Re: bug#54667: 29.0.50; posix_spawn breaks emacs-gdb

Re: bug#54667: 29.0.50; posix_spawn breaks emacs-gdb

[Robert Pluim]

>>>>> On Mon, 04 Apr 2022 16:11:51 +0200, Robert Pluim <rpluim@gmail.com> said:
    Robert> In emacs-28 we only use posix_spawn on macOS because its vork is
    Robert> sub-optimal. I donʼt remember the rationale for switching to using it
    Robert> everywhere, itʼs undoubtedly in the archives somewhere.

    Robert> Closing.
    Robert> Committed as 8103b060d8

Actually, do we want this in emacs-28 as well?

Robert
-- 

Re: bug#54667: 29.0.50; posix_spawn breaks emacs-gdb

[Eli Zaretskii]

> From: Robert Pluim <rpluim@gmail.com>
> Date: Wed, 06 Apr 2022 11:02:12 +0200
> 
> >>>>> On Mon, 04 Apr 2022 16:11:51 +0200, Robert Pluim <rpluim@gmail.com> said:
>     Robert> In emacs-28 we only use posix_spawn on macOS because its vork is
>     Robert> sub-optimal. I donʼt remember the rationale for switching to using it
>     Robert> everywhere, itʼs undoubtedly in the archives somewhere.
> 
>     Robert> Closing.
>     Robert> Committed as 8103b060d8
> 
> Actually, do we want this in emacs-28 as well?

The fix installed on master isn't macOS-specific, AFAICT.  Can it be
reasonably installed on emacs-28 as macOS-specific?  If yes, I think
we should indeed backport it.

Thanks.

Re: bug#54667: 29.0.50; posix_spawn breaks emacs-gdb

[Robert Pluim]

>>>>> On Wed, 06 Apr 2022 14:09:04 +0300, Eli Zaretskii <eliz@gnu.org> said:

    >> From: Robert Pluim <rpluim@gmail.com>
    >> Date: Wed, 06 Apr 2022 11:02:12 +0200
    >> 
    >> >>>>> On Mon, 04 Apr 2022 16:11:51 +0200, Robert Pluim <rpluim@gmail.com> said:
    Robert> In emacs-28 we only use posix_spawn on macOS because its vork is
    Robert> sub-optimal. I donʼt remember the rationale for switching to using it
    Robert> everywhere, itʼs undoubtedly in the archives somewhere.
    >> 
    Robert> Closing.
    Robert> Committed as 8103b060d8
    >> 
    >> Actually, do we want this in emacs-28 as well?

    Eli> The fix installed on master isn't macOS-specific, AFAICT.  Can it be
    Eli> reasonably installed on emacs-28 as macOS-specific?  If yes, I think
    Eli> we should indeed backport it.

USABLE_POSIX_SPAWN is 0 on emacs-28 for every platform except macOS,
so applying the same patch as on master is all thatʼs needed.

Robert
-- 

Re: bug#54667: 29.0.50; posix_spawn breaks emacs-gdb

[Eli Zaretskii]

> From: Robert Pluim <rpluim@gmail.com>
> Cc: emacs-devel@gnu.org
> Date: Wed, 06 Apr 2022 14:04:39 +0200
> 
> >>>>> On Wed, 06 Apr 2022 14:09:04 +0300, Eli Zaretskii <eliz@gnu.org> said:
> 
>     >> From: Robert Pluim <rpluim@gmail.com>
>     >> Date: Wed, 06 Apr 2022 11:02:12 +0200
>     >> 
>     >> >>>>> On Mon, 04 Apr 2022 16:11:51 +0200, Robert Pluim <rpluim@gmail.com> said:
>     Robert> In emacs-28 we only use posix_spawn on macOS because its vork is
>     Robert> sub-optimal. I donʼt remember the rationale for switching to using it
>     Robert> everywhere, itʼs undoubtedly in the archives somewhere.
>     >> 
>     Robert> Closing.
>     Robert> Committed as 8103b060d8
>     >> 
>     >> Actually, do we want this in emacs-28 as well?
> 
>     Eli> The fix installed on master isn't macOS-specific, AFAICT.  Can it be
>     Eli> reasonably installed on emacs-28 as macOS-specific?  If yes, I think
>     Eli> we should indeed backport it.
> 
> USABLE_POSIX_SPAWN is 0 on emacs-28 for every platform except macOS,
> so applying the same patch as on master is all thatʼs needed.

