sysdeps/unix/sysv/linux/clone-pidfd-support.c

   1 /* Check if kernel supports PID file descriptors.
   2    Copyright (C) 2023-2024 Free Software Foundation, Inc.
   3    This file is part of the GNU C Library.
   4
   5    The GNU C Library is free software; you can redistribute it and/or
   6    modify it under the terms of the GNU Lesser General Public
   7    License as published by the Free Software Foundation; either
   8    version 2.1 of the License, or (at your option) any later version.
   9
  10    The GNU C Library is distributed in the hope that it will be useful,
  11    but WITHOUT ANY WARRANTY; without even the implied warranty of
  12    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  13    Lesser General Public License for more details.
  14
  15    You should have received a copy of the GNU Lesser General Public
  16    License along with the GNU C Library; if not, see
  17    <https://www.gnu.org/licenses/>.  */
  18
  19 #include <atomic.h>
  20 #include <sys/wait.h>
  21 #include <sysdep.h>
  22
  23 /* The PID file descriptors was added during multiple releases:
  24    - Linux 5.2 added CLONE_PIDFD support for clone and __clone_pidfd_supported
  25      syscall.
  26    - Linux 5.3 added support for poll and CLONE_PIDFD for clone3.
  27    - Linux 5.4 added P_PIDFD support on waitid.
  28
  29    For internal usage on spawn and fork, it only make sense to return a file
  30    descriptor if caller can actually waitid on it.  */
  31
  32 static int __waitid_pidfd_supported = 0;
  33
  34 bool
  35 __clone_pidfd_supported (void)
  36 {
  37   int state = atomic_load_relaxed (&__waitid_pidfd_supported);
  38   if (state == 0)
  39     {
  40       /* Linux define the maximum allocated file descriptor value as
  41          0x7fffffc0 (from fs/file.c):
  42
  43          #define __const_min(x, y) ((x) < (y) ? (x) : (y))
  44          unsigned int sysctl_nr_open_max =
  45            __const_min(INT_MAX, ~(size_t)0/sizeof(void *)) & -BITS_PER_LONG;
  46
  47          So we can detect whether kernel supports all pidfd interfaces by
  48          using a valid but never allocated file descriptor: if is not
  49          supported waitid will return EINVAL, otherwise EBADF.
  50
  51          Also the waitid is a cancellation entrypoint, so issue the syscall
  52          directly.  */
  53       int r = INTERNAL_SYSCALL_CALL (waitid, P_PIDFD, INT_MAX, NULL,
  54                                      WEXITED | WNOHANG);
  55       state = r == -EBADF ? 1 : -1;
  56       atomic_store_relaxed (&__waitid_pidfd_supported, state);
  57     }
  58
  59   return state > 0;
  60 }