[0/4] Add pidfd_spawn, pidfd_spawnp, pidfd_fork, and pidfd_getpid

Message ID 20230418213505.3834934-1-adhemerval.zanella@linaro.org
Headers
Series Add pidfd_spawn, pidfd_spawnp, pidfd_fork, and pidfd_getpid |

Message

Adhemerval Zanella Netto April 18, 2023, 9:35 p.m. UTC
  The glibc 2.36 added wrappers for Linux syscall pidfd_open, pidfd_getfd,
and pidfd_send_signal, and exported the P_PIDFD to use with waitid.
However, although the pidfd is a race free interface, the pidfd_open 
is subject to TOCTOU if the file descriptor is not obtained directly
from the clone or clone3 syscall (there is still a small window between
the clone return and the pidfd_getfd where the process can be reaped
and the process ID reused).

A fully race free interface with posix_spawn interface is being discussed
by GNOME [1] [2], and Qt already uses on its QtProcess implementation [3].
The Qt code has some pitfalls by not using a libc provided symbol:

  - It calls clone through the syscall symbol, which does not run the
    pthread_atfork handlers even though it really intends to use the
    clone semantic for fork (by only using CLONE_PIDFD | SIGCHLD).

  - It also does not reset any internal state, such as internal IO,
    malloc, loader, etc. locks.

  - It does not set the TCB tid field nor the robust list, used by
    pthread code.

  - It does not optimize process creation by using CLONE_VM and CLONE_VFORK.

The pidfd_spawn and pidfd_spawnp handles all these cases by using the
same internal implementation used by posix_spawn:

  int pidfd_spawn (int *restrict pidfd,
 		   const char *restrict file,
  		   const posix_spawn_file_actions_t *restrict facts,
  		   const posix_spawnattr_t *restrict attrp,
  		   char *const argv[restrict],
  		   char *const envp[restrict])

  int pidfd_spawnp (int *restrict pidfd,
 		    const char *restrict path,
  		    const posix_spawn_file_actions_t *restrict facts,
  		    const posix_spawnattr_t *restrict attrp,
  		    char *const argv[restrict_arr],
  		    char *const envp[restrict_arr]);

The implementation makes sure that kernel must support the complete
pidfd interface, meaning that waitid (P_PIDFD) should be supported.  It
ensure that non racy workaround is required (such as reading procfs
fdinfo pid to use along with old wait interfaces).  If kernel does not
have the required support the interface returns ENOSYS.

A new symbol is used instead of a posix_spawn extension to avoid possible
issue with language bindings that might track the argument lifetime.
Although for Linux pid_t and int are interchangeable, POSIX only state
that pid_t should be a signed interger.

Both symbols reuse the posix_spawn posix_spawn_file_actions_t and
posix_spawnattr_t, to either avoid rehash posix_spawn API or add a new
one.  It also mean that both interfaces support the same attribute and
file actions, and a new flag or file actions on posix_spawn is also
added automatically for pidfd_spawn.

Along with the spawn interface, a fork like one is also provided:

  int pidfd_fork (unsigned int flags)

The kernel already sets O_CLOEXEC as default and it follow fork/_Fork
convention on returning a positive or negative value to the parent
(with negative indicating an error) and zero to the child.

Different than fork, pidfd_fork does not run the pthread_atfork handlers
(similar to _Fork).  It can be change by using PIDFD_FORK_RUNATFORK with
flags.

To have a way to interop between process IDs and process file descriptors,
the pidfd_getpid is also provided.  It just read the procps fdinfo entry
from the file descriptor to get the process ID.