I didn't see USABLE_POSIX_SPAWN in the patch.  Maybe I was looking at
the wrong patch.  Can you show it?

Re: bug#54667: 29.0.50; posix_spawn breaks emacs-gdb

[Robert Pluim]

>>>>> On Wed, 06 Apr 2022 15:21:16 +0300, Eli Zaretskii <eliz@gnu.org> said:

    >> From: Robert Pluim <rpluim@gmail.com>
    >> Cc: emacs-devel@gnu.org
    >> Date: Wed, 06 Apr 2022 14:04:39 +0200
    >> 
    >> >>>>> On Wed, 06 Apr 2022 14:09:04 +0300, Eli Zaretskii <eliz@gnu.org> said:
    >> 
    >> >> From: Robert Pluim <rpluim@gmail.com>
    >> >> Date: Wed, 06 Apr 2022 11:02:12 +0200
    >> >> 
    >> >> >>>>> On Mon, 04 Apr 2022 16:11:51 +0200, Robert Pluim <rpluim@gmail.com> said:
    Robert> In emacs-28 we only use posix_spawn on macOS because its vork is
    Robert> sub-optimal. I donʼt remember the rationale for switching to using it
    Robert> everywhere, itʼs undoubtedly in the archives somewhere.
    >> >> 
    Robert> Closing.
    Robert> Committed as 8103b060d8
    >> >> 
    >> >> Actually, do we want this in emacs-28 as well?
    >> 
    Eli> The fix installed on master isn't macOS-specific, AFAICT.  Can it be
    Eli> reasonably installed on emacs-28 as macOS-specific?  If yes, I think
    Eli> we should indeed backport it.
    >> 
    >> USABLE_POSIX_SPAWN is 0 on emacs-28 for every platform except macOS,
    >> so applying the same patch as on master is all thatʼs needed.

    Eli> I didn't see USABLE_POSIX_SPAWN in the patch.  Maybe I was looking at
    Eli> the wrong patch.  Can you show it?

USABLE_POSIX_SPAWN is 0 on emacs-28 for all platforms except macOS,
and 1 on master for all platforms (modulo availability of some other
required features) => posix_spawn can only be used by macOS on
emacs-28.

emacs-28 callproc.c:

/* In order to be able to use `posix_spawn', it needs to support some
   variant of `chdir' as well as `setsid'.  */
#if defined DARWIN_OS					    \
  && defined HAVE_SPAWN_H && defined HAVE_POSIX_SPAWN	    \
  && defined HAVE_POSIX_SPAWNATTR_SETFLAGS                  \
  && (defined HAVE_POSIX_SPAWN_FILE_ACTIONS_ADDCHDIR        \
      || defined HAVE_POSIX_SPAWN_FILE_ACTIONS_ADDCHDIR_NP) \
  && defined HAVE_DECL_POSIX_SPAWN_SETSID                   \
  && HAVE_DECL_POSIX_SPAWN_SETSID == 1
# include <spawn.h>
# define USABLE_POSIX_SPAWN 1
#else
# define USABLE_POSIX_SPAWN 0
#endif

master callproc.c:

