SNAP2EXE aims to generate an ELF executable from a run-time snapshot of a process so that user can continue that process from where it was snapshoted by simply executing the genereated executable. We call such an executable as a snapshot executable.

We provide a library and a binary tool:

• The library mainly contains two interfaces:
  • snap2exe(pid, save_dir): snapshot the process at run-time by pid and generate a snapshot executable to the save_dir.
  • s2e_checkpoint(save_dir, policy): a process can call it to snapshot itself to generate a snapshot executable according to some policy.
• The binary tool snap2exe encapsulates the snap2exe(pid, save_dir) interface, which attaches to a process at run-time by pid and snapshot it to an executable.

The snapshot executable generated by the library and the tool will continue to execute from where it was snapshoted.

Just use make, and the binary snap2exe and the library libsnap2exe.a will be output to the ./build directory.

Use PREFIX=path/to/snap2exe make install to install them. The default PREFIX is $HOME/local/snap2exe.

Add -I path/to/snap2exe/include to your compile system and link path/to/snap2exe/lib/libsnap2exe.a.

Check the headers for the details of interface.

snap2exe <pid> <save_dir>
    <pid>: the process to snapshot.
    <save_dir>: the directory to save snapshot executables.

An example (test/test-ckpt.c) using the s2e_checkpoint() interface:

$ make
$ make test
$ ./test-ckpt
$ ./snapshots/test-ckpt/<id>/cont

An example (test/test-tool.c) using the snap2exe tool:

$ make
$ make test
$ ./test-tool &
$ ./build/snap2exe $! snapshots/test-tool  # need sudo
$ cd snapshots/test-tool/<id>
$ chmod a+x cont && ./cont  # need sudo

• Can be only used on x86_64-linux.
• Cannot recover kernel states execpt opened regular files.
• KASLR should be closed.