Awesome article! Did you try to run something like this?

char *argv[] = {"/bin/sh", NULL};
char *envp[] = {NULL};
long ret;

__asm__ volatile (
    "syscall"
    : "=a"(ret)
    : "0"(59L),   // SYS_execve
      "D"(argv[0]),
      "S"(argv),
      "d"(envp)
    : "rcx", "r11", "memory"
);

As for how to mitigate this, the sandbox could try to use seccomp. Ideally it should allow only a subset of the syscalls (i.e. not execve)

score 1 · canonical