[TRTLLM-17005][refactor] Component-local C++ Tests (Part 5)

cpp/tensorrt_llm/CMakeLists.txt

@@ -192,11 +192,6 @@ endif()
 find_package(Threads REQUIRED)
 add_subdirectory(common)
 
-add_library(ipc_nvls_memory_src STATIC runtime/ipcNvlsMemory.cu)
-set_target_properties(
-  ipc_nvls_memory_src PROPERTIES POSITION_INDEPENDENT_CODE ON
-                                 CUDA_RESOLVE_DEVICE_SYMBOLS ON)
-
 add_subdirectory(kernels)
 add_subdirectory(runtime)
 
@@ -254,7 +249,6 @@ set(TRTLLM_LINK_LIBS
     compressorKernels_src
     fusedSamplingKernels_src
     mhcKernels_src
-    userbuffers_src
     ${DECODER_SHARED_TARGET_0}
     ${DECODER_SHARED_TARGET_1})
 if(USING_OSS_CUTLASS_LOW_LATENCY_GEMM)
@@ -318,15 +312,48 @@ endif()
 
 target_link_libraries(${SHARED_TARGET} PUBLIC ${TRTLLM_LINK_LIBS})
 
-target_link_libraries(
-  ${SHARED_TARGET}
-  PRIVATE
-    $<LINK_LIBRARY:WHOLE_ARCHIVE,tllm_common_error,tllm_common_logging,tllm_common_cuda,tllm_executor_types,tllm_executor_serialization,tllm_runtime_memory,tllm_runtime_buffers,tllm_runtime_distributed,tllm_runtime_topology>
-    $<LINK_LIBRARY:WHOLE_ARCHIVE,tllm_common_environment,tllm_executor_values,tllm_executor_transfer_agent,tllm_kv_cache_host_memory,tllm_kv_cache,tllm_kv_cache_legacy,tllm_batch_scheduler,tllm_kernel_kv_cache_copy>
-    $<LINK_LIBRARY:WHOLE_ARCHIVE,${BATCH_MANAGER_TARGET}>
-    $<LINK_LIBRARY:WHOLE_ARCHIVE,${EXECUTOR_TARGET}>
-    $<LINK_LIBRARY:WHOLE_ARCHIVE,fp8_blockscale_gemm_src>
-    $<LINK_LIBRARY:WHOLE_ARCHIVE,${INTERNAL_CUTLASS_KERNELS_TARGET}>)
+set(TLLM_FACADE_COMPONENT_TARGETS
+    ipc_nvls_memory_src
+    tllm_batch_request_info
+    tllm_batch_scheduler
+    tllm_common_cuda
+    tllm_common_environment
+    tllm_common_error
+    tllm_common_logging
+    tllm_common_profiler
+    tllm_context_transfer_consensus
+    tllm_executor_agent_name
+    tllm_executor_memory_desc
+    tllm_executor_serialization
+    tllm_executor_transfer_agent
+    tllm_executor_types
+    tllm_executor_values
+    tllm_kernel_kv_cache_copy
+    tllm_kv_cache
+    tllm_kv_cache_host_memory
+    tllm_kv_cache_legacy
+    tllm_peft_cache
+    tllm_runtime_buffers
+    tllm_runtime_distributed
+    tllm_runtime_gdr
+    tllm_runtime_host_accessible_memory
+    tllm_runtime_locality
+    tllm_runtime_lora
+    tllm_runtime_lora_cache
+    tllm_runtime_memory
+    tllm_runtime_numpy
+    tllm_runtime_topology
+    tllm_runtime_utils
+    tllm_runtime_worker_pool
+    userbuffers_src)
+
+foreach(
+  component
+  ${TLLM_FACADE_COMPONENT_TARGETS} ${BATCH_MANAGER_TARGET} ${EXECUTOR_TARGET}
+  fp8_blockscale_gemm_src ${INTERNAL_CUTLASS_KERNELS_TARGET})
+  target_link_libraries(${SHARED_TARGET}
+                        PRIVATE $<LINK_LIBRARY:WHOLE_ARCHIVE,${component}>)
+endforeach()
 
 foreach(component IN LISTS TLLM_KERNEL_COMPONENT_TARGETS)
   target_link_libraries(${SHARED_TARGET}
@@ -336,29 +363,7 @@ foreach(component IN LISTS TLLM_KERNEL_COMPONENT_TARGETS)
 endforeach()
 
