Generate syscall info for Lurk

README.md

• get-syscalls.py generates syscall informations.
• syscalls-table-riscv.rs is an example of using get-syscalls.py to generate ones for RISC-V, with manual modifications noted.

get-syscalls.py

"""
Extracts syscall parameter info from Linux source headers.

Note that there might be missing some syscalls in the final result, or there
may be false positives. Manual check is needed after generation.
"""

import sys
import requests
import re

syscalls = requests.get(
  "https://github.com/torvalds/linux/raw/master/include/linux/syscalls.h")
syscalls_regex = re.compile(
  r'asmlinkage long sys_([\w\d_]+?)\(([^"]+?)\);', re.MULTILINE)
syscalls_matches = re.findall(syscalls_regex, syscalls.text)

defines = requests.get(
  "https://github.com/torvalds/linux/raw/master/include/uapi/asm-generic/unistd.h")
defines_regex = re.compile(r'#define __NR[^_]*_([\w\d_]+) (\d+)')
defines_matches = re.findall(defines_regex, defines.text)

integers = ("int", "long", "uint32_t", "unsigned", "u32", "u64", "size_t", "pid_t", "s32",
            "mqd_t", "loff_t", "clockid_t", "timer_t", "umode_t", "uid_t", "gid_t", "key_t",
            "aio_context_t", "rwf_t", "qid_t", "__s32", "key_serial_t", "__u32", "off_t")
structs = ("struct", "enum", "union", "cap_user_header_t", "cap_user_data_t")

table = []

for a, b in defines_matches:
  for c, d in syscalls_matches:
    if c == a:
      params = []
      for p in d.split(','):
        p = p.strip().removeprefix("const").strip()
        if '*' not in p and p.startswith(integers):
          params.append("INT")
        elif '*' in p and p.startswith("char"):
          params.append("STR")
        elif '*' in p or p.startswith(structs):
          params.append("ADDR")
        elif p.startswith("void"):
          params.append("VOID")
        else:
          print("UNKNOWN:", p, file=sys.stderr)
          params.append("UNKNOWN")

      if len(params) == 1 and params[0] == "VOID":
        params = []

      table.append((b, a, params))
      break
  else:
    table.append((b, a, None))

prev = -999
for num, name, params in table:
  num = int(num)
  if num - prev > 1:
    print(f"// {num}")
  if params:
    print(f"syscall!({name}, {', '.join(params)}),")
  elif params != None:
    print(f"syscall!({name}),")
  else:
    print(f"// TODO: syscall!({name})")
  prev = num