/* In order to be able to use `posix_spawn', it needs to support some
   variant of `chdir' as well as `setsid'.  */
#if defined HAVE_SPAWN_H && defined HAVE_POSIX_SPAWN        \
  && defined HAVE_POSIX_SPAWNATTR_SETFLAGS                  \
  && (defined HAVE_POSIX_SPAWN_FILE_ACTIONS_ADDCHDIR        \
      || defined HAVE_POSIX_SPAWN_FILE_ACTIONS_ADDCHDIR_NP) \
  && defined HAVE_DECL_POSIX_SPAWN_SETSID                   \
  && HAVE_DECL_POSIX_SPAWN_SETSID == 1
# include <spawn.h>
# define USABLE_POSIX_SPAWN 1
#else
# define USABLE_POSIX_SPAWN 0
#endif

and all uses of posix_spawn are protected by USABLE_POSIX_SPAWN

Robert
-- 

Re: bug#54667: 29.0.50; posix_spawn breaks emacs-gdb

[Eli Zaretskii]

> From: Robert Pluim <rpluim@gmail.com>
> Cc: emacs-devel@gnu.org
> Date: Wed, 06 Apr 2022 14:33:50 +0200
> 
>     >> USABLE_POSIX_SPAWN is 0 on emacs-28 for every platform except macOS,
>     >> so applying the same patch as on master is all thatʼs needed.
> 
>     Eli> I didn't see USABLE_POSIX_SPAWN in the patch.  Maybe I was looking at
>     Eli> the wrong patch.  Can you show it?
> 
> USABLE_POSIX_SPAWN is 0 on emacs-28 for all platforms except macOS,
> and 1 on master for all platforms (modulo availability of some other
> required features) => posix_spawn can only be used by macOS on
> emacs-28.
> 
> emacs-28 callproc.c:
> 
> /* In order to be able to use `posix_spawn', it needs to support some
>    variant of `chdir' as well as `setsid'.  */
> #if defined DARWIN_OS					    \
>   && defined HAVE_SPAWN_H && defined HAVE_POSIX_SPAWN	    \
>   && defined HAVE_POSIX_SPAWNATTR_SETFLAGS                  \
>   && (defined HAVE_POSIX_SPAWN_FILE_ACTIONS_ADDCHDIR        \
>       || defined HAVE_POSIX_SPAWN_FILE_ACTIONS_ADDCHDIR_NP) \
>   && defined HAVE_DECL_POSIX_SPAWN_SETSID                   \
>   && HAVE_DECL_POSIX_SPAWN_SETSID == 1
> # include <spawn.h>
> # define USABLE_POSIX_SPAWN 1
> #else
> # define USABLE_POSIX_SPAWN 0
> #endif
> 
> master callproc.c:
> 
> /* In order to be able to use `posix_spawn', it needs to support some
>    variant of `chdir' as well as `setsid'.  */
> #if defined HAVE_SPAWN_H && defined HAVE_POSIX_SPAWN        \
>   && defined HAVE_POSIX_SPAWNATTR_SETFLAGS                  \
>   && (defined HAVE_POSIX_SPAWN_FILE_ACTIONS_ADDCHDIR        \
>       || defined HAVE_POSIX_SPAWN_FILE_ACTIONS_ADDCHDIR_NP) \
>   && defined HAVE_DECL_POSIX_SPAWN_SETSID                   \
>   && HAVE_DECL_POSIX_SPAWN_SETSID == 1
> # include <spawn.h>
> # define USABLE_POSIX_SPAWN 1
> #else
> # define USABLE_POSIX_SPAWN 0
> #endif
> 
> and all uses of posix_spawn are protected by USABLE_POSIX_SPAWN

We are mis-communicating, because that's not what I asked.

But never mind, I've looked at the sources, and I think it's safe to
backport this to the emacs-28 branch.

Thanks.

Re: bug#54667: 29.0.50; posix_spawn breaks emacs-gdb

[Robert Pluim]

>>>>> On Wed, 06 Apr 2022 18:48:05 +0300, Eli Zaretskii <eliz@gnu.org> said:

    Eli> But never mind, I've looked at the sources, and I think it's safe to
    Eli> backport this to the emacs-28 branch.

Done as 98abf01fd6

Robert
--