[1] https://gitlab.gnome.org/GNOME/glib/-/issues/1866
[2] https://sourceware.org/bugzilla/show_bug.cgi?id=30349
[3] https://codebrowser.dev/qt6/qtbase/src/3rdparty/forkfd/forkfd_linux.c.html

Adhemerval Zanella (4):
  posix: Re-flow and sort multiline definitions
  posix: Add pidfd_spawn and pidfd_spawnp (BZ# 30349)
  posix: Add pidfd_fork
  linux: Add pidfd_getpid

 NEWS                                          |  16 +
 bits/spawn_ext.h                              |  21 +
 include/clone_internal.h                      |   4 +
 manual/process.texi                           |  22 +-
 posix/Makefile                                | 556 ++++++++++++++----
 posix/fork-internal.c                         | 125 ++++
 posix/fork-internal.h                         |  29 +
 posix/fork.c                                  |  98 +--
 posix/spawn.h                                 |   2 +
 posix/spawn_int.h                             |   3 +-
 posix/tst-posix_spawn-setsid.c                | 168 ++++--
 posix/tst-spawn-chdir.c                       |  15 +-
 posix/tst-spawn.c                             |  24 +-
 posix/tst-spawn.h                             |  36 ++
 posix/tst-spawn2.c                            |  17 +-
 posix/tst-spawn3.c                            | 100 ++--
 posix/tst-spawn4.c                            |   7 +-
 posix/tst-spawn5.c                            |  14 +-
 posix/tst-spawn6.c                            |  15 +-
 posix/tst-spawn7.c                            |  13 +-
 sysdeps/nptl/_Fork.c                          |   2 +-
 sysdeps/unix/sysv/linux/Makefile              |  35 +-
 sysdeps/unix/sysv/linux/Versions              |   6 +
 sysdeps/unix/sysv/linux/aarch64/libc.abilist  |   4 +
 sysdeps/unix/sysv/linux/alpha/libc.abilist    |   4 +
 sysdeps/unix/sysv/linux/arc/libc.abilist      |   4 +
 sysdeps/unix/sysv/linux/arch-fork.h           |  16 +-
 sysdeps/unix/sysv/linux/arm/be/libc.abilist   |   4 +
 sysdeps/unix/sysv/linux/arm/le/libc.abilist   |   4 +
 sysdeps/unix/sysv/linux/bits/spawn_ext.h      |  45 ++
 sysdeps/unix/sysv/linux/clone-pidfd-support.c |  58 ++
 sysdeps/unix/sysv/linux/csky/libc.abilist     |   4 +
 sysdeps/unix/sysv/linux/hppa/libc.abilist     |   4 +
 sysdeps/unix/sysv/linux/i386/libc.abilist     |   4 +
 sysdeps/unix/sysv/linux/ia64/libc.abilist     |   4 +
 .../sysv/linux/loongarch/lp64/libc.abilist    |   4 +
 .../sysv/linux/m68k/coldfire/libc.abilist     |   4 +
 .../unix/sysv/linux/m68k/m680x0/libc.abilist  |   4 +
 .../sysv/linux/microblaze/be/libc.abilist     |   4 +
 .../sysv/linux/microblaze/le/libc.abilist     |   4 +
 .../sysv/linux/mips/mips32/fpu/libc.abilist   |   4 +
 .../sysv/linux/mips/mips32/nofpu/libc.abilist |   4 +
 .../sysv/linux/mips/mips64/n32/libc.abilist   |   4 +
 .../sysv/linux/mips/mips64/n64/libc.abilist   |   4 +
 sysdeps/unix/sysv/linux/nios2/libc.abilist    |   4 +
 sysdeps/unix/sysv/linux/or1k/libc.abilist     |   4 +
 sysdeps/unix/sysv/linux/pidfd_fork.c          |  76 +++
 sysdeps/unix/sysv/linux/pidfd_getpid.c        |  70 +++
 sysdeps/unix/sysv/linux/pidfd_spawn.c         |  30 +