syscalls-table-riscv.rs

// 0
syscall!(io_setup, INT, ADDR),
syscall!(io_destroy, INT),
syscall!(io_submit, INT, INT, ADDR),
syscall!(io_cancel, INT, ADDR, ADDR),
syscall!(io_getevents, INT, INT, INT, ADDR, ADDR),
syscall!(setxattr, STR, STR, ADDR, INT, INT),
syscall!(lsetxattr, STR, STR, ADDR, INT, INT),
syscall!(fsetxattr, INT, STR, ADDR, INT, INT),
syscall!(getxattr, STR, STR, ADDR, INT),
syscall!(lgetxattr, STR, STR, ADDR, INT),
syscall!(fgetxattr, INT, STR, ADDR, INT),
syscall!(listxattr, STR, STR, INT),
syscall!(llistxattr, STR, STR, INT),
syscall!(flistxattr, INT, STR, INT),
syscall!(removexattr, STR, STR),
syscall!(lremovexattr, STR, STR),
syscall!(fremovexattr, INT, STR),
syscall!(getcwd, STR, INT),
syscall!(lookup_dcookie, INT, STR, INT),
syscall!(eventfd2, INT, INT),
syscall!(epoll_create1, INT),
syscall!(epoll_ctl, INT, INT, INT, ADDR),
syscall!(epoll_pwait, INT, ADDR, INT, INT, ADDR, INT),
syscall!(dup, INT),
syscall!(dup3, INT, INT, INT),
syscall!(fcntl, INT, INT, INT),
syscall!(inotify_init1, INT),
syscall!(inotify_add_watch, INT, STR, INT),
syscall!(inotify_rm_watch, INT, INT),
syscall!(ioctl, INT, INT, INT),
syscall!(ioprio_set, INT, INT, INT),
syscall!(ioprio_get, INT, INT),
syscall!(flock, INT, INT),
syscall!(mknodat, INT, STR, INT, INT),
syscall!(mkdirat, INT, STR, INT),
syscall!(unlinkat, INT, STR, INT),
syscall!(symlinkat, STR, INT, STR),
syscall!(linkat, INT, STR, INT, STR, INT),
syscall!(renameat, INT, STR, INT, STR),
syscall!(umount2, STR, INT), // manually added
syscall!(mount, STR, STR, STR, INT, ADDR),
syscall!(pivot_root, STR, STR),
syscall!(nfsservctl, INT, ADDR, ADDR), // manually added
syscall!(statfs, STR, ADDR),
syscall!(fstatfs, INT, ADDR),
syscall!(truncate, STR, INT),
syscall!(ftruncate, INT, INT),
syscall!(fallocate, INT, INT, INT, INT),
syscall!(faccessat, INT, STR, INT),
syscall!(chdir, STR),
syscall!(fchdir, INT),
syscall!(chroot, STR),
syscall!(fchmod, INT, INT),
syscall!(fchmodat, INT, STR, INT),
syscall!(fchownat, INT, STR, INT, INT, INT),
syscall!(fchown, INT, INT, INT),
syscall!(openat, INT, STR, INT, INT),
syscall!(close, INT),
syscall!(vhangup),
syscall!(pipe2, ADDR, INT),
syscall!(quotactl, INT, STR, INT, ADDR),
syscall!(getdents64, INT, ADDR, INT),
syscall!(lseek, INT, INT, INT),
syscall!(read, INT, STR, INT),
syscall!(write, INT, STR, INT),
syscall!(readv, INT, ADDR, INT),
syscall!(writev, INT, ADDR, INT),
syscall!(pread64, INT, STR, INT, INT),
syscall!(pwrite64, INT, STR, INT, INT),
syscall!(preadv, INT, ADDR, INT, INT, INT),
syscall!(pwritev, INT, ADDR, INT, INT, INT),
syscall!(sendfile, INT, INT, ADDR, INT),
syscall!(pselect6, INT, ADDR, ADDR, ADDR, ADDR, ADDR),
syscall!(ppoll, ADDR, INT, ADDR, ADDR, INT),
syscall!(signalfd4, INT, ADDR, INT, INT),
syscall!(vmsplice, INT, ADDR, INT, INT),
syscall!(splice, INT, ADDR, INT, ADDR, INT, INT),
syscall!(tee, INT, INT, INT, INT),
syscall!(readlinkat, INT, STR, STR, INT),
syscall!(fstatat, INT, STR, ADDR, INT),  // manually added
syscall!(fstat, INT, ADDR),
syscall!(sync),
syscall!(fsync, INT),
syscall!(fdatasync, INT),
syscall!(sync_file_range2, INT, INT, INT, INT),
syscall!(sync_file_range, INT, INT, INT, INT),
syscall!(timerfd_create, INT, INT),
syscall!(timerfd_settime, INT, INT, ADDR, ADDR),
