【OpenCL】paddleDetection model s2anet_conv segfault with Opencl backend

[QShiX]

-系统信息    1）Paddle Lite 版本：develop bb0260034b4f1ac77f502021dee587c6eeddff6c     2）Host 环境：Ubuntu 22.04.1 LTS 5.15.0-47-generic    3）运行设备环境：cpu: x86, gpu:GPU AMD Radeon RX 6900 XT

• 预测信息    1）预测 API：C++    2）预测选项信息：./build.lite.linux.x86.gcc.opencl/lite/api/tools/benchmark/benchmark_bin --model_file=./models/s2anet_conv_2x_dota/model.pdmodel --param_file=./models/s2anet_conv_2x_dota/model.pdiparams --input_shape=1,2:1,3,1,q:1,2 --warmup=10 --repeats=20 --backend=opencl,x86    3）预测库来源：paddleDetection 导出

错误信息： Segmentation fault at /home/qingchuan/work/gitlab/paddle-lite/lite/backends/opencl/cl_wrapper.cc:252

stack： libamdocl64.so!clSetKernelArg (Unknown Source:0) libOpenCL.so!clSetKernelArg(cl_kernel kernel, cl_uint arg_index, size_t arg_size, const void * arg_value) (/home/qingchuan/work/ROCm/ROCm-OpenCL-Runtime/khronos/icd/loader/icd_dispatch.c:684) clSetKernelArg(cl_kernel kernel, cl_uint arg_index, size_t arg_size, const void * arg_value) (/home/qingchuan/work/gitlab/paddle-lite/lite/backends/opencl/cl_wrapper.cc:252) cl::Kernel::setArgcl::Image2D(cl::Kernel * const this, cl_uint index, const cl::Image2D & value) (/home/qingchuan/work/gitlab/paddle-lite/third-party/opencl/include/CL/cl2.hpp:5858) paddle::lite::kernels::opencl::ElementwiseImageCompute::Run(paddle::lite::kernels::opencl::ElementwiseImageCompute * const this) (/home/qingchuan/work/gitlab/paddle-lite/lite/kernels/opencl/elementwise_image_compute.cc:287) paddle::lite::KernelBase::Launch(paddle::lite::KernelBase * const this) (/home/qingchuan/work/gitlab/paddle-lite/lite/core/kernel.h:125) paddle::lite::Instruction::Run(paddle::lite::Instruction * const this) (/home/qingchuan/work/gitlab/paddle-lite/lite/core/program.cc:669) paddle::lite::RuntimeProgram::Run(paddle::lite::RuntimeProgram * const this) (/home/qingchuan/work/gitlab/paddle-lite/lite/core/program.cc:461) paddle::lite::LightPredictor::Run(paddle::lite::LightPredictor * const this) (/home/qingchuan/work/gitlab/paddle-lite/lite/api/light_api.h:67) paddle::lite::LightPredictorImpl::Run(paddle::lite::LightPredictorImpl * const this) (/home/qingchuan/work/gitlab/paddle-lite/lite/api/light_api_impl.cc:133) paddle::lite_api::RunImpl(std::shared_ptrpaddle::lite_api::PaddlePredictor predictor, paddle::lite_api::PerfData * perf_data) (/home/qingchuan/work/gitlab/paddle-lite/lite/api/tools/benchmark/benchmark.cc:82) paddle::lite_api::Run(const std::string & model_file, const std::vector<std::vector<long, std::allocator >, std::allocator<std::vector<long, std::allocator > > > & input_shapes) (/home/qingchuan/work/gitlab/paddle-lite/lite/api/tools/benchmark/benchmark.cc:193) paddle::lite_api::Benchmark(int argc, char ** argv) (/home/qingchuan/work/gitlab/paddle-lite/lite/api/tools/benchmark/benchmark.cc:60) main(int argc, char ** argv) (/home/qingchuan/work/gitlab/paddle-lite/lite/api/tools/benchmark/benchmark.cc:36)

导出的s2anet_conv_2x_dota 太大无法上传，可以用别的途径发过来。

[paddle-bot[bot]]

您好，我们已经收到了您的问题，会安排技术人员尽快解答您的问题，请耐心等待。请您再次检查是否提供了清晰的问题描述、复现代码、环境&版本、报错信息等。同时，您也可以通过查看官网文档、常见问题、历史Issue来寻求解答。祝您生活愉快～

Hi! We've received your issue and please be patient to get responded. We will arrange technicians to answer your questions as soon as possible. Please make sure that you have posted enough message to demo your request. You may also check out the API，FAQ and Github Issue to get the answer.Have a nice day!

[QShiX]

closed 与https://github.com/PaddlePaddle/Paddle-Lite/issues/9476重复