运行StyleGAN2时遇NVCC编译错误及未定义符号问题求助

初始错误：TensorFlow符号未找到 #

运行StyleGAN2时触发以下错误：

NotFoundError: /home/chengjun/stylegan2-master/dnnlib/tflib/_cudacache/fused_bias_act_347e82e8919aeb0d5c7dc989e996c091.so: undefined symbol: _ZN10tensorflow12OpDefBuilder4AttrENSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEE

使用ldd命令检查该.so文件，发现其未链接tensorflow_framework.so.1：

linux-vdso.so.1 (0x00007ffdbe136000)
    _pywrap_tensorflow_internal.so => not found
    libstdc++.so.6 => /home/chengjun/anaconda3/lib/libstdc++.so.6 (0x00007f1ba7000000)
    libgcc_s.so.1 => /home/chengjun/anaconda3/lib/libgcc_s.so.1 (0x00007f1ba72df000)
    libc.so.6 => /lib/x86_64-linux-gnu/libc.so.6 (0x00007f1ba6c00000)
    /lib64/ld-linux-x86-64.so.2 (0x00007f1ba73d5000)
    libm.so.6 => /lib/x86_64-linux-gnu/libm.so.6 (0x00007f1ba6f19000)

添加路径后出现新的NVCC编译错误 #

将TensorFlow相关路径添加到bashrc：

export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/chengjun/anaconda3/envs/python3/lib/python3.6/site-packages/tensorflow/python/

再次运行后触发NVCC编译错误：

raise RuntimeError('NVCC returned an error. See below for full command line and output log:\n\n%s\n\n%s' % (cmd, output))
RuntimeError: NVCC returned an error. See below for full command line and output log:

nvcc "/home/chengjun/anaconda3/envs/python3/lib/python3.6/site-packages/tensorflow/python/_pywrap_tensorflow_internal.so" --compiler-options '-fPIC -D_GLIBCXX_USE_CXX11_ABI= 0' --gpu-architecture=sm_86 --use_fast_math --disable-warnings --include-path "/home/chengjun/anaconda3/envs/python3/lib/python3.6/site-packages/tensorflow/include" --include-path "/home/chengjun/anaconda3/envs/python3/lib/python3.6/site-packages/tensorflow/include/external/protobuf_archive/src" --include-path "/home/chengjun/anaconda3/envs/python3/lib/python3.6/site-packages/tensorflow/include/external/com_google_absl" --include-path "/home/chengjun/anaconda3/envs/python3/lib/python3.6/site-packages/tensorflow/include/external/eigen_archive" 2>&1 "/home/chengjun/stylegan2-master/dnnlib/tflib/ops/fused_bias_act.cu" --shared -o "/tmp/tmp8e0pgiod/fused_bias_act_tmp.so" --keep --keep-dir "/tmp/tmp8e0pgiod"

gcc-11: fatal error: cannot specify ‘-o’ with ‘-c’, ‘-S’ or ‘-E’ with multiple files
compilation terminated.

待解决问题 #

• 上述错误的根本原因是什么？
• 如何修复当前的编译及链接问题？
• 目前即使删除添加的LD_LIBRARY_PATH路径，错误仍存在，该如何处理？

内容的提问来源于stack exchange，提问作者Chengjun Duan