syscall!(timerfd_gettime, INT, ADDR),
syscall!(utimensat, INT, STR, ADDR, INT),
syscall!(acct, STR),
syscall!(capget, ADDR, ADDR),
syscall!(capset, ADDR, ADDR),
syscall!(personality, INT),
syscall!(exit, INT),
syscall!(exit_group, INT),
syscall!(waitid, INT, INT, ADDR, INT, ADDR),
syscall!(set_tid_address, ADDR),
syscall!(unshare, INT),
syscall!(futex, ADDR, INT, INT, ADDR, ADDR, INT),
syscall!(set_robust_list, ADDR, INT),
syscall!(get_robust_list, INT, ADDR, ADDR),
syscall!(nanosleep, ADDR, ADDR),
syscall!(getitimer, INT, ADDR),
syscall!(setitimer, INT, ADDR, ADDR),
syscall!(kexec_load, INT, INT, ADDR, INT),
syscall!(init_module, ADDR, INT, STR),
syscall!(delete_module, STR, INT),
syscall!(timer_create, INT, ADDR, ADDR),
syscall!(timer_gettime, INT, ADDR),
syscall!(timer_getoverrun, INT),
syscall!(timer_settime, INT, INT, ADDR, ADDR),
syscall!(timer_delete, INT),
syscall!(clock_settime, INT, ADDR),
syscall!(clock_gettime, INT, ADDR),
syscall!(clock_getres, INT, ADDR),
syscall!(clock_nanosleep, INT, INT, ADDR, ADDR),
syscall!(syslog, INT, STR, INT),
syscall!(ptrace, INT, INT, INT, INT),
syscall!(sched_setparam, INT, ADDR),
syscall!(sched_setscheduler, INT, INT, ADDR),
syscall!(sched_getscheduler, INT),
syscall!(sched_getparam, INT, ADDR),
syscall!(sched_setaffinity, INT, INT, ADDR),
syscall!(sched_getaffinity, INT, INT, ADDR),
syscall!(sched_yield),
syscall!(sched_get_priority_max, INT),
syscall!(sched_get_priority_min, INT),
syscall!(sched_rr_get_interval, INT, ADDR),
syscall!(restart_syscall),
syscall!(kill, INT, INT),
syscall!(tkill, INT, INT),
syscall!(tgkill, INT, INT, INT),
syscall!(sigaltstack, ADDR, ADDR),
syscall!(rt_sigsuspend, ADDR, INT),
syscall!(rt_sigaction, INT, ADDR, ADDR, INT),
syscall!(rt_sigprocmask, INT, ADDR, ADDR, INT),
syscall!(rt_sigpending, ADDR, INT),
syscall!(rt_sigtimedwait, ADDR, ADDR, ADDR, INT),
syscall!(rt_sigqueueinfo, INT, INT, ADDR),
syscall!(rt_sigreturn), // manually added
syscall!(setpriority, INT, INT, INT),
syscall!(getpriority, INT, INT),
syscall!(reboot, INT, INT, INT, ADDR),
syscall!(setregid, INT, INT),
syscall!(setgid, INT),
syscall!(setreuid, INT, INT),
syscall!(setuid, INT),
syscall!(setresuid, INT, INT, INT),
syscall!(getresuid, ADDR, ADDR, ADDR),
syscall!(setresgid, INT, INT, INT),
syscall!(getresgid, ADDR, ADDR, ADDR),
syscall!(setfsuid, INT),
syscall!(setfsgid, INT),
syscall!(times, ADDR),
syscall!(setpgid, INT, INT),
syscall!(getpgid, INT),
syscall!(getsid, INT),
syscall!(setsid),
syscall!(getgroups, INT, ADDR),
syscall!(setgroups, INT, ADDR),
syscall!(uname, ADDR),
syscall!(sethostname, STR, INT),
syscall!(setdomainname, STR, INT),
syscall!(getrlimit, INT, ADDR),
syscall!(setrlimit, INT, ADDR),
syscall!(getrusage, INT, ADDR),
syscall!(umask, INT),
syscall!(prctl, INT, INT, INT, INT, INT),
syscall!(getcpu, ADDR, ADDR, ADDR),
syscall!(gettimeofday, ADDR, ADDR),
syscall!(settimeofday, ADDR, ADDR),
syscall!(adjtimex, ADDR),
syscall!(getpid),
syscall!(getppid),
syscall!(getuid),
syscall!(geteuid),
syscall!(getgid),
syscall!(getegid),
syscall!(gettid),
syscall!(sysinfo, ADDR),
syscall!(mq_open, STR, INT, INT, ADDR),
syscall!(mq_unlink, STR),
syscall!(mq_timedsend, INT, STR, INT, INT, ADDR),
