Re: [PATCH v2 1/4] posix_spawn.3: Document implementation using clone() since glibc 2.24

Re: [PATCH v2 1/4] posix_spawn.3: Document implementation using clone() since glibc 2.24

[Michael Kerrisk (man-pages)]

Hello Olivier

On 10/16/18 9:37 PM, Olivier Gayot wrote:
> Since glibc 2.24, the use of posix_spawn (2) makes an unconditional call
> to clone(CLONE_VM | CLONE_VFORK ...) rather than relying on fork (2) or
> vfork (2).
> 
> As a consequence, the statements regarding the use of the flag
> POSIX_SPAWN_USEVFORK and how the function decides whether it should use
> fork (2) or vfork (2) are obsolete since glibc 2.24.
> 
> This patch makes a distinction in the manual page between glibc 2.24 and
> older versions.
> 
> Upstream commit in glibc.git:
> 
>   9ff72da471 posix: New Linux posix_spawn{p} implementation
> 
> Signed-off-by: Olivier Gayot <olivier.gayot@sigexec.com>

My apologies for the delay. This patch is now applied. Thanks for
sending it.

Adhemerval, thank you for your reviews.

Cheers,

Michael

> ---
>  man3/posix_spawn.3 | 26 ++++++++++++++++++--------
>  1 file changed, 18 insertions(+), 8 deletions(-)
> 
> diff --git a/man3/posix_spawn.3 b/man3/posix_spawn.3
> index 3e1e9011c..e8c0de172 100644
> --- a/man3/posix_spawn.3
> +++ b/man3/posix_spawn.3
> @@ -155,9 +155,16 @@ and the
>  .BR exec ()
>  step (executed in the child).
>  .SS fork() step
> -The
> +Since glibc 2.24, the
>  .BR posix_spawn ()
>  function commences by calling
> +.BR clone (2)
> +with
> +.BR CLONE_VM
> +and
> +.BR CLONE_VFORK
> +flags.
> +Older implementations use
>  .BR fork (2),
>  or possibly
>  .BR vfork (2)
> @@ -176,7 +183,7 @@ If the child fails in any of the housekeeping steps described below,
>  or fails to execute the desired file,
>  it exits with a status of 127.
>  .PP
> -The child process is created using
> +Before glibc 2.24, the child process is created using
>  .BR vfork (2)
>  instead of
>  .BR fork (2)
> @@ -480,14 +487,16 @@ The
>  and
>  .BR posix_spawnp ()
>  functions fail only in the case where the underlying
> -.BR fork (2)
> -or
> +.BR fork (2),
>  .BR vfork (2)
> +or
> +.BR clone (2)
>  call fails;  in these cases, these functions return an error number,
>  which will be one of the errors described for
> -.BR fork (2)
> +.BR fork (2),
> +.BR vfork (2)
>  or
> -.BR vfork (2).
> +.BR clone (2).
>  .PP
>  In addition, these functions fail if:
>  .TP
> @@ -527,13 +536,14 @@ only the POSIX-specified functions.
>  although these objects may be implemented as structures containing fields,
>  portable programs must avoid dependence on such implementation details.)
>  .PP
> -According to POSIX, it unspecified whether fork handlers established with
> +According to POSIX, it is unspecified whether fork handlers established with
>  .BR pthread_atfork (3)
>  are called when
>  .BR posix_spawn ()
>  is invoked.
> -On glibc,
> +Since glibc 2.24, the fork handlers are not executed in any case.
>  .\" Tested on glibc 2.12
> +On older implementations,
>  fork handlers are called only if the child is created using
>  .BR fork (2).
>  .PP
> 


-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
Linux/UNIX System Programming Training: http://man7.org/training/

Re: [PATCH v2 1/4] posix_spawn.3: Document implementation using clone() since glibc 2.24

[Michael Kerrisk (man-pages)]

Florian,

On 10/20/18 11:35 AM, Florian Weimer wrote:
> * Olivier Gayot:
> 
>> As a consequence, the statements regarding the use of the flag
>> POSIX_SPAWN_USEVFORK and how the function decides whether it should use
>> fork (2) or vfork (2) are obsolete since glibc 2.24.
> 
> Maybe we should add that it is always never correct to use
> POSIX_SPAWN_USEVFORK?
> 
> Either you are on a glibc which ignores it, then it is pointless, or
> on a glibc which has buggy semantics when vfork is used, so you
> probably do not want to use it.

Could you formulate a sentence for me that I could drop into
the current manual page?

Are there any bug reports about the older buggy semantics that I
can add as part of a commit message?

Thanks,

Michael



-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
Linux/UNIX System Programming Training: http://man7.org/training/

Re: [PATCH v2 2/4] posix_spawn.3: Document POSIX_SPAWN_USEVFORK

[Michael Kerrisk (man-pages)]

Hello Olivier,

On 10/16/18 9:37 PM, Olivier Gayot wrote:
> Added a few lines about POSIX_SPAWN_USEVFORK so that it appears clearly
> that since glibc 2.24, the flag has no effect.
> 
> Signed-off-by: Olivier Gayot <olivier.gayot@sigexec.com>

Thanks. Patch applied.

