Linux uses patchelf tool

1. Download patchelf tool

github address: https://github.com/NixOS/patchelf

Download link: https://github.com/NixOS/patchelf/releases

I downloaded PatchELF 0.12 directly

patchelf-0.12.tar.gz

2. Install patchelf tool

Unzip:

$ tar -axf patchelf-0.12.tar.gz
$ cd patchelf-0.12/
$ ls
bootstrap.sh  COPYING     Makefile.am       README.md    tests
BUGS          flake.lock  patchelf.1        release.nix  version
configure.ac  flake.nix   patchelf.spec.in  src

installation:

$ ./bootstrap.sh 
./bootstrap.sh: 2: ./bootstrap.sh: autoreconf: not found

Find solutions based on the error:

$ sudo apt-get install autoconf automake libtool

Then continue execution:

$ sh bootstrap.sh
autoreconf: Entering directory `.'
autoreconf: configure.ac: not using Gettext
autoreconf: running: aclocal --force --warnings=all 
autoreconf: configure.ac: tracing
autoreconf: configure.ac: creating directory build-aux
autoreconf: configure.ac: not using Libtool
autoreconf: running: /usr/bin/autoconf --force --warnings=all
autoreconf: configure.ac: not using Autoheader
autoreconf: running: automake --add-missing --copy --force-missing --warnings=all
configure.ac:7: installing 'build-aux/compile'
configure.ac:5: installing 'build-aux/install-sh'
configure.ac:5: installing 'build-aux/missing'
src/Makefile.am: installing 'build-aux/depcomp'
parallel-tests: installing 'build-aux/test-driver'
autoreconf: Leaving directory `.'

$ ./configure 
checking for a BSD-compatible install... /usr/bin/install -c
checking whether build environment is sane... yes
checking for a thread-safe mkdir -p... /bin/mkdir -p
checking for gawk... no
checking for mawk... mawk
checking whether make sets $(MAKE)... yes
checking whether make supports nested variables... yes
checking for style of include used by make... GNU
checking for gcc... gcc
checking whether the C compiler works... yes
checking for C compiler default output file name... a.out
checking for suffix of executables... 
checking whether we are cross compiling... no
checking for suffix of object files... o
checking whether we are using the GNU C compiler... yes
checking whether gcc accepts -g... yes
checking for gcc option to accept ISO C89... none needed
checking whether gcc understands -c and -o together... yes
checking dependency style of gcc... gcc3
checking for g++... g++
checking whether we are using the GNU C++ compiler... yes
checking whether g++ accepts -g... yes
checking dependency style of g++... gcc3
checking that generated files are newer than configure... done
configure: creating ./config.status
config.status: creating Makefile
config.status: creating src/Makefile
config.status: creating tests/Makefile
config.status: creating patchelf.spec
config.status: executing depfiles commands

After that:

make
make install

3. Verify the installation

patchelf --version

4. Use

(1) View dependent libraries

Assuming that the name of the dynamic library or executable program compiled by yourself is demo, here we take demo as an example.