syscall!(mq_timedreceive, INT, STR, INT, ADDR, ADDR),
syscall!(mq_notify, INT, ADDR),
syscall!(mq_getsetattr, INT, ADDR, ADDR),
syscall!(msgget, INT, INT),
syscall!(msgctl, INT, INT, ADDR),
syscall!(msgrcv, INT, ADDR, INT, INT, INT),
syscall!(msgsnd, INT, ADDR, INT, INT),
syscall!(semget, INT, INT, INT),
syscall!(semctl, INT, INT, INT, INT),
syscall!(semtimedop, INT, ADDR, INT, ADDR),
syscall!(semop, INT, ADDR, INT),
syscall!(shmget, INT, INT, INT),
syscall!(shmctl, INT, INT, ADDR),
syscall!(shmat, INT, STR, INT),
syscall!(shmdt, STR),
syscall!(socket, INT, INT, INT),
syscall!(socketpair, INT, INT, INT, ADDR),
syscall!(bind, INT, ADDR, INT),
syscall!(listen, INT, INT),
syscall!(accept, INT, ADDR, ADDR),
syscall!(connect, INT, ADDR, INT),
syscall!(getsockname, INT, ADDR, ADDR),
syscall!(getpeername, INT, ADDR, ADDR),
syscall!(sendto, INT, ADDR, INT, INT, ADDR, INT),
syscall!(recvfrom, INT, ADDR, INT, INT, ADDR, ADDR),
syscall!(setsockopt, INT, INT, INT, STR, INT),
syscall!(getsockopt, INT, INT, INT, STR, ADDR),
syscall!(shutdown, INT, INT),
syscall!(sendmsg, INT, ADDR, INT),
syscall!(recvmsg, INT, ADDR, INT),
syscall!(readahead, INT, INT, INT),
syscall!(brk, INT),
syscall!(munmap, INT, INT),
syscall!(mremap, INT, INT, INT, INT, INT),
syscall!(add_key, STR, STR, ADDR, INT, INT),
syscall!(request_key, STR, STR, STR, INT),
syscall!(keyctl, INT, INT, INT, INT, INT),
syscall!(clone, INT, INT, ADDR, INT, ADDR),
syscall!(execve, STR, STR, STR),
syscall!(mmap, ADDR, INT, INT, INT, INT, INT), // manually added
syscall!(fadvise64, INT, INT, INT, INT),
syscall!(swapon, STR, INT),
syscall!(swapoff, STR),
syscall!(mprotect, INT, INT, INT),
syscall!(msync, INT, INT, INT),
syscall!(mlock, INT, INT),
syscall!(munlock, INT, INT),
syscall!(mlockall, INT),
syscall!(munlockall),
syscall!(mincore, INT, INT, ADDR),
syscall!(madvise, INT, INT, INT),
syscall!(remap_file_pages, INT, INT, INT, INT, INT),
syscall!(mbind, INT, INT, INT, ADDR, INT, INT),
syscall!(get_mempolicy, ADDR, ADDR, INT, INT, INT),
syscall!(set_mempolicy, INT, ADDR, INT),
syscall!(migrate_pages, INT, INT, ADDR, ADDR),
syscall!(move_pages, INT, INT, ADDR, ADDR, ADDR, INT),
syscall!(rt_tgsigqueueinfo, INT, INT, INT, ADDR),
syscall!(perf_event_open, ADDR, INT, INT, INT, INT),
syscall!(accept4, INT, ADDR, ADDR, INT),
syscall!(recvmmsg, INT, ADDR, INT, INT, ADDR),
// 260
syscall!(wait4, INT, ADDR, INT, ADDR),
syscall!(prlimit64, INT, INT, ADDR, ADDR),
syscall!(fanotify_init, INT, INT),
syscall!(fanotify_mark, INT, INT, INT, INT, STR),
syscall!(name_to_handle_at, INT, STR, ADDR, INT, INT), // manually added
syscall!(open_by_handle_at, INT, ADDR, INT), // manually added
syscall!(clock_adjtime, INT, ADDR),
syscall!(syncfs, INT),
syscall!(setns, INT, INT),
syscall!(sendmmsg, INT, ADDR, INT, INT),
syscall!(process_vm_readv, INT, ADDR, INT, ADDR, INT, INT),
syscall!(process_vm_writev, INT, ADDR, INT, ADDR, INT, INT),
syscall!(kcmp, INT, INT, INT, INT, INT),
syscall!(finit_module, INT, STR, INT),
syscall!(sched_setattr, INT, ADDR, INT),
syscall!(sched_getattr, INT, ADDR, INT, INT),
syscall!(renameat2, INT, STR, INT, STR, INT),
syscall!(seccomp, INT, INT, ADDR),
syscall!(getrandom, STR, INT, INT),
syscall!(memfd_create, STR, INT),
syscall!(bpf, INT, ADDR, INT),
