATUS_OK(OrtApis::SetSessionExecutionMode(ortso, execution_mode));
-      ASSERT_ORT_STATUS_OK(OrtApis::SetSessionLogId(ortso, ToUTF8String(test_case_name).c_str()));
-      ASSERT_ORT_STATUS_OK(OrtApis::SetSessionLogSeverityLevel(ortso, ORT_LOGGING_LEVEL_ERROR));
+      ortso.SetExecutionMode(execution_mode);
+      ortso.SetLogId(ToUTF8String(test_case_name).c_str());
+      ortso.SetLogSeverityLevel(ORT_LOGGING_LEVEL_ERROR);
       if (provider_name == "cuda") {
         OrtCUDAProviderOptionsV2* cuda_options = nullptr;
         ASSERT_ORT_STATUS_OK(OrtApis::CreateCUDAProviderOptions(&cuda_options));
         std::unique_ptr<OrtCUDAProviderOptionsV2, decltype(&OrtApis::ReleaseCUDAProviderOptions)> rel_cuda_options(
             cuda_options, &OrtApis::ReleaseCUDAProviderOptions);
-        ASSERT_ORT_STATUS_OK(OrtApis::SessionOptionsAppendExecutionProvider_CUDA_V2(ortso, cuda_options));
+        ortso.AppendExecutionProvider_CUDA_V2(*cuda_options);
       } else if (provider_name == "rocm") {
         OrtROCMProviderOptions ep_options;
-        ASSERT_ORT_STATUS_OK(OrtApis::SessionOptionsAppendExecutionProvider_ROCM(ortso, &ep_options));
+        ortso.AppendExecutionProvider_ROCM(ep_options);
       }
 #ifdef USE_DNNL
       else if (provider_name == "dnnl") {
@@ -703,26 +702,26 @@ TEST_P(ModelTest, Run) {
           OrtTensorRTProviderOptionsV2 params{0, 0, nullptr, 1000, 1, 1 << 30,
                                               1,  // enable fp16
                                               0, nullptr, 0, 0, 0, 0, 0, nullptr, 0, nullptr, 0, 0, 0};
-          ASSERT_ORT_STATUS_OK(OrtApis::SessionOptionsAppendExecutionProvider_TensorRT_V2(ortso, &params));
+          ortso.AppendExecutionProvider_TensorRT_V2(params);
         } else {
-          OrtTensorRTProviderOptionsV2* ep_option;
+          OrtTensorRTProviderOptionsV2* ep_option = nullptr;
           ASSERT_ORT_STATUS_OK(OrtApis::CreateTensorRTProviderOptions(&ep_option));
           std::unique_ptr<OrtTensorRTProviderOptionsV2, decltype(&OrtApis::ReleaseTensorRTProviderOptions)>
               rel_cuda_options(ep_option, &OrtApis::ReleaseTensorRTProviderOptions);
-          ASSERT_ORT_STATUS_OK(OrtApis::SessionOptionsAppendExecutionProvider_TensorRT_V2(ortso, ep_option));
+          ortso.AppendExecutionProvider_TensorRT_V2(*ep_option);
         }
         // Enable CUDA fallback
         OrtCUDAProviderOptionsV2* cuda_options = nullptr;
         ASSERT_ORT_STATUS_OK(OrtApis::CreateCUDAProviderOptions(&cuda_options));
         std::unique_ptr<OrtCUDAProviderOptionsV2, decltype(&OrtApis::ReleaseCUDAProviderOptions)> rel_cuda_options(
             cuda_options, &OrtApis::ReleaseCUDAProviderOptions);
-        ASSERT_ORT_STATUS_OK(OrtApis::SessionOptionsAppendExecutionProvider_CUDA_V2(ortso, cuda_options));
+        ortso.AppendExecutionProvider_CUDA_V2(*cuda_options);
       } else if (provider_name == "migraphx") {
         OrtMIGraphXProviderOptions ep_options;
-        ASSERT_ORT_STATUS_OK(OrtApis::SessionOptionsAppendExecutionProvider_MIGraphX(ortso, &ep_options));
+        ortso.AppendExecutionProvider_MIGraphX(ep_options);
       } else if (provider_name == "openvino") {
         OrtOpenVINOProviderOptions ep_options;
-        ASSERT_ORT_STATUS_OK(OrtApis::SessionOptionsAppendExecutionProvider_OpenVINO(ortso, &ep_options));
+        ortso.AppendExecutionProvider_OpenVINO(ep_options);