Cheers,

Michael


> ---
>  man3/posix_spawn.3 | 12 ++++++++++++
>  1 file changed, 12 insertions(+)
> 
> diff --git a/man3/posix_spawn.3 b/man3/posix_spawn.3
> index e8c0de172..8badb43bf 100644
> --- a/man3/posix_spawn.3
> +++ b/man3/posix_spawn.3
> @@ -382,6 +382,18 @@ the child's process group ID is made the same as its process ID.
>  If the
>  .B POSIX_SPAWN_SETPGROUP
>  flag is not set, the child inherits the parent's process group ID.
> +.TP
> +.B POSIX_SPAWN_USEVFORK
> +Since glibc 2.24, this flag has no effect.
> +On older implementations, setting this flag forces the
> +.BR fork()
> +step to use
> +.BR vfork (2)
> +instead of
> +.BR fork (2).
> +The
> +.B _GNU_SOURCE
> +feature test macro must be defined to obtain the definition of this contant.
>  .PP
>  If
>  .I attrp
> 


-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
Linux/UNIX System Programming Training: http://man7.org/training/

Re: [PATCH v2 3/4] posix_spawn.3: Clarify by using name of steps rather than syscalls

[Michael Kerrisk (man-pages)]

Hello Olivier,

On 10/16/18 9:37 PM, Olivier Gayot wrote:
> The implementation of the fork() step in posix_spawn(2) relies on either
> fork(2), vfork(2) or clone(2) depending on the version of the
> glibc and the arguments passed to posix_spawn(2).
> 
> It is sometimes ambiguous whether, when we are mentioning "fork(2)", we
> are referring to the fork() step or the actual fork(2) syscall.
> 
> This patch hopefully avoids the ambiguity by replacing confusing
> occurrences by "the xxx() step" where appropriate.
> 
> Signed-off-by: Olivier Gayot <olivier.gayot@sigexec.com>

Thanks. Patch applied.

Cheers,

Michael

> ---
>  man3/posix_spawn.3 | 10 +++++-----
>  1 file changed, 5 insertions(+), 5 deletions(-)
> 
> diff --git a/man3/posix_spawn.3 b/man3/posix_spawn.3
> index 8badb43bf..72e7509d8 100644
> --- a/man3/posix_spawn.3
> +++ b/man3/posix_spawn.3
> @@ -220,10 +220,10 @@ the requested file.
>  .PP
>  .SS pre-exec() step: housekeeping
>  In between the
> -.BR fork (2)
> +.BR fork()
>  and the
> -.BR exec (3),
> -a child process may need to perform a set of housekeeping actions.
> +.BR exec()
> +steps, a child process may need to perform a set of housekeeping actions.
>  The
>  .BR posix_spawn ()
>  and
> @@ -478,8 +478,8 @@ and
>  place the PID of the child process in
>  .IR pid ,
>  and return 0.
> -If there is an error before or during the
> -.BR fork (2),
> +If there is an error during the
> +.BR fork() step,
>  then no child is created,
>  the contents of
>  .IR *pid
> 


-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
Linux/UNIX System Programming Training: http://man7.org/training/

Re: [PATCH v2 4/4] posix_spawn.3: Document the POSIX_SPAWN_SETSID attribute

[Michael Kerrisk (man-pages)]

Hello Olivier,

On 10/16/18 9:37 PM, Olivier Gayot wrote:
> Since glibc 2.26, posix_spawn (2) function accepts the
> POSIX_SPAWN_SETSID flag. This flag has been accepted by POSIX and should
> be added to the next major revision. The current support can be enabled
> with _GNU_SOURCE.
> 
> Upstream commit in glibc.git:
> 
>   daeb1fa2e1 [BZ 21340] add support for POSIX_SPAWN_SETSID

Thanks. Patch applied.

Cheers,

Michael

> Signed-off-by: Olivier Gayot <olivier.gayot@sigexec.com>
> ---
>  man3/posix_spawn.3 | 14 ++++++++++++++
>  1 file changed, 14 insertions(+)
> 
> diff --git a/man3/posix_spawn.3 b/man3/posix_spawn.3
> index 72e7509d8..93f1124a1 100644
> --- a/man3/posix_spawn.3
> +++ b/man3/posix_spawn.3
> @@ -394,6 +394,20 @@ instead of
>  The
>  .B _GNU_SOURCE
>  feature test macro must be defined to obtain the definition of this contant.
> +.TP
> +.BR POSIX_SPAWN_SETSID " (since glibc 2.26)"
> +If this flag is set,
> +the child process shall create a new session and become the session leader.
> +The child process shall also become the process group leader of the new process
> +group in the session (see
> +.BR setsid (2)).
> +The
> +.B _GNU_SOURCE
> +feature test macro must be defined to obtain the definition of this contant.
> +.\" This flag has been accepted in POSIX, see:
> +.\" http://austingroupbugs.net/view.php?id=1044
> +.\" and has been implemented in glibc since version 2.26
> +.\" commit daeb1fa2e1b33323e719015f5f546988bd4cc73b
>  .PP
>  If
>  .I attrp




-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
Linux/UNIX System Programming Training: http://man7.org/training/