 sysdeps/unix/sysv/linux/pidfd_spawnp.c        |  30 +
 .../linux/powerpc/powerpc32/fpu/libc.abilist  |   4 +
 .../powerpc/powerpc32/nofpu/libc.abilist      |   4 +
 .../linux/powerpc/powerpc64/be/libc.abilist   |   4 +
 .../linux/powerpc/powerpc64/le/libc.abilist   |   4 +
 sysdeps/unix/sysv/linux/procutils.c           |  99 ++++
 sysdeps/unix/sysv/linux/procutils.h           |  37 ++
 .../unix/sysv/linux/riscv/rv32/libc.abilist   |   4 +
 .../unix/sysv/linux/riscv/rv64/libc.abilist   |   4 +
 .../unix/sysv/linux/s390/s390-32/libc.abilist |   4 +
 .../unix/sysv/linux/s390/s390-64/libc.abilist |   4 +
 sysdeps/unix/sysv/linux/sh/be/libc.abilist    |   4 +
 sysdeps/unix/sysv/linux/sh/le/libc.abilist    |   4 +
 .../sysv/linux/sparc/sparc32/libc.abilist     |   4 +
 .../sysv/linux/sparc/sparc64/libc.abilist     |   4 +
 sysdeps/unix/sysv/linux/spawni.c              |  20 +-
 sysdeps/unix/sysv/linux/sys/pidfd.h           |  17 +
 sysdeps/unix/sysv/linux/tst-pidfd.c           |   7 +
 sysdeps/unix/sysv/linux/tst-pidfd_fork.c      | 150 +++++
 .../sysv/linux/tst-posix_spawn-setsid-pidfd.c |  20 +
 .../unix/sysv/linux/tst-spawn-chdir-pidfd.c   |  20 +
 sysdeps/unix/sysv/linux/tst-spawn-pidfd.c     |  20 +
 sysdeps/unix/sysv/linux/tst-spawn-pidfd.h     |  63 ++
 sysdeps/unix/sysv/linux/tst-spawn2-pidfd.c    |  20 +
 sysdeps/unix/sysv/linux/tst-spawn3-pidfd.c    |  20 +
 sysdeps/unix/sysv/linux/tst-spawn4-pidfd.c    |  20 +
 sysdeps/unix/sysv/linux/tst-spawn5-pidfd.c    |  20 +
 sysdeps/unix/sysv/linux/tst-spawn6-pidfd.c    |  20 +
 sysdeps/unix/sysv/linux/tst-spawn7-pidfd.c    |  20 +
 .../unix/sysv/linux/x86_64/64/libc.abilist    |   4 +
 .../unix/sysv/linux/x86_64/x32/libc.abilist   |   4 +
 80 files changed, 1975 insertions(+), 387 deletions(-)
 create mode 100644 bits/spawn_ext.h
 create mode 100644 posix/fork-internal.c
 create mode 100644 posix/fork-internal.h
 create mode 100644 posix/tst-spawn.h
 create mode 100644 sysdeps/unix/sysv/linux/bits/spawn_ext.h
 create mode 100644 sysdeps/unix/sysv/linux/clone-pidfd-support.c
 create mode 100644 sysdeps/unix/sysv/linux/pidfd_fork.c
 create mode 100644 sysdeps/unix/sysv/linux/pidfd_getpid.c
 create mode 100644 sysdeps/unix/sysv/linux/pidfd_spawn.c
 create mode 100644 sysdeps/unix/sysv/linux/pidfd_spawnp.c
 create mode 100644 sysdeps/unix/sysv/linux/procutils.c
 create mode 100644 sysdeps/unix/sysv/linux/procutils.h
 create mode 100644 sysdeps/unix/sysv/linux/tst-pidfd_fork.c
 create mode 100644 sysdeps/unix/sysv/linux/tst-posix_spawn-setsid-pidfd.c
 create mode 100644 sysdeps/unix/sysv/linux/tst-spawn-chdir-pidfd.c
 create mode 100644 sysdeps/unix/sysv/linux/tst-spawn-pidfd.c
 create mode 100644 sysdeps/unix/sysv/linux/tst-spawn-pidfd.h
 create mode 100644 sysdeps/unix/sysv/linux/tst-spawn2-pidfd.c
 create mode 100644 sysdeps/unix/sysv/linux/tst-spawn3-pidfd.c
 create mode 100644 sysdeps/unix/sysv/linux/tst-spawn4-pidfd.c
 create mode 100644 sysdeps/unix/sysv/linux/tst-spawn5-pidfd.c
 create mode 100644 sysdeps/unix/sysv/linux/tst-spawn6-pidfd.c
 create mode 100644 sysdeps/unix/sysv/linux/tst-spawn7-pidfd.c