       }
 #ifdef USE_NNAPI
       else if (provider_name == "nnapi") {
@@ -760,8 +759,7 @@ TEST_P(ModelTest, Run) {
 #ifndef USE_DNNL  // potential crash for DNNL pipeline
       if (data_count > 1 && tests_run_parallel.find(l->GetTestCaseName()) != tests_run_parallel.end()) {
         LOGS_DEFAULT(ERROR) << "Parallel test for " << l->GetTestCaseName();  // TODO(leca): change level to INFO or even delete the log once verified parallel test working
-        Ort::SessionOptions ort_session_options(ortso);
-        std::shared_ptr<TestCaseResult> results = TestCaseRequestContext::Run(tp.get(), *l, *ort_env, ort_session_options, data_count, 1 /*repeat_count*/);
+        std::shared_ptr<TestCaseResult> results = TestCaseRequestContext::Run(tp.get(), *l, *ort_env, ortso, data_count, 1 /*repeat_count*/);
         for (EXECUTE_RESULT res : results->GetExcutionResult()) {
           EXPECT_EQ(res, EXECUTE_RESULT::SUCCESS) << "is_single_thread:" << is_single_thread << ", execution_mode:" << execution_mode << ", provider_name:"
                                                   << provider_name << ", test name:" << results->GetName() << ", result: " << res;
@@ -769,8 +767,6 @@ TEST_P(ModelTest, Run) {
         continue;
       }
 #endif  // !USE_DNNL
-      std::unique_ptr<OrtSessionOptions, decltype(&OrtApis::ReleaseSessionOptions)> rel_ort_session_option(
-          ortso, &OrtApis::ReleaseSessionOptions);
       // TODO(leca): leverage TestCaseRequestContext::Run() to make it short
       auto default_allocator = std::make_unique<MockedOrtAllocator>();
 

tools/ci_build/github/azure-pipelines/post-merge-jobs.yml

@@ -1,143 +1,158 @@
-jobs:
-- template: templates/win-ci-vs-2019.yml
-  parameters:
-    BuildConfig: 'Debug'
-    EnvSetupScript: setup_env.bat
-    buildArch: x64
-    additionalBuildFlags: --disable_memleak_checker --use_mimalloc
-    msbuildPlatform: x64
-    isX86: false
-    job_name_suffix: x64_mimalloc
-    RunOnnxRuntimeTests: true
-    RunStaticCodeAnalysis: false
-    isTraining: false
-    ORT_EP_NAME: CPU
-    GenerateDocumentation: false
-    EnablePython: false
-    MachinePool: 'Win-CPU-2021'
+stages:
+- stage: Mimalloc
+  dependsOn: []
+  jobs:
+  - template: templates/win-ci-vs-2019.yml
+    parameters:
+      BuildConfig: 'Debug'
+      EnvSetupScript: setup_env.bat
+      buildArch: x64
+      additionalBuildFlags: --disable_memleak_checker --use_mimalloc
+      msbuildPlatform: x64
+      isX86: false
+      job_name_suffix: x64_mimalloc
+      RunOnnxRuntimeTests: true
+      RunStaticCodeAnalysis: false
+      isTraining: false
+      ORT_EP_NAME: CPU
+      GenerateDocumentation: false
+      EnablePython: false
+      MachinePool: 'Win-CPU-2021'
 
-- template: templates/win-ci-vs-2019.yml
-  parameters:
-    BuildConfig: 'Debug'
-    EnvSetupScript: setup_env.bat
-    buildArch: x64
-    additionalBuildFlags: --cmake_extra_defines onnxruntime_DISABLE_ABSEIL=ON
-    msbuildPlatform: x64
-    isX86: false
-    job_name_suffix: x64_no_absl
-    RunOnnxRuntimeTests: true
-    RunStaticCodeAnalysis: false
-    isTraining: false
-    ORT_EP_NAME: CPU
-    GenerateDocumentation: false
-    EnablePython: false
-    MachinePool: 'Win-CPU-2021'
+- stage: NoAbseil
+  dependsOn: []
+  jobs:
+  - template: templates/win-ci-vs-2019.yml
+    parameters:
+      BuildConfig: 'Debug'
+      EnvSetupScript: setup_env.bat