$ readelf -d demo  
Dynamic section at offset 0x7e3c30 contains 35 entries:
  Tag        Type                         Name/Value
 0x0000000000000001 (NEEDED)             Shared library: [libtorch.so]
 0x0000000000000001 (NEEDED)             Shared library: [libc10.so]
 0x0000000000000001 (NEEDED)             Shared library: [libIlmImf-2_2.so.22]
 0x0000000000000001 (NEEDED)             Shared library: [libdl.so.2]
 0x0000000000000001 (NEEDED)             Shared library: [librt.so.1]
 0x0000000000000001 (NEEDED)             Shared library: [libtorch_cpu.so]
 0x0000000000000001 (NEEDED)             Shared library: [libpthread.so.0]
 0x0000000000000001 (NEEDED)             Shared library: [libstdc++.so.6]
 0x0000000000000001 (NEEDED)             Shared library: [libm.so.6]
 0x0000000000000001 (NEEDED)             Shared library: [libgcc_s.so.1]
 0x0000000000000001 (NEEDED)             Shared library: [libc.so.6]
 0x000000000000000f (RPATH)              Library rpath: [/home/wmz/Desktop/mypractice/libtorch/lib]
 0x000000000000000c (INIT)               0x40c668
 0x000000000000000d (FINI)               0xa1745c
 0x0000000000000019 (INIT_ARRAY)         0xdd3348
 0x000000000000001b (INIT_ARRAYSZ)       200 (bytes)
 0x000000000000001a (FINI_ARRAY)         0xdd3410
 0x000000000000001c (FINI_ARRAYSZ)       8 (bytes)
 0x000000006ffffef5 (GNU_HASH)           0x400298
 0x0000000000000005 (STRTAB)             0x4040b0
 0x0000000000000006 (SYMTAB)             0x400a08
 0x000000000000000a (STRSZ)              22774 (bytes)
 0x000000000000000b (SYMENT)             24 (bytes)
 0x0000000000000015 (DEBUG)              0x0
 0x0000000000000003 (PLTGOT)             0xde4000
 0x0000000000000002 (PLTRELSZ)           8232 (bytes)
 0x0000000000000014 (PLTREL)             RELA
 0x0000000000000017 (JMPREL)             0x40a640
 0x0000000000000007 (RELA)               0x409ff8
 0x0000000000000008 (RELASZ)             1608 (bytes)
 0x0000000000000009 (RELAENT)            24 (bytes)
 0x000000006ffffffe (VERNEED)            0x409e38
 0x000000006fffffff (VERNEEDNUM)         7
 0x000000006ffffff0 (VERSYM)             0x4099a6
 0x0000000000000000 (NULL)               0x0

(2) Modify rpath

patchelf --set-rpath 'ORIGIN' demo

This adds the path where the executable program demo is located as the dependent library search path.

Reference: https://www.cnblogs.com/ar-cheng/p/13225342.html

Instructions for the official website:

README.md

PatchELF is a simple utility for modifying existing ELF executables and libraries. In particular, it can do the following:

• Change the dynamic loader ("ELF interpreter") of executables:

  $ patchelf --set-interpreter /lib/my-ld-linux.so.2 my-program

• Change the RPATH of executables and libraries:

  $ patchelf --set-rpath /opt/my-libs/lib:/other-libs my-program

• Shrink the RPATH of executables and libraries:

  $ patchelf --shrink-rpath my-program

  This removes from the RPATH all directories that do not contain a library referenced by DT_NEEDED fields of the executable or library. For instance, if an executable references one library libfoo.so, has an RPATH /lib:/usr/lib:/foo/lib, and libfoo.so can only be found in /foo/lib, then the new RPATH will be /foo/lib.

  In addition, the --allowed-rpath-prefixes option can be used for further rpath tuning. For instance, if an executable has an RPATH /tmp/build-foo/.libs:/foo/lib, it is probably desirable to keep the /foo/lib reference instead of the /tmp entry. To accomplish that, use:

  $ patchelf --shrink-rpath --allowed-rpath-prefixes /usr/lib:/foo/lib my-program

• Remove declared dependencies on dynamic libraries (DT_NEEDED entries):

  $ patchelf --remove-needed libfoo.so.1 my-program

  This option can be given multiple times.

• Add a declared dependency on a dynamic library (DT_NEEDED):

  $ patchelf --add-needed libfoo.so.1 my-program

  This option can be give multiple times.

• Replace a declared dependency on a dynamic library with another one (DT_NEEDED):

  $ patchelf --replace-needed liboriginal.so.1 libreplacement.so.1 my-program

  This option can be give multiple times.

• Change SONAME of a dynamic library:

  $ patchelf --set-soname libnewname.so.3.4.5 path/to/libmylibrary.so.1.2.3

Usage supported by patchelf:

$ patchelf -h
syntax: patchelf
  [--set-interpreter FILENAME]
  [--page-size SIZE]
  [--print-interpreter]
  [--print-soname]		Prints 'DT_SONAME' entry of .dynamic section. Raises an error if DT_SONAME doesn't exist
  [--set-soname SONAME]		Sets 'DT_SONAME' entry to SONAME.
  [--set-rpath RPATH]
  [--remove-rpath]
  [--shrink-rpath]
  [--allowed-rpath-prefixes PREFIXES]		With '--shrink-rpath', reject rpath entries not starting with the allowed prefix
  [--print-rpath]
  [--force-rpath]
  [--add-needed LIBRARY]
  [--remove-needed LIBRARY]
  [--replace-needed LIBRARY NEW_LIBRARY]
  [--print-needed]
  [--no-default-lib]
  [--clear-symbol-version SYMBOL]
  [--output FILE]
  [--debug]
  [--version]
  FILENAME...

In actual application, when setting the runtime library path, you can set multiple at one time, separated by a colon, as follows, this command specifies three paths at the same time, the current path, the lib path under the upper level, and the one under cuda-10.2. lib64 path:

$ patchelf --set-rpath 'ORIGIN:../lib/:/usr/local/cuda-10.2/lib64' dbnet_demo

After checking, these dependent libraries can be found correctly:

$ ldd dbnet_demo 
	linux-vdso.so.1 (0x00007ffc96ce1000)
	libtorch.so => ../lib/libtorch.so (0x00007fb7014f9000)
	libc10.so => ../lib/libc10.so (0x00007fb701275000)
	libdl.so.2 => /lib/x86_64-linux-gnu/libdl.so.2 (0x00007fb701071000)
	libpthread.so.0 => /lib/x86_64-linux-gnu/libpthread.so.0 (0x00007fb700e52000)
	librt.so.1 => /lib/x86_64-linux-gnu/librt.so.1 (0x00007fb700c4a000)
	libtorch_cpu.so => ../lib/libtorch_cpu.so (0x00007fb6f0a89000)
	libtorch_cuda.so => ../lib/libtorch_cuda.so (0x00007fb6b5e89000)
	libnvToolsExt.so.1 => /usr/local/cuda-10.2/lib64/libnvToolsExt.so.1 (0x00007fb6b5c80000)
	libcudart.so.10.2 => /usr/local/cuda-10.2/lib64/libcudart.so.10.2 (0x00007fb6b5a02000)
	libstdc++.so.6 => /usr/lib/x86_64-linux-gnu/libstdc++.so.6 (0x00007fb6b5679000)
	libm.so.6 => /lib/x86_64-linux-gnu/libm.so.6 (0x00007fb6b52db000)
	libgcc_s.so.1 => /lib/x86_64-linux-gnu/libgcc_s.so.1 (0x00007fb6b50c3000)
	libc.so.6 => /lib/x86_64-linux-gnu/libc.so.6 (0x00007fb6b4cd2000)
	libgomp-7c85b1e2.so.1 => /home/wmz/Documents/pytorch-dbnet/build2/../lib/libgomp-7c85b1e2.so.1 (0x00007fb6b4aa8000)
	/lib64/ld-linux-x86-64.so.2 (0x00007fb702206000)
	libtensorpipe.so => /home/wmz/Documents/pytorch-dbnet/build2/../lib/libtensorpipe.so (0x00007fb6b45df000)
	libcudart-80664282.so.10.2 => /home/wmz/Documents/pytorch-dbnet/build2/../lib/libcudart-80664282.so.10.2 (0x00007fb6b435e000)
	libc10_cuda.so => /home/wmz/Documents/pytorch-dbnet/build2/../lib/libc10_cuda.so (0x00007fb6b412d000)
	libnvToolsExt-3965bdd0.so.1 => /home/wmz/Documents/pytorch-dbnet/build2/../lib/libnvToolsExt-3965bdd0.so.1 (0x00007fb6b3f23000)

Run the executable program, you can also get the correct result:

 

$ ./dbnet_demo 
ok
 file_names.size():5
0001004.jpg
./data/0001004.jpg
[W TensorIterator.cpp:918] Warning: Mixed memory format inputs detected while calling the operator. The operator will output contiguous tensor even if some of the inputs are in channels_last format. (function operator())
The run time is: 0.408824s
CLOCKS_PER_SEC is: 1000000
0001001.jpg
./data/0001001.jpg
The run time is: 0.112061s
CLOCKS_PER_SEC is: 1000000
0001002.jpg
./data/0001002.jpg
The run time is: 0.100809s
CLOCKS_PER_SEC is: 1000000
0001003.jpg
./data/0001003.jpg
The run time is: 0.101038s
CLOCKS_PER_SEC is: 1000000
0000000.jpg
./data/0000000.jpg
The run time is: 0.095049s
CLOCKS_PER_SEC is: 1000000