 # Components also link each other normally; retain all component exports here.
-set_property(TARGET ${SHARED_TARGET}
-             PROPERTY LINK_LIBRARY_OVERRIDE_tllm_common_error WHOLE_ARCHIVE)
-set_property(TARGET ${SHARED_TARGET}
-             PROPERTY LINK_LIBRARY_OVERRIDE_tllm_common_logging WHOLE_ARCHIVE)
-set_property(TARGET ${SHARED_TARGET}
-             PROPERTY LINK_LIBRARY_OVERRIDE_tllm_executor_types WHOLE_ARCHIVE)
-
-foreach(
-  component
-  tllm_common_cuda
-  tllm_runtime_memory
-  tllm_runtime_buffers
-  tllm_runtime_distributed
-  tllm_runtime_topology
-  tllm_executor_serialization
-  tllm_common_environment
-  tllm_executor_values
-  tllm_executor_transfer_agent
-  tllm_kv_cache_host_memory
-  tllm_kv_cache
-  tllm_kv_cache_legacy
-  tllm_batch_scheduler
-  tllm_kernel_kv_cache_copy)
+foreach(component IN LISTS TLLM_FACADE_COMPONENT_TARGETS)
   set_property(TARGET ${SHARED_TARGET}
                PROPERTY LINK_LIBRARY_OVERRIDE_${component} WHOLE_ARCHIVE)
 endforeach()

cpp/tensorrt_llm/batch_manager/CMakeLists.txt

@@ -27,7 +27,6 @@ set(SRCS
     kvCacheManagerV2Utils.cpp
     kvCacheManagerV2Utils.cu
     mlaCacheFormatter.cpp
-    peftCacheManager.cpp
     rnnCacheFormatter.cpp
     rnnCacheTransBuffer.cpp)
 
@@ -117,6 +116,38 @@ target_link_libraries(tllm_batch_scheduler PUBLIC tllm::kv_cache_legacy)
 target_include_directories(tllm_batch_scheduler
                            PRIVATE ${maybe_nvtx_includedir})
 
+add_library(tllm_context_transfer_consensus STATIC
+            contextTransferVoteReducer.cpp)
+add_library(tllm::context_transfer_consensus ALIAS
+            tllm_context_transfer_consensus)
+target_link_libraries(tllm_context_transfer_consensus PUBLIC tllm::common_error)
+target_compile_features(tllm_context_transfer_consensus PUBLIC cxx_std_20)
+set_target_properties(
+  tllm_context_transfer_consensus PROPERTIES CXX_EXTENSIONS OFF
+                                             POSITION_INDEPENDENT_CODE ON)
+
+add_library(tllm_batch_request_info STATIC requestInfo.cpp)
+add_library(tllm::batch_request_info ALIAS tllm_batch_request_info)
+target_link_libraries(
+  tllm_batch_request_info
+  PUBLIC tllm::kv_cache_legacy
+  PRIVATE ${Python3_LIBRARIES})
+target_compile_features(tllm_batch_request_info PUBLIC cxx_std_20)
+set_target_properties(
+  tllm_batch_request_info PROPERTIES CXX_EXTENSIONS OFF
+                                     POSITION_INDEPENDENT_CODE ON)
+
+add_library(tllm_peft_cache STATIC peftCacheManager.cpp)
+add_library(tllm::peft_cache ALIAS tllm_peft_cache)
+target_link_libraries(
+  tllm_peft_cache
+  PUBLIC tllm::runtime_lora_cache tllm::kv_cache_legacy
+         tllm::runtime_worker_pool
+  PRIVATE tllm::runtime_numpy tllm::runtime_utils)
+target_compile_features(tllm_peft_cache PUBLIC cxx_std_20)
+set_target_properties(tllm_peft_cache PROPERTIES CXX_EXTENSIONS OFF
+                                                 POSITION_INDEPENDENT_CODE ON)
+
 foreach(component tllm_kv_cache_host_memory tllm_kv_cache tllm_kv_cache_legacy
                   tllm_batch_scheduler)
   target_compile_features(${component} PUBLIC cxx_std_20)
@@ -129,8 +160,10 @@ target_compile_options(
 add_cuda_architectures(tllm_kv_cache 89)
 
