python - 如何列出在 python 脚本运行时加载的 .so 文件？

Question

如何列出在 python 脚本运行时加载的所有 .so 文件(完整路径)？

是否可以获得(例如)libcudart.so.10.1 的完整路径？是否可以获取已加载的 .so(完整路径)列表？

例如我正在运行:

python -c "import tensorflow as tf; tf.test.is_gpu_available()"

输出:

2019-09-27 15:02:27.186029: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcuda.so.1
2019-09-27 15:02:27.657901: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1640] Found device 0 with properties:
name: TITAN RTX major: 7 minor: 5 memoryClockRate(GHz): 1.77
pciBusID: 0000:17:00.0
2019-09-27 15:02:27.658580: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1640] Found device 1 with properties:
name: TITAN RTX major: 7 minor: 5 memoryClockRate(GHz): 1.77
pciBusID: 0000:65:00.0
2019-09-27 15:02:27.658766: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcudart.so.10.1
2019-09-27 15:02:27.659868: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcublas.so.10
2019-09-27 15:02:27.661073: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcufft.so.10
2019-09-27 15:02:27.661305: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcurand.so.10
2019-09-27 15:02:27.662477: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcusolver.so.10
2019-09-27 15:02:27.663054: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcusparse.so.10
2019-09-27 15:02:27.665455: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcudnn.so.7
2019-09-27 15:02:27.667986: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1763] Adding visible gpu devices: 0, 1
2019-09-27 15:02:27.668031: I tensorflow/stream_executor/platform/default/dso_loader.cc:42] Successfully opened dynamic library libcudart.so.10.1
2019-09-27 15:02:27.669523: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1181] Device interconnect StreamExecutor with strength 1 edge matrix:
2019-09-27 15:02:27.669535: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1187]      0 1
2019-09-27 15:02:27.669542: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1200] 0:   N N
2019-09-27 15:02:27.669546: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1200] 1:   N N
2019-09-27 15:02:27.674204: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1326] Created TensorFlow device (/device:GPU:0 with 22845 MB memory) -> physical GPU (device: 0, name: TITAN RTX, pci bus id: 0000:17:00.0, compute capability: 7.5)
2019-09-27 15:02:27.675838: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1326] Created TensorFlow device (/device:GPU:1 with 22823 MB memory) -> physical GPU (device: 1, name: TITAN RTX, pci bus id: 0000:65:00.0, compute capability: 7.5)

Answer 1

在 Linux 下运行时，根据您的环境，您可以使用以下替代方法之一获取信息:

1. ld.so 调试标志
2. 跟踪

使用 ld.so 调试标志，您可以在运行 python 代码(取决于您的 shell:export LD_DEBUG=libs，或setenv LD_DEBUG 库)。动态加载的库搜索路径将显示到 stderr。

查看 ld.so 手册页以了解其他选项。

示例输出 - 来自 LD_DEBUG=libs(使用 python 'import json')

owner@vm1:~/Project/stackoverflow/58152351$ python
     28796: find library=libc.so.6 [0]; searching
     28796:  search cache=/etc/ld.so.cache
     28796:   trying file=/lib/x86_64-linux-gnu/libc.so.6
     28796: 
     28796: find library=libpthread.so.0 [0]; searching
     28796:  search cache=/etc/ld.so.cache
     28796:   trying file=/lib/x86_64-linux-gnu/libpthread.so.0
     28796: 
     28796: find library=libdl.so.2 [0]; searching
     28796:  search cache=/etc/ld.so.cache
     28796:   trying file=/lib/x86_64-linux-gnu/libdl.so.2
...
>>> import json
     28796: 
     28796: calling init: /usr/lib/python2.7/lib-dynload/_json.x86_64-linux-gnu.so
     28796: 

作为替代方案，可以使用 strace，但解析它的输出需要更多的努力。 “stat”行可以提供有关每个库的搜索路径的信息。