+      buildArch: x64
+      additionalBuildFlags: --cmake_extra_defines onnxruntime_DISABLE_ABSEIL=ON
+      msbuildPlatform: x64
+      isX86: false
+      job_name_suffix: x64_no_absl
+      RunOnnxRuntimeTests: true
+      RunStaticCodeAnalysis: false
+      isTraining: false
+      ORT_EP_NAME: CPU
+      GenerateDocumentation: false
+      EnablePython: false
+      MachinePool: 'Win-CPU-2021'
 
-- template: templates/win-ci-vs-2019.yml
-  parameters:
-    BuildConfig: 'Debug'
-    EnvSetupScript: setup_env.bat
-    buildArch: x64
-    additionalBuildFlags: --build_shared_lib --minimal_build --disable_exceptions
-    msbuildPlatform: x64
-    isX86: false
-    job_name_suffix: x64_minimal_no_exception
-    RunOnnxRuntimeTests: true
-    RunStaticCodeAnalysis: false
-    isTraining: false
-    ORT_EP_NAME: CPU
-    GenerateDocumentation: false
-    EnablePython: false
-    MachinePool: 'Win-CPU-2021'
+- stage: MinimalBuildWithNoExceptions
+  dependsOn: []
+  jobs:
+  - template: templates/win-ci-vs-2019.yml
+    parameters:
+      BuildConfig: 'Debug'
+      EnvSetupScript: setup_env.bat
+      buildArch: x64
+      additionalBuildFlags: --build_shared_lib --minimal_build --disable_exceptions
+      msbuildPlatform: x64
+      isX86: false
+      job_name_suffix: x64_minimal_no_exception
+      RunOnnxRuntimeTests: true
+      RunStaticCodeAnalysis: false
+      isTraining: false
+      ORT_EP_NAME: CPU
+      GenerateDocumentation: false
+      EnablePython: false
+      MachinePool: 'Win-CPU-2021'
 
-- template: templates/win-ci-vs-2019.yml
-  parameters:
-    BuildConfig: 'Debug'
-    EnvSetupScript: setup_env.bat
-    buildArch: x64
-    additionalBuildFlags: --build_shared_lib --cmake_extra_defines onnxruntime_DEBUG_NODE_INPUTS_OUTPUTS=ON
-    msbuildPlatform: x64
-    isX86: false
-    job_name_suffix: x64_debug_node_input_output
-    RunOnnxRuntimeTests: true
-    RunStaticCodeAnalysis: false
-    isTraining: false
-    ORT_EP_NAME: CPU
-    GenerateDocumentation: false
-    EnablePython: false
-    MachinePool: 'Win-CPU-2021'
+- stage: DebugNodeInputsOutputs
+  dependsOn: []
+  jobs:
+  - template: templates/win-ci-vs-2019.yml
+    parameters:
+      BuildConfig: 'Debug'
+      EnvSetupScript: setup_env.bat
+      buildArch: x64
+      additionalBuildFlags: --build_shared_lib --cmake_extra_defines onnxruntime_DEBUG_NODE_INPUTS_OUTPUTS=ON
+      msbuildPlatform: x64
+      isX86: false
+      job_name_suffix: x64_debug_node_input_output
+      RunOnnxRuntimeTests: true
+      RunStaticCodeAnalysis: false
+      isTraining: false
+      ORT_EP_NAME: CPU
+      GenerateDocumentation: false
+      EnablePython: false
+      MachinePool: 'Win-CPU-2021'
 
-- job: CodeCoverage
-  workspace:
-    clean: all
-  timeoutInMinutes:  150
-  variables:
-    skipComponentGovernanceDetection: true
-  pool: 'Linux-CPU'
-  steps:
-    - template: templates/set-version-number-variables-step.yml
+- stage: CodeCoverage
+  dependsOn: []
+  jobs:
+  - job: CodeCoverage
+    workspace:
+      clean: all
+    timeoutInMinutes:  150
+    variables:
+      skipComponentGovernanceDetection: true
+    pool: 'Linux-CPU'
+    steps:
+      - template: templates/set-version-number-variables-step.yml
 
-    - task: CmdLine@2
+      - task: CmdLine@2
+        inputs:
+          script: |
+            set -e
+            ln -s /data/models .
