virtual function override intended?解决

最新推荐文章于 2023-03-01 12:18:24 发布

AI算法加油站

最新推荐文章于 2023-03-01 12:18:24 发布

阅读量1.3k

点赞数

文章标签： yolov5

原文链接：https://blog.csdn.net/XCCCCZ/article/details/117128710

版权

TensorRT的IPluginV2IOExt 和 IPluginV2Ext 的configurePlugin() 的virtual function override问题

virtual function override intended?

有类似如下代码:

namespace nvinfer1
{
    class YoloLayerPlugin : public IPluginV2IOExt {
        ...
        void configurePlugin(const PluginTensorDesc* in, int nbInput, const PluginTensorDesc* out, int nbOutput) override;
        ...
    }
    ...
}

在编译时总有如下警告:

yololayer.h(85): warning: function "nvinfer1::IPluginV2Ext::configurePlugin(const nvinfer1::Dims *, int, const nvinfer1::Dims *, int, const nvinfer1::DataType *, const nvinfer1::DataType *, const __nv_bool *, const __nv_bool *, nvinfer1::PluginFormat, int)" is hidden by "nvinfer1::YoloLayerPlugin::configurePlugin" -- virtual function override intended?
yololayer.h(85): warning: function "nvinfer1::IPluginV2Ext::configurePlugin(const nvinfer1::Dims *, int, const nvinfer1::Dims *, int, const nvinfer1::DataType *, const nvinfer1::DataType *, const bool *, const bool *, nvinfer1::PluginFormat, int)" is hidden by "nvinfer1::YoloLayerPlugin::configurePlugin" -- virtual function override intended?

为了消除警告，翻看了一下tensorrt相关代码，明白了产生这些警告是怎么回事:

类YoloLayerPlugin 继承IPluginV2IOExt，并且实现了IPluginV2IOExt的虚函数configurePlugin():

class YoloLayerPlugin : public IPluginV2IOExt

从/usr/include/aarch64-linux-gnu/NvInferRuntimeCommon.h里代码看，IPluginV2IOExt继承自IPluginV2Ext，而IPluginV2Ext也有一个同名虚函数configurePlugin():

class IPluginV2IOExt : public IPluginV2Ext
{
public:
    //!
    //! \brief Configure the layer.
    //!
    //! This function is called by the builder prior to initialize(). It provides an opportunity for the layer to make
    //! algorithm choices on the basis of I/O PluginTensorDesc and the maximum batch size.
    //!
    //! \param in The input tensors attributes that are used for configuration.
    //! \param nbInput Number of input tensors.
    //! \param out The output tensors attributes that are used for configuration.
    //! \param nbOutput Number of output tensors.
    //!
    virtual void configurePlugin(const PluginTensorDesc* in, int nbInput, const PluginTensorDesc* out, int nbOutput) TRTNOEXCEPT = 0;
...
 
class IPluginV2Ext : public IPluginV2
{
public:
  ...
   virtual void configurePlugin(const Dims* inputDims, int nbInputs, const Dims* outputDims,
                                 int nbOutputs, const DataType* inputTypes, const DataType* outputTypes,
                                 const bool* inputIsBroadcast, const bool* outputIsBroadcast, PluginFormat floatFormat, int maxBatchSize) TRTNOEXCEPT = 0;

...
这样的话，IPluginV2Ext::configurePlugin()就被hidden了，要消除这个警告，在类YoloLayerPlugin的定义代码里增加个using IPluginV2Ext::configurePlugin即可解决：

namespace nvinfer1
{
    class YoloLayerPlugin : public IPluginV2IOExt
        ...
        void configurePlugin(const PluginTensorDesc* in, int nbInput, const PluginTensorDesc* out, int nbOutput) override;
 
        using IPluginV2Ext::configurePlugin;
        ...
     }
     ...
 }

原文链接：https://blog.csdn.net/XCCCCZ/article/details/117128710

我是在yololayer.h加上这个代码也有效：

        void configurePlugin(const PluginTensorDesc* in, int nbInput, const PluginTensorDesc* out, int nbOutput) TRT_NOEXCEPT override;
        using IPluginV2Ext::configurePlugin;