 add_library(${BATCH_MANAGER_STATIC_TARGET} STATIC ${SRCS})
-target_link_libraries(${BATCH_MANAGER_STATIC_TARGET}
-                      PUBLIC tllm::kv_cache tllm::batch_scheduler)
+target_link_libraries(
+  ${BATCH_MANAGER_STATIC_TARGET}
+  PUBLIC tllm::kv_cache tllm::batch_scheduler tllm::context_transfer_consensus
+         tllm::peft_cache tllm::batch_request_info)
 target_include_directories(
   ${BATCH_MANAGER_STATIC_TARGET}
   PUBLIC ${KV_CACHE_MANAGER_V2_INCLUDE_DIR}

cpp/tensorrt_llm/batch_manager/contextTransferCoordinator.cpp

@@ -32,79 +32,6 @@
 namespace tensorrt_llm::batch_manager
 {
 
-ContextTransferVoteReducer::ContextTransferVoteReducer(int const participantCount)
-    : mParticipantCount(participantCount)
-{
-    TLLM_CHECK_WITH_INFO(participantCount > 0, "Context-transfer consensus requires at least one participant.");
-}
-
-void ContextTransferVoteReducer::recordVote(
-    int const participantRank, std::uint64_t const requestId, ContextTransferVote const vote)
-{
-    TLLM_CHECK_WITH_INFO(participantRank >= 0 && participantRank < mParticipantCount,
-        "Context-transfer consensus participant rank is out of range.");
-    TLLM_CHECK_WITH_INFO(vote == ContextTransferVote::kCompleted || vote == ContextTransferVote::kFailed,
-        "Context-transfer consensus received an invalid vote.");
-
-    auto [requestIt, inserted] = mRequestVotes.try_emplace(requestId, mParticipantCount);
-    static_cast<void>(inserted);
-    auto& requestVotes = requestIt->second;
-    auto& recordedVote = requestVotes.votes.at(static_cast<std::size_t>(participantRank));
-    auto const packedVote = static_cast<std::uint64_t>(vote);
-    if (recordedVote != 0)
-    {
-        TLLM_CHECK_WITH_INFO(
-            recordedVote == packedVote, "Context-transfer participant changed its terminal vote for a request.");
-        return;
-    }
-
-    recordedVote = packedVote;
-    ++requestVotes.terminalCount;
-    requestVotes.failed = requestVotes.failed || vote == ContextTransferVote::kFailed;
-}
-
-void ContextTransferVoteReducer::recordTimeout(std::uint64_t const requestId)
-{
-    auto [requestIt, inserted] = mRequestVotes.try_emplace(requestId, mParticipantCount);
-    static_cast<void>(inserted);
-    auto& requestVotes = requestIt->second;
-    if (!requestVotes.timedOut)
-    {
-        requestVotes.timedOut = true;
-        requestVotes.timeoutPending = true;
-    }
-}
-
-ContextTransferConsensusResult ContextTransferVoteReducer::takeReady()
-{
-    ContextTransferConsensusResult result;
-    for (auto requestIt = mRequestVotes.begin(); requestIt != mRequestVotes.end();)
-    {
-        auto& requestVotes = requestIt->second;
-        if (requestVotes.timeoutPending)
-        {
-            result.timedOutRequestIds.insert(requestIt->first);
-            requestVotes.timeoutPending = false;
-        }
-        if (requestVotes.terminalCount != mParticipantCount)
-        {
-            ++requestIt;
-            continue;
-        }
-
-        auto& terminalRequestIds
-            = (requestVotes.failed || requestVotes.timedOut) ? result.failedRequestIds : result.completedRequestIds;
-        terminalRequestIds.insert(requestIt->first);
-        requestIt = mRequestVotes.erase(requestIt);
-    }
-    return result;
-}
-
-void ContextTransferVoteReducer::clear() noexcept
-{
-    mRequestVotes.clear();
-}
-
 class ContextTransferCoordinator::Impl
 {
 public:

cpp/tensorrt_llm/batch_manager/contextTransferVoteReducer.cpp

@@ -0,0 +1,98 @@
+/*
+ * SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "tensorrt_llm/batch_manager/contextTransferCoordinator.h"
+
+#include "tensorrt_llm/common/assert.h"
+
+namespace tensorrt_llm::batch_manager
+{
+
+ContextTransferVoteReducer::ContextTransferVoteReducer(int const participantCount)
+    : mParticipantCount(participantCount)
+{
+    TLLM_CHECK_WITH_INFO(participantCount > 0, "Context-transfer consensus requires at least one participant.");
+}
+
+void ContextTransferVoteReducer::recordVote(
+    int const participantRank, std::uint64_t const requestId, ContextTransferVote const vote)
+{
+    TLLM_CHECK_WITH_INFO(participantRank >= 0 && participantRank < mParticipantCount,
+        "Context-transfer consensus participant rank is out of range.");
+    TLLM_CHECK_WITH_INFO(vote == ContextTransferVote::kCompleted || vote == ContextTransferVote::kFailed,
+        "Context-transfer consensus received an invalid vote.");
+
+    auto [requestIt, inserted] = mRequestVotes.try_emplace(requestId, mParticipantCount);
+    static_cast<void>(inserted);
+    auto& requestVotes = requestIt->second;
+    auto& recordedVote = requestVotes.votes.at(static_cast<std::size_t>(participantRank));
+    auto const packedVote = static_cast<std::uint64_t>(vote);
+    if (recordedVote != 0)
+    {
+        TLLM_CHECK_WITH_INFO(
+            recordedVote == packedVote, "Context-transfer participant changed its terminal vote for a request.");
+        return;
+    }
+
+    recordedVote = packedVote;
+    ++requestVotes.terminalCount;
+    requestVotes.failed = requestVotes.failed || vote == ContextTransferVote::kFailed;
+}
+
+void ContextTransferVoteReducer::recordTimeout(std::uint64_t const requestId)
+{
+    auto [requestIt, inserted] = mRequestVotes.try_emplace(requestId, mParticipantCount);
+    static_cast<void>(inserted);
+    auto& requestVotes = requestIt->second;
+    if (!requestVotes.timedOut)
+    {
+        requestVotes.timedOut = true;
+        requestVotes.timeoutPending = true;
+    }
+}
+
+ContextTransferConsensusResult ContextTransferVoteReducer::takeReady()
+{
+    ContextTransferConsensusResult result;
+    for (auto requestIt = mRequestVotes.begin(); requestIt != mRequestVotes.end();)
+    {
+        auto& requestVotes = requestIt->second;
+        if (requestVotes.timeoutPending)
+        {
+            result.timedOutRequestIds.insert(requestIt->first);
+            requestVotes.timeoutPending = false;
+        }
+        if (requestVotes.terminalCount != mParticipantCount)
+        {
+            ++requestIt;
+            continue;
+        }
+
+        auto& terminalRequestIds
+            = (requestVotes.failed || requestVotes.timedOut) ? result.failedRequestIds : result.completedRequestIds;
+        terminalRequestIds.insert(requestIt->first);
+        requestIt = mRequestVotes.erase(requestIt);
+    }
+    return result;
+}
+
+void ContextTransferVoteReducer::clear() noexcept
+{
+    mRequestVotes.clear();
+}
+
+} // namespace tensorrt_llm::batch_manager