+            #Build onnxruntime and run the instrumented program(unitests)
+            LLVM_PROFILE_FILE="%p.profraw" CFLAGS="-g -fprofile-instr-generate -fcoverage-mapping" CXXFLAGS="-g -fprofile-instr-generate -fcoverage-mapping" CC=clang CXX=clang++  python3 $(Build.SourcesDirectory)/tools/ci_build/build.py --build_dir=$(Build.BinariesDirectory) --config Debug --parallel --skip_submodule_sync --build_shared_lib --enable_onnx_tests --cmake_extra_defines RUN_MODELTEST_IN_DEBUG_MODE=ON
+
+            cd Debug
+            ./onnxruntime_mlas_test
+            #Merge the mulitple prof data into a single indexed profile data file
+            llvm-profdata merge -sparse -o ort.profdata *.profraw
+            #Create coverage report, output the result to 'report.json'
+            llvm-cov export -summary-only -instr-profile=ort.profdata onnxruntime_test_all -object onnxruntime_mlas_test -object onnxruntime_api_tests_without_env -object onnx_test_runner -object onnxruntime_shared_lib_test -object onnxruntime_global_thread_pools_test -object onnxruntime_api_tests_without_env $(Build.SourcesDirectory)/include/onnxruntime $(Build.SourcesDirectory)/onnxruntime/core $(Build.SourcesDirectory)/onnxruntime/contrib_ops > $(Build.BinariesDirectory)/report.json
+
+            llvm-cov show -instr-profile=ort.profdata onnxruntime_test_all -object onnxruntime_mlas_test -object onnxruntime_api_tests_without_env -object onnx_test_runner -object onnxruntime_shared_lib_test -object onnxruntime_global_thread_pools_test -object onnxruntime_api_tests_without_env $(Build.SourcesDirectory)/include/onnxruntime $(Build.SourcesDirectory)/onnxruntime/core $(Build.SourcesDirectory)/onnxruntime/contrib_ops --format=html -output-dir=$(Build.ArtifactStagingDirectory)
+          workingDirectory: $(Build.BinariesDirectory)
+
+      - task: AzureCLI@2
+        displayName: 'Azure CLI'
+        inputs:
+          azureSubscription: AIInfraBuildOnnxRuntimeOSS
+          scriptType: bash
+          scriptPath: $(Build.SourcesDirectory)/tools/ci_build/github/linux/upload_code_coverage_data.sh
+          arguments: '"$(Build.BinariesDirectory)/report.json" "https://aiinfra.visualstudio.com/Lotus/_build/results?buildId=$(Build.BuildId)" x64 linux default'
+          workingDirectory: '$(Build.BinariesDirectory)'
+
+      - task: PublishPipelineArtifact@1
+        displayName: 'Publish Pipeline Artifact'
+        inputs:
+          targetPath: '$(Build.ArtifactStagingDirectory)'
+          artifact: html
+
+  - job: Upload
+    workspace:
+      clean: all
+    pool: 'Win-CPU-2021'
+    variables:
+    - name: runCodesignValidationInjection
+      value: false
+    - name: skipComponentGovernanceDetection
+      value: true
+    dependsOn:
+    - CodeCoverage
+    condition: and (succeeded(), eq(variables['Build.SourceBranch'], 'refs/heads/main'))
+    steps:
+    - checkout: self
+      submodules: false
+
+    - task: DownloadPipelineArtifact@2
+      displayName: 'Download Pipeline Artifact'
       inputs:
-        script: |
-          set -e
-          ln -s /data/models .
-          #Build onnxruntime and run the instrumented program(unitests)
-          LLVM_PROFILE_FILE="%p.profraw" CFLAGS="-g -fprofile-instr-generate -fcoverage-mapping" CXXFLAGS="-g -fprofile-instr-generate -fcoverage-mapping" CC=clang CXX=clang++  python3 $(Build.SourcesDirectory)/tools/ci_build/build.py --build_dir=$(Build.BinariesDirectory) --config Debug --parallel --skip_submodule_sync --build_shared_lib --enable_onnx_tests --cmake_extra_defines RUN_MODELTEST_IN_DEBUG_MODE=ON
+        buildType: 'current'
+        artifactName: 'html'
+        targetPath: '$(Build.BinariesDirectory)'
 
-          cd Debug
-          ./onnxruntime_mlas_test
-          #Merge the mulitple prof data into a single indexed profile data file
-          llvm-profdata merge -sparse -o ort.profdata *.profraw
-          #Create coverage report, output the result to 'report.json'
-          llvm-cov export -summary-only -instr-profile=ort.profdata onnxruntime_test_all -object onnxruntime_mlas_test -object onnxruntime_api_tests_without_env -object onnx_test_runner -object onnxruntime_shared_lib_test -object onnxruntime_global_thread_pools_test -object onnxruntime_api_tests_without_env $(Build.SourcesDirectory)/include/onnxruntime $(Build.SourcesDirectory)/onnxruntime/core $(Build.SourcesDirectory)/onnxruntime/contrib_ops > $(Build.BinariesDirectory)/report.json
-
-          llvm-cov show -instr-profile=ort.profdata onnxruntime_test_all -object onnxruntime_mlas_test -object onnxruntime_api_tests_without_env -object onnx_test_runner -object onnxruntime_shared_lib_test -object onnxruntime_global_thread_pools_test -object onnxruntime_api_tests_without_env $(Build.SourcesDirectory)/include/onnxruntime $(Build.SourcesDirectory)/onnxruntime/core $(Build.SourcesDirectory)/onnxruntime/contrib_ops --format=html -output-dir=$(Build.ArtifactStagingDirectory)
-        workingDirectory: $(Build.BinariesDirectory)
-
-    - task: AzureCLI@2
-      displayName: 'Azure CLI'
+    - task: AzureFileCopy@4
+      displayName: 'AzureBlob File Copy'
       inputs:
+        SourcePath: '$(Build.BinariesDirectory)\*'
         azureSubscription: AIInfraBuildOnnxRuntimeOSS
-        scriptType: bash
-        scriptPath: $(Build.SourcesDirectory)/tools/ci_build/github/linux/upload_code_coverage_data.sh
-        arguments: '"$(Build.BinariesDirectory)/report.json" "https://aiinfra.visualstudio.com/Lotus/_build/results?buildId=$(Build.BuildId)" x64 linux default'
-        workingDirectory: '$(Build.BinariesDirectory)'
-
-    - task: PublishPipelineArtifact@1
-      displayName: 'Publish Pipeline Artifact'
-      inputs:
-        targetPath: '$(Build.ArtifactStagingDirectory)'
-        artifact: html
-
-- job: Upload
-  workspace:
-    clean: all
-  pool: 'Win-CPU-2021'
-  variables:
-  - name: runCodesignValidationInjection
-    value: false
-  - name: skipComponentGovernanceDetection
-    value: true
-  dependsOn:
-  - CodeCoverage
-  condition: and (succeeded(), eq(variables['Build.SourceBranch'], 'refs/heads/main'))
-  steps:
-  - checkout: self
-    submodules: false
-
-  - task: DownloadPipelineArtifact@2
-    displayName: 'Download Pipeline Artifact'
-    inputs:
-      buildType: 'current'
-      artifactName: 'html'
-      targetPath: '$(Build.BinariesDirectory)'
-
-  - task: AzureFileCopy@4
-    displayName: 'AzureBlob File Copy'
-    inputs:
-      SourcePath: '$(Build.BinariesDirectory)\*'
-      azureSubscription: AIInfraBuildOnnxRuntimeOSS
-      Destination: AzureBlob
-      storage: onnxruntime
-      ContainerName: '$web'
+        Destination: AzureBlob
+        storage: onnxruntime
+        ContainerName: '$web'