syscall!(execveat, INT, STR, STR, STR, INT),
syscall!(userfaultfd, INT),
syscall!(membarrier, INT, INT, INT),
syscall!(mlock2, INT, INT, INT),
syscall!(copy_file_range, INT, ADDR, INT, ADDR, INT, INT),
syscall!(preadv2, INT, ADDR, INT, INT, INT, INT),
syscall!(pwritev2, INT, ADDR, INT, INT, INT, INT),
syscall!(pkey_mprotect, INT, INT, INT, INT),
syscall!(pkey_alloc, INT, INT),
syscall!(pkey_free, INT),
syscall!(statx, INT, STR, INT, INT, ADDR),
syscall!(io_pgetevents, INT, INT, INT, ADDR, ADDR, ADDR),
syscall!(rseq, ADDR, INT, INT, INT),
syscall!(kexec_file_load, INT, INT, INT, STR, INT),
// 403
// These time64 syscalls are the same as the original variant on 64-bit systems.
syscall!(clock_gettime64, INT, ADDR), // manually added
syscall!(clock_settime64, INT, ADDR), // manually added
syscall!(clock_adjtime64, INT, ADDR), // manually added
syscall!(clock_getres_time64, INT, ADDR), // manually added
syscall!(clock_nanosleep_time64, INT, INT, ADDR, ADDR), // manually added
syscall!(timer_gettime64, INT, ADDR), // manually added
syscall!(timer_settime64, INT, INT, ADDR, ADDR), // manually added
syscall!(timerfd_gettime64, INT, ADDR), // manually added
syscall!(timerfd_settime64, INT, INT, ADDR, ADDR), // manually added
syscall!(utimensat_time64, INT, STR, ADDR, INT), // manually added
syscall!(pselect6_time64, INT, ADDR, ADDR, ADDR, ADDR, ADDR), // manually added
syscall!(ppoll_time64, ADDR, INT, ADDR, ADDR, INT), // manually added
// 416
syscall!(io_pgetevents_time64, INT, INT, INT, ADDR, ADDR, ADDR), // manually added
syscall!(recvmmsg_time64, INT, ADDR, INT, INT, ADDR), // manually added
syscall!(mq_timedsend_time64, INT, STR, INT, INT, ADDR), // manually added
syscall!(mq_timedreceive_time64, INT, STR, INT, ADDR, ADDR), // manually added
syscall!(semtimedop_time64, INT, ADDR, INT, ADDR), // manually added
syscall!(rt_sigtimedwait_time64, ADDR, ADDR, ADDR, INT), // manually added
syscall!(futex_time64, ADDR, INT, INT, ADDR, ADDR, INT), // manually added
syscall!(sched_rr_get_interval_time64, INT, ADDR), // manually added
syscall!(pidfd_send_signal, INT, INT, ADDR, INT),
syscall!(io_uring_setup, INT, ADDR),
syscall!(io_uring_enter, INT, INT, INT, INT, ADDR, INT),
syscall!(io_uring_register, INT, INT, ADDR, INT),
syscall!(open_tree, INT, STR, INT),
syscall!(move_mount, INT, STR, INT, STR, INT),
syscall!(fsopen, STR, INT),
syscall!(fsconfig, INT, INT, STR, ADDR, INT),
syscall!(fsmount, INT, INT, INT),
syscall!(fspick, INT, STR, INT),
syscall!(pidfd_open, INT, INT),
syscall!(clone3, ADDR, INT),
syscall!(close_range, INT, INT, INT),
syscall!(openat2, INT, STR, ADDR, INT),
syscall!(pidfd_getfd, INT, INT, INT),
syscall!(faccessat2, INT, STR, INT, INT),
syscall!(process_madvise, INT, ADDR, INT, INT, INT),
syscall!(epoll_pwait2, INT, ADDR, INT, ADDR, ADDR, INT),
syscall!(mount_setattr, INT, STR, INT, ADDR, INT),
syscall!(quotactl_fd, INT, INT, INT, ADDR),
syscall!(landlock_create_ruleset, ADDR, INT, INT),
syscall!(landlock_add_rule, INT, ADDR, ADDR, INT),
syscall!(landlock_restrict_self, INT, INT),
syscall!(memfd_secret, INT),
syscall!(process_mrelease, INT, INT),
syscall!(futex_waitv, ADDR, INT, INT, ADDR, INT),
syscall!(set_mempolicy_home_node, INT, INT, INT, INT),