Minor improvements

Author: acvictor

cpp/velox/cudf/GpuLock.cc

@@ -45,7 +45,6 @@ void lockGpu() {
         return;
     }
 
-
     // Wait until the GPU lock becomes available
     getGpuLockState().gGpuCv.wait(lock, [] {
         return !getGpuLockState().gGpuOwner.has_value();
@@ -72,12 +71,12 @@ void unlockGpu() {
     std::thread::id tid = std::this_thread::get_id();
     std::unique_lock<std::mutex> lock(getGpuLockState().gGpuMutex);
     if (!getGpuLockState().gGpuOwner.has_value()) {
-        LOG(INFO) <<"unlockGpu() called by non-owner thread!"<< std::endl;
+        LOG(INFO) << "unlockGpu() called when no thread holds the lock!";
         return;
     }
 
-    if (!getGpuLockState().gGpuOwner.has_value() || getGpuLockState().gGpuOwner != tid) {
-        throw std::runtime_error("unlockGpu() called by other-owner thread!");
+    if (getGpuLockState().gGpuOwner != tid) {
+        throw std::runtime_error("unlockGpu() called by other thread!");
     }
 
     // Release ownership
@@ -88,5 +87,4 @@ void unlockGpu() {
     getGpuLockState().gGpuCv.notify_one();
 }
 
-
 } // namespace gluten

cpp/velox/shuffle/VeloxGpuShuffleReader.cc

@@ -75,7 +75,6 @@ bool VeloxGpuHashShuffleReaderDeserializer::resolveNextBlockType() {
     default:
       throw GlutenException(fmt::format("Unsupported block type: {}", static_cast<int32_t>(blockType)));
   }
-  return true;
 }
 
 void VeloxGpuHashShuffleReaderDeserializer